Data Sources and Data-Linking Strategies to Support Research to Address the Opioid Crisis

U.S. Department of Health and Human Services

Assistant Secretary for Planning and Evaluation

Office of Health Policy

Data Sources and Data-Linking

Strategies to Support Research to

Address the Opioid Crisis

FINAL REPORT

September 2018

The Office of the Assistant Secretary for Planning and Evaluation (ASPE) is the

principal advisor to the Secretary of the Department of Health and Human Services

(HHS) on policy development issues, and is responsible for major activities in the areas

of legislative and budget development, strategic planning, policy research and

evaluation, and economic analysis.

The Office of Health Policy (HP), within ASPE, provides a cross-cutting policy perspective

that bridges Departmental programs, public and private sector activities, and the research

community, in order to develop, analyze, coordinate and provide leadership on health

policy issues for the Secretary.

This report was prepared under contract # HHSP23320095649WC. The task order number

for the current Time & Materials umbrella contract is: HHSP23337038T between HHS’s

ASPE/HP and the RAND Corporation.

The opinions and views expressed in this report are those of the authors. They do not necessarily

reflect the views of the Department of Health and Human Services, the contractor or any other

funding organization.

September 2018

Data Sources and Data-Linking Strategies to

Support Research to Address the Opioid Crisis

Final Report

Submitted to

Office of Health Policy

Assistant Secretary for Planning and Evaluation

U.S. Department of Health and Human Services

Hubert H. Humphrey Building

200 Independence Avenue SW

Washington, DC 20201

Submitted by

Rosanna Smart, Courtney Ann Kase, Amanda Meyer, and Bradley D. Stein

RAND Corporation

1776 Main Street

P.O. Box 2138

Santa Monica, CA 90407-2138

About This Report

This report presents findings from a scoping study to assess the types of data sources and

data-linkage efforts that are currently being used or could potentially be leveraged to support

research and evaluations relevant to the U.S. Department of Health and Human Services

Strategic Priorities to combat the opioids crisis. Based on an environmental scan of the literature

and interviews with opioid policy and research efforts, the purpose of the project is to provide an

overview of the types of secondary data sources and data linkages commonly used in opioid-

related research to highlight some of the key gaps or challenges for existing data-collection and

analysis efforts and to outline potential steps that could be taken to overcome these challenges.

The initial scoping study was conducted in summer 2017, with an update to the scan of the

literature conducted in February 2018.

We would like to acknowledge the participation and assistance of all researchers and federal

program officials who participated in the stakeholder interviews. This effort would not have been

possible without their generosity in providing their time and expertise on challenges and

opportunities for the use of secondary data in research relevant to the opioids crisis. We also

thank Hilary Peterson and Mary Vaiana for their keen attention to detail and for providing

excellent assistance in the creation of this report. Finally, we would like to acknowledge the

contributions of Susan Lumsden and Scott R. Smith from the Office of the Assistant Secretary

for Planning and Evaluation, as well as the valuable insights we received from the peer reviewers

of the report, Erin Taylor of RAND and Brendan Saloner of Johns Hopkins University.

The research reported here was undertaken within RAND Health, a division of the RAND

Corporation, and funded by the Office of the Assistant Secretary for Planning and Evaluation,

Department of Health and Human Services. A profile of RAND Health, abstracts of its

publications, and ordering information can be found at www.rand.org/health. Questions and

comments about this report should be sent to the project leader, Bradley Stein ([email protected]).

About the Authors

Bradley Stein is a senior physician policy researcher at the RAND Corporation and an adjunct

associate professor of psychiatry at the University of Pittsburgh School of Medicine. A

practicing psychiatrist and health services and policy researcher, his research is focused on better

understanding and improving care for individuals with mental health and substance use disorders

in community settings.

ii

Rosanna Smart is an associate economist at the RAND Corporation whose research centers on

studying the public health and policy implications of licit and illicit substance use, drug markets

and drug policy, and issues related to the criminal justice system.

Courtney A. Kase is a policy analyst at the RAND Corporation whose prior research includes

evaluations of service integration within community-based behavioral health centers, approaches

to reducing health disparities, and approaches for technology use and collaboration in rural

educational settings.

Amanda Meyer is a research assistant at the RAND Corporation with research interests in

tobacco control and regulation, mental health policy and interventions, trauma, and school

health.

iii

Contents

About This Report ......................................................................................................................ii

Tables......................................................................................................................................... v

Abbreviations ............................................................................................................................ vi

1. Introduction ............................................................................................................................ 1

2. Background on the U.S. Department of Health and Human Services’ Strategic Priorities........ 3

Better Practices for Pain Management............................................................................................. 3

Better Addiction Prevention, Treatment, and Recovery Services...................................................... 4

Better Targeting of Overdose-Reversing Drugs ............................................................................... 4

Better Data...................................................................................................................................... 5

Better Research............................................................................................................................... 5

3. Current State of the Evidence: Findings from the Environmental Scan.................................... 8

Better Practices for Pain Management............................................................................................. 8

Better Addiction Prevention, Treatment, and Recovery Services.................................................... 11

Better Targeting of Overdose-Reversing Drugs ............................................................................. 13

Better Data.................................................................................................................................... 15

4. Sources of Secondary Data: Data Inventory Findings............................................................ 17

National Surveys........................................................................................................................... 19

Electronic Health Records and Claims Data .................................................................................. 20

Mortality Records ......................................................................................................................... 22

Prescription Drug–Monitoring Data .............................................................................................. 22

Contextual and Policy Data........................................................................................................... 23

Other National, State, and Local Sources....................................................................................... 24

5. High-Priority Research Needs and Data Efforts: Findings from the

Stakeholder Discussions ..................................................................................................... 26

Better Practices for Pain Management........................................................................................... 26

Better Addiction Prevention, Treatment, and Recovery Services.................................................... 30

Better Targeting of Overdose-Reversing Drugs ............................................................................. 34

Better Data.................................................................................................................................... 36

6. Challenges and Opportunities for Implementing Successful Data-Linking Strategies ............ 40

Summary ...................................................................................................................................... 49

References................................................................................................................................ 52

iv

Appendix Overview of Types of Secondary Data Sources and Data Inventory Content 72— ..........

Tables

v

Table 3.1. Commonly Used Data Sources and Measures in Research to Advance Better Pain

Management Practices ......................................................................................................... 9

Table 3.2. Contextual Data Sources and Measures Commonly Linked to Opioid Outcome Data in

Research Related to the Five-Point HHS Strategy.............................................................. 10

Table 3.3. Commonly Used Data Sources and Measures in Research to Improve Addiction

Prevention, Treatment, and Recovery Services .................................................................. 12

Table 4.1. Data Source Categories Identified ............................................................................. 18

Table 4.2. Comparison of Electronic Health Record and Administrative Claims Data ............... 21

Table 5.1. Commonly Referenced Data Sources for Understanding Better Practices for Pain

Management...................................................................................................................... 28

Table 5.2. Commonly Referenced Data Sources for Understanding Treatment Need and Access

.......................................................................................................................................... 32

Table 5.3. Commonly Referenced Data Sources for Understanding Naloxone Access ............... 35

Table 5.4. Commonly Referenced Data Sources for Understanding the Epidemic Through Better

Public Health Surveillance ................................................................................................ 38

Table 6.1. Time Frame for Potential Approaches to Implementing Successful Data-Linking

Strategies .......................................................................................................................... 50

Table A.1. National Survey Data ............................................................................................... 75

Table A.2. Claims and Electronic Health Records Secondary Data Sources .............................. 78

Table A.3. Mortality Records .................................................................................................... 84

Table A.4. Prescription Monitoring Secondary Data Sources .................................................... 86

Table A.5. Contextual and Policy Data Sources ........................................................................ 89

Table A.6. Other National, State, and Local Secondary Data Sources ........................................ 91

Abbreviations

ADAM Arrestee Drug Abuse Monitoring

AHRQ Agency for Healthcare Research and Quality

ARCOS Automation of Reports and Consolidated Orders System

CDC Centers for Disease Control and Prevention

CMS Centers for Medicare and Medicaid Services

DAWN Drug Abuse Warning Network

DEA Drug Enforcement Agency

DEA ACSA Drug Enforcement Agency Active Controlled Substances Act Registrants

Database

EHR electronic health record

EMS Emergency medical services

HHS Department of Health and Human Services

MEPS Medical Expenditure Panel Survey

NAMSDL National Alliance for Model State Drug Laws

NAVIPPRO National Addictions Vigilance Intervention and Prevention Program

NDI National Death Index

NEMSIS National Emergency Medical Services Information System

NESARC National Epidemiologic Survey on Alcohol and Related Conditions

NPDS National Poison Data System

NSDUH National Survey on Drug Use and Health

N-SSATS National Survey of Substance Abuse Treatment Services

NVSS MCOD National Vital Statistics System Multiple Cause of Death

OEND overdose education and naloxone distribution

PBSS Prescription Behavior Surveillance System

PDAPS Prescription Drug Abuse Policy System

PDMP prescription drug monitoring program

RADARS Researched Abuse, Diversion and Addiction-Related Surveillance System

SAMHSA Substance Abuse and Mental Health Administration

STRIDE System to Retrieve Information from Drug Evidence

TEDS Treatment Episodes Data Set

vi

1. Introduction

The Department of Health and Human Services

(HHS) has a five-point strategy for addressing the

significant social and public costs associated with

the opioid crisis (see Box 1) (HHS, undated).

Numerous efforts are underway to implement these

strategies, which are intended to address key

contributors and harms related to the opioid crisis,

enhance the ability of public health officials and

policymakers to monitor the crisis as it evolves,

and facilitate more-informed policymaking.

However, progress will also be made by identifying

which research questions to prioritize, data sources

to support such research, and approaches that can be used to leverage or link multiple

complementary data sources. Much of the research on the opioid crisis relies on information

drawn from sources outside of clinical research settings. Researchers can leverage “real-world

evidence” to enhance the field’s ability to address the crisis and generate new evidence to inform

decisions.

Box 1. HHS Strategic Priorities

! Better practices for pain

management

! Better addiction prevention,

treatment, and recovery services

! Better targeting of overdose-

reversing drugs

! Better data

! Better research.

The ability to link data—combining data from two or more sources to study the same

individual, facility, organization, e vent, or geographic area—often makes it possible to enhance

the value of the information obtained beyond what is available from any single source. Data sets

that contain unique individual identifiers make it possible to link information from different

sources at the individual level. Linkages at a more-aggregate level include analyses that merge

two or more data sources at the state or county level or at a finer geographic level. Finally, while

they do not directly “link” data sources, many studies analyze multiple complementary data

sources (e.g., geographic spatial analyses of heroin-related emergency department visits and

heroin-related deaths) to provide more-robust or comprehensive evidence of policy or program

impact (Hudson, Klekamp, and Matthews, 2017). Each method has strengths and limitations, but

all can contribute toward informing evidence-based policymaking (Commission on Evidence-

Based Policymaking, 2017).

This report provides an overview of the types of secondary data sources currently being used

or that could potentially be used to evaluate interventions or conduct other analyses that address

the five-part HHS strategy. The report highlights key research questions in each area and

identifies opportunities to use existing data sources and implement data-linking strategies that

can support assessments of the HHS strategy. Findings are based on interviews with 16

experts—academic researchers, federal researchers, and federal program officials—

1

complemented by an environmental scan of the literature. This report does not address all the

strengths and limitations of these data sources; rather, it is intended to provide sufficient

information to serve as a resource to researchers in the field of opioids and opioid use disorder.

This report is organized as follows:

• Chapter 2 provides background information on each of the HHS Strategic Priorities.

• Chapter 3 informs the Strategic Priority of better research by presenting an overview of

existing research related to the first four HHS Strategic Priorities as identified through an

environmental scan, including commonly used data sources and common approaches to

linking or merging data sources.

• Chapter 4 broadly categorizes the types of secondary data sources used in research

related to the Strategic Priorities and provides examples of specific data sources and data

elements.

• Chapter 5 describes findings identified through stakeholder discussions on key research

needs and the opportunities and challenges for using secondary data sources to address

those needs.

• Chapter 6 summarizes key challenges facing researchers and policymakers in studying

and responding to the opioid crisis and suggests potential solutions.

2

2. Background on the U.S. Department of Health and Human

Services’ Strategic Priorities

Addressing the opioid crisis is one of HHS’s top priorities. Therefore, HHS has developed a

comprehensive strategy to empower local communities on the frontlines. In 2017, HHS unveiled

a five-point strategy, encompassing (1) better pain management; (2) better treatment, prevention,

and recovery services; (3) better targeting of overdose-reversing drugs; (4) better data on the

crisis; and (5) better research to inform strategies to combat the crisis. In this chapter, we provide

an overview of information needs and research considerations underlying each component of the

strategy.

Better Practices for Pain Management

An estimated 20 percent of noncancer outpatients with pain receive opioid analgesics

(Daubresse et al., 2013); those who receive such medications chronically are at significant risk of

developing an opioid use disorder (Boscarino et al., 2010), characterized by persistent use that is

functionally impairing (American Psychiatric Association, 2013). Growth in opioid analgesic

prescribing has occurred alongside increasing rates of opioid-related misuse, emergency

department visits, and deaths (HHS, 2013; Rudd et al., 2016). Efforts to minimize opioid-

prescribing practices that likely lead to misuse or opioid-related harms must be balanced with

maintaining appropriate, high-quality pain management for patients (Interagency Pain Research

Coordinating Committee, 2015).

In recent years, federal agencies such as the Centers for Disease Control and Prevention

(CDC) and Centers for Medicare and Medicaid Services (CMS) have worked with private

insurers, medical educators, and other stakeholders to promote safe opioid use while limiting

addiction risk (Price, 2017). National medical organizations, states, and large health systems

have published clinical practice guidelines for prescribing opioids for chronic pain (Nuckols et

al., 2014; Haegerich et al., 2014; Mai et al., 2015). Likewise, efforts by the Interagency Pain

Research Coordinating Committee (created by HHS) and CDC have worked toward providing

clinicians, researchers, and the public with recommendations concerning the prescribing and use

of opioids for pain management (Interagency Pain Research Coordinating Committee, 2015;

Dowell, Haegerich, and Chou, 2016). Federal agencies have also called for research and science

to improve the effectiveness of existing alternative pain treatments, including nonpharmacologic

options (e.g., physical or behavioral therapy) and nonopioid pharmacotherapies, and to develop

treatments for pain that are safer and more effective than opioid analgesics (Volkow and Collins,

2017). While research in this area continues to develop, important questions remain about how

3

pain can be treated more effectively while minimizing potential unintended consequences such

as dependence and overdose.

Better Addiction Prevention, Treatment, and Recovery Services

Opioid use disorders, which, in 2016, affected over 2.1 million people in the United States

(Amhsbrak et al., 2017), contribute to medical morbidity, can promote risky behaviors, and often

complicate treatment for human immunodeficiency virus (HIV) and other comorbid conditions

(Becker et al., 2007; Becker et al., 2008; Johnson et al., 2013; Broz and Ouellet, 2008; CDC,

2012; Hall et al., 2008; Estrada, 2005). The availability of medication-assisted therapies has been

substantially improved in part because of collaborations between HHS agencies and public and

private stakeholders (Volkow et al., 2014), however, substantial gaps persist between the need

for treatment and the capacity to provide it (Saloner and Karthikeyan, 2015; Jones et al., 2015;

Feder, Krawczyk, and Saloner, 2017; Morgan et al., 2018; Hadland, Wharam, and Schuster,

2017). Thus, there is a critical need to better understand and address existing provider, patient,

and systemic barriers to treatment (Chou, Korthuis, and Weimer, 2016; Rinaldo and Rinaldo,

2013; Shen and Zuckerman, 2005; Cunningham and Nichols, 2005; Bradley, Dahman, and

Given, 2009; Schuur et al., 2009; Yoo et al., 2010; Kwiatkowski et al., 2000; Maddux and

Desmond, 1997; Clark et al., 2011; Burns et al., 2016) to improve access to treatment (Watkins

et al., 2017) and recovery services, and to ensure high-quality care (Chou, Korthuis, and Weimer,

2016; Gordon et al., 2016). To promote evidence-based prevention and treatment activities, $485

million in grants were distributed in 2017 to states through the 21st Century Cures Act, with

additional grants forthcoming based on further assessment of effective strategies and community

needs (Price, 2017).

Better Targeting of Overdose-Reversing Drugs

In 2016, more than 42,000 overdose deaths involved opioids; nearly 40 percent involved

heroin (Rudd et al., 2016; National Institute on Drug Abuse, 2017; CDC, 2017) and almost 45

percent involved synthetic opioids (e.g., fentanyl) (CDC, 2017). Overdose deaths often involved

multiple opioids or other medications such as benzodiazepines. Overdose-reversing drugs, such

as naloxone, play a critical role in preventing opioid overdose death. With the emergence of new

formulations of naloxone that can more easily be administered by individuals without medical

training (Merlin et al., 2015; Gupta, Shah, and Ross, 2016), efforts to encourage naloxone access

and use have grown rapidly, generally through three broad mechanisms: (1) community-based

distribution programs to expand community access to naloxone (Wheeler et al., 2015; Fairbairn,

Coffin, and Walley, 2017), (2) state laws and protocols encouraging bystanders to summon first

responders in the event of an overdose (Davis and Carr, 2015) and broadening the authority of

emergency services personnel and other first responders (e.g., law enforcement) to administer

naloxone (Davis, Southwell et al., 2014; Davis, Ruiz et al., 2014), and (3) policies to encourage

4

retail pharmacy dispensing of naloxone (Davis and Carr, 2017). Given the continued growth in

opioid-overdose mortality and influx of lethal synthetic opioids, promoting access to and use of

overdose-reversing drugs is essential to combat this public health crisis (Price, 2017).

Better Data

To understand effective strategies to reduce opioid misuse and associated harms and monitor

the evolving crisis, data are needed that can capture trends in opioid use, risk or protective

factors that influence the transition to risky use or opioid use disorder, and the risk among opioid

users of experiencing mortality or other harms. Given the rapidity with which opioid use and

markets have evolved over the past decade, developing and using public health surveillance

systems that offer near-real-time information have become essential. Historically, death

certificate and hospitalization data have been used to monitor drug use trends, but these sources

often suffer from data availability lags of one or two years. Variation in medical examiner and

coroner procedures in determining manner of death and the specific drugs involved in overdose

deaths also presents challenges for understanding the drug overdose crisis (Ruhm, 2017; Warner

et al., 2013).

Some states (e.g., Rhode Island) have made strides in improving the timeliness of reporting

for overdose deaths (Rhode Island Department of Health, 2015). Improved timeliness and

consistency of death certificate data can enable states and local communities to more rapidly

identify and respond to overdose spikes, facilitating timelier and more appropriately tailored

interventions (Houry, 2017). Federal programs, such as the CDC’s Data-Driven Prevention

Initiative (CDC, 2017) and Enhanced State Opioid Overdose Surveillance System (CDC, 2017),

are supporting the efforts of states and local authorities to track developments in the opioid crisis

and implement rapid and targeted responses (Price, 2017).

Additionally, better public health surveillance tools for monitoring medical and nonmedical

use of prescription opioids can promote public health and safety. Prescription drug monitoring

programs (PDMPs) are increasingly used to identify opioid analgesic prescribing trends (Katz et

al., 2010; HHS, 2013; O’Kane et al., 2016) and apply risk indicators for inappropriate prescriber

behavior (Ringwalt et al., 2015; Kreiner et al., 2017; Porucznik et al., 2014). Other large

databases, such as all-payers claims databases, are also valuable resources for understanding the

crisis, particularly if they are able to accurately link individuals over time and/or link to other

relevant data sources. However, the usefulness of such systems for analyses requires a data

infrastructure and legal authority for creating linked health databases that are not always

available.

Better Research

Data can be linked at various levels (e.g., individual, county, state, or multilevel linkages);

each approach offers benefits and challenges. Individual-level linkages and analyses are most

5

appropriate for inferring individual-level

relationships (Greenland, 2002; Robinson,

1950; Finney et al., 2011) and longitudinal

data can support analyses of individual-

level prescribing or treatment trajectories

as well as pathways that precede opioid

harms (e.g., overdose) or entry into

treatment. However, very few national data

sources can be linked at the person level,

and efforts to develop such linked data

sources and make them more accessible

must address statistical issues in generating

matches when unique identifiers or full

personal identifiable information are not

universally available across data sets

(Winkler, 2006; Winkler, 1999; Kum et al., 2014; Desetzina et al., 2014; Fellegi and Sunter,

1969). Potential benefits from individual-level analyses must also be balanced with potential

privacy concerns (Doshi et al., 2016; Kho et al., 2015; Ross and Krumholz, 2013). The need for

data owners to maintain protections for individual privacy may also limit the ability to create

person-level linked data files for research. Linking or analyzing data sources at more aggregate

levels is less resource-intensive, but such analyses may be more limited in their potential to

identify many key factors influencing the opioid crisis.

Box 2. General Steps for Conducting Data

Linkages

! Identify the necessary data sets.

! Obtain required approvals from regulatory

authorities, funding sources, and

institutional review boards.

! Select the data elements that will be used to

link across data sources.

! Determine the most appropriate method and

matching algorithms for linking.

! If a gold standard validation method is

available, assess match quality through

metrics such as sensitivity, specificity,

positive and negative predictive value.

The general steps for conducting data linkages are outlined in Box 2 (Bradley et al., 2010;

Dusetzina, Tyree, and Meyer, 2014; Dusetzina et al., 2014). Each step poses potential challenges,

and the most pronounced challenges generally arise in linking data at the individual level. These

include several institutional challenges for obtaining required data approvals. Linking and

obtaining approvals to use data sources hosted by different agencies, which may differ in their

legal obligations, interests, and resource capacities, can be burdensome, time-intensive, and

costly. Even when approval is obtained, there can be substantial statistical challenges in

conducting the linkages, exacerbated in data sets that lack common data elements. Choices must

be made regarding how to define unique person identifiers and to determine the best method(s)

for linking (e.g., deterministic or probabilistic matching, Bayesian approaches, or machine-

learning techniques; see Dusetzina et al. [2014] for a recent overview); and these choices will

influence the quality of matches (Campbell et al., 2008; Clark, 2004; Méray et al., 2007; Sayers

et al., 2016; Asnsolabehere and Hersh, 2017). Errors that may occur during this process, such as

errors of incorrectly linking records that do not belong to the same person (false positive) and

errors of incorrectly failing to link records that belong to the same person (false negative)

influence the rigor of subsequent analyses (Méray et al., 2007; Tromp et al., 2011).

6

In the following sections, we document the more-common types of data and linkages that

researchers are using to advance our understanding of the opioid crisis.

7

3. Current State of the Evidence: Findings from the Environmental

Scan

To gather information about data sets currently used in empirical studies, we conducted an

environmental scan, with special focus on research relevant to the HHS Strategic Priorities. We

also sought to identify common ways in which these data sources are being linked in existing

research.

We searched the peer-reviewed and grey literature for relevant articles, complemented by a

snowball approach, in which we reviewed citations and references in the articles we identified to

identify additional relevant materials that may not have been captured in the original search. As

part of this initial stage of work, we also conducted telephone conversations with five opioid

researchers currently using secondary data sources, several of whom also participated in the

stakeholder discussions described in Chapter 5, to ensure that the literature review did not miss

key data sources. These conversations confirmed the use of data sources identified in the

literature scan but did not identify any additional data sources. In total, we identified 278

documents that we reviewed for the scan, of which 250 were peer-reviewed publications; the

remainder were largely reports, working papers, and newspaper or internet articles.

Below, we summarize the environmental scan’s main findings, grouping research topics,

variables, and data sources by HHS Strategic Priority. The discussion focuses on highlighting

more-common research questions evaluated in the existing literature, as well as the more-

common specific secondary data sources and measures used to answer such questions. Chapter 4

categorizes the types of secondary data sources used in research related to HHS Strategic

Priorities, with more general discussion of differences across data source types. Other important

but less commonly used data sources are described in Chapter 5.

Better Practices for Pain Management

Research has improved the understanding of opioid analgesic prescribing patterns,

prescription fill behavior, and prescription characteristics predictive of misuse or opioid-related

harms. Research has also improved the understanding of the effectiveness of states’ efforts to

advance better pain management practices. PDMPs are the most commonly studied state

initiatives, with more limited research examining the effects of laws-regulating “pill mills,” (i.e.,

clinics prescribing high volumes of opioids with limited clinical oversight), abuse-deterrent

opioid formulations, pain management education, and prescribing guidelines. Table 3.1 lists data

sources and measures commonly used in research related to pain management practices

identified through the environmental scan.

8

The measures identified in Table 3.1 can be used to evaluate how PDMP implementation

affects opioid-related consequences. The measures can also be used to evaluate the trends in

opioid analgesic prescribing and associations with risky prescribing or opioid-related harms.

!"#$%&'()(&*+,,+-$.&/0%1&2"3"&4+567%0&"-1&8%"056%0&9-&:%0%"67;&3+&<1="-7%&>%33%6&?"9-&

8"-"@%,%-3&?6"7397%0&

Data Type Commonly Used Sources Commonly Used Measures

Commercial

insurance

claims

&

• IQVIA

• Marketscan

• Health Care Cost Institute

• Opioid analgesic prescriptions

• Prescribing patterns or prescription-fill behavior

indicative of misuse

• Morphine equivalent daily dose (MEDD)

• Payment type (e.g., Medicare Part D, cash)

Medicaid

claims

• Medicaid State Drug

Utilization file

• State Medicaid data

sources

• Opioid analgesic prescriptions

• Prescribing patterns or prescription-fill behavior

indicative of misuse

• MEDD

• Diagnostic codes for nonfatal overdose

• Payment type

Medicare

claims

• Medicare Prescription Drug

Event data linked to

Medicare Beneficiary

Summary File

• Opioid analgesic prescriptions

• Prescribing patterns or prescription-fill behavior

indicative of misuse

• MEDD

• Diagnostic codes for nonfatal overdose

• Payment type

Electronic

health

records

(EHRs) and

claims data

• National or regional

Veterans Health

Administration (VHA) data

warehouses

• Opioid analgesic prescriptions

• MEDD

• Indicators of prescription opioid abuse or dependence

• Clinical diagnoses (e.g., pain conditions)

PDMP data

• State PDMPs • Opioid analgesic prescriptions

• MEDD

• Prescribing patterns or prescription-fill behavior

indicative of misuse

Mortality data

• National Death Index (NDI)

• National Vital Statistics

System Multiple Cause of

Death (NVSS MCOD)

• CDC WONDER

• State death certificate data

• Opioid overdose fatality

• Injury intent (e.g., suicide, accidental)

Policy data

• Prescription Drug Abuse

Policy System (PDAPS)

• National Alliance for Model

State Drug Laws

(NAMSDL)

•

PDMP enactment

• PDMP design features

Several common research questions can be addressed using a single data source. For

example, research evaluating time trends or geographic variation in opioid analgesic prescribing

9

A

among the general population has used information from state-specific PDMPs or from

commercial insurance claims such as IQVIA. Other studies have assessed prescribing practices

within the Medicaid, Medicare, or veteran populations using administrative claims or EHR data

sets specific to those populations. Five opioid-related indicators and their respective algorithms

developed by CMS for researchers to use with Medicaid and Medicare administrative claims

data were recently made available for public comment (CMS, 2018); these indicators are planned

for inclusion in the CMS Chronic Conditions Data Warehouse.

However, other research questions rely on linked data sets. Research evaluating the effects of

PDMP implementation on opioid-related consequences commonly merges state-level policy data

with state- or county-level data on opioid prescription claims or rates of fatal opioid overdose

from the NVSS MCOD microdata, CDC WONDER, or state-specific death certificate data.

These analyses also generally control for state- or county-level factors linked from other data

sources, such as those noted in Table 3.2. The commonly used state- or county-level measures in

Table 3.2 can be linked with data on opioid-related consequences and state policy data to control

for potential time-varying community-level confounders correlated with opioid outcomes of

interest. These measures can also be used to estimate how community-level factors relate to

opioid analgesic use and associated harms. Community-level factors of interest generally include

socioeconomic factors (e.g., unemployment rate), demographics (e.g., percentage population

male), or measures of health care infrastructure (e.g., physicians per capita).

!"#$%&'(B(&*+-3%C35"$&2"3"&4+567%0&"-1&8%"056%0&*+,,+-$.&D9-E%1&3+&FG9+91&F537+,%&2"3"&9-&

:%0%"67;&:%$"3%1&3+&3;%&H9=%



?+9-3&II4&436"3%@.&

Data Sources Commonly Used Measures

Bureau of Economic Analysis

• Unemployment rate

• Per capita income

Area Resource Files or Health

Resources Files

• Unemployment rate, per capita income, urban-rural status

• Demographics (e.g., age, sex, race/ethnicity distribution)

• Number of hospital beds per capita, physician density

American Community Survey

• Poverty rates, unemployment rate, education distribution

• Median home prices, median age of housing stock

• Demographics (e.g., age, sex, race/ethnicity distribution)

• Rates of public and private health insurance coverage

Current Population Survey

• Rates of health insurance coverage

• Demographics (e.g., age, sex, race/ethnicity, marital status)

•

Unemployment rate; poverty rates

CMS

• Rates of Medicaid and/or Medicare coverage

Studies evaluating the association of opioid analgesic prescribing patterns or prescription-fill

behavior with opioid-related harms often require data sources linked at the individual level.

10

Noted data-linkage strategies include linking state-specific PDMP data with other data sources,

such as Medicaid administrative claims, hospital discharge data, or vital records; using multiple

linked VHA databases, which have also been linked at the individual level to mortality data from

the NDI; linking Medicaid claims with state vital records data; and using Medicare Prescription

Drug Event data linked with the Medicare Beneficiary Summary file. While not commonly used

in existing opioid-related research, information from the Medicare Current Beneficiary Survey, a

survey of a nationally representative sample of Medicare beneficiaries released three times

annually, has been linked at the patient-level to Medicare billing claims (Wright et al., 2014).

Better Addiction Prevention, Treatment, and Recovery Services

Researchers commonly evaluate how policies intended to expand the number of waivered

buprenorphine prescribers (i.e., prescribers who have received a waiver from the Drug

Enforcement Agency (DEA) allowing them to prescribe buprenorphine for the treatment of

opioid use disorder) relate to buprenorphine prescribing, factors that predict the availability of

waivered prescribers, and factors associated with the monthly patient censuses of waivered

prescribers. Some studies investigate patterns of buprenorphine use among those receiving

opioid use disorder treatment. Data sources and measures commonly used in research related to

opioid use disorder and treatment are shown in Table 3.3.

The measures in Table 3.3 may be used to evaluate trends and geographic variation in

treatment need and opioid agonist treatment capacity, as well as associations between individual-

level characteristics, opioid analgesic use, and opioid use disorder. They can also be used to

evaluate trends, geographic variation, and factors associated with buprenorphine physician

supply. Lastly, they can be used to evaluate national trends and patient trajectories in treatment

for opioid use disorder.

11

!"#$%&'('(&*+,,+-$.&/0%1&2"3"&4+567%0&"-1&8%"056%0&9-&:%0%"67;&3+&J,G6+=%&<119739+-&

?6%=%-39+-K&!6%"3,%-3K&"-1&:%7+=%6.&4%6=97%0&&

Data Type Commonly Used Sources& Commonly Used Measures

Commercial

insurance

claims

• IQVIA&

• Symphony Health

• Buprenorphine prescriptions

• Patient censuses of buprenorphine prescribers

PDMP data

• State-specific PDMPs • Buprenorphine prescriptions

• Patient censuses of buprenorphine prescribers

Medicaid

claims

• National or state Medicaid

data sources

• Buprenorphine prescriptions

• Patient censuses of buprenorphine prescribers

• Opioid use disorder diagnoses

EHR

• HealthCore Integrated

Research Database&

• Group Health Cooperative

• National or regional VHA

data warehouses

• Prescription opioid abuse or dependence

• Diagnostic measures of pain

• Opioid analgesic prescriptions

• Other clinical diagnoses, comorbidities, demographic

characteristics

Household

surveys

• National Survey on Drug

Use and Health (NSDUH)&

• National Epidemiologic

Survey on Alcohol and

Related Conditions

(NESARC)

• Opioid use disorder treatment need

• Treatment source or source of payment

• Opioid use disorder

• Nonmedical prescription opioid misuse

• Other substance use disorders, mental health

conditions, and demographic characteristics

Treatment

facility

surveys

• Treatment Episodes Data

Set-Admissions (TEDS-A)&

• National Survey of

Substance Abuse

Treatment Services (N-

SSATS)

• Number of patients receiving methadone in opioid

treatment programs (OTPs)

• Outpatient operating capacity of OTPs

• Number of substance abuse treatment programs

providing methadone and/or buprenorphine

• Substance abuse treatment services offered

• Number of t

reatment admissions for opioid use disorder

Provider

census

• Substance Abuse and

Mental Health Services

Administration (SAMHSA)

database&

• DEA Active Controlled

Substances Act

Registrants Database

(ACSA)

• Number of buprenorphine providers

• Waiver limits

• Buprenorphine treatment capacity

Policy data

• RAND/National

Conference of State

Legislators Survey

• State Medicaid reimbursement policies for

buprenorphine

Research studying associations between individual-level characteristics, opioid analgesic use,

and opioid use disorder leverages data sources that contain person-level information on these

measures within the same data set. Relevant data sources include household surveys such as the

NSDUH series managed by SAMHSA, NESARC sponsored by the National Institute on Alcohol

12

Abuse and Alcoholism, as well as EHR and claims data from various sources (Table 3.3).

Research examining trends or geographic variation in demand or capacity for opioid use disorder

treatment instead often uses measures from treatment facility surveys, such as the TEDS-A or N-

SSATS, both of which are maintained by SAMHSA.

While studies assessing trends or geographic variation in treatment need and treatment

capacity may advance research using measures from a single data source, a more comprehensive

picture of the relationship between demand for and supply of treatment has been obtained by

linking data sources. For example, studies estimating treatment shortage areas commonly merge

information on treatment need with information on treatment capacity at the state- or county-

level.

Researchers have also used data linkages to better understand factors associated with

buprenorphine prescriber supply and buprenorphine utilization. Information on buprenorphine

prescriber locations is available through two commonly used sources: SAMHSA’s

Buprenorphine Waiver Notification System or the Drug Enforcement Agency Active Controlled

Substances Act Registrants database (DEA ACSA). Information on buprenorphine prescriptions

often comes from insurance claims data or PDMP data. By linking information on buprenorphine

prescribers or prescriptions with state-level policy and county-level contextual factors relevant

for opioid use disorder treatment, research can improve the understanding of factors associated

with buprenorphine treatment capacity and utilization.

Better Targeting of Overdose-Reversing Drugs

The most commonly studied interventions promoting use of overdose reversing drugs are

community-based overdose education and naloxone distribution (OEND) programs. Emerging

evidence focuses on state laws intended to increase naloxone access through retail pharmacy

distribution channels (Naloxone Access Laws) or to encourage community bystanders to

summon emergency aid or administer naloxone in the event of witnessing an overdose (Good

Samaritan Laws). Table 3.4 lists the most commonly used variables and secondary data sources

identified in research related to overdose-reversing drugs.

The measures noted in Table 3.4 can be used to evaluate trends or geographic variation in the

distribution of naloxone through retail pharmacies, presence of community-based OEND

programs, and naloxone administrations by emergency medical services (EMS) personnel. They

can also be used to study how state naloxone policies influence opioid overdose mortality or the

role of OEND programs in impacting knowledge about how to respond to a witnessed overdose,

distribution of naloxone kits and naloxone administrations, and overdose reversals.

13

!"#$%&'(L(&*+,,+-$.&/0%1&2"3"&4+567%0&"-1&8%"056%0&9-&:%0%"67;&3+&J-M+6,&>%33%6&!"6@%39-@&+M&

F=%61+0%-:%=%609-@&265@0&

Data Type Commonly used sources Commonly used measures

Commercial

insurance claims

• IQVIA • Naloxone prescriptions through retail

pharmacy channels

• Prescriber specialty

• Patient age, gender

Mortality data

• CDC WONDER

• NVSS MCOD

• Opioid analgesic overdose deaths

• Heroin overdose deaths

• Synthetic opioid overdose deaths

OEND program data

• Massachusetts Opioid

Overdose Prevention Pilot

Program

• Harm Reduction Coalition

• Reported overdose reversals

• Number of naloxone administrations

• Number persons trained and naloxone kits

distributed

• Knowledge about how to respond to a

witnessed overdose and administer naloxone

EMS data

• NEMSIS

• EMS naloxone administration

Policy data

• PDAPS

• Network of Public Health

Law (NPHL)

• Legal databases

• Good Samaritan laws

• Naloxone access laws

Research on policies or programs to expand naloxone use often rely on data from a single

source. Studies of the effects of community-based OEND programs on overdose knowledge and

outcomes generally rely on case studies using surveys of OEND program participants or other

data collected by the specific OEND programs. Other research has documented the evolution of

state laws governing naloxone access and use, drawing on review of legal databases to obtain

information about state policies related to naloxone access and use for community bystanders or

first responders. Finally, some studies have described trends in naloxone distribution through

different channels using retail pharmacy naloxone distribution (IQVIA) or EMS naloxone

administration (National Emergency Medical Services Information System [NEMSIS]).

Data linkages are most commonly used to examine the effects of state naloxone policies or

OEND programs on opioid overdose. Such research commonly merges state- or county-level

mortality data from the NVSS MCOD microdata or CDC WONDER with state-level information

on naloxone access policies or Good Samaritan Laws compiled by the Prescription Drug Abuse

Policy System (PDAPS) or the NPHL program. Studies of state naloxone policy effects also

commonly control for other state- or county-level contextual factors as described in Table 3.2.

Other state-specific analyses use multiple complementary data sources to examine whether

implementation of a community OEND program (Albert et al., 2011) influences trends in

emergency department visits for substance abuse and accidental poisonings, opioid overdose

mortality, and outpatient-dispensed controlled substances.

14

Better Data

Researchers concerned with surveillance often use multiple complementary data sources to

better understand trends and disparities related to the opioid crisis, develop methods to improve

monitoring through existing public health surveillance systems (e.g., EHR, emergency

department encounter data), identify patients at high risk of prescription opioid misuse or abuse,

and promote improved opioid toxicosurveillance (i.e., rapid analysis of drug exposure data).

Below we briefly describe the data sources and measures most commonly used to strengthen

public health surveillance research.

Much public health surveillance research uses near-real time surveillance tools to better

understand product-specific abuse and emerging trends. Three databases have been designed to

provide near-real-time surveillance data on opioid misuse: the Researched Abuse, Diversion and

Addiction-Related Surveillance System (RADARS), the National Addictions Vigilance

Intervention and Prevention Program (NAVIPPRO), and the Prescription Behavior Surveillance

System (PBSS). The RADARS and NAVIPPRO compile information on opioid use,

consequences, and markets from multiple sources; the PBSS compiles state-specific PDMP

information from several states. In addition, opioid overdose information collected from poison

control centers through the National Poison Data System (NPDS) has been used by research and

surveillance efforts to capture product-specific opioid overdose events that may not result in

death.

Data costs or other barriers to access may limit widespread use of these systems in existing

research; however, they are increasingly used in studies related to problematic opioid use and

product-specific abuse trends. Data collected through online social media has also been

increasingly used to monitor illicit or problem opioid use (Parker et al., 2017; Katsuki et al.,

2015; Anderson et al., 2017).

Significant progress has been made in developing metrics and leveraging existing

surveillance systems to better detect opioid misuse or potentially inappropriate prescribing. As

detailed in the prior sections, information on opioid prescriptions and opioid misuse indicators

are available through multiple data sources, including claims and EHR data. State-specific

PDMP data and all-payers claims databases (APCDs) are also emerging as useful data sources to

better understand opioid prescribing and potential misuse. While we identified fewer studies

examining illicit opioids, some studies have used local law enforcement data on drug seizures or

arrests to better understand heroin markets, illicit opioid analgesic markets, and illicit markets for

synthetic opioids. Other research using RADARS, NAVIPPRO, and the NSDUH has examined

sources of prescription opioids and measures of prescription opioid diversion.

A common data-linking strategy for public health surveillance is to leverage multiple data

sets and conduct complementary analyses of state- or county-level information to better

understand the evolution of the opioid crisis. For example, studies have linked individual-level

15

prescription data from PDMPs or Medicaid claims with state death certificate data to examine

trends in prescribing behavior preceding overdose death.

States are also implementing strategies to better link and analyze data across state agencies.

For example, with Chapter 55 of the Acts of 2015, Massachusetts’ Department of Public Health

has connected ten data sources managed by five state agencies to develop a data warehouse

structure. These data sources include the state APCD; the Massachusetts PDMP; death certificate

records and toxicology results; substance abuse treatment information; hospital, emergency

department, and outpatient records; incarceration and criminal justice system treatment records;

and emergency medical service incident data from licensed ambulance services. Chapter 55 is

discussed further in Chapter 6 of this report.

16

4. Sources of Secondary Data: Data Inventory Findings

In Chapter 3, we provided an overview of

the more commonly identified research

questions that secondary data sources have

been used to examine, organized by HHS

Strategic Priorities. However, our

environmental scan uncovered a broader array

of existing data resources relevant to the HHS

Strategic Priorities. In Table 4.1, we

categorize and describe the types of additional

secondary data sources and provide examples

of common data sources and variables within each type.

Box 3. Major Sources of Secondary Data

! National surveys

! Claims and EHR data sources

! Mortality record data sources

! Prescription drug monitoring data

sources

! Contextual and policy data sources

! Other national, state, or local data

sources

Box 3 highlights the six broad sources of data we identified: (1) national surveys, (2) EHR

and claims data, (3) mortality records, (4) prescription drug-monitoring data, (5) contextual and

policy data, and (6) other national, state, or local data sources (e.g., national poison control

center data, state arrest records). The full data inventory provided in the appendix to this report

contains more-detailed information on each identified data set within these broader categories.

This information includes the agency hosting the data and type of data; a high-level summary of

data content, including geographic coverage, timing of collection or data availability, and

important measures; information on accessing the data, including a link to the website,

information on access costs, and other restrictions; a link to any available analytics; and

information on linking capability.

17

Table 4.1. Data Source Categories Identified

Data Description Summary Examples of Important Measures Data Source Examples

National surveys

Description: Generally household or school4based surveys with self4

reported information on drug use and health; other surveys are of

hospitals, treatment facilities, or other medical service providers

Geographic coverage: National

Timing: Generally collected and available annually

Prescription opioid use, heroin use,

opioid use disorder, medical

conditions, health care utilization

National Survey on Drug Use and

Health, National Ambulatory Medical

Care Survey, National Survey of

Substance Abuse Treatment Services

Data, Medical Expenditure Panel

Survey

EHR Description: An EHR contains the medical and treatment histories of

patients. However, it often contains more than standard clinical data,

and may also include a broader view of a patient’s care. EHRs may

contain a

patient’s medical history, diagnoses, medications, treatment

plans, allergies, radiology images, and laboratory and test results

Geographic coverage: Varies by source

Timing: Near4real time or real4time collection

Previously prescribed opioids or

other medications; patient history,

medications, clinical conditions,

treatment plans, and lab/test

results; may include clinician notes

Stanford Translational Research

Integrated Database, HealthCore

Integrated Research Database, Group

Health Cooperative in Washington

State

Claims data

Description: Patient4level claims data for reimbursement for services

submitted by health care providers and pharmacies to insurance

companies. Validated algorithms to identify opioid misuse or abuse

from claims data are being developed

Geographic coverage: Varies by source

Timing: Varies by source

Prescription drug utilization; service

utilization

IQVIA, Symphony Health, Truven

Marketscan data, Medicaid claims,

Medicare Part D Prescription Drug

Event data

Mortality records

Description: Death rates and causes of death by drug compound

and/or International Classification of Diseases code. Additional

information can include toxicology reports

Geographic coverage: National or single state

Timing: Generally available annually

Rates of opioid4involved deaths;

drugs involved in overdose deaths

CDC WONDER Multiple4cause4of death

data; Fatal Accident Reporting System;

NDI

Prescription

monitoring data

Description: Data systems to track and monitor the distribution or

prescription of controlled substances

Geographic coverage: Varies by source

Timing: Varies by source

Opioid prescribing rates (by type);

indicators of "doctor shopping,"

coprescribing of opioids and other

controlled drugs, geographic

variation in opioid distribution

Automation of Reports and

Consolidated Orders System (ARCOS);

state prescription drug–monitoring

programs

Contextual and

policy data

Description: Causal analyses of the effects of policy changes on

opioid4related outcomes generally use data on state laws from these

sources and/or includes controls for state or county characteristics to

support causal interpretation

Geographic coverage: National

Timing: Varies, but generally semiannually

State opioid policies, state and

county demographic and

socioeconomic factors, state and

county health care variables

Area Health Resources Files, Policy

Surveillance System, PDAPS

Other national,

state, and local

sources

Description: Includes data collected through law enforcement,

national public health surveillance systems (e.g., poison control center

data, emergency department visit data), OEND program data, other

hospitalization and emergency department data

Geographic coverage: Varies by source

Timing: Varies by source

Law enforcement drug seizures,

nonfatal opioid overdose, opioid4

related emergency department visits

and hospitalizations, naloxone

distribution through community

organizations

NEMSIS, NPDS, HCUP emergency

department and hospitalization data

18

National Surveys

National survey data sources, often collected annually, include population-based surveys,

such as household surveys or school-based surveys, as well as surveys of medical providers,

hospitals, emergency departments, and treatment facilities. Population-based surveys often

include self-reported information on lifetime or current heroin or opioid analgesic use, symptoms

of opioid use disorder, and treatment or unmet treatment need for opioid use disorder; as well as

a variety of measures describing respondent demographics, socioeconomics, and other mental

health or substance use behaviors. Systematic data collection over time supports trend analyses at

the national and sometimes state or local level; however, significant changes to survey design or

implementation may limit longitudinal comparisons.

One caveat with regard to many national population-based surveys is that they restrict their

sample to the civilian, noninstitutionalized population, thus excluding some high-risk groups,

such as homeless individuals not residing in shelters and incarcerated individuals. However, a

few national surveys, such as the Arrestee Drug Abuse Monitoring System (ADAM) and the

National HIV Behavioral Surveillance System, have focused specifically on high-risk

populations, arrestees, and persons at risk for HIV infection.

Other national survey data-collection efforts gather information from hospitals, emergency

departments, and outpatient departments. These data sources offer information on prescriptions

received through various health care settings as well as acute health care visits attributable to

opioid use or misuse; data from three of these surveys have been integrated into the National

Hospital Care Survey (CDC, 2015). Finally, national surveys of mental health or substance abuse

treatment facilities collect information relevant to treatment utilization and treatment capacity for

opioid use disorder.

While most national survey data sources (with some exceptions, see Table A.1 in the

appendix) allow public access at no cost, access to certain data elements may be restricted.

Restricted data elements often include geocoded variables that would allow analyses or linkages

at the state or substate level. Obtaining access to these geocoded variables typically involves an

application process; use of such information is often only allowed through a Research Data

Center (U.S. Census Bureau, 2015) or other secure access data portal and, in some cases, is

restricted to use by federal employees. Similarly, while several national surveys permit person-

level linkages with other national data sources (e.g., the National Health Interview Survey [CDC,

2017] supports person-level linkages with the NDI, Medicare data sources, and AHRQ’s Medical

Expenditure Panel Survey) upon approval of the research project, access to the linked files is

typically only permitted through secure Research Data Centers. Currently, national survey data

from substance use treatment facilities may not be linked to units below the county level.

19

Electronic Health Records and Claims Data

An EHR is an electronic version of a patient’s medical history. It may include a variety of

key clinical data, including demographics, medical history, medications, progress notes,

problems, and other physician or nurse documentation. Efforts to expand the adoption and use of

EHRs have been focused primarily on improving the quality of health care (Appari et al., 2013;

Blumenthal and Tavenner, 2010; Campanella et al., 2016). However, there has been growing

interest in using EHR data for public health surveillance and response efforts (Friedman, Parrish,

and Ross, 2013; Coorevits et al., 2013). EHRs have been proposed as a tool to help practitioners

implement better pain assessment and management practices (Anderson et al., 2016; Harle et al.,

2014), as well as a potential data resource to better identify factors associated with opioid

misuse, adverse events, or development of opioid use disorder (Lingren et al, 2018; Hser et al.,

2017; Green et al., 2017; Carrell et al., 2017). Typically available in real time, EHR systems may

contain a variety of measures, such as health behaviors indicative of opioid misuse, that may not

be needed for billing purposes and thus would not be captured in claims data. For example,

EHRs may contain relevant laboratory values, such as urine drug screens, as well as allowing a

calculation of abandoned opioid analgesic prescriptions (prescriptions that are written but never

filled by patients).

However, there are several challenges to using EHR data, including issues with fragmented

or incomplete data, the need for text note processing and validation, and a lack of consistency in

methods to assess EHR data quality (Madden et al., 2016; Weiskopf and Weng, 2013; Häyrinen,

Saranto, and Nykänen, 2008; Raghupathi and Raghupathi, 2014). Data-quality concerns can

generate serious issues in determining unique patient identifiers, which in turn creates errors in

person-level record linkage with other data sources (McCoy et al., 2013; Murray, 2014).

Challenges with gaining approvals and access to EHR data may also restrict the use of EHR data

in secondary research (Russo et al., 2016).

Table 4.2 compares EHR and administrative claims data sources. Because claims data are

intended to support reimbursement for services submitted by health care providers and

pharmacies to insurance companies, they tend to have fewer data-quality issues, have a more-

standardized structure and method for entering data, and assign standardized definitions for data-

point entry. Claims records can come from data sources hosted by a single federal insurer, single

state insurer, integrated database of a privately insured population, multipayer claims database

owned by a private agency, or state all-payer claims database. While access restrictions are often

not as burdensome as those for EHR data, the required approval process and costs of obtaining

person-level claims data may be a barrier to use for research purposes.

20

Table 4.2. Comparison of Electronic Health Record and Administrative Claims Data

EHR Data Insurance Claims Data

Coverage or source of data set

(examples)

● Single institution (private)

● Health information exchanges or group health

network

● National or regional VHA systems

● Commercial claims from private payers

● Federal and state claims (Medicaid, Medicare)

● Integrated databases with medical and pharmacy

claims

Potential scope of patients All patients, including those with no insurance

coverage (in systems that have adopted an

EHR)

Insured patients, may be restricted to single payer

population

Breadth of data Richer data but greater variability in data

element availability

More limited set of data elements but more

standardized collection

Prescription data Information on whether medication was

prescribed, not whether it was filled or refilled

Detailed information on filled prescriptions and

refilled prescriptions (assuming there was a claim)

Data structure and quality Data format, completeness, and overall quality

can vary greatly. Researcher may need to

operationalize how variables of interest are

defined, and this may look different with different

EHRs

Fairly standardized claim data formats, although data

warehouse structures can vary by payer. Variables

(e.g., diagnostic codes, drug dispensing) typically

well-defined and complete when required for

payment

Data access May require on-site access, remote access may

be restricted to limited data set, security

protocols, costs unclear

Costs vary depending on request. Some data must

be requested and approved. Varying privacy levels

for some CMS Medicaid and Medicare files

21

Mortality Records

Information on opioid overdose mortality from death records and postmortem toxicology

data can be obtained from state-specific death certificate or from national data sets that compile

death certificate data submitted by states into a single source. There are three primary sources for

national data on mortality, one of which (CDC WONDER [CDC, 2018]) is publicly available,

does not carry fees, and supports readily downloadable data files through an easy-to-use online

system. However, the public version of the multiple-cause-of-death files provided through CDC

WONDER masks subnational estimates in which fewer than ten deaths occurred. Thus, for

county- or state-level analyses stratified by demographic variables—where cell sizes may

become quite small—obtaining access to the underlying NVSS MCOD microdata may be

necessary (national opioid mortality data analytics are available online [CDC, 2017]). While one

limitation of mortality data is the long lag time for data to become available, the Vital Statistics

Rapid Release Provision Drug Overdose Death Counts (CDC, 2018) is an effort by the National

Center for Health Statistics to provide timelier information on drug overdose mortality based on

provisional fatality counts from the NVSS MCOD.

While both CDC WONDER and NVSS MCOD support linkages and county-level analyses,

person-level linkages with national geographic coverage are only supported through the NDI, a

centralized national database of death records that is not available to the general public, has a fee

schedule with charges per record requested, and entails costs to obtain cause-of-death

information. The NDI can be linked at the individual level to multiple other data sources,

including national surveys, VHA health care data, and other national or state sources. State death

records, while not publicly available, can also be linked at the person level to other state-specific

databases, including PDMP data.

Prescription Drug–Monitoring Data

Prescription drug–monitoring data sources are those designed to monitor controlled

substance prescribing, distribution, or dispensation. These include a federal database monitoring

national distribution of controlled substances from manufacture to sale (i.e., ARCOS) as well as

state PDMP systems, electronic databases generally hosted by a state licensing, health, or

criminal justice agency and intended to track controlled prescription drugs dispensed to patients

within the state (Pardo, 2017). The lag time for data reporting, degree of coverage, ability to

identify providers, and specific measures captured within a given PDMP system vary across

states depending on the state law regulating the PDMP (Greenwood-Ericksen et al., 2016;

Manasco et al., 2016).

States also vary in the degree to which their state PDMP system allows interstate information

sharing, authorizes access for research and public health purposes, and/or permits person-level

linkage to other state-owned data sources. As of December 5, 2017, 48 states and U.S. territories

22

are authorized to provide de-identified PDMP data to researchers, and 25 of these states have

released PDMP data for research, epidemiological, or educational purposes (PDMP, 2017).

Contextual and Policy Data

Contextual data sources are generally used in opioid research to assess state- or county-level

factors associated with opioid-related outcomes or to account for time-varying state- or county-

level demographic, health care, or socioeconomic factors that may confound estimation in

analyses of policies targeting opioid use, treatment, or opioid-related harms. When used in

research related to the HHS strategic areas, measures derived from contextual data sources are

generally obtained at more aggregate levels (e.g., state, county) or are aggregated up from

person-level data sources to the state or county level.

Most contextual data sources are hosted by federal agencies, although some private

organizations (e.g., Kaiser Family Foundation) and some federal entities (e.g., the Health

Resources and Services Administration) compile information from several federally hosted

contextual data sources into a single location and also maintain their own data sources.

Depending on the source, data may be representative at the state or substate level, with supported

linkage or unit of analysis as finely geographically detailed as the ZIP level (e.g., the U.S.

Census Bureau Zip Code Business Patterns data) (Cerdá et al,. 2017), although this level of detail

is generally not available in public data sets. Additionally, contextual information compiled from

national person-level survey data sources (e.g., the Current Population Survey) is less likely to be

representative at the substate level (Blewett and Davern, 2006) or to provide microdata for all

counties. Reviewing all contextual data sources identified through the environmental scan was

outside the scope of this project. However, we highlight a few of the most commonly used data

sources in Table A.5 in the appendix.

Policy data sources capture information on state opioid policies and thus are generally

analyzed and linked using state as the unit of analysis. A variety of agencies, including federal,

federally funded, and private organizations, collect information on state opioid policies.

Information on state PDMP policies, naloxone access laws, and Good Samaritan laws have been

compiled by several sources, including PDAPS and NAMSDL, although these sources often vary

in the exact classification they use to define the components and timing of such laws. In many

cases, policy data are publicly available at no cost. However, free and publicly available policy

data are often not provided in analytic formats or as a historical data set; instead, they often

represent a “snapshot” of current policies. Additionally, few data sources are available that

systematically track and provide information on how state opioid policies are being

implemented, note changes in local efforts related to the opioid crisis (e.g., law enforcement

carrying naloxone), or describe large-scale opioid policies or guidelines implemented by payers

or health care systems to address opioid prescribing.

23

Other National, State, and Local Sources

Several data sources relevant to the HHS opioid strategies do not directly fit within any of the

aforementioned categories. These include national censuses of waivered buprenorphine

providers; national proprietary data systems, such as RADARS, that combine information from

various sources to describe and surveil misuse, abuse, and diversion of prescription drugs; and

national data on emergency medical services utilization such as NEMSIS, drugs seized by law

enforcement, and calls to poison control centers.

This data source category also includes a suite of national- and state-level data products

capturing hospital inpatient stays and hospital-based emergency department visits available

through the Healthcare Cost and Utilization Project (HCUP), managed by the Agency for

Healthcare Research and Quality (AHRQ). Access to the state or national HCUP data files must

be applied for and purchased; however, the HCUP website offers a publicly available online

query system (Agency for Healthcare Research and Quality, 2018) and a limited set of user-

friendly graphics and tables showing state and national trends in opioid-related inpatient stays

and emergency department visits (Healthcare Cost and Utilization Project, 2018). Finally,

increased public attention to the opioid crisis has led to the emergence of online state opioid

dashboards; new opioid data-compilation efforts; as well as increased attention to data sources

that may capture the complex role of clinical conditions, health care delivery and access,

prescribing, and opioid misuse or development of opioid use disorder (see Box 4 for examples).

24

Box 4. Other Data Sources Relevant to the HHS Strategic Priorities

The data inventory was intended to provide an overview of commonly used secondary

data sources in research related to the HHS strategic areas. It is not an exhaustive list of

secondary data currently or potentially available to further our understanding of the opioid

crisis. We here note several data sources that are not commonly used in existing research, but

may be of interest.

! State opioid dashboards provide state statistics related to the opioid crisis. Examples

include

• Arizona Department of Health Service’s Arizona’s Real-Time Opioid Data

(2017–2018)

• Minnesorta Department of Public Health’s Opioid Dashboard (undated)

• Tennessee Department of Health’s Drug Overdose Dashboard (undated)

! National opioid data collections compile or support the compilation of relevant data

from a variety of sources into a single location. Examples include

• Opioid and Health Indicators Database by amfAR (undated), the Foundation

for AIDS Research

• Opioid Mapping Initiative (undated), an open-data project with several

participating local governments and local agencies

! PCORNet Clinical Data Research Networks include a range of participating health

care–based networks (pcornet, undated) engaged in partnering to link claims and EHR

data. These include resources such as the Chicago Area Patient Centered Outcomes

Research Network (Capricorn, undated) and OCHIN’s Data Warehouse (OCHIN,

2014–2018)

! The Health Resources and Services Administration (HRSA)’s Health Center Program

offers several resources, including

• HRSA’s Uniform Data System (HRSA, 2018) provides publicly available

aggregate data on patients who have opioid use disorder diagnoses or who are

receiving medication-assisted treatment through HRSA-funded health center

grantees and lookalikes.

• The Health Center Patient Survey (HCPS) data, made available with support

from Assistant Secretary for Planning and Evaluation, provides information on

health center patients’ conditions and demographics, health behaviors, service

use, and satisfaction (HRSA, undated).

25

5. High-Priority Research Needs and Data Efforts:

Findings from the Stakeholder Discussions

To assess high-priority research areas and data efforts relevant to the HHS strategy, we

conducted a set of stakeholder discussions to gather insights into opportunities to enhance data

collection and data linkages. In consultation with staff within the Office of the Assistant

Secretary for Planning and Evaluation, we identified 25 key stakeholders with particular

expertise or research experience related to the HHS strategy, 16 of whom participated in phone

discussions. Each discussion was tailored and focused on the HHS strategy about which the

stakeholder was most knowledgeable.

In this section, we highlight themes that emerged from stakeholder discussions of research

opportunities using secondary data sources to support the HHS strategy. We also provide a table

summarizing strengths and limitations of data sources that stakeholders referenced with respect

to each Strategic Priority. The appendix to this report provides additional data source details.

Better Practices for Pain Management

Common themes emerging from discussions related to key research aims for advancing

better practices for pain management include:

• Opioid prescribing guidelines and clinician education: Better documentation of

opioid-prescribing guidelines and clinician education requirements, linked with outcome

data at the prescriber or patient level, would shed light on how variation in these

protocols relates to variation in treatment for pain, and how this in turn impacts patient

outcomes.

• Nonopioid treatments for pain: Opioid analgesics may not be more effective than other

treatments in the management of many tyes of long-term pain (Krebs et al., 2010; Krebs

et al., 2018). More evidence is needed regarding the full range of long-term effective

treatments for chronic pain, including combinations that might be more effective than

opioid analgesics.

• Patient trajectories: Longitudinal patient-level data linking prescriptions with outcomes

can enhance better understanding of the pathways and sequences of events leading to

adverse outcomes such as hospitalization and overdose death. Medicaid and commercial

claims data can be useful, but each provide information on only one population and often

cannot track individuals when they transition across different types of insurance (Table

5.1). APCDs (in states that have them) provide a comprehensive picture of health care

claims across a state’s insured population to track utilization and compare rates across

different populations with different types of insurance, although the ability to track

patients across changes in insurance varies by state (The Commonwealth of

Massachusetts, Executive Office of Human Services, Department of Public Health,

2017).

26

Table 5.1 highlights common data source strengths and limitations noted during stakeholder

discussions regarding better practices for pain management. Key takeaways regarding the

advantages and limitations of various data source types include the following:

• Overall, EHR, PDMP, and claims data can provide detailed information on prescription

characteristics and payment, but the systems may not allow longitudinal follow-up of a

given individual across longer periods of time or across insurance coverage transitions.

• While commercial claims and PDMP data may have strengths in capturing information

from multiple payers, Medicaid claims and VHA data warehouses appear to better

support individual-level linkages with other national-level data sources, such as national

mortality records.

• The ability to conduct cross-state analyses may bolster research examining the effects of

interventions on prescribing outcomes, and the compilation of historical information on

PDMP enactment in several data sources has supported such research.

• Other efforts to target opioid prescribing (e.g., guidelines, prescribing limits) have not yet

been systematically collected in a way that facilitates research on their effects.

27

Table 5.1. Commonly Referenced Data Sources for Understanding Better Practices for Pain Management

Data Type and Example

Sources

Strengths Limitations

Commercial claims

• IQVIA

• Truven

• Multipayer; may include cash payments (e.g., IQVIA)

• Captures detail on opioid analgesic prescription

characteristics and other prescriptions filled

• Data systems are not set up to track people long-

term given insurance coverage transitions

• Limited information on diagnoses or other health

care utilization

• Difficult to link to outcomes (e.g., mortality)

Medicaid claims

• National or state Medicaid

data sources

• Can link hospital and pharmacy claims

• Can look at prescription histories of patients who

make it to the hospital or emergency department for

fatal or nonfatal overdose

• Captures detail on opioid analgesic prescription

characteristics and other prescriptions filled

• Only provides information on one population

(Medicaid enrollees)

• Data systems are not set up to track people long-

term given insurance coverage transitions

• Cannot measure opioid mortality: dates of death

commonly not available and cause of death not

included

EHR and claims data

• National or regional VHA

data warehouses

• VHA data warehouse enables linkages across

multiple VHA data sources

• VHA data have been linked with NDI to connect

prescribing to mortality

• Captures detail on opioid analgesic prescription

characteristics and other prescriptions filled

• Access is highly limited

• Findings from veteran population may not be

directly generalizable to other populations

PDMP data

• State PDMPs

• PBSS

• Not restricted to one payer

• Can be used to develop measures around patient,

prescriber, and pharmacist risky behaviors

• Detail on scheduled substance prescriptions

(coverage varies across states)

• Access barriers

• Many states have capacity issues that limit ability to

link PDMP data with other data sources

• Many PDMPs do not collect unique identiers or

have errors in entry, creating technical issues in

matching at the individual level

28

Data Type and Example

Sources

Strengths Limitations

Mortality data

• NDI

• NVSS MCOD

• CDC WONDER

• State death certificate

data

• Information on cause of death and drugs involved

• NDI has been linked at person-level to other data

sources

• State vital records can offer detail on cause of death

• CDC WONDER publicly available

• Generally updated annually; up to 11-month delay

• Data request and approval can take up to three

months

• For NDI, cause of death codes are an additional

cost

Policy data

• PDAPS

• NAMSDL

• Information on PDMPs, pain clinic laws, education

requirements, prescribing limits

• Can be linked with outcome data to examine impact

of state policies

• Some data not provided in analyzable format

• Some policy information not provided available

historically (e.g., only provides a snapshot)

29

Better Addiction Prevention, Treatment, and Recovery Services

Common themes emerging from discussions related to improving access to treatment and

recovery services include the following:

• Supply of treatment: Understanding how policies and initiatives are influencing access

to treatment and recovery services requires access to treatment supply and capacity data.

Claims data and data on Drug Addiction Treatment Act–waivered physicians have been

used to examine buprenorphine treatment capacity (Table 5.2) (Rosenblatt et al., 2015;

Knudsen et al., 2015; Stein et al., 2015; Stein et al., 2015; Dick et al., 2015). However,

developing a fully comprehensive picture of the treatment landscape is challenging: We

lack data on individuals receiving methadone from opioid treatment programs or

receiving treatment provided under state block grants, federal grants provided to support

substance abuse treatment services that are not tied to public or private insurance.

• Treatment demand and utilization: Better understanding the size and characteristics of

the population with opioid use disorder, and who gets treatment, could inform efforts to

close the treatment gap. Analyses of national cross-sectional surveys and claims data

have been useful, but longitudinal data with unique patient identifiers would allow

longer-term analyses of treatment patterns, identifying gaps or limited access points,

events leading to induction or dropout, and processes to improve continued abstinence.

• Treatment processes and quality: Understanding the quality of opioid use disorder care

could benefit from the development of a set of standard performance measures with

respect to quality of opioid use disorder treatment and specifically for medication-

assisted treatment, potentially by leveraging information from EHRs, as well as the more

commonly used services and pharmacy claims. Standardized or systematic reporting of

treatment process measures (e.g., frequency of urinalysis, drug screens, dosing) or

patient-reported outcomes (e.g., abstinence, craving, illicit drug use) would be valuable.

• Treatment and outcomes for criminal justice populations: Linking criminal justice

and treatment services data sources can clarify the treatments being used in the criminal

justice system and continuity of care for individuals who leave the criminal justice

system. For instance, under Chapter 55, Massachusetts has aimed to link person-level

data on substance abuse treatment received by prisoners with mortality data to understand

whether treatment during incarceration reduces likelihood of experience a fatal opioid-

related overdose (The Commonwealth of Massachusetts, Executive Office of Health and

Human Services, Department of Public Health, 2017).

Table 5.2 highlights common data source strengths and limitations noted during stakeholder

discussions regarding opioid use disorder treatment. Key takeaways regarding the advantages

and limitations of various data source types include:

• Many national data sources, including claims data, EHR data, and national surveys, offer

insights into treatment need, treatment utilization, and treatment supply. Each source uses

different measures to assess these outcomes.

• Information on buprenorphine prescriptions and buprenorphine-waivered prescribers is

available through several data sources, but using these data may entail costs.

30

Furthermore, these data provide information on only one type of treatment for opioid use

disorder.

• State-level information on treatment admissions for opioid use disorder and facilities

providing treatment for opioid use disorder are publicly available through national

treatment facility surveys. The quality of admissions data varies across states and over

time.

• Many of these data sets can be triangulated at the county- or state-level to better assess

the overall picture of how treatment need aligns with treatment capacity. However, none

supports person-level linkages across different potential sources of treatment for the

general population.

31

Table 5.2. Commonly Referenced Data Sources for Understanding Treatment Need and Access

Data Type and

Example Sources

Strengths Limitations

!"##$%&'()*&)('#+*

• ,-.,/

• 0%12$3

4(%5$6+&(3

• 78#9:"38*;$()6:

• 41)6'9(8$%*(3<*'3&)1<$+*&(+:*9(8#$36

• =%$+&%'96'"3*<(6(*&(3*&(961%$*6:$*9"91)(6'"3*6%$(6$<

>'6:*?19%$3"%9:'3$

• !(3*$@(#'3$*9%"2'<$%A9(6'$36*&$3+1+$+

• ,3B"%#(6'"3*"3*&"#"%?'<'6'$+*(3<*"6:$%*9%$+&%'96'"3+

C$DEDF*"9'"'<+G

• H'#'6$<*'3B"%#(6'"3*"3*<'(E3"+$+F*"6:$%*:$()6:&(%$*16')'I(6'"3

• J$K1'%$+*6%'(3E1)(6'3E*"6:$%*<(6(*+"1%&$+*6"*(++$++*"9'"'<*1+$

<'+"%<$%*(3<*6%$(6#$36*(&&$++

• ,++1$+*6%(&5'3E*'3<'2'<1()+*"2$%*6'#$

• L9'"'<*1+$*<'+"%<$%*6%$(6#$36*'+*"B6$3*9%'2(6$*&(+:*9(8*(3<

6:1+*3"6*(99%"9%'(6$)8*&(961%$<*'3*&)('#+*(3<*'+*3"6*&(961%$<*(6

())*'3*9:(%#(&8*&)('#+

• !"+6+*6"*"?6('3

4$<'&('<*&)('#+*

• M(6'"3()*"%*+6(6$

4$<'&('<*<(6(

+"1%&$+

• !(3*)'35*:"+9'6()*(3<*9:(%#(&8*&)('#+

• 7"#$*+'3E)$A+6(6$*(3()8+$+*:(2$*)'35$<*6"*<$(6:

&$%6'B'&(6$*<(6(

• !(3*$@(#'3$*"9'"'<*1+$*<'+"%<$%*<'(E3"+'+

• ,3B"%#(6'"3*"3*&"#"%?'<'6'$+*(3<*"6:$%*9%$+&%'96'"3+

C$DEDF*"9'"'<+G

• L3)8*9%"2'<$+*'3B"%#(6'"3*"3*"3$*9"91)(6'"3*C4$<'&('<G

• N(6(*+8+6$#+*3"6*+$6*19*6"*6%(&5*9$"9)$*)"3EA6$%#*E'2$3

'3+1%(3&$*&"2$%(E$*6%(3+'6'"3+

• !(33"6*+$$*'B*%$&$'2'3E*"6:$%*91?)'&)8*B13<$<*+1?+6(3&$*(?1+$

6%$(6#$36

• N'(E3"+'+*&"<$+*?'))$<*B"%*<"*3"6*3$&$++(%')8*%$B)$&6*(&61()

<'(E3"+'+

O;J*

• ;$()6:!"%$

,36$E%(6$<

J$+$(%&:

N(6(?(+$

• P%"19*;$()6:

!""9$%(6'2$

• !(3*&(961%$*3"3#$<'&(6'"3*6%$(6#$36*C$DEDF

9+8&:"+"&'()*6:$%(98G

• /?)$*6"*%$2'$>*9(6'$36*6$+6*%$+1)6+F*:'+6"%8F*<'(E3"+$+F

(3<*9)(3+*B"%*6%$(6#$36

• !)'3'&()*6$@6*:(+*%'&:*<(6(*"3*9%"E%$++*(3<*9%"?)$#+

6:(6*"B6$3*)(&5*,36$%3(6'"3()*!)(++'B'&(6'"3*"B*N'+$(+$+

&"<$+

• N$6(')$<*'3B"%#(6'"3*"3*9('3F*&"#"%?'<'6'$+F

+8#96"#+

• 41)6'9)$*)(>+*%$E(%<'3E*&"3B'<$36'()'68Q9%'2(&8*9%$&)1<$*(&&$++

6"*&)'3'&'(3*3"6$+

• N(6(*K1()'68*'+*6'$<*6"*<(6(*$36%8F*(3<*6:$%$*(%$*2(%'"1+*<(6(A

$36%8*'++1$+

• ,3*#(38*6%$(6#$36*+8+6$#+F*"3)8*:()B*"B*())*9%"2'<$%+*:(2$

(<"96$<*O;J+

32

Data Type and

Example Sources

Strengths Limitations

;"1+$:")<*+1%2$8+*

• M7NR;

• MO7/J!

• M(6'"3()*<(6(

• J'&:*'3B"%#(6'"3*"3*#$36()*:$()6:*(3<*+1?+6(3&$*1+$F

'3&)1<'3E*"9'"'<*#'+1+$*(3<*1+$*<'+"%<$%

• M7NR;*&"))$&6$<*(331())8

• M7NR;*"BB$%+*+"#$*'3+'E:6+*"3*<'2$%+'"3

• M7NR;*STUV*%$<$+'E3*:(+*'3B"%#(6'"3*"3*(38

9%$+&%'96'"3*9('3*%$)'$2$%*1+$*C3"6*"3)8*#'+1+$G

• ;'+6"%'&())8*:(2$*3"6*'3&)1<$<*#$<'&(6'"3+*1+$<*B"%*6%$(6#$36

• 7&%$$3*B"%*1+$*<'+"%<$%*+8#96"#+F*?16*<"*3"6*(+5*(?"16

B"%#()*<'(E3"+'+

• M"*#$(+1%$*"B*&(%$*K1()'68F*6%$(6#$36*%$6$36'"3

• 7(#9)$*#(8*#'++*:'E:A%'+5*E%"19+*C$DEDF*:"#$)$++F*(%%$+6$$+G

• MO7/J!*3"6*&"))$&6$<*(331())8*"%*#(<$*%$(<')8*(2(')(?)$*6"

%$+$(%&:$%+

• 76(6$*'<$36'B'$%+*%$+6%'&6$<

0%$(6#$36*B(&')'68*

+1%2$8+*

• 0ON7A/

• MA77/07

• M(6'"3()*<(6(*"3*(<#'6+*6"*6%$(6#$36*(3<*91?)'&

+$&6"%*+9$&'()68*&(%$

• R9*6"*6:%$$*<%1E+*"B*(?1+$*)'+6$<

• ,3B"%#(6'"3*"3*%$B$%%()*+"1%&$*C$DEDF*&%'#'3()*W1+6'&$

+8+6$#G

• MA77/07*'3&)1<$+*91?)'&*(3<*9%'2(6$*B(&')'6'$+*(3<

#$(+1%$+*"B*&(9(&'68

•

4$<'&(6'"3(++'+6$<*6%$(6#$36*'+*%$9"%6$<*(+*(*+'3E)$*2(%'(?)$

'3*0ON7*+"*&(33"6*<'BB$%$36'(6$*?$6>$$3*?19%$3"%9:'3$*(3<

#$6:(<"3$*0ON7*"3)8*'3&)1<$+*(E"3'+6*6%$(6#$36+

• H'#'6$<*'3B"%#(6'"3*"3*9(8#$36

• -1()'68*&"36%")*'++1$+*>'6:*0ON7F*(+*+6(6$+*#(8*3"6

&"3+'+6$36)8*(++$++*<(6(*K1()'68*"%*%$9"%6*"3*+'#')(%*9(6'$36+

• 0ON7*<"*3"6*'3&)1<$*9%'2(6$*B"%A9%"B'6*B(&')'6'$+

=%"2'<$%*&$3+1+*

• 7/4;7/

<(6(?(+$

• NO/*/!7/

• 4$(+1%$+*+199)8Q&(9(&'68*"B*>('2$%$<*9:8+'&'(3+

• !(3*)'35*6"*6:$*/#$%'&(3*4$<'&()*/++"&'(6'"3

=:8+'&'(3*4(+6$%B')$

• P$"E%(9:'&*<$6(')

• !"+6+*6"*"?6('3*NO/*/!7/

• 7/4;7/X+*91?)'&)8*(2(')(?)$*<(6(*+$6*&(961%$+*(%"13<*VV

9$%&$36*"B*9:8+'&'(3+

33

Better Targeting of Overdose1Reversing Drugs

Common themes emerging from discussions related to promoting use of overdose-reversing

drugs include the following:

• Naloxone distribution: Data about naloxone distributed outside of standard outpatient

pharmacy channels would help to identify capacity problems and ways to get naloxone to

the right individuals. There have been several case studies of OEND programs (Doyon et

al., 2016), but data on naloxone distribution through such programs are not systematically

collected or made publicly available.

• Naloxone effectiveness: Better data on the circumstances surrounding overdoses and

naloxone reversals would improve our understanding of under what circumstances and

how frequently naloxone fails to reverse an overdose. These data could also inform

efforts to modify naloxone use in communities facing increased fentanyl or carfentanil

overdoses. EMS data may be of particular value in this area.

• Treatment for individuals receiving naloxone: Linking individual-level naloxone

administration data with health care utilization data would improve our understanding of

the emergency department services and subsequent opioid use disorder treatment

provided to individuals receiving naloxone.

Table 5.3 highlights common data source strengths and limitations of data sources noted

during stakeholder discussions regarding naloxone access and use. Key takeaways regarding the

advantages and limitations of various data source types include the following:

• Commercial claims data may help in understanding trends and geographic variation in

naloxone distribution through retail pharmacy channels; however, pharmacies are just

one of the sources through which naloxone is distributed.

• A national data source containing information on community-based OEND programs is

managed by the Harm Reduction Council, but these data are not publicly available.

• While EMS data through NEMSIS can offer valuable insights regarding EMS

administration of naloxone, these data cannot include state identifiers, serving as a barrier

to analyses of the effects of state policy on EMS use of naloxone.

• There are some sources of systematically collected data on state naloxone policies, which

enhances assessment of how such policies affect outcomes such as mortality; however,

few data sources capture policy implementation or variation in local regulations or

protocols.

• Opioid-related mortality is an important outcome to evaluate in this area but greater use

of EMS or hospitalization data—particularly if the sources could be linked—would offer

value in understanding the trajectories of individuals treated with naloxone.

34

Table 5.3. Commonly Referenced Data Sources for Understanding Naloxone Access

Data Type and Example

Sources

Strengths Limitations

Commercial claims

• IQVIA

• Measures pharmacy distribution of naloxone

• Information on prescriber specialty

• Data on formulation

• Only captures the distribution of naloxone via pharmacy

channel

• Does not capture purchase and distribution via state or

community programs

• Costs to obtain

Mortality data

• CDC WONDER

• NVSS MCOD

• National data on opioid overdose mortality

• Information on opioid type

• CDC WONDER is readily downloadable

• Lags in data availability

• Variation in quality of reporting detail on drug

involvement

OEND program data

• MA Opioid Overdose

Prevention Pilot Program

• Harm Reduction

Coalition

• Fills in some data gaps regarding naloxone

distributed via state or community programs

• Information on where sites located, number of kits

distributed, etc.

• Not standardized

• National data not systematically collected or updated

• Not publicly available

EMS data

• NEMSIS

• Naloxone administration is reportedly a fairly high-

quality variable, and NEMSIS offers a Public

Naloxone Administration Dashboard (NEMSIS,

undated)

• Standardized collection of 911 call, incident, and

transport information across multiple EMS agencies

• Can do small-area analysis

• Not a registry of patients receiving care

• Data quality differs across agencies/states

• Some measures restricted

• No diagnosis information

• Barriers to linking or accessing geographic identifiers

Policy data

• PDAPS

• NPHL

• Legal databases

• Information on state policies to increase naloxone

access or use

• Can be merged at the state level with other data on

opioid-related outcomes

• Variation in naloxone-related regulations between

states may not be fully captured

• Data on EMS protocols not readily available

• Some historical data may not be provided in readily

analyzable formats

35

Better Data

Common themes emerging from discussions related to strengthening data to improve public

health surveillance include the following:

• Understanding the dynamic opioid ecosystem: The opioid crisis is a dynamic system

with multiple agents and networks of interacting individuals and agencies (Wakeland et

al., 2015; Burke, 2016). Greater efforts are needed to model and understand the dynamics

of the crisis, network patterns (e.g., prescriber, patient) at play, as well as macro-level

factors (e.g., sociological, economic, technological) involved. Such analyses would

require leveraging multiple data sources, including data about users of illicit opioids and

the illicit drug market. For instance, drug-seizure and drug-testing data from the National

Forensic Laboratory Information System (NFLIS) or System to Retrieve Information

from Drug Evidence (STRIDE), both managed by the DEA, contain product-specific data

on substances secured in law enforcement operations (see Table 5.4).

• Early warning signs of problematic use or problematic prescribing: Linking PDMP

data with outcomes data (e.g., hospital discharge, emergency department visit, treatment,

death, or criminal justice data) can facilitate development and validation of risk indicators

for opioid analgesic misuse, diversion, and/or potential overdose. For instance, one study

validated prescriber risk indicators derived using PDMP data by linking prescriber-level

data from Maine’s PDMP with data on medical board actions to assess how well their

prescriber risk indicators predicted likelihood of receiving a disciplinary action (Kreiner

et al., 2017).

• Detail on drugs involved in overdoses: Improved standardization across local

jurisdictions regarding testing for and recording specific drugs and drug types during

autopsies would enhance the consistency, validity, and reliability of information about

drug-related overdose deaths (Ruhm, 2017). Data about nonfatal overdose may also help

fill gaps in knowledge, although data costs are a potential barrier.

• Near-real-time data collection and access: Timely collection and access to data are

necessary to keep pace with the rapid evolution of the crisis, would facilitate

understanding emerging developments and local variation in the illicit supply of opioids,

and may facilitate timely responses. Other opportunities for surveillance could include

ways to leverage novel data sources (e.g., analysis of social media, the Dark Web,

wastewater analysis) to produce near-real-time insights (Kalyanam and Mackey, 2017;

Kalyanam et al., 2017).

Table 5.4 highlights common data source strengths and limitations noted during stakeholder

discussions regarding public health surveillance. Key takeaways regarding the advantages and

limitations of various data source types include the following:

• Each data source has notable strengths in identifying product-specific abuse or risk,

understanding interactions between licit and illicit markets for opioids and providing

timely information for surveillance and monitoring.

• Stakeholders noted common challenges that may limit the use of such data sources by

researchers. These include barriers to access (e.g., high costs, no explicit documentation

36

on how to access) and barriers to analyses (e.g., data files not provided in computable

formats, absence of unique identifiers).

• Some data sources, such as ADAM, that could offer insights on drug use and treatment

among high-risk populations are no longer fully operational.

37

Table 5.4. Commonly Referenced Data Sources for Understanding the Epidemic Through Better Public Health Surveillance

Data Type and

Example

Sources

Strengths Limitations

Mortality data

• NVSS MCOD

• Detail on drugs involved in overdose death

• Information on cause of death

• Complete census of deaths over time

• Access to microdata is limited

• Can be difficult and cumbersome to download

• Variation in quality of reporting detail on drug involvement

• Reporting delays

Prescription drug

monitoring data

• PDMP

• PBSS

• ARCOS

• Comprehensive data on distribution (ARCOS) and

prescribing (PDMP)

• Not restricted to one payer

• PDMPs can be used to develop measures around patient,

prescriber, and pharmacist risky behaviors

• Access barriers

• ARCOS not available in computable formats (i.e., only in

PDF form)

• Many states have capacity issues that limit ability to link

PDMP data with other data sources

• Many PDMPs do not collect unique IDs or have errors in

ID entry, creating technical issues in matching at the

individual level

National surveys

• ADAM

• Captures a high-risk population (arrestees)

• Has urinalysis results in addition to self-reported drug use

• Collects drug market information (e.g., drug acquisition

and payment)

• Collects information on substance abuse treatment history

• No longer fully operational

• Limited to few sites collecting data

• Recent data limited to adult male arrestees

Drug arrest data

• Criminal justice

agencies

• Could be used to examine network patterns of co-arrests

• If linked with other data, could be used to examine

systematic histories leading to arrest or indications of

diversion-related behaviors

• Often not available in electronic form that is usable

• Often difficulties in obtaining permissions to use data

Nonfatal

overdose data

• NPDS

• RADARS

• Captures broader set of overdose incidents than fatalities

• Detailed product- and drug-specific information

• Near-real time data

• Can analyze at local level

• RADARS has additional programs capturing measures of

diversion, use, street price

• Must be requested and purchased

• Data availability lags may vary by poison center

• High costs to obtain

38

Data Type and

Example

Sources

Strengths Limitations

Drug seizure and

drug testing data

• National

Forensic

Laboratory

Information

System (NFLIS)

• STRIDE

• Data on illicit drug supply, prices (STRIDE), and purity

• Product-specific information

• Seizure data generally available with less lag time

• Useful for assessing prevalence and location of emerging

drugs

• Access barriers

• Summary data may be available but are not generally

provided at the substate level

• Some drugs seized by law enforcement are not analyzed

by participating laboratories

39

6. Challenges and Opportunities for Implementing Successful

Data-Linking Strategies

Most of our findings from the environmental scan, data inventory, and stakeholder

discussions were applicable across the five-part HHS strategy. Thus, we do not structure our

discussion in this section around Strategic Priorities; rather, we identify general opportunities to

improve data quality and data linkages to enhance the ability of researchers to answer questions

related to the opioid crisis.

In the next section, we present nine key observations about challenges to data linkage or

analyses that emerged from our study. After each, we describe approaches that could potentially

help to reduce the challenge(s).

Key Observation 1

To advance research studying the effects of changes in state policies related to opioids, the

absence of national data collected in a standardized manner across states can limit the rigor and

robustness of potential analyses. While there are various state-based data initiatives aimed at

synthesizing data from different agencies into one data warehouse, national standards that align

states’ reporting in existing data systems would allow for nationally representative policy

studies. Barriers to research could also be lowered by ensuring that collected data are recorded

and made available in usable formats that support empirical analyses. There is thus a benefit to

be gained from standardizing how data currently being collected are recorded, reported, and

made available.

• Approach 1.1: Establish national standards on data collection and reporting for

currently available data sources. Challenges identified by stakeholders included limited

information on individuals who overdose and are attended to by EMS personnel but

decline transport to the hospital so are generally not captured in administrative claims

data. Thus, one approach may be to further encourage high-quality reporting by EMS

providers of a standardized set of information (Becknell and Simon, 2016) that would

ultimately flow up to state health systems and systems such as NEMSIS. Another

challenge identified was variation in reporting quality to TEDS across states and, over

time, such as variation in what states determine are eligible reporting facilities, what

counts as a treatment episode, and what data elements are required for reporting. This

variation may indicate a need to promote standardized high-quality reporting by states

and to establish improved documentation of potential differences across states and over

time in reporting to TEDS. Data transparency can be further enhanced by supporting the

development and dissemination of a data inventory for opioid research, accompanied by

appropriate technical documentation outlining the contents, characteristics, quality, and

potential limitations of individual data sets. This could be modeled similarly to the new

U.S. Census Bureau Data Repository (U.S. Census Bureau, 2017).

40

• Approach 1.2: Enhance data usability by ensuring that available data are provided

in readily analyzable formats. A substantial barrier noted by stakeholders was that

useful data sources are sometimes provided in formats (e.g., PDF formats) that do not

readily support empirical analyses. Examples include the ARCOS data, as well as state

data made available in a PDF table even though it had originally been created in Excel.

Having to work with such formats creates a cost for researchers, who must translate data

into a format that can be analyzed with statistical programs. The translation creates

unneeded risk of further data-entry errors. For data that are already being made available,

and particularly for data that may already exist in formats that facilitate incorporation into

analytic software, costs to researchers can be reduced by ensuring that data are provided

in files that support analysis is a straightforward way.

• Approach 1.3: Establish standardized performance measures for quality of

treatment processes and outcomes and encourage state treatment programs to

report on these measures. There are several challenges in developing performance

measures, including the need for rigorous assessment of their importance, feasibility, and

validity. One potential opportunity comes from EHRs which, depending on the quality of

the information contained within, may provide an opportunity to collect more in-depth

information, facilitate text mining of clinicians’ notes (e.g., through natural language

programming), and provide ongoing data collection during the course of treatment

(Garnick et al., 2012). Developing such standardized measures or guidelines for quality

of opioid use disorder treatment would facilitate assessment of which efforts effectively

improve access to treatment and recovery services while maintaining high-quality care.

Systematically reporting on the measures would enhance provider accountability and

provide evidence on treatment quality. Some states have already taken steps in this area.

For instance, Vermont has created a public dashboard that includes comparative

reporting on each of its treatment service “spokes,” and Rhode Island requires that

medical homes within opioid treatment programs track performance data (Boss, 2017)

and report data to the state to receive an enhanced payment rate (Chalk and Mark,

2017). Process-related measures of care and patient-centered outcomes data would be

valuable for understanding not just treatment utilization but quality of care.

Key Observation 2

Stakeholders consistently noted the particular value of state all-payer claims databases and

criminal justice data.

• Approach 2.1: Enhance researcher use of all-payer claims databases. While all-payer

claims data are not available for all states, stakeholders highlighted their benefits in

potentially capturing health care claims across an entire state’s population, allowing

studies to track utilization and compare rates across different populations with different

types of insurance. By making these data accessible and comparable in a single source,

all-payers claims data may be less costly to obtain or burdensome to analyze, compared

with obtaining and analyzing data from many different claims data sources. Furthermore,

some states (e.g., Massachusetts) have expended significant resources to enable record

linkages across payers (The Commonwealth of Massachusetts, Executive Office of

Health and Human Services, Department of Public Health, 2017), which potentially

offers a key advantage over other claims data sources (Dworsky, 2017). It is also worth

41

noting that Massachusetts’ all-payer claims database forms the spine of their Chapter 55

data system (discussed further in Approach 9.1) to enable linkages across multiple

interagency data sources (The Commonwealth of Massachusetts, Executive Office of

Health and Human Services, Department of Public Health, 2017).

Research could potentially be enhanced by promoting awareness of the benefits of such data

sources, socializing best practices for their creation and use in research, and making resources

available to increase awareness and prompt greater use by the research community. Many

discussants believed that significant benefits could be gained by encouraging more states to

create such databases and to make them more available to researchers while maintaining fidelity

to confidentiality and privacy requirements. However, self-insured plans can opt out of APCDs,

a significant limitation in examining the employer insured market (U.S. Supreme Court, 2016).

• Approach 2.2: Encourage incorporation of criminal justice data into public health

research. Person-level linkages of public health data sources (e.g., death records,

PDMPs, treatment facility data) with criminal justice data on arrests, incarcerations, or

treatment within the criminal justice system could be of value. Prior research has

obtained de-identified data that link state administrative data on clients receiving publicly

funded substance abuse treatment in specialty settings to arrest and incarceration data

from state criminal justice agencies (Acevedo et al., 2015; Garnick et al., 2014). Small-

area analyses of drug-seizure data complemented by analyses of detailed drug-overdose

data could also inform our understanding of illicit drug markets and supply-side

dynamics and may be less challenging to implement than person-level linkages.

One way to implement this approach would be a research partnership to develop data

systems focused on the criminal justice system and the opioid crisis and to potentially

provide researchers with de-identified files that would support analyses at the level of

fine geographic detail. At the state level, a recently published study (DeHart and Shapiro,

2016) offers insights into the implementation and use of integrated criminal justice and

public health data in South Carolina (DeHart, 2015). Further efforts in this area could

advance our understanding of treatments being used in the criminal justice system and

continuity of care for individuals who leave the criminal justice system; factors that

precede or follow criminal justice involvement related to opioids; and the evolution or

dynamics of illicit opioid markets and illicit opioid use.

Key Observation 3

Stakeholders noted that some data that were useful in strengthening public health

surveillance or capturing high-risk populations are no longer being collected (e.g., Drug Abuse

Warning Network [DAWN], ADAM). In addition, there are current data-collection efforts that

are well-positioned to collect measures relevant to the opioid crisis but historically have not

captured that information or are currently not making the information readily available to

researchers.

• Approach 3.1: Support reinstitution of useful data sets no longer being collected.

Stakeholders noted that the arrestee interview and drug-testing data collected through

42

ADAM provided insights not offered through other household surveys. Furthermore,

ADAM provided a national data source on individual users’ consumption and

expenditures, which offered valuable information on illicit drug markets. DAWN

provided a vital source of information on emergency room visits at the local level.

Bringing back and improving these data sets could help fill gaps in our understanding of

the opioid crisis. SAMHSA is planning to release an improved replacement of DAWN

(i.e., SAMHSA’s Emergency Department Surveillance System) (SAMHSA, 2016), and

research could be enhanced by promoting awareness of its value and supporting its

analysis.

• Approach 3.2: Augment existing federal data collections to capture information

relevant to the opioid crisis and facilitate researcher access to such data. Federally

funded surveys that are collected annually could incorporate new data elements or new

modules relevant to the opioid crisis. For instance, the NSDUH could begin collecting

information on pain, pain treatment, or diagnosis of opioid use disorder. TEDS could be

modified to include whether pharmacotherapy (and what types) is planned or offered at

discharge (Thomas et al., 2011). Interested researchers or other individuals could be

invited to propose new elements or modules to be incorporated into existing systems. To

maximize the benefits of these secondary data sources, there is a concurrent need to

facilitate researcher access to important but sensitive data elements (e.g., state identifiers

in the NSDUH are collected but are not widely available to researchers, and even

researchers with permission to use restricted NSDUH state identifiers have experienced

lengthy disruptions in access over the last several years).

Key Observation 4

An accessible source of consistent national data on opioid policies and strategies being

implemented is essential for evaluating the impacts of policies and initiatives. The Alcohol

Policy Information System (undated) is an exemplar of a rich source of policy data, providing

detailed state-by-state information for a variety of alcohol policies (and more recently for

cannabis policies). PDAPS (undated) offers an excellent source of policy data for state laws

related to PDMPs, naloxone access laws, Good Samaritan laws, and pain management clinic

laws. Continuing to expand and support such efforts in light of the rapidly evolving policy

environment offers a vital benefit to researchers evaluating the impact of state opioid policies

and initiatives.

• Approach 4.1: Support the construction and dissemination of a national database of

state policy and initiatives. Efforts in this area could include expanding the scope of

policies currently collected by systems such as PDAPS or developing new systems that

provide consistent information on state policies or efforts that have received less

evaluation (e.g., opioid prescribing policies, clinician education efforts, insurance policies

regarding reimbursement for pain treatment). Making data available on the timing of

policy enactment or implementation would also help support evaluations of the impact of

these interventions.

43

Key Observation 5

In the areas of treatment for opioid use disorder and use of naloxone, data currently being

used to understand and address these issues offer an incomplete picture. With respect to

treatment, there is not a strong set of measures that captures the extent to which treatment for

opioid use disorder is occurring through state block grants. With respect to naloxone, there is not

a strong set of measures available to track the distribution of naloxone through nonretail

pharmacies. This limits evaluation of the impact of naloxone programs and policies.

• Approaches 5.1: Support the systematic collection and availability of data on

individuals being treated through state block grants. Several data sources capture

information on prescriptions for buprenorphine and availability of treatment with

buprenorphine. However, stakeholders noted that administrative claims data fail to

capture individuals receiving treatment outside of the payment system (e.g., through state

block grants and community treatment programs). Additionally, better information is

needed about the population receiving treatment outside of the public sector (e.g., full

private facilities, which may still be subject to public credentialing). One approach to

begin filling this gap could be exploring ways to incentivize substance abuse treatment

programs to report to Medicaid using information about services provided for individuals

receiving capitated services (e.g., shadow claims) as occurs currently in some

jurisdictions. Careful consideration of patient privacy and confidentiality would be

critical in these efforts.

• Approach 5.2: Support reporting of and access to data on naloxone distribution

through nonpharmacy channels. While commercial claims data (e.g., IQVIA) can offer

insights on access to naloxone through outpatient retail channels, other important

distribution channels for naloxone are not captured in these data. These include naloxone

being provided to first responders directly through hospitals, grants, or other sources, as

well as naloxone being provided directly to the public through OEND programs and other

entities. Identifying methods to track naloxone being distributed through such channels,

such as working with manufacturers, is critical to developing a more comprehensive

understanding of the effectiveness of policies and initiatives seeking to enhance naloxone

distribution.

Key Observation 6

A fundamental need for linking data at the individual level is collection of individual

identifiers. Unique identifiers (e.g., social security numbers) have traditionally not been collected

or made available because of a number of regulatory and privacy concerns (Dokholyan et al.,

2009). Instead, linkages often rely on indirect identifiers (e.g., some combination of age, sex,

date of birth, geography). To accurately link data based on indirect identifiers, it is critical to

have matching algorithms that allow for the accurate extraction and utilization of meaningful

information, given the quantity and quality of the data elements available to link (Dusetzina,

Tyree, and Meyer, 2014).

44

• Approach 6.1: Support methodological research to develop improved algorithms for

matching individuals across and within data sources. Stakeholders noted a key barrier

for data linkages is that we cannot match as well as we need to across databases (or in

some cases, within databases). Identifying reasons for insufficient matching and

developing and validating improved matching algorithms is key to supporting data

linkages.

Key Observation 7

Mortality data are a key resource for both researchers and policymakers, but existing

collection and reporting efforts need to be improved. Up to 25 percent of all death certificates

fail to note the specific drug responsible for fatal overdose, and there are substantial geographic

disparities in rates of missingness (Ruhm, 2017). This reporting variation complicates both

research efforts and targeted enforcement or treatment efforts. Furthermore, stakeholders noted

that there are particular opportunities for linking mortality data with other sources, given the

more-limited confidentiality violations and hence lower privacy barriers in linking data once

someone is deceased (Code of Federal Regulations, 2009). The National Center for Health

Statistics has linked several surveys (e.g., the National Health Interview Survey) with death

certificate data from the NDI (CDC, 2018), although stakeholders noted that these data are

underused. Improving mortality data, leveraging these linking opportunities, and making linked

data more readily available could offer substantial progress toward better understanding opioid-

related harms.

• Approach 7.1: Support improved toxicology studies and reporting. The CDC has

expanded funding to help states and medical examiners improve data collection and

reporting for nonfatal surveillance and fatal overdose data and has funded states to

increase comprehensive toxicology testing (with 60 percent of this funding going toward

medical examiners and coroners). Facilitating access to these data will enhance the ability

of researchers and policymakers to better understand and respond to the rapidly evolving

opioid crisis by understanding trends such as the use of adulterants in illicit opioids.

• Approach 7.2: Support universal and timely reporting of overdose deaths by states and

encourage states to leverage interagency partnerships. Partnerships between departments

of public health, local police departments, emergency medical services, hospitals, and

other agencies could enhance the ability of states to obtain complete and timely

information on overdose deaths in the community. Combined with improved toxicology

studies and reporting, these efforts could support targeted interventions to aid community

organizations, law enforcement, public health agencies, and the broader general public.

• Approach 7.3: Enhance linkage mortality data to other data sources and promote their

use by researchers. Stakeholders noted that a key opportunity for advancing our

understanding of the opioid crisis is linking Medicaid claims with mortality data. One

approach could be to develop standards and requirements for data sharing by state

agencies overseeing Medicaid data and mortality data. Stakeholders mentioned prior

efforts to link CMS claims data with mortality data that were supported by the American

Recovery and Reinvestment Act of 2009, but it is unclear to what extent such initiatives

have been undertaken in more recent years. Given the scope of the opioid crisis, there

45

may be greater value in supporting such linkages than there has been historically. Such

linkages can enhance policy and program evaluations; for instance, the U.S. Census

Bureau Center for Administrative Records Research and Application, in partnership with

Chapin Hall at the University of Chicago and supported by the Laura and John Arnold

Foundation, are promoting research studies and methods for combining data across

agencies and levels of government to advance evidence-based policymaking (Goerge,

Gjertson, and De La Cruz, 2017).

Key Observation 8

Effective responses to the rapidly evolving opioid crisis rely on the timely collection,

reporting, and analyses of crucial health information. Near-real time data collection at fine

geographic detail can support identification of high-risk locations and help inform timely and

effective community interventions. Several states have made great strides toward improving the

speed at which data on nonfatal overdose are collected and analyzed.

However, there are several challenges with near-real time surveillance systems. Substantial

costs and resources are required to implement and manage such systems, and the costs of data

management and analysis increase as systems receive increasing amounts of data with increasing

speed and diversity. Stakeholders also mentioned that laws governing the process by which data

collection occurs, such as the Paperwork Reduction Act, may also create substantial lags in

starting up new data collection efforts for surveillance, as the time to obtain the requisite

permissions often exceeds six months. Near-real time data collection also suffers from greater

data-quality challenges compared with data collection that occurs over a longer time frame, and

potential issues with record completeness and accurate processing and transmission necessitate

ongoing monitoring and communication (Ising et al., 2016). Despite these challenges, there is

significant public health value to be gained by supporting the collection and analysis of such

systems.

• Approach 8.1: Use evidence on innovative state or local approaches to develop and

utilize near-real time surveillance systems to advance the use and operations of such

systems more broadly. Many existing state approaches to near-real-time surveillance

systems leverage data on nonfatal overdoses (Box 5). Evidence on how states have used

these systems, challenges faced in their implementation and use, and insights about how

challenges have been overcome can be used to support the development of near-real-time

surveillance tools in other jurisdictions.

• Approach 8.2: Support innovative research on the use of nontraditional data

sources (e.g., social media, the Dark Web) to inform public health action. There has

been increasing research interest in methods to mine and analyze nontraditional data

resources to bolster public health surveillance. Studies have analyzed Twitter messages

and web forum postings to understand various forms of opioid misuse and prescription

drug diversion (Katsuki, Mackey, and Cuomo, 2015; Anderson et al., 2017; Chan, Lopez,

and Sakar, 2015), used Google trends data to forecast state-level mortality (Parker et al.,

2017), and used information from cryptomarket forums on the Dark Web to assess

emerging trends in new psychoactive substances (Van Hout, Claire, and Hearne, 2017).

46

While cryptomarkets represent only a slice of the total illicit drug trade, studies have used

web crawlers to scrape cryptomarket listings, vendor profiles, and forum discussions to

map online illicit drug distribution networks, assess prevalence and trends in the illicit

online sales and prices of different drug types, and identify emerging drug trends (see

Barratt and Aldridge, 2016, for an overview of challenges and opportunities in

cryptomarket research) (Van Hout, Claire, and Hearne, 2017; Bhaskar, Linacre, and

Machin, forthcoming, Ladegaard, 2017; Broséus et al., 2016). Advancing methods to

harness these data sources as a public health surveillance tool can offer a key resource for

identifying risks and emerging trends (Brownstein et al., 2009).

Box 5. Examples of State Efforts to Develop and Use Near-Real Time Surveillance

systems

Rhode Island’s Opioid Overdose Reporting System is a flexible near-real time

surveillance system that compiles information on cases of opioid overdose from the state’s

hospitals and emergency departments, although noted challenges have included incomplete

compliance with reporting requirements (McCormick, Koziol, and Sanchez, 2017).

North Carolina’s statewide syndromic surveillance system (the North Carolina Disease

Event Tracking and Epidemiologic Collection Tool) provides near-real time collection and

analysis of statewide emergency department data, poison center call data, and emergency

medical services data (Ising et al., 2016).

Key Observation 9

PDMPs are a valuable resource for understanding the opioid crisis, and there is significant

value to be gained by linking PDMPs with a variety of other data sources. All states now operate

a PDMP system (PDMP, Training and Technical Assistance Center, 2017), but they vary

substantially in data collection, reporting, and interoperability (Pardo, 2017; Manasco et al.,

2016). Developing a complete and consistent PDMP data set for analysis is essential for studying

prescription drug abuse. Linking these data with public health and criminal justice data sources

would support public health surveillance of opioid-related problems. Several state-level efforts,

often organized around PDMPs, are underway toward developing individual-level data linkages

across multiple data sources, and there is an opportunity for further partnerships between federal

and state agencies to support such efforts and invest in making these state linkages more useful

to research and practice. An exemplar of such an approach is the Bureau of Justice Assistance’s

Harold Rogers PDMP, which supports local, state, or regional collaborative efforts to collect and

analyze multiple sources of data. The program’s goal is to enhance understanding of the opioid

crisis and develop data-driven strategies to support surveillance, treatment, and prevention

efforts for at-risk individuals (Paulozzi, Kilbourne, and Desai, 2011).

47

• Approach 9.1: Use evidence on innovative state approaches to leveraging and

linking PDMP systems to publish guidance and recommendations on how states can

support linking PDMP data with other data sources. Given regulatory and

confidentiality concerns, it may be helpful to explore if a directive could be issued

indicating that there should be no attempts to subpoena PDMP data (or associated data

linked to PDMPs) for federal investigations. Establishing guidance for allowing

researchers controlled access to de-identified linked data could further promote the value

of linking PDMP with other sources, particularly if de-identified statistical data from

multiple states could be made available through a single federal or federally supported

source, such as the Brandeis PBSS. Some examples of state PDMP data-linkage efforts

are highlighted in Box 6.

There is also substantial interest in linking PDMP data with social services data (e.g., child

welfare data) to better understand how opioid misuse affect child welfare outcomes. While our

stakeholder discussion did not identify states that are currently making these linkages, a recent

study linked county-level data on controlled substance prescriptions rates from Florida’s Drug-

Related Outcomes Surveillance and Tracking System with county-level data on child removal

rates (Quast, Storch and Yampolskaya, 2018), and developing broader data linkages to support

analyses of the effects of the opioid crisis on children and families is an area to consider

supporting.

Box 6. Examples of Approaches to Linking PDMP Data with Other State Sources

Washington state links PDMP data to the state’s Medicaid and Worker’s Compensation

claims data through Washington State’s Data Sharing Initiative with Medicaid and Workers’

Compensation (PMDP, 2013).

Massachusetts is a noted example of state success in linking PDMP data to a broad range

of other public health and criminal justice data sources. Chapter 55 of the Acts of 2015

permitted the linkage and analysis of several government data sources to inform

programmatic decisions, guide the development of policies, and advance understanding of

the opioid crisis. Under Chapter 55, Massachusetts’ Department of Public Health has

connected (in most cases, at the individual level) ten data sources managed by five state

agencies to develop a data warehouse structure. The system also collects community-level

data on naloxone (e.g., enrollments, refills, and rescues through the Massachusetts

Department of Public Health Naloxone program), drug seizures, and socioeconomic and

demographic characteristics (The Commonwealth of Massachusetts, Executive Office of

Health and Human Services, Department of Public Health, 2017).

Maryland is another example of a state that has overcome interpretational challenges of

42 CFR Part II (establishing special privacy protections for health care records related to the

treatment of substance use disorders) and is currently advancing efforts to link person-level

data from the PDMP, drug use and alcohol treatment admissions, hospital admissions,

fatalities investigated by the medical examiner, and criminal justice data (Saloner, 2016;

Lyons, 2017).

48

• Approach 9.2: Encourage states to improve PDMP systems to ensure data

compatibility with other states. Standardization of electronic data collection for key

elements for all state PDMPs would facilitate cross-state sharing and collaboration with

other agencies (e.g., Medicaid, Department of Veterans Affairs). The American Society

for Automation in Pharmacy guidelines created a PDMP standard for reporting, most

recently updated in 2016 (American Society for Automation in Pharmacy, 2016).

Encouraging states to use the most recent version of the guidelines could support

interoperability and comprehensive data analysis (Greenwood-Ericksen et al., 2016). To

enhance interstate accessibility of PDMPs, one approach could include legislation

enabling sharing between PDMPs in all states.

Summary

Significant work is being done at the federal, state, and local level to combat the opioid

crisis. There has also been a substantial increase in research that has improved our understanding

of the complex and multidimensional nature of the opioid crisis, and that has advanced the

evidence base regarding the effectiveness of opioid policies and initiatives to reduce opioid-

related harms. There are significant resources within reach for the use and analysis of secondary

data, but not all are being taken advantage of. This report outlines a range of strategies that can

improve and promote available data to better understand the crisis.

Meaningful progress can be made on many of our potential approaches; doing so would

likely provide significant value to opioid policy researchers and inform policy developments.

Potential approaches that can be taken in the short-term include the following:

• Enhance data usability by ensuring available data are provided in readily

analyzable formats (Approach 1.2, Section 6): Progress over the short term can be made

by ensuring that publicly available data that are already electronically compiled in

analyzable formats are made readily accessible in machine-readable formats (e.g., CSV,

XML, ASCII) and by providing adequate technical documentation about important

aspects of the data. Entities publishing data in graphical format could enhance data

transparency by including links to analyzable formats of the data underlying the graphs.

• Support universal and timely reporting of overdose deaths by states and improve

toxicology reporting (Approach 7.2, Section 6): Progress over the short term can be

made in this area by updating best practices for coroners and medical examiners to report

overdose fatalities by disseminating such best practices and by continuing to support

improved toxicology reporting. Studies have supported that centralized medical examiner

systems have more-complete recording of specific drugs involved in drug intoxication

deaths compared with states with a decentralized county coroner system (Warner et al.,

2013).

• Use evidence on innovative state approaches to leverage and link PDMP systems to

publish guidelines and recommendations for states to support linking PDMP data

with other data sources (Approach 9.1, Section 6): There are several innovative state

approaches currently underway to link PDMP systems with other data sources—in some

cases, at the individual level. Short-term progress can be made in this area by developing

49

guidelines based on the challenges states have faced and how they have been overcome

and investing in making these linkages more useful to research and practice.

In Table 6.1, we provide an overview of timelines, based on conversations with stakeholders,

in which it may be possible to achieve meaningful progress toward the approaches discussed

above. We recognize, however, that there may be a range of complexities that stakeholders are

unaware of that may challenge meeting such aggressive timelines. Nevertheless, given the

human and societal toll of the opioid crisis and the potential benefits from additional high-quality

research that these approaches could support, we believe it is a public health imperative to create

and make available improved data assets that will support more informed efforts to address the

opioid crisis.

Table 6.1. Time Frame for Potential Approaches to Implementing Successful Data-Linking

Strategies

Approach Approach Description

Short

Term*

Intermediate

Term*

Long

Term*

1.1

Establish national standards on data collection and

reporting for currently available data sources.

X X

1.2

Enhance data usability by ensuring available data are

provided in readily analyzable formats.

X

1.3

Establish standardized performance measures for

quality of treatment.

X X

1.4

Encourage state treatment programs to report on

treatment processes and outcomes.

X X

2.1 Enhance researcher use of all-payer claims databases. X

2.2

Encourage incorporation of criminal justice data into

public health research.

X X

3.1

Support reinstitution of useful data sources no longer

being collected.

X X

3.2

Augment existing federal data collections to capture

information relevant to the opioid crisis and facilitate

researcher access to such data.

X X

4.1

Support the construction and dissemination of a

national database of state policy and initiatives.

X X

5.1

Support the systematic collection and availability of data

on individuals being treated through state block grants.

X X

5.2

Support reporting of and access to data on naloxone

distribution through nonpharmacy channels.

X X

6.1

Support methodological research to develop improved

algorithms for matching individuals across and within

data sources.

X X

7.1 Support improved toxicology studies and reporting. X X

7.2

Support universal and timely reporting of overdose

deaths by states and encourage states to leverage

interagency partnerships.

X X

7.3

Enhance linking mortality data to other data sources

and promote their use by researchers.

X X

50

Approach Approach Description

Short

Term*

Intermediate

Term*

Long

Term*

8.1

Use evidence on innovative state or local approaches to

develop and utilize near-real time surveillance systems

to advance the use and operations of such systems

more broadly.

X X

8.2

Support innovative research on the use of nontraditional

data sources (e.g., social media, the Dark Web) to

inform public health action.

X

9.1

Use evidence on innovative state approaches to

leverage and link PDMP systems to publish guidelines

and recommendations on how states can support

linking PDMP data with other data sources.

X

9.2

Encourage states to improve PDMP systems to ensure

data compatibility with other states.

X X

* Short term = meaningful progress within six months; intermediate term = meaningful progress within 12 months;

long term = meaningful progress may take more than 12 months

51

References

Acevedo, Amdrea, Deborah W. Garnick, Robert Dunigan, Constance M. Horgan, Grant A.

Ritter, Margaret T. Lee, Lee Panas, Kevin Campbell, Karin Haberlin, Dawn Lambert-Wacey,

Tracy Leeper, Mark Reynolds, and David Wright, “Performance Measures and Racial/Ethnic

Disparities in the Treatment of Substance Use Disorders,” Journal of Studies on Alcohol and

Drugs, Vol. 76, No. 1, January 2015, pp. 57–67.

Agency for Healthcare Research and Quality, “Medical Expenditure Panel Survey,” webpage,

undated. As of July 6, 2018:

https://meps.ahrq.gov/mepsweb/

Agency for Healthcare Research and Quality, “Healthcare Cost and Utilization Project (HCUP),”

webpage, last updated June 2018. As of July 1, 2018:

https://hcupnet.ahrq.gov/#setup

Ahmsbrak, Rebecca, Jonaki Bose, Sarra L. Hedden, Rachel N. Lipari, and Eunice Park-Lee E,

Key Substance Use and Mental Health Indicators in the United States: Results from the 2016

National Survey on Drug Use and Health, Rockville, Md.: Center for Behavioral Health

Statistics and Quality, Substance Abuse and Mental Health Services Administration,

September 2017.

Albert, Su, Fred W. Brason II, Catherine K. Sanford, Nabarun Dasgupta, Jim Graham, and Beth

Lovette, “Project Lazarus: Community-Based Overdose Prevention in Rural North Carolina,”

Pain Medicine, Vol. 12, Supplement 2, June 2011, pp. S77–S85.

Alcohol Policy Information System, “Welcome to the Alcohol Policy Information System,”

webpage, undated. As of May 17, 2018:

https://alcoholpolicy.niaaa.nih.gov

American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders:

DSM-5, 5th ed. Arlington, Va.: American Psychiatric Publishing; 2013.

American Society for Automation in Pharmacy, “2016 ASAP Version 4.2A Standard for

Prescription Drug Monitoring Programs,” webpage, 2016. As of September 18, 2017:

https://www.asapnet.org/pmp-implementation-guides.html

amfAR, “Opioid and Health Indicators Database,” webpage, undated. As of May 17, 2018:

http://opioid.amfar.org/

Anderson, Daren R., Ianita Zlateva, Emil N. Coman, Khushbu Khatri, Terrence Tian, and Robert

D. Kerns, “Improving Pain Care Through Implementation of the Stepped Care Model at a

52

Multisite Community Health Center,” Journal of Pain Research, Vol. 201, No. 9, 2016, pp.

1021–1029.

Anderson, Laurie S., Heidi G. Bell, Michael Gilbert, Julie E. Davidson, Christina Winter,

Monica J. Barratt, Beta Win, Jeffery L. Painter, Christopher Menone, Jonathan Sayegh, and

Nabarun Dasgupta, “Using Social Listening Data to Monitor Misuse and Nonmedical Use of

Bupropion: A Content Analysis,” JMIR Public Health Surveillance, Vol. 3, No. 1, February

1, 2017, p. e6.

Asnsolabehere, Stephen, and Eitan D. Hersh, “ADGN: An Algorithm for Record Linkage Using

Address, Date of Birth, Gender, and Name,” Statistics and Public Policy, Vol. 4, No. 1,

2017, pp. 1–10.

Appari, Ajit, M. Eric Johnson, and Denise L. Anthony, “Meaningful Use of Electronic Health

Record Systems and Process Quality of Care: Evidence from a Panel Data Analysis of U.S.

Acute-Care Hospitals,” Health Services Research, Vol. 48, Part 1, April 2013, pp. 354–375.

Arizona Department of Health Services, “Opioid Epidemic: Real Time Opioid Data,” webpage,

date range June 15–2017–June 14, 2018. As of June 27, 2018:

https://www.azdhs.gov/prevention/womens-childrens-health/injury-prevention/opioid-

prevention/index.php

Baehren, David F., Catherine A. Marco, Danna E. Droz, Sameer Sinha, E. Megan Callan, and

Peter Akpuonu, “A Statewide Prescription Monitoring Program Affects Emergency

Department Prescribing Behaviors,” Annals of Emergency Medicine, Vol. 56, No. 1, July

2010, pp. 19–23.

Barratt, Monica J., and Judith Aldridge, “Everything You Always Wanted to Know About Drug

Cryptomarkets* (*but Were Afraid to Ask),” Internal Journal Drug Policy, Vol. 35,

September 2016, pp. 1–6.

Baumblatt, Jane A. Gwira, Caleb Wiedeman, John R. Dunn, William Schaffner, Leonard J.

Paulozzi, and Timothy F. Jones, “High-Risk Use by Patients Prescribed Opioids for Pain and

Its Role in Overdose Deaths,” JAMA Internal Medicine, Vol. 174, No. 5, 2014, pp. 796–801.

Becker, William C., Brenda T. Fenton, Cynthia A. Brandt, Erin L. Doyle, Joseph Francis, Joseph

L. Goulet, Brent A. Moore, Virginia Torrise, Robert D. Kerns, and Peter W. Kreiner,

“Multiple Sources of Prescription Payment and Risky Opioid Therapy Among Veterans,”

Medical Care, Vol. 55, July 2017, pp. S33–S36.

Becker, William C., David A. Fiellin, and Rani A. Desai, “Non-Medical Use, Abuse, and

Dependence on Sedatives and Tranquilizers Among U.S. Adults: Psychiatric and Socio-

Demographic Correlates,” Drug and Alcohol Dependence, Vol. 90, Nos. 2–3, October 8,

2007, pp. 280–287.

53

Becker, William C., Lynn E. Sullivan, Jeanette M. Tetrault, Rani A. Desai, and David A. Fiellin,

“Non-Medical Use, Abuse, and Dependence on Prescription Opioids Among U.S. Adults:

Psychiatric, Medical, and Substance Use Correlates,” Drug and Alcohol Dependence, Vol.

94, Nos. 1–3, April 1, 2008, pp. 38–47.

Becknell, John, and Lauren Simon, “Beyond EMS Data Collection: Envisioning an Information-

Driven Future for Emergency Medical Services,” Washington, D.C.: National Highway

Traffic Safety Administration, Report No. DOT HS 812 361, December 2016. As of April

13, 2018:

https://www.ems.gov/pdf/Beyond_EMS_Data_Collection.pdf

Bhaskar, V., Robin Linacre, Stephen Machin, “The Economic Functioning of Online Drugs

Markets,” Journal of Economic Behavior and Organization, forthcoming.

Blewett, Lynn A., and Michael Davern, “Meeting the Need for State-Level Estimates of Health

Insurance Coverage: Use of State and Federal Survey Data,” Health Services Research, Vol.

41, Part 1, June 2006, pp. 946–975.

Blondell, Richard D., Heather N. Dodds, Monica N. Blondell, and Danna C. Droz, “Is the

Kentucky Prescription Reporting System Useful in the Care of Hospitalized Patients?” The

Journal of the Kentucky Medical Association, Vol. 102, No. 1, p. 15-9.

Blumenthal, David, and Marilyn Tavenner, “The ‘Meaningful Use’ Regulation for Electronic

Health Records,” New England Journal of Medicine, Vol. 363, August 5, 2010, pp. 501–504.

Boscarino, Joseph A., Margaret Rukstalis, Stuart N. Hoffman, John J. Han, Porat M. Erlich,

Glenn S. Gerhard, and Walter F. Stewart, “Risk Factors for Drug Dependence Among Out-

Patients on Opioid Therapy in a Large U.S. Health-Care System,” Addiction, Vol. 105, No.

10, October 2010, pp. 1776–1782.

Boss, Rebecca, “Planning and Implementing Comprehensive MAT Service Deliver Models: A

Vision for Substance Use Disorder Treatment,” slide presentation, October 26, 2017. As of

April 13, 2018:

http://www.dhss.delaware.gov/dhss/dph/hsp/files/visconboss.pdf

Bradley, Cathy J., Bassam Dahman, and Charles W. Given “Inadequate Access to Surgeons:

Reason for Disparate Cancer Care?” Medical Care, Vol. 47, No. 7, July 2009, pp. 758–764.

Bradley, Cathy J., Lynne Penberthy, Kelly J. Devers, and Debra J. Holden, “Health Services

Research and Data Linkages: Issues, Methods, and Directions for the Future,” Health

Services Research, Vol. 45, Part 2, October 2010, pp. 1468–1488.

Brady, Joanne E., Hannah Wunsch, Charles DiMaggio, Barbara H. Lang, James Giglio, and

Guohua Li, “Prescription Drug Monitoring and Dispensing of Prescription Opioids,” Public

Health Reports, Vol. 129, No. 2, March–April 2014, pp. 139–147.

54

Broséus, J., Damien Rhumorbarbe, Caro Mireault, V. Ouellette, Frank Crispino, and David

Décart-Hétu, “Studying Illicit Drug Trafficking on Darknet Markets: Structure and

Organisation from a Canadian Perspective,” Forensic Science International, Vol. 264, July

2016, pp. 7–14.

Brownstein, John S. Clark C. Freifeld, and Lawrence C. Madoff, “Digital Disease Detection—

Harnessing the Web for Public Health Surveillance,” New England Journal of Medicine, Vol.

360, No. 21, May 21, 2009, pp. 2153–2155, 2157.

Broz, Dita, and Lawrence J. Ouellet, “Racial and Ethnic Changes in Heroin Injection in the

United States: Implications for the HIV/AIDS Epidemic,” Drug and Alcohol Dependence,

Vol. 94, Nos. 1–3, April 1, 2008, pp. 221–233.

Burke, Donald S., “Forecasting the Opioid Epidemic,” Science, Vol. 354, No. 6312, November

4, 2016, p. 529.

Burns, Rachel M., Rosalie L. Pacula, Sebastian Bauhoff, Adam J. Gordon, Hollie Hendrikson,

Douglas L. Leslie, and Bradley D. Stein, “Policies Related to Opioid Agonist Therapy for

Opioid Use Disorders: The Evolution of State Policies from 2004 to 2013,” Substance Abuse,

Vol. 37, No. 1, January–March 2016, pp. 63–69.

Campanella, Paolo, Emanuela Lovato, Claudio Marone, Lucia Fallacara, Agostino Mancuso,

Walter Ricciardi, and Maria Lucia Specchia, “The Impact of Electronic Health Records on

Healthcare Quality: A Systematic Review and Meta-Analysis,” European Journal of Public

Health, Vol. 26, No. 1, February 2016, pp. 60–64.

Campbell, Kevin M., Dennis Deck, and Antoinette Krupski, “Record Linkage Software in the

Public Domain: A Comparison of Link Plus, The Link King, and a ‘Basic’ Deterministic

Algorithm,” Health Informatics Journal, Vol. 14, No. 1, March 2008, pp. 5–15.

Capricorn, “Chicago Area Patient-Centered Outcomes Research Network,” website, undated. As

of May 17, 2018:

http://capricorncdrn.org

Carrell, David, Jack Mardekian, David Cronkite, Arvind Ramaprasan, Kristina Hansen, David E.

Gross, Roy E. Palmer, Elizabeth Masters, and Michael Von Korff, “A Fully Automated

Algorithm for Identifying Patients with Problem Prescription Opioid Use Using Electronic

Health Record Data,” Drug and Alcohol Dependence, Vol. 171, February 1, 2017, p. e36.

Centers for Disease Control and Prevention, “HIV Infection and HIV-Associated Behaviors

Among Injecting Drug Users—20 Cities, United States, 2009,” Morbidity and Mortality

Weekly Report (MMWR), Vol, 61, No. 8, March 2, 2012, pp. 133–138.

Centers for Disease Control and Prevention, “National Hospital Care Survey,” webpage, last

updated September 10, 2015. As of May 17, 2018:

https://www.cdc.gov/nchs/nhcs/about_nhcs.htm

55

Centers for Disease Control and Prevention, “Opioid Overdose: Opioid Data Analysis,”

webpage, last updated February 9, 2017. As of May 17, 2018:

https://www.cdc.gov/drugoverdose/data/analysis.html

Centers for Disease Control and Prevention, “Data-Driven Prevention Initiative (DDPI),”

webpage, last updated October 3, 2017. As of May 16, 2018:

https://www.cdc.gov/drugoverdose/foa/ddpi.html

Centers for Disease Control and Prevention, “National Health Interview Survey,” PDF, updated

November 9, 2017. As of May 17, 2018:

https://www.cdc.gov/nchs/data/datalinkage/LinkageTable.pdf

Centers for Disease Control and Prevention “Drug Overdose Death Data,” December 19, 2017.

As of April 9, 2018:

https://www.cdc.gov/drugoverdose/data/statedeaths.html

Centers for Disease Control and Prevention, “National Center for Health Statistics: Vital

Statistics Rapid Release, Provisional Drug Overdose Death Counts,” webpage, last updated

June 13, 2018. As of July 1, 2018:

https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm

Centers for Disease Control and Prevention, “CDC WONDER,” webpage, updated June 27,

2018. As of July 1, 2018:

https://wonder.cdc.gov

Centers for Disease Control and Prevention, “NCHS Data Linked to NDI Mortality Files,”

webpage, June 28, 2018. As of July 1, 2018:

https://www.cdc.gov/nchs/data-linkage/mortality.htm

Centers for Medicare and Medicaid Services, “Chronic Conditions Data Warehouse (CCW),”

webpage, 2018. As of May 30, 2018:

https://www.ccwdata.org/web/guest/condition-categories#Proposed_OUD_Indicator

Cerdá, Magdalena, Andrew Gaidus, Katherine M. Keyes, William Ponicki, Silvia Martins,

Sandro Galea, and Paul Gruenewald, “Prescription Opioid Poisoning Across Urban and

Rural Areas: Identifying Vulnerable Groups and Geographic Areas,” Addiction, Vol. 112,

No. 1, January 2017, pp. 103–112.

Chalk, Mady, and Tami L. Mark, “Deploying the Cascade of Care Framework to Address the

Opioid Epidemic Means Taking a Closer Look at Quality Measures,” Health Affairs Blog,

June 21, 2017. As of September 18, 2017:

http://healthaffairs.org/blog/2017/06/21/deploying-the-cascade-of-care-framework-to-

address-the-opioid-epidemic-means-taking-a-closer-look-at-quality-measures

56

Chan, Brian, Andrea Lopez, and Urmimala Sarkar, “The Canary in the Coal Mine Tweets: Social

Media Reveals Public Perceptions of Non-Medical Use of Opioids,” PLoS One, Vol. 10, No.

8, August 7, 2015, p. e0135072.

Chou, Roger, P. Todd Korthuis, Melissa Weimer, Christina Bougatsos, Ian Blazina, Bernadette

Zakher, Sarah Grusing, Beth Devine, and Dennis McCarty, Medication-Assisted Treatment

Models of Care for Opioid Use Disorder in Primary Care Settings, Rockville, Md.: Agency

for Healthcare Research and Quality, AHRQ Publication No. 16(17)-EHC039-EF, December

2016.

Clark, D. E., “Practical Introduction to Record Linkage for Injury Research,” Injury Prevention,

Vol. 10, No. 3, June 2004, pp. 186–191.

Clark, Robin E., Mihail Samnaliev, Jeffrey D. Baxter, and Gary Y. Leung, “The Evidence

Doesn’t Justify Steps by State Medicaid Programs to Restrict Opioid Addiction Treatment

with Buprenorphine,” Health Affairs, Vol. 30, No. 8, August 2011, pp. 1425–1433.

Code of Federal Regulations, Title 45, Public Welfare, Department of Health and Human

Services, Part 46, Protection of Human Subjects, revised January 15, 2009, effective July 14,

2009.

Commission on Evidence-Based Policymaking, “The Promise of Evidence-Based Policymaking:

Report of the Commission on Evidence-Based Policymaking,” Washington, D.C., September

2017. As of June 27, 2018:

https://www.cep.gov/content/dam/cep/report/cep-final-report.pdf

The Commonwealth of Massachusetts, Executive Office of Health and Human Services,

Department of Public Health, An Assessment of Opioid-Related Deaths in Massachusetts

(2013–2014), September 2016.

The Commonwealth of Massachusetts, Executive Office of Health and Human Services,

Department of Public Health, “An Assessment of Fatal and Nonfatal Opioid Overdoses in

Massachusetts (2011–2015),” August 16, 2017. As June 27, 2018:

http://www.mass.gov/eohhs/docs/dph/stop-addiction/legislative-report-chapter-55-aug-

2017.pdf

Coorevits, Pascal, Mats Sundgren, Gunnar O. Klein, Anne Bahr, Brecht Claerhout, Christel

Daniel, Martin Dugas, Danielle Dupont, Andreas Schmidt, Peter Singleton, Georges De

Moor, and Dipak Kalra, “Electronic Health Records: New Opportunities for Clinical

Research,” Journal of Internal Medicine, Vol. 274, No. 6, December 2013, pp. 547–560.

Cunningham, Peter J., and Len M. Nichols, “The Effects of Medicaid Reimbursement on the

Access to Care of Medicaid Enrollees: A Community Perspective,” Medical Care Research

and Review, Vol. 62, No. 6, December 2005, pp. 676–696.

57

Dasgupta, Nabarun, Michele Jonsson Funk, Scott Proescholdbell, Annie Hirsch, and Kurst M.

Ribisl, “Cohort Study of the Impact of High-Dose Opioid Analgesics on Overdose

Mortality,” Pain Medicine, Vol. 17, No. 1, January 1, 2016, pp. 85–98.

Daubresse, Matthew, Hsien-Yen Chang, Yuping Yu, Shilpa Viswanathan, Nilay D. Shah,

Randall S. Stafford, Stefan P. Kruszewski, and G. Caleb Alexander, “Ambulatory Diagnosis

and Treatment of Nonmalignant Pain in the United States, 2000–2010,” Medical Care, Vol.

51, No. 10, October 2013, pp. 870–878.

Davis, Corey S., and Derek Carr, “Legal Changes to Increase Access to Naloxone for Opioid

Overdose Reversal in the United States,” Drug and Alcohol Dependence, Vol. 157,

December 1, 2015, pp. 112–120.

Davis, Corey, and Derek Carr, “State Legal Innovations to Encourage Naloxone Dispensing,”

Journal of the American Pharmacists Association, Vol. 57, No. 2, March–April 2017, pp.

S180–S184.

Davis, Corey S., Sarah Ruiz, Patrick Glynn, Gerald Picariello, and Alexander Y. Walley,

“Expanded Access to Naloxone Among Firefighters, Police Officers, and Emergency

Medical Technicians in Massachusetts,” American Journal of Public Health, Vol. 104, No. 8,

August 1, 2014, pp. e7–e9.

Davis, Corey S., Jessica K. Southwell, Virginia Radford Niehaus, Alexander Y. Walley, and

Michael W. Dailey, “Emergency Medical Services Naloxone Access: A National Systematic

Legal Review,” Academic Emergency Medicine, Vol. 21, No. 10, October 2014, pp. 1173–

1177.

DeHart, Dana D., and Cheri J. Shapiro, “Integrated Administrative Data and Criminal Justice

Research,” American Journal of Criminal Justice, Vol. 42, No. 2, July 2016, pp. 255–274.

Delcher, Chris, Alexander C. Wagenaar, Bruce A. Goldberger, Robert L. Cook, and Mildred M.

Maldonado-Molina, “Abrupt Decline in Oxycodone-Caused Mortality After Implementation

of Florida's Prescription Drug Monitoring Program,” Drug and Alcohol Dependence, Vol.

150, May 1, 2016, pp. 63–68.f

Deyo, Richard A., Sara E. Hallvik, Christi Hildebran, Miguel Marino, Eve Dexter, Jessica M.

Irvine, Nicole O’Kane, Joshua Van Otterloo, Dagan A. Wright, Gillian Leichtling, and Lisa

M. Millet, “Association Between Initial Opioid Prescribing Patterns and Subsequent Long-

Term Use Among Opioid-Naïve Patients: A Statewide Retrospective Cohort Study,” Journal

of General Internal Medicine, Vol. 32, No. 1, January 2017, pp. 21–27.

Dick, Andrew W., Rosalie Liccardo Pacula, Adam J. Gordon, Mark Sorbero, Rachel M. Burns,

Douglas L. Leslie, and Bradley D. Stein, “Growth in Buprenorphine Waivers for Physicians

Increased Potential Access to Opioid Agonist Treatment, 2002–11,” Health Affairs, Vol. 34,

No. 6, 2015, pp. 1028–1034.

58

Dokholyan, Rachel S., Lawrence H. Muhlbaier, John M. Falletta, Jeffrey P. Jacobs, David

Shahian, Constance K. Haan, and Eric D. Peterson, “Regulatory and Ethical Considerations

for Linking Clinical and Administrative Databases,” American Heart Journal, Vol. 157, No.

6, June 2009, pp. 971–982.

Doshi, Jalpa A., Franklin B. Hendrick, Jennifer S. Graff, and Bruce C. Stuart, “Data, Data

Everywhere, but Access Remains a Big Issue for Researchers: A Review of Access Policies

for Publicly Funded Patient-Level Health Care Data in the United States,” EGEMS (Wash

DC), Vol. 4, No. 2, 2016, p. 1204.

Dowell, Deborah, Tamara M. Haegerich, and Roger Chou, “CDC Guideline for Prescribing

Opioids for Chronic Pain—United States, 2016,” Journal of the American Medical

Association, Vol 65, No. 1, March 18, 2016, pp. 1624–1645.

Doyon, Suzanne, Carleigh Benton, Bruce A. Anderson, Michael Baier, Erin Haas, Lisa Hadley,

Jennifer Maehr, Kathleen Rebbert-Franklin, Yngvild Olsen, and Christopher Welsh,

“Incorporation of Poison Center Services in a State-Wide Overdose Education and Naloxone

Distribution Program,” American Journal on Addictions, Vol. 25, No. 4, June 2016, pp. 301–

306.

Dusetzina, Stacie B., Seth Tyree, and Anne-Marie Meyer, Linking Data for Health Services

Research: A Framework and Instructional Guide [Internet], webpage, Rockville, Md.:

Agency for Healthcare Resarch and Quality, September 4, 2014. As of July 1, 2018:

https://www.ncbi.nlm.nih.gov/books/NBK253312/

Dusetzina, Stacie B., Seth Tyree, Anne-Marie Meyer, Adrian Meyer, Laura Green, and William

R. Carpenter, Linking Data for Health Services Research: A Framework and Instructional

Guide, Rockville, Md.: Agency for Healthcare Research and Quality, report no. 14-EHC033-

EF, September 2014.

Dworsky, Michael, “Using All-Payer Claims Databases to Study Insurance and Health Care

Utilization Dynamic,” Journal of General Internal Medicine, Vol. 32, No. 10, October 2017,

pp. 1069–1070.

Estrada, Antonio L., “Health Disparities Among African-American and Hispanic Drug

Injectors—HIV, AIDS, Hepatitis B Virus and Hepatitis C Virus: A Review,” AIDS, Vol. 19,

Supplement 3, October 2005, pp. S47–S52.

Fairbairn, Nadia, Phillip O. Coffin, and Alexander Y. Walley, “Naloxone for Heroin,

Prescription Opioid, and Illicitly Made Fentanyl Overdoses: Challenges and Innovations

Responding to a Dynamic Epidemic,” Internal Journal of Drug Policy, Vol. 46, August

2017, pp. 172–179.

59

Feder, Kenneth A., Noa Krawczyk, and Brendan Saloner, “Medication-Assisted Treatment for

Adolescents in Specialty Treatment for Opioid Use Disorder,” Journal of Adolescent Health,

Vol. 60, No. 6, June 2017, pp. 747–750.

Fellegi, Ivan P., and Alan B. Sunter, “A Theory for Record Linkage,” Journal of the American

Statistical Association, Vol. 64, No. 328, 1969, pp. 1183–1210.

Finney, John W., Keith Humphreys, Daniel R. Kivlahan, and Alex H. S. Harris, “Why Health

Care Process Performance Measures Can Have Different Relationships to Outcomes for

Patients and Hospitals: Understanding the Ecological Fallacy,” American Journal of Public

Health, Vol. 101, No. 9, September 2011, pp. 1635–1642.

Friedman, Daniel J., Gibson Parrish, and David A. Ross, “Electronic Health Records and U.S.

Public Health: Current Realities and Future Promise,” American Journal of Public Health,

Vol. 103, No. 9, September 2013, pp. 1560–1567.

Garnick, Deborah W., Constance M. Horgan, Andrea Acevedo, Margaret T. Lee, Lee Panas,

Grant A. Ritter, Robert Dunigan, Alfred Bidorini, Kevin Campbell, Karin Haberlin, Alice

Huber, Dawn Lambert-Wacey, Tracy Leeper, Mark Reynolds, and David Wright, “Criminal

Justice Outcomes After Engagement in Outpatient Substance Abuse Treatment,” Journal of

Substance Abuse Treatment, Vol. 46, No. 3, March 2014, pp. 295–305.

Garnick, Deborah W., Constance M. Horgan, Andrea Acevedo, Frank McCorry, and Constance

Weisner, “Performance Measures for Substance Use Disorders--What Research Is Needed?”

Addiction Science and Clinical Practice, Vol. 7, No. 18, 2012.

Gilson, Aaron M., Scott M. Fishman, Barth L. Wilsey, Carlos Casamalhuapa, and Hassan Baxi,

“Time Series Analysis of California’s Prescription Monitoring Program: Impact on

Prescribing and Multiple Provider Episodes,” The Journal of Pain, Vol. 13, No. 2, February

2012, pp. 103–111.

Goerge, Robert, Leah Gjertson, and Ella De La Cruz, Administrative Data for the Public Good:

Opportunities for Advancing Evidence-Based Policymaking Using Data Held by the U.S.

Census Bureau, Chicago, Ill.: Chapin Hall at the University of Chicago, 2017.

Gordon, Adam J., Wei-Hsuan Lo-Ciganic, Gerald Cochran, Walid F. Gellad, Terri Cathers, and

Julie M. Donohue, “Treatment Quality for Buprenorphine Care: The Pot at the End of the

Rainbow [In Reply to ‘Measurement Quality of Buprenorphine Care’], Journal of Addiction

Medicine, Vol. 10, No. 3, May–June 2016, pp. 210–211.

Green, Carla A., Nancy A. Perrin, Shannon L. Janoff, Cynthia I. Campbell, Howard D. Chilcoat,

and Paul M. Coplan, “Assessing the Accuracy of Opioid Overdose and Poisoning Codes in

Diagnostic Information from Electronic Health Records, Claims Data, and Death Records,”

Pharmacoepidemiology and Drug Safety, Vol. 26, No. 5, May 2017, pp. 509–517.

60

Greenland, Sander, “A Review of Multilevel Theory for Ecologic Analyses,” Statistics in

Medicine, Vol. 21, No. 3, February 15, 2002, pp. 389–395.

Greenwood-Ericksen, Margaret B., Sabrina J. Poon, Lewis S. Nelson, Scott G. Weiner, and

Jeremiah D. Schuur, “Best Practices for Prescription Drug Monitoring Programs in the

Emergency Department Setting: Results of an Expert Panel,” Annals of Emergency Medicine,

Vol. 67, No. 6, June 2016, pp. 755–764.e4.

Gupta, Ravi, Nilay D. Shah, and Joseph S. Ross, “The Rising Price of Naloxone—Risks to

Efforts to Stem Overdose Deaths,” New England Journal of Medicine, Vol. 375, December

8, 2016, pp. 2213–2215.

Hadland, Scott E., J. Frank Wharam, and Mark A. Schuster “Trends in Receipt of Buprenorphine

and Naltrexone for Opioid Use Disorder Among Adolescents and Young Adults, 2001–

2014,” JAMA Pediatrics, Vol. 171, No. 8, August 1, 2017, pp. 747–755.

Haegerich, Tamara M., Leonard J. Paulozzi, Brian J. Manns, and Christopher M. Jones, “What

We Know, and Don’t Know, About the Impact of State Policy and Systems-Level

Interventions on Prescription Drug Overdose,” Drug and Alcohol Dependence, Vol. 145,

December 1, 2014, pp. 34–47.

Hall, H. Irene, Ruiguang Song, Philip Rhodes, Joseph Prejean, Qian An, Lisa M. Lee, John

Karon, Ron Brookmeyer, Edward H. Kaplan, Matthew T. McKenna, and Robert S. Janssen,

“Estimation of HIV Incidence in the United States,” Journal of the American Medical

Association, Vol. 300, No. 5, August 6, 2008, pp. 520–529.

Han, Huijun, Philip H. Kass, Barth L. Wilsey, and Chin-Shang Li, “Increasing Trends in

Schedule II Opioid Use and Doctor Shopping During 1999–2007 in California,”

Pharmacoepidemiology and Drug Safety, Vol. 23, No. 1, January 2014, pp. 26–35.

Harle, Christopher A., Robert L. Cook, Heidi S. Kinsell, and Jeffrey S. Harman, “Opioid

Prescribing by Physicians with and Without Electronic Health Records,” Journal of Medical

Systems, Vol. 38, No. 11, November 2014, p. 138.

Hartung, Daniel M., Sharia M. Ahmed, Luke Middleton, Joshua Van Otterloo, Kun Zhang,

Shellie Keast, Hyunjee Kim, Kirbee Johnston, and Richard A. Deyo, “Using Prescription

Monitoring Program Data o Characterize Out!Oof!Pocket Payments for Opioid Prescriptions

in a State Medicaid Program,” Pharmacoepidemiology and Drug Safety, Vol. 26, No. 9,

September 2017, pp. 1053–1060.

Häyrinen, Kristiina, Kaija Saranto, and Pirkko Nykänen, “Definition, Structure, Content, Use

and Impacts of Electronic Health Records: A Review of the Research Literature,”

International Journal of Medical Informatics, Vol. 77, No. 5, May 2008, pp. 291–304.

61

Healthcare Cost and Utilization Project, “HCUP Fast Stats—Opioid-Related Hospital Use,”

webpage, June 2018. As of July 1, 2018:

https://www.hcup-us.ahrq.gov/faststats/OpioidUseServlet

Health Resources and Services Administration, “Health Center Patient Survey,” webpage,

undated. As of May 30, 2018:

https://bphc.hrsa.gov/datareporting/research/hcpsurvey/index.html

Health Resources and Services Administration, “Uniform Data System (UDS) Resources,”

webpage, June 2018. As of May 17, 2018:

https://bphc.hrsa.gov/datareporting/reporting/index.html

HHS—See U.S. Department of Health and Human Services.

Hirsch, Anne, Scott K. Proescholdbell, William Bronson, and Nabarun Dasgupta, “Prescription

Histories and Dose Strengths Associated with Overdose Deaths,” Pain Medicine, Vol. 15,

No. 7, July 2014, pp. 1187–1195.

Houry, Debra, “Testimony from Debra Houry, M.D. on Fentanyl: The Next Wave of the Opioid

Crisis before Committee on Energy and Commerce,” testimony before the Committee on

Energy and Commerce, Subcommittee on Oversight and Investigations, Washington, D.C.,

March 21, 2017. As of July 1, 2018:

https://www.hhs.gov/about/agencies/asl/testimony/2017-03/fentanyl-next-wave-opioid-

crisis.html

Hser, Yih-Ing, Larissa J. Mooney, Andrew J. Saxon, Karen Miotto, Douglas S. Bell, and David

Huang, “Chronic Pain Among Patients with Opioid Use Disorder: Results from Electronic

Health Records Data,” Journal of Substance Abuse Treatment, Vol. 77, June 2017, pp. 26–

30.

HRSA—See Health Resources and Services Administration.

Hudson, Toni-Marie, Benjamin G. Klekamp, and Sarah D. Matthews, “Local Public Health

Surveillance of Heroin-Related Morbidity and Mortality, Orange County, Florida, 2010–

2014,” Public Health Reports, Vol. 132, Supplement 1, July–August 2017, pp. 80S–87S.

Interagency Pain Research Coordinating Committee, National Pain Strategy: A Comprehensive

Population Health-Level Strategy for Pain, Washington, D.C.: Department of Health and

Human Services, 2015. As of June 27, 2018:

https://iprcc.nih.gov/docs/HHSNational_Pain_Strategy.pdf

Ising, Amy, Scott Proescholdbell, Katherine J. Harmon, Nidhi Sachdeva, Stephen W. Marshall,

and Anna E. Waller, “Use of Syndromic Surveillance Data to Monitor Poisonings and Drug

Overdoses in State and Local Public Health Agencies,” Injury Prevention, Vol. 22,

Supplement 1, 2016, pp. i43–i49.

62

Johnson, Kristen M., Meghan Fibbi, Debra Langer, Karol Silva, and Stephen E. Lankenau,

“Prescription Drug Misuse and Risk Behaviors Among Young Injection Drug Users,”

Journal of Psychoactive Drugs, Vol. 45, No. 2, April–June 2013, pp. 112–121.

Jones, Christopher M., Melinda Campopiano, Grant Baldwin, and Elinore McCance-Katz,

“National and State Treatment Need and Capacity for Opioid Agonist Medication-Assisted

Treatment,” American Journal of Public Health, Vol. 105, No. 8, August 2015, pp. e55–e63.

Kalyanam, Janani, Takeo Katsuki, Gert R. G. Lanckriet, and Tim K. Mackey, “Exploring Trends

of Nonmedical Use of Prescription Drugs and Polydrug Abuse in the Twittersphere Using

Unsupervised Machine Learning,” Addictive Behaviors, Vol. 65, February 2017, pp. 289–

295.

Kalyanam, Janani, and Tim K. Mackey, “A Review of Digital Surveillance Methods and

Approaches to Combat Prescription Drug Abuse,” Current Addiciton Reports, Vol. 4, No. 4,

December 2017, pp. 1–13.

Katsuki, Takeo, Tim Ken Mackey, and Raphael Cuomo, “Establishing a Link Between

Prescription Drug Abuse and Illicit Online Pharmacies: Analysis of Twitter Data,” Journal of

Medical Internet Research, Vol. 17, No. 12, December 16, 2015, p. e280.

Katz, Nathaniel, Lee Panas, MeeLee Kim, Adele D. Audet, Arnold Bilansky, John Eadie, Peter

Kreiner, Florence C. Paillard, Cindy Thomas, and Grant Carrow, “Usefulness of Prescription

Monitoring Programs for Surveillance—Analysis of Schedule II Opioid Prescription Data in

Massachusetts, 1996–2006,” Pharmacoepidemiology and Drug Safety, Vol. 19, No. 2,

February 2010, pp. 115–123.

Kennedy-Hendricks, Alene, Matthew Richey, Emma E. McGinty, Elizabeth A. Stuart, Colleen

L. Barry, and Daniel W. Webster, “Opioid Overdose Deaths and Florida’s Crackdown on Pill

Mills,” American Journal of Public Health, Vol. 106, No. 2, February 2016, pp. 291–297.

Kho, Abel N., John P. Cashy, Kathryn L. Jackson, Adam R. Pah, Satyender Goel, Jörn Boehnke,

John Eric Humphries, Scott Duke Kominers, Bala N. Hota, Shannon A. Sims, Bradley A.

Malin, Dustin D. French, Theresa L. Walunas, David O. Meltzer, Erin O. Kaleba, Roderick

C. Jones, and William L. Galanter, “Design and Implementation of a Privacy Preserving

Electronic Health Record Linkage Tool in Chicago,” Journal of the American Medical

Informatics Association, Vol. 22, No. 5, September 2015, pp. 1072–1080.

Knudsen, Hannah K., Michelle R. Lofwall, Jennifer R. Havens, and Sharon L. Walsh, “States’

Implementation of the Affordable Care Act and the Supply of Physicians Waivered to

Prescribe Buprenorphine for Opioid Dependence,” Drug and Alcohol Dependence, Vol. 157,

December 1, 2015, pp. 36–43.

Krebs, Erin E., Amy Gravely, Sean Nugent, Agnes C. Jensen, Beth DeRonne, Elizabeth S.

Goldsmith, Kurt Kroenke, Matthew J. Bair, and Siamak Noorbaloochi, “Effect of Opioid vs.

63

Nonopioid Medications on Pain-Related Function in Patients with Chronic Back Pain or Hip

or Knee Osteoarthritis Pain: The SPACE Randomized Clinical Trial,” Journal of the

American Medical Association, Vol. 319, No. 9, March 6, 2018, pp. 872–882.

Krebs, Erin E., Jon D. Lurie, Gilbert Fanciullo, Tor D. Tosteson, Emily A. Blood, Timothy S.

Carey, and James N. Weinstein, “Predictors of Long-Term Opioid Use Among Patients with

Painful Lumbar Spine Conditions,” Journal of Pain, Vol. 11, No. 1, January 2010, pp. 44–

52.

Kreiner, Peter W., Gail K. Strickler, Eduardo A. Undurraga, Maria E. Torres, Ruslan V. Nikitin,

and Anne Rogers, “Validation of Prescriber Risk Indicators Obtained from Prescription Drug

Monitoring Program Data,” Drug and Alcohol Dependence, Vol. 173, Supplement 1, April 1,

2017, pp. S31–S38.

Kum, Hye-Chung, Ashok Krishnamurthy, Ashwin Machanavajjhala, Michael K. Reiter, and

Stanley Ahalt, “Privacy Preserving Interactive Record Linkage (PPIRL),” Journal of the

American Medical Informatics Association, Vol. 21, No. 2, March–April 2014, pp. 212–220.

Kwiatkowski, Carol F., Robert E. Booth, Laura V. Lloyd, “The Effects of Offering Free

Treatment to Street-Recruited Opioid Injectors,” Addiction, Vol. 95, No. 5, May 2000, pp.

697–704.

Ladegaard, Isak, “Instantly Hooked? Freebies and Samples of Opioids, Cannabis, MDMA, and

Other Drugs in an Illicit E-Commerce Market,” Journal of Drug Issues, Vol. 48, No. 2,

December 2017.

Lin, Dora H., Eleanor Lucas, Irene B. Murimi, Katherine Jackson, Michael Baier, Shannon

Frattaroli, Andrea C. Gielen, Patience Moyo, Linda Simoni-Wastila, and G. Caleb

Alexander, “Physician Attitudes and Experiences with Maryland's Prescription Drug

Monitoring Program (PDMP),” Addiction, Vol. 112, No. 2, February 2017, pp. 311–319.

Lingren, Todd, Senthilkumar Sadhasivam, Xue Zhang, and Keith Marsolo, “Electronic Medical

Records as a Replacement for Prospective Research Data Collection in Postoperative Pain

and Opioid Response Studies,” International Journal of Medical Informatics, Vol. 111,

March 2018, pp. 45–50.

Lyons, B. Casey, and Kirsten Madison, Predictive Risk Evaluation to Combat Overdose Grant

(PRECOG),” Maryland Department of Health and Mental Hygiene, March 31, 2017. As of

September 18, 2017:

http://www.pdmpassist.org/pdf/31-A-4_Maryland.pdf

Madden, Jeanne M., Matthew D. Lakoma, Donna Rusinak, Christine Y. Lu, and Stephen B.

Soumerai, “Missing Clinical and Behavioral Health Data in a Large Electronic Health

Record (EHR) System,” Journal of American Medical Informatics Association, Vol. 23, No.

6, November 2016, pp. 1143–1149.

64

Maddux, James F., and David P. Desmond, “Outcomes of Methadone Maintenance 1 Year After

Admission,” Journal of Drug Issues, Vol. 27, No. 2, April 1, 1997, pp. 225–238.

Mai, Jaymie, Gary Franklin, and David Tauben, “Guideline for Prescribing Opioids to Treat Pain

in Injured Workers,” Physical Medicine and Rehabilation Clinics of North America, Vol. 26,

No. 3, August 2015, pp. 453–465.

Manasco, A. Travis, Christopher Griggs, Rebecca Leeds, Breanne K. Langlois, Alan H. Breaud,

Patricia M. Mitchell, and Scott G. Weiner, “Characteristics of State Prescription Drug

Monitoring Programs: A State-by-State Survey,” Pharmacoepidemiology and Drug Safety,

Vol. 25, No. 7, July 2016, pp. 847–851.

McCormick, Meghan, Jennifer Koziol, and Kelly Sanchez, “Development and Use of a New

Opioid Overdose Surveillance System, 2016,” Substance Abuse, No. 1091, 2017, pp. 71–76.

McCoy, Allison B., Adam Wright, Michael G. Kahn , Jason S. Shapiro, Elmer Victor Bernstam,

and Dean F. Sittig, “Matching Identifiers in Electronic Health Records: Implications for

Duplicate Records and Patient Safety,” BMJ Quality and Safety, Vol. 22, No. 3, March 2013,

pp. 219–224.

Méray, Nora, Johannes B. Reitsma, Anita C. J. Ravelli, and Gouke J. Bonsel, “Probabilistic

Record Linkage Is a Valid and Transparent Tool to Combine Databases Without a Patient

Identification Number,” Journal of Clinical Epidemiology, Vol. 60, No. 9, September 2007,

pp. 883–891.

Merlin, Mark A., Navin Ariyaprakai, and Faizan H. Arshad, “Assessment of the Safety and Ease

of Use of the Naloxone Auto-Injector for the Reversal of Opioid Overdose,” Open Access

Emergency Medicine, Vol. 7, June 8, 2015, pp. 21–24.

Minnesota Department of Health, “Opioid Dashboard,” webpage, undated. As of June 27, 2018:

http://www.health.state.mn.us/divs/healthimprovement/opioid-dashboard

Morgan, Jake R., Bruce R. Schackman, Jared A. Leff, Benjamin P. Linas, and Alexander Y.

Walley, “Injectable Naltrexone, Oral Naltrexone, and Buprenorphine Utilization and

Discontinuation Among Individuals Treated for Opioid Use Disorder in a United States

Commercially Insured Population,” Journal of Substance Abuse Treatment, Vol. 85,

February 2018, pp. 90–86.

Murray, Michael D., “Use of Data from Electronic Health Records for Pharmacoepidemiology,”

Current Epidemiology Reports, Vol. 1, No. 4, December 2014, pp. 186–193.

National Institute on Drug Abuse, “Overdose Death Rates,” webpage, revised September 2017.

As of July 1, 2018:

https://www.drugabuse.gov/related-topics/trends-statistics/overdose-death-rates

65

National Institute of Justice, “NIJ’s Drugs and Crime Research: Arrestee Drug Abuse

Monitoring Programs,” webpage, June 18, 2014. As of July 6, 2018:

https://www.nij.gov/topics/drugs/markets/adam/pages/welcome.aspx

NEMSIS, “Public Naloxone Administration Dashboard,” webpage, undated. As of May 17,

2018:

https://nemsis.org/view-reports/public-reports/version-3-public-dashboards/public-naloxone-

administration-dashboard

Nuckols, Teryl K., Laura Anderson, Ioana Popescu, Allison L. Diamant, Brian Doyle, Paul Di

Capua, and Roger Chou, “Opioid Prescribing: ASystematic Review and Critical Appraisal of

Guidelines for Chronic Pain,” Annals of Internal Medicine, Vol. 160, No. 1, January 2014,

pp. 38–47.

OCHIN, “ADVANCE: Accelerating Data Value Across a National Community Health Center

Network,” webpage, March 1, 2014–August 1, 2018. As of May 17, 2018:

https://ochin.org/ochin-research/advance-research-summary

O’Kane, Nicole, Sara E. Hallvik, Miguel Marino, Joshua Van Otterloo, Christi Hildebran,

Gillian Leichtling, and Richard A. Deyo, “Preparing a Prescription Drug Monitoring

Program Data Set for Research Purposes,” Pharmacoepidemiology and Drug Safety, Vol. 26,

No. 9, September 2016, pp. 993–997.

“Opioid Mapping Initiative,” webpage, undated. As of May 17, 2018:

http://opioidmappinginitiative-opioidepidemic.opendata.arcgis.com

Pardo, Bryce, “Do More Robust Prescription Drug Monitoring Programs Reduce Prescription

Opioid Overdose?” Addiction, Vol. 112, No. 10, October 2017, pp. 1773–1783.

Parker, Jason, Courtney Cuthbertson, Scott Loveridge, Mark Skidmore, and Will Dyar,

“Forecasting State-Level Premature Deaths from Alcohol, Drugs, and Suicides Using Google

Trends Data,” Journal of Affective Disorders, Vol. 213, April 15, 2017, pp. 9–15.

Paulozzi, Leonard J., Edwin M. Kilbourne, and Hema A. Desai, “Prescription Drug Monitoring

Programs and Death Rates from Drug Overdose,” Pain Medicine, Vol. 12, No. 5, May 2011,

pp. 747–754.

pcornet, “Partner Networks,” webpage, undated. As of May 17, 2018:

http://www.pcornet.org/participating-networks

PDAPS—See Prescription Drug Abuse Policy System.

PDMP—See Prescription Drug Monitoring Program.

Piper, Brian J., Claire E. Desrosiers, John W. Lipovsky, Matthew A. Rodney, Robert P. Baker,

Kenneth L. McCall, Stephanie D. Nichols, and Sarah L. Martin, “Use and Misuse of Opioids

in Maine: Results From Pharmacists, the Prescription Monitoring, and the Diversion Alert

66

Programs,” Journal of Studies on Alcohol and Drugs, Vol. 77, No. 4, July 2016, pp. 556–

565.

Porucznik, Christina A., Erin M. Johnson, Robert T. Rolfs, and Brian C. Sauer, “Specialty of

Prescribers Associated with Prescription Opioid Fatalities in Utah, 2002–2010,” Pain

Medicine, Vol. 15, No. 1, January 1, 2014, pp. 73–78.

Prescription Drug Abuse Policy System, homepage, undated. As of May 17, 2018:

http://www.pdaps.org

Prescription Drug Monitoring Program Center of Excellence at Brandeis, Using PDMPs to

Improve Medical Care: Washington State’s Data Sharing Initiative with Medicaid and

Workers’ Compensation, Notes from the Field series, April 2013. As of June 27, 2018:

http://www.pdmpassist.org/pdf/COE_documents/Add_to_TTAC/washington_nff_final.pdf

Prescription Drug Monitoring Program, Training and Technical Assistance Center, “Status of

Prescription Drug Monitoring Programs (PDMPs),” webpage, August 24, 2017. As of

September 18, 2017:

http://www.pdmpassist.org/pdf/PDMP_Program_Status_20170824.pdf

Prescription Drug Monitoring Program, “Release of PDMP Data for Research, Epidemiological,

or Educational Purposes,” webpage, December 5, 2017. As of May 17, 2018:

http://www.pdmpassist.org/pdf/Data_Use_Res_Epi_Educ_20171205.pdf

Price, Thomas E., “Secretary Price Announces HHS Strategy for Fighting Opioid Crisis,”

Atlanta, Ga., National Rx Drug Abuse and Heroin Summit, April 19, 2017. As of June 27,

2018:

https://www.hhs.gov/about/leadership/secretary/speeches/2017-speeches/secretary-price-

announces-hhs-strategy-for-fighting-opioid-crisis/index.html

Quast, Troy, Eric A. Storch, and Svetlana Yampolskaya, “Opioid Prescription Rates and Child

Removals: Evidence from Florida,” Health Affairs, Vol. 37, No. 1, January 2018, pp. 134–

139.

Raghupathi, Wullianallur, and Viju Raghupathi, “Big Data Analytics in Healthcare: Promise and

Potential,” Health Information Science and Systems, Vol. 2, No. 3, 2014.

Rhode Island Department of Health, “Policy: Identifying and Reporting Confirmed Accidental

Drug-Related Overdose Deaths, June 2015,” 2015. As of September 21, 2017:

http://www.health.ri.gov/publications/policies/IdentifyingAndReportingConfirmedAccidenta

lDrugRelatedOverdoseDeaths.pdf

Rinaldo, Suzanne Gelber, and David W. Rinaldo, “Report I: Availability Without Accessability?

State Medicaid Coverage and Authorization Requirements for Opioid Dependence

Medications: Implications for Opioid Addiction Treatment,” in Advancing Access to

67

Addiction Medications: Implications for Opioid Addiction Treatment, Chevy Chase, Md.:

American Society of Addiction Medicine, 2013.

Ringwalt, Christopher, Sharon Schiro, Meghan Shanahan, Scott Proescholdbell, Harold Meder,

Anna Austin, and Nidhi Sachdeva, “The Use of a Prescription Drug Monitoring Program to

Develop Algorithms to Identify Providers with Unusual Prescribing Practices for Controlled

Substances,” Journal of Primary Prevention, Vol. 36, No. 5, October 2015, pp. 287–299.

Robert, Andrew W., Joel F. Farley, G. Mark Holmes, Christine U. Oramasionwu, Chris

Ringwalt, Betsy Sleath, and Asheley C. Skinner, “Controlled Substance Lock-In Programs:

Examining An Unintended Consequence Of A Prescription Drug Abuse Policy,” Health

Affairs, Vol. 35, No. 10, October 1, 2016, pp. 1884–1892.

Robinson, W. S., “Ecological Correlations and the Behavior of Individuals,” American

Sociological Review, Vol. 15, No. 3, June 1950, pp. 351–357.

Rosenblatt, Roger A., C. Holly A. Andrilla, Mary Catlin, and Eric H. Larson, “Geographic and

Specialty Distribution of U.S. Physicians Trained to Treat Opioid Use Disorder,” Annals of

Family Medicine, Vol. 13, No. 1, January–February 2015, pp. 23–26.

Ross, Joseph S., and Harlan M. Krumholz, “Ushering in a New Era of Open Science Through

Data Sharing: The Wall Must Come Down,” Journal of the American Medical Association,

Vol. 309, No. 13, April 3, 2013, pp. 1355–1356.

Rudd, Rose A., Puja Seth, Felicita David, and Lawrence Scholl, “Increases in Drug and Opioid-

Involved Overdose Deaths—United States, 2010–2015,” Morbidity and Mortality Weekly

Report (MMWR), Vol. 65, No. 50–51, December 2016, pp. 1445–1452.

Ruhm, Christopher J., “Geographic Variation in Opioid and Heroin Involved Drug Poisoning

Mortality Rates,” American Journal of Preventive Medicine, Vol. 53, No. 6, August 1, 2017,

pp. 745–753

Russo, Elise, Dean F. Sittig, Daniel R. Murphy, and Hardeep Singh, “Challenges in Patient

Safety Improvement Research in the Era of Electronic Health Records,” Healthcare, Vol. 4,

No. 4, December 2016, pp. 285–290.

Saloner, Brendan, “Using Data Science to Identify Individuals at High Risk of Opioid Overdose:

A Multiyear Data Linkage Project in Maryland,” paper presented at the Association for

Public Policy Analysis and Management 38th Annual Fall Research Conference: The Role of

Research in Making Government More Effective, Washington, D.C., November 4, 2016.

Saloner, Brendan, and Shankar Karthikeyan, “Changes in Substance Abuse Treatment Use

Among Individuals with Opioid Use Disorders in the United States, 2004–2013,” Journal of

the American Medical Association, Vol. 314, No. 14, October 13, 2015, pp. 1515–1517.

68

Sayers, Adrian, Yoav Ben-Shlomo, Ashley W. Blom, and Fiona Steele, “Probabilistic Record

Linkage,” Internal Journal of Epidemiology, Vol. 45, No. 3, June 2016, pp. 954–964.

Schuur, Jeremiah D., Akash Shah, Zheyang Wu, Howard P. Forman, and Cary P. Gross, “The

Impact of Medicaid Coverage and Reimbursement on Access to Diagnostic Mammography,”

Cancer, Vol. 115, No. 23, December 1, 2009, pp. 5566–5578.

Shen, Yu-Chu, and Stephen Zuckerman, “The Effect of Medicaid Payment Generosity on Access

and Use Among Beneficiaries,” Health Services Research, Vol. 40, No. 3, June 2005, pp.

723–744.

Slavova, Svetla, Julia F.Costich, Terry L.Bunn, Huong Luu, Michael Singleton, Sarah L.

Hargrove, Jeremy S. Triplett, Dana Quesinberry, William Ralston, and Van Ingram, “Heroin

and Fentanyl Overdoses in Kentucky: Epidemiology and Surveillance,” International

Journal of Drug Policy, Vol. 46, August 2017, pp. 120–129.

Stein, Bradley D., Adam J. Gordon, Andrew W. Dick, Rachel M. Burns, Rosalie Liccardo

Pacula, Carrie M. Farmer, Douglas L. Leslie, and Mark Sobero, “Supply of Buprenorphine

Waivered Physicians: The Influence of State Policies,” Journal of Substance Abuse

Treatment, Vol. 48, No. 1, January 2015, pp. 104–111.

Stein, Bradley D., Rosalie Liccardo Pacula, Adam J. Gordon, Rachel M. Burns, Douglas L.

Leslie, Mark J. Sorbero, Sebastian Bauhoff, Todd W. Mandell, and Andrew W. Dick,

“Where Is Buprenorphine Dispensed to Treat Opioid Use Disorders? The Role of Private

Offices, Opioid Treatment Programs, and Substance Abuse Treatment Facilities in Urban and

Rural Counties,” The Milbank Quarterly, Vol. 93, No. 3, September 2015, pp. 561–583.

Substance Abuse and Mental Health Services Administration, “DAWN: Drug Abuse Warning

Network,” webpage, 2018,. As of July 16, 2018:

https://www.samhsa.gov/data/data-we-collect/dawn-drug-abuse-warning-network

Tennessee Department of Health, “Data Dashboard,” webpage, undated. As of May 17, 2018:

https://www.tn.gov/health/health-program-areas/pdo/pdo/data-dashboard.html

Thomas, Cindy Parks, Deborah W. Garnick, Constance M. Horgan, Frank McCorry, Amanda

Gmyrek, Mady Chalk, David R. Gastfriend, Suzanne Gelber Rinaldo, Joann Albright, Victor

A. Capoccia, Alex H. S. Harris, Henrick J. Harwood, Pamela Greenberg, Tami L. Mark,

Huong Un, Marla Oros, Mark Stringer, and James Thatcher, “Advancing Performance

Measures for Use of Medications in Substance Abuse Treatment,” Journal of Substance

Abuse Treatment, Vol. 40, No. 1, January 2011, pp. 35–43.

Tromp, Miranda, Anita C. J. Ravelli, Gouke J. Bonsel, Arie Hasman, and Johannes B. Reitsma,

“Results from Simulated Data Sets: Probabilistic Record Linkage Outperforms Deterministic

Record Linkage,” Journal of Clinical Epidemiology, Vol. 54, No. 5, May 2011, pp. 565–572.

69

U.S. Census Bureau, “Federal Statistical Research Data Centers,” webpage, last revised January

14, 2015. As of May 17, 2018:

https://www.census.gov/about/adrm/fsrdc/about.html

U.S. Census Bureau, “United States Census Bureau Data Repository,” webpage, 2017. As of

May 30, 2018:

https://census.icpsr.umich.edu/census

U.S. Department of Health and Human Services, “Help, Resources, and Information: National

Opioid Crisis,” webpage, undated. As of March 28, 2018:

https://www.hhs.gov/opioids

U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services

Administration, Center for Behavioral Health Statistics and Quality, Drug Abuse Warning

Network, 2011: National Estimates of Drug-Related Emergency Department Visits,

Rockville, Md., May 2013.

U.S. Department of Health and Human Services, Behavioral Health Coordinating Committee,

Prescription Drug Abuse Subcommittee, Addressing Prescription Drug Abuse in the United

States: Current Activities and Future Opportunities, Washington, D.C., September 2013. As

of September 18, 2017:

http://www.cdc.gov/drugoverdose/pdf/hhs_prescription_drug_abuse_report_09.2013.pdf

U.S. Supreme Court, Gobeille v. Liberty Mutual Insurance Co. 577 U.S., 2016. As of June 11,

2018:

https://supreme.justia.com/cases/federal/us/577/14-181

Wakeland, Wayne, Alexandra Nielsen, and Peter Geissert, “Dynamic Model of Nonmedical

Opioid Use Trajectories and Potential Policy Interventions,” American Journal of Drug and

Alcohol Abuse, Vol. 41, No. 6, November 2015, pp. 508–518.

Walley, Alexander Y., Ziming Xuan, H. Holly Hackman, Emily Quinn, Maya Doe-Simkins,

Amy Sorensen-Alawad, Sarah Ruiz, and Al Ozonoff, “Opiod Overdose Rates and

Implementation of Overdose Education and Nasal Naloxone Distribution in Massachusetts:

Interrupted Time Series Analysis,” BMJ, Vol. 346, 2013.

Warner, Margaret, Leonard J. Paulozzi, Kurt B. Nolte, Gregory G. Davis, and Lewis S. Nelson,

“State Variation in Certifying Manner of Death and Drugs Involved in Drug Intoxication

Deaths,” Academic Forensic Pathology, Vol. 3, No. 2, June 2013, pp. 231–237.

Watkins, Katherine E., Allison J. Ober, Karen Lamp, Mimi Lind, Claude Setodji, Karen Chan

Osilla, Sarah B. Hunter, Colleen M. McCullough, Kirsten Becker, Praise O. Iyiewuare,

Allison Diamant, Keith Heinzerling, and Harold Alan Pincus, “Collaborative Care for Opioid

and Alcohol Use Disorders in Primary Care: The SUMMIT Randomized Clinical Trial,”

JAMA Internal Medicine, Vol. 177, No. 10, August 28, 2017, pp. 1480–1488.

70

Weiskopf, Nicole Gray, and Chunhua Weng, “Methods and Dimensions of Electronic Health

Record Data Quality Assessment: Enabling Reuse for Clinical Research,” Journal of

American Medical Informatics Association, Vol. 20, January 1, 2013, pp. 144–151.

Weiner, Scott G., Olesya Baker, Sabrina J. Poon, Ann F. Rodgers, Chad Garner, Lewis S.

Nelson, and Jeremiah D. Schuur, “The Effect of Opioid Prescribing Guidelines on

Prescriptions by Emergency Physicians in Ohio,” Annals of Emergency Medicine, Vol. 70,

No. 6, December 2017, pp. 799–808.

Wheeler, Eliza, T. Stephen Jones, Michael K. Gilbert, and Peter J. Davidson, “Opioid Overdose

Prevention Programs Providing Naloxone to Laypersons—United States, 2014,” Morbidity

and Mortality Weekly Report (MMWR), Vol. 64, No. 23, June 19, 2015, pp. 631–635.

Winkler, William E., The State of Record Linkage and Current Research Problems, Washington,

D.C.: Statistical Research Division, U.S. Census Bureau, 1999.

Winkler, William E., Overview of Record Linkage and Current Research Directions,

Washington, D.C.: Statistical Research Division, U.S. Census Bureau, research report series

(stiastics #2006-2), February 8, 2006.

Wilsey, Barth L., Scott M. Fishman, Aaron M. Gilson, Carlos Masamalhuapa, Hassan Baxi, Tzu-

Chun Lin, and Chin-Shang Li, “An Analysis of the Number of Multiple Prescribers for

Opioids Utilizing Data from the California Prescription Monitoring Program*,”

Pharmacoepidemiology and Drug Safety, Vol. 20, No. 12, December 2011, pp. 1262–1268.

Wright, Elizabeth A., Jeffrey N. Katz, Stanley Abrams, Daniel H. Solomon, and Elena Losina,

“Trends in Prescription of Opioids from 2003–2009 in Persons with Knee Osteoarthritis,”

Arthritis Care and Research, Vol. 66, No. 10, October 2014, pp. 1489–1495.

Van Hout, Marie Claire, and Evelyn Hearne, “New Psychoactive Substances (NPS) on

Cryptomarket Fora: An Exploratory Study of Characteristics of Forum Activity Between

NPS Buyers and Vendors,” Internal Journal of Drug Policy, Vol. 40, February 2017, pp.

102–110.

Volkow, Nora D., and Francis S. Collins, “The Role of Science in Addressing the Opioid Crisis,”

New England Journal of Medicine, Vol. 377, No. 4, July 27, 2017, pp. 391–394.

Volkow, Nora D., Thomas R. Frieden, Pamela S. Hyde, and Stephen S. Cha, “Medication-

Assisted Therapies—Tackling the Opioid-Overdose Epidemic,” New England Journal of

Medicine, Vol. 370, No. 22, May 29, 2014, pp. 2063–2066.

Yoo, Byung-Kwang, Andrea Berry, Megumi Kasajima, and Peter G. Szilagyi, “Association

Between Medicaid Reimbursement and Child Influenza Vaccination Rates,” Pediatrics, Vol.

126, No. 5, November 2010, pp. e998–e1010.

71

72

Appendix—Overview of Types of Secondary Data Sources and

Data Inventory Content



73

Appendix Table of Contents

Overview of Types of Secondary Data Sources and Data Inventory Content ...................................................................................................................74

Table A.1. National Survey Data ......................................................................................................................................................................................75

Table A.2. Claims and Electronic Health Records Secondary Data Sources ....................................................................................................................78

Table A.3. Mortality Records ............................................................................................................................................................................................84

Table A.4. Prescription Monitoring Secondary Data Sources ...........................................................................................................................................86

Table A.5. Contextual and Policy Data Sources ................................................................................................................................................................89

Table A.6. Other National, State, and Local Secondary Data Sources ..............................................................................................................................91

Data Description Summary Examples of Important Measures Data Source Examples

Information on Linking

Capability

National surveys Description: Generally household or school-based surveys with self- Prescription opioid use, heroin use, National Survey on Drug State, substate, and

(Table A.1) reported information on drug use and health; other surveys are of

hospitals, treatment facilities, or of other medical service providers

Geographic coverage: National

Timing: Generally collected and available annually

opioid use disorders, medical

conditions, health care utilization

Use and Health, National

Ambulatory Medical Care

Survey, National Survey of

Substance Abuse

Treatment Services Data

person-level linkages

possible. See Table A.1

for details.

Electronic health Description: An EHR is a digital version of a patient’s paper chart. Previously prescribed opioids or Stanford Translational State, substate, and

records (EHRs) While an EHR contains the medical and treatment histories of other medications; patient history, Research Integrated person-level linkages

(Table A.2) patients, an EHR system is built to go beyond standard clinical data

collected in a provider’s office and can be inclusive of a broader view

of a patient’s care. EHRs contain

a patient’s medical history,

diagnoses, medications, treatment plans, allergies, radiology images,

laboratory, and test results.

Geographic coverage: Varies by source

Timing: Near-real time or real-time collection

medications, clinical conditions,

treatment plans, and lab/test

results; may include clinician notes

Database, HealthCore

Integrated Research

Database, Group Health

Cooperative in

Washington State

possible. See Table A.2

for details.

Claims data Description: Patient-level claims data for reimbursement for services Prescription drug utilization; service IMS, Symphony Health, State, substate, and

(Table A.2) submitted by health care providers and pharmacies to insurance

companies. Validated algorithms to identify opioid misuse or abuse

from claims data are being developed.

Geographic coverage: Varies by source

Timing: Varies by source

utilization Truven Marketscan data,

Medicaid claims, Medicare

Part D Prescription Drug

Event data

person-level linkages

possible. See Table A.2

for details.

Mortality Description: Death rates and causes of death by drug compound Rates of opioid-involved deaths, Centers for Disease State, substate, and

records and/or International Classification of Diseases (ICD) code. Additional drugs involved in overdose deaths Control and Prevention person-level linkages

(Table A.3) information can include toxicology reports.

Geographic coverage: National or single state

Timing: Generally available annually

(CDC) WONDER Multiple

Cause of Death data, Fatal

Accident Reporting

System, National Death

Index (NDI)

possible. See Table A.3

for details.

Prescription Description: Data systems to track and monitor the distribution or Opioid prescribing rates (by type), Automation of Reports State, substate, and

monitoring data prescription of controlled substances indicators of "doctor shopping," and Consolidated Orders person-level linkages

(Table A.4) Geographic coverage: Varies by source

Timing: Varies by source

geographic variation in opioid

distribution

System; state prescription

drug-monitoring programs

possible. See Table A.4

for details.

Contextual and Description: Causal analyses of the effects of policy changes on State opioid policies, state and Area Health Resource Typically merged at the

policy data opioid-related outcomes generally use data on state laws from these county demographic and Files, Policy Surveillance state or county level

(Table A.5) sources and/or includes controls for state or county characteristics to

support causal interpretation.

Geographic coverage: National

Timing: Varies, but generally semiannually

socioeconomic factors, state and

county health care variables

System, Prescription Drug

Abuse Policy System

with other data on

opioid-related

outcomes

Other national, Description: Includes data collected through law enforcement, Law enforcement drug seizures, NEMSIS, National Poison State, substate, and

state, and local national public health surveillance systems (e.g., poison control nonfatal opioid overdose, opioid- Data System, Healthcare person-level linkages

sources centers, emergency department visits), overdose education and related emergency department Cost and Utilization possible. See Table A.6

(Table A.6) naloxone distribution programs, and hospitalization and emergency

departments

Geographic coverage: Varies by source

Timing: Varies by source

visits and hospitalizations, naloxone

distribution through community

organizations

Project (HCUP) emergency

department and

hospitalization data

for details.

74

Overview of Types of Secondary Data Sources and Data Inventory Content

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Arrestee Drug Abuse Monitoring System (ADAM): Urinalysis results and self-reported drug use and drug use–treatment history collected from adult male arrestees shortly

after their arrests; prevalence estimates are annualized to analyze trends. https://www.nij.gov/topics/drugs/markets/adam/pages/welcome.aspx

National National Annual, Urine screen results, self-reported drug use, No costs stated, some data available only None identified. See No individual-level

Institute of (subset of 1998– self-reported lifetime treatment history, to users at Inter-university Consortium recent report here. linkages identified

Justice jurisdictions 2003 and some information on illicit drug markets for Political and Social Research (ICPSR)

(federal) varies by

year)

2007–

2013

member institutions or upon signing a

Restricted Data Use Agreement

Medical Expenditure Panel Survey (MEPS): A set of large-scale surveys of families and individuals, their medical providers, and employers across the United States. MEPS is

the most-complete source of data on the cost and use of health care (including prescription medications) and health insurance coverage. Data on MEPS participants from

health care providers and facilities are cross-referenced with survey responses from the participants. https://meps.ahrq.gov/mepsweb/

Agency for National Annual Health care visits, use, events, and No costs stated. Researchers and users See online query system MEPS link files to

Healthcare expenditures, names of any prescription with approved projects can access here. National Health

Research medications, and the name and location of restricted data or state/county Interview Survey

and the pharmacy where they obtained the identifiers that have not been publicly person-level public use

Quality prescription. Data on pharmacy-filled released for reasons of confidentiality at data files

(AHRQ) prescription include type, dosage, and the AHRQ Data Center in Rockville,

(federal) payment Maryland, or through the U.S. Census

Research Data Center (RDC) network.

Monitoring the Future Survey (MTF): Nationally representative survey of self-reported drug use by 8th, 10th, and 12th graders. Longitudinal data collection (designed to be

nationally and not state representative). http://www.monitoringthefuture.org/

University National Annual Opioid misuse rates. Contains specific No costs stated. Geographic identifiers None identified. See No individual-level

of questions for OxyContin and Vicodin are not public access recent figures provided linkages identified. Has

Michigan here. been linked with other

(private) state-level information

National Ambulatory Medical Care Survey (NAMCS): Information about the provision and use of ambulatory medical care services based on a sample of visits to non–federally

employed office-based physicians primarily engaged in direct patient care and, starting in 2006, a separate sample of visits to community health centers. Estimates generally

only representative at national or Census region levels (depends on year). https://www.cdc.gov/nchs/ahcd/index.htm

CDC National Annual Utilization of physician, hospital outpatient, No costs stated. Some restricted items Online query system No individual-level

(federal) and emergency department services; the

conditions most often treated; and the

diagnostic and therapeutic services

rendered, including medications prescribed

can only be accessed through National

Center for Health Statistics (NCHS) RDC

available here; other

research tools here

linkages identified

75

Table A.1. National Survey Data

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

National Epidemiologic Survey on Alcohol and Related Conditions (NESARC): This is an epidemiological survey conducted to provide information on mental health, substance,

and psychiatric disorders. Wave I and Wave II were a longitudinal panel. Wave III is a new sample. Sampling is designed to be nationally representative.

https://www.niaaa.nih.gov/research/nesarc-iii

National

Institute

on Alcohol

Abuse and

Alcoholis

(NIAAA)

(federal)

National 2001–

2002,

2004–

2005,

2012–

2013

Nonmedical prescription opioid use and

opioid disorder; mood and anxiety

disorders; other substance use, alcohol

disorder, and other drug use disorder

No costs stated. Only provided to

investigators who agree in advance to

adhere to established policies for

distribution

None identified Access to geocode

identifiers may permit

linkage at state level

National Hospital Ambulatory Medical Care Survey (NHAMCS): Information on the utilization and provision of ambulatory care services in hospital emergency and outpatient

departments and ambulatory surgery locations based on a national sample of visits to the emergency departments, outpatient departments, and ambulatory surgery locations

of noninstitutional general and short-stay hospitals. Estimates only representative at national or Census region. https://www.cdc.gov/nchs/ahcd/index.htm

CDC

(federal)

National Annual Receipt of opioid prescription; visits

specifically for chronic pain condition;

utilization and provision of ambulatory care

services in hospital emergency department,

outpatient departments, and ambulatory

surgery locations

No costs stated. Some restricted items

can only be accessed through NCHS RDC

Online query system

available here; other

research tools here

No individual-level

linkages identified

National Hospital Care Survey (NHCS): NHCS integrates inpatient data formerly collected by the National Hospital Discharge Survey, emergency department and outpatient

department data collected by NHAMCS, and substance-involved visit data previously collected by the Drug-Abuse Warning Network (DAWN). The integration of these three

surveys allows examination of care provided across treatment settings. https://www.cdc.gov/nchs/nhcs/about_nhcs.htm

CDC

(federal)

National,

participating

hospitals

Annual Emergency department

visits for substance abuse and/or resulting

from substance misuse or abuse, adverse

reactions to medications taken as prescribed

or directed, accidental ingestion of drugs,

and drug-related suicide attempts

No costs stated, but fees may apply for

use of the RDC. Access to the data is

allowed through a proposal submission

process and is accessed through the

NCHS RDCs.

Analytics for some

components of the

NHCS available through

ICPSR

Can link with the NDI,

MedPAR, and Medicaid

Statistical Information

System data sets

National Health Interview Survey (NHIS): Data on a broad range of health topics (medical conditions, health insurance, doctor’s office visits, physical activity, and other health

behaviors) are collected through personal household interviews. https://www.cdc.gov/nchs/nhis/about_nhis.htm

CDC

(federal)

National Annual Medical conditions, health insurance,

doctor’s office visits, physical activity, and

other health behaviors

No costs stated. Some variables are

considered restricted data (including

some linkages and geocoded variables)

Online analysis provided

through ICPSR with

account

Can be linked to MEPS,

NDI, Medi-care

enrollment and claims

data, and Social Security

benefit history data

76

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

National HIV Behavioral Surveillance System (NHBSS): NHBSS collects data relating to behavioral risk factors for human immunodeficiency virus (HIV) (e.g. sexual behaviors,

drug use), HIV testing behaviors, the receipt of prevention services, and use of prevention strategies (e.g. condoms, PrEP). In addition to these interview data, all NHBSS

participants are offered an HIV test. Findings from NHBSS are published in annual reports and other scientific publications.

https://www.cdc.gov/hiv/statistics/systems/nhbs/index.html

CDC National (22 Annual; HIV behavioral risk factors (e.g., sexual No costs stated. Not publicly available; None identified None identified

(federal) city “subject

areas”)

populatio

n cycle

rotation

behaviors, drug use), HIV testing behaviors,

injection drug use, receipt of prevention

services, use of prevention strategies (e.g.

condoms, PrEP)

as a component of HIV/acquired

immunodeficiency syndrome (AIDS)

surveillance, NHBSS data are protected

by the Assurance of Confidentiality

(Section 308[d] of the Public Health

Service Act, 42 U.S.C. 242 m[d]), which

prohibits the disclosure of any

information that could be used to

directly or indirectly identify individuals.

National Survey on Drug Use and Health (NSDUH): Self-reported information on drug use and abuse or dependence, mental health, and substance use disorder treatment

among respondents ages 12 and older. Results available at the national level and for some metropolitan statistical areas and sub-state areas. Designed to be representative at

the national and state levels. https://nsduhweb.rti.org/respweb/homepage.cfm

Substance National Annual Lifetime nonmedical opioid, heroin use; No costs stated. Geographic identifiers Online analysis provided Merged at the state

Abuse and first-time nonmedical opioid use, heroin are restricted access. Restricted access through ICPSR with level with other data

Mental initiates; past-year, past-month heroin, data elements must be applied for and account sets

Health nonmedical opioid, and opioid use by approved, with access to data provided

Administra therapeutic drug class; treatment for opioid through the Substance Abuse and

tion use disorder; self-reported unmet treatment Mental Health Data Archive (SAMHDA)

(SAMHSA) need data portal.

(federal)

National Survey of Substance Abuse Treatment Services Data (N-SSATS): N-SSATS is an annual survey of participating substance use treatment facilities to collect information

on location, characteristics, services offered, and utilization. Information from N-SSATS is used to compile and update the National Directory of Drug and Alcohol Abuse

Treatment Programs and the online Substance Abuse Treatment Facility Locator. https://www.dasis.samhsa.gov/dasis2/nssats.htm

SAMHSA National Annual Type of care provided, including detox and No costs stated. Publicly available State profiles accessible Merged at the county or

(federal) opioid treatment programs, substance

abuse problem treated, types of services

offered, facility funding and capacity

here. state level with other

data sets

Treatment Episodes Dataset (TEDS): Admissions to publicly funded treatment programs and opioid substitution programs by primary, secondary, and tertiary drug, route of

administration, and demographics. Data are available at the national and state levels. Data are submitted from state and local treatment agencies.

SAMHSA

(federal)

National Annual Admissions to treatment facilities (by type;

source of referral) for opioid analgesics or

heroin

No costs stated. Publicly available Data are available online

here.

Linked at the state level

with other information

77

https://wwwdasis.samhsa.gov/webt/information.htm

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Medicaid State Drug Utilization Data: Drug utilization for sStates are available for covered outpatient drugs paid for by sState Medicaid agencies since the start of the

Medicaid Drug Rebate Program. States are required to report numbers of prescriptions for Medicaid-covered outpatient drugs as well as Medicaid expenditures on the drugs

through Medicaid fee-for-serviceFFS and managed care. https://www.medicaid.gov/medicaid/prescription-drugs/state-drug-utilization-data/index.html

Centers for National Quarterly Drug name, National Drug Code (can Medicaid open data publicly available View data by state Linked at the state level

Medicare and identify specific opioid analgesics), here. Because of privacy restrictions, all online or access CMS with other data sets

Medicaid number of prescriptions, and dollars direct identifiers are removed in the drug spending

Services (CMS) reimbursed public data and aggregate data fewer dashboard here.

(federal) than 11 counts are suppressed.

Medicare Data Files: The Master Beneficiary Summary File includes several segments including enrollment information, chronic conditions data (e.g., mental health, substance

use conditions), service utilization, Medicare payment amounts, and place of residence at the zip-code level. Other notable databases include the Medicare Carrier File (final

action fee-for-service claims submitted on a CMS-1500 claim form); Medicare Outpatient Standard Analytic File (claims and treatment codes); Medicare Provider Analysis and

Review files (hospital inpatient services), and Part D Prescription Drug Event data (contains prescription drug cost and payment data). https://www.resdac.org/cms

-

data/search?f%5B0%5D=im_field_data_file_category%3A46

Research National Annual or Notably, Medicare enrollment, mental May include costs. Varying privacy levels See statistics, trends, Linked at the state,

Data semiannual health and substance use conditions, service for CMS files; requires data use and reports here. county, or zip-code level

Assistance utilization and Medicare payment amounts, agreements to various data sets.

Center death information (only through 2008), and Linked at the person

(ResDAC), prescription drug information level with other

CMS Medicare files; Veterans

(federal) Health Administration

(VHA) data; or Medicaid

claims for Medicare-

Medicaid enrollees

Medicare Part D Prescription Drug Event Data: Every time a beneficiary fills a prescription under Medicare Part D, a prescription drug plan sponsor must submit a summary

record called the prescription drug event (PDE) data to CMS. The PDE record contains prescription drug cost and payment data that enables CMS to make payments to plans

and otherwise administer the Part D benefit. https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovGenIn/PartDData.html

CMS (federal) National Annual Prescription drug costs, payment data,

identifiers, coverage information, and

prescription information

Includes request fee; must be

requested; certain data elements may

be encrypted and/or unavailable

depending on the particular requestor

entity and the demonstrated need for

an element

Medicare Part D Opioid

Mapping Tool

Linked at the state,

county, or zip-code level

to various data sets.

Linked at the person-

level with other

Medicare files; VHA

data; or Medicaid claims

for Medicare-Medicaid

enrollees

78

Table A.2. Claims and Electronic Health Records Secondary Data Sources

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Veterans’ Health Administration (VHA) Corporate Data Warehouse (CDW): The CDW and four Regional Data Warehouses (RDW 1–4) were built by the U.S. Department of

Veterans Affairs (VA) Office of Information and Technology to provide a high-performance business intelligence infrastructure through standardization, consolidation, and

streamlining of clinical data systems. https://www.hsrd.research.va.gov/for_researchers/vinci/cdw.cfm

VA (federal) National Near real-

time

Patient-level data on prescriptions and

health care utilization

This data set is not for public access or

use. Research requests must go through

the Data Access Request Tracker

application. With approval, data access

to CDW can be obtained from CDW

through approved SQL tables delivered

to a research project or accessed

through SAS Proc SQL.

Maps of VA opioid

prescribing data here.

Links data across

multiple VHA data

source system

VHA National Patient Care Database: The National Patient Care Database (NPCD), which is housed at the Austin Information Technology Center, is part of the National

Medical Information Systems (NMIS). The NPCD collects integrated patient care data from all Veterans Health Information Systems and Technology Architecture (VistA)

information technology systems. https://www.data.va.gov/dataset/national-patient-care-database-npcd

VA (federal) National Updated

daily

Clinical data resulting from ambulatory

care patient encounters; primary care

patient to provider assignments and

provider utilization data

This data set is not for public access or

use. Research requests must go through

the Data Access Request Tracker

application. With approval, data access

to CDW can be obtained from CDW

through approved SQL tables delivered

to research project or accessed through

SAS Proc SQL.

Maps of VA opioid-

prescribing data here.

Can be linked with other

VHA patient-level data

systems and across

years to generate

episodes of care for

individuals; can be

linked with mortality

data by Social Security

Number

Clinical Data Base/Resource Manager of Vizient Inc.: The Vizient Clinical Data Base and Resource Manager™ (formerly University Helathsystem Consortium) is an

administrative, clinical, and financial database providing clinical, discharge, procedure, and outcome data for hospital encounters from a consortium of hospitals and academic

health centers.

Vizient Inc.

(private)

National

(across

network)

Not stated Patient outcome data including

mortality, length of stay, complication

rates, and readmission rates—can

categorize by opioid use (does not

appear to collect dose information)

Costs not stated; may need to be a

Vizient member to access data

None identified May be linked at the zip-

code tabulation area or

more-aggregate level;

supports linkage with

American Hospital

Association survey data

EHRs from Group Health Cooperative (GHC, now Kaiser Permanente): Information from EHRs of patients in the GHC network. https://www.ghc.org/

Kaiser (private) Washington

state

Near–real

time

Prescribing of opioids, past prescribing,

and reason for admission (opioids

related)

Costs not stated; unclear if data are

available for online analysis

None identified

79

https://www.vizientinc.com/Our-solutions/Clinical Solutions/Clinical-Data Base

Has been linked at the

individual level to

mortality data and

traffic accident data

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Stanford Translational Research Integrated Database: The Stanford Translational Research Integrated Database has three integrated components: a clinical data warehouse,

based on the HL7 Reference Information Model, containing clinical information on over 1.3 million pediatric and adult patients since 1995; an application development

framework for building research data management applications on the data platform; and a biospecimen data management system. http://med.stanford.edu/researchit.html

Replaced by the STAnford medicine Research data Repository in 2017: http://med.stanford.edu/researchit/infrastructure/clinical-data-warehouse/starr-faq.html

Stanford

University

(private)

Stanford

University

Medical

Center

Real time Prescribing information; has been used

to identify patient research cohorts by

condition

Costs not stated. Identified clinical data

in the CDW is only released to

Institutional Review Board (IRB)–

approved research studies that have

received the appropriate IRB approval.

De-identified data are made available

for Stanford research projects that

qualify as a nonhuman subject research

study. Answers to data-

access questions

are available here.

Online access to the

cohort tool only through

connection to Stanford

network or virtual

private network

The Stanford

Translational Research

Integrated Database

exists in part as a tool

for data linkages,

although no linkages

specific to opioids

identified.

HealthCore Integrated Research Database: Integrated database of commercially insured population. Contains medical and pharmacy administrative claims data plus health

plan eligibility information on enrollees in large commercial insurance plans (Blue Cross/Blue Shield) across 14 states. https://www.healthcore.com/database/

HealthCore

(private)

National

(subset of

states)

Not stated Insurance holder demographics, claims

data relevant for opioid use, including

emergency department visits and

adverse drug events; prescription

information

Costs not stated. Data primarily

available only through consultants.

HealthCore does not sell data to third

parties for their independent use or

otherwise.

None identified Linked with hospital,

local, and federal data

MarketScan commercial claims database: The MarketScan commercial claims and encounters database consists of employer- and health plan–sourced data containing

medical and drug data for several million individuals annually. Health care for these individuals is provided under a variety of fee-for-service, fully capitated, and partially

capitated health plans, including preferred and exclusive provider organizations (PPOs and EPOs), point-of-service plans, indemnity plans, health maintenance organizations

(HMOs), and consumer-directed health plans. Medical claims are linked to outpatient prescription drug claims and person-level enrollment information.

https://truvenhealth.com/your-healthcare-focus/analytic-research/marketscan-research-databases

Truven Health

Analytics

(private)

National Quarterly Prescribing trends, rates of opioid

prescribing

Costs vary. Customized data sets and

licensing agreements available.

Accessing the data requires data

management software. DataProbe® and

MarketScan Online Tools (e.g., Sample

Select, Sample Select Prevalence,

Inpatient View, Outpatient View,

Disease Profiler, Treatment Pathways)

can facilitate access.

Online access available,

but must be purchased

Can be linked with other

MarketScan databases

80

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

MarketScan Multi-State Medicaid Database: The MarketScan Medicaid database contains standardized, fully integrated, enrollee-level de-identified claims across inpatient,

outpatient, and prescription drug services for both fee-for-services and capitation plans. Data on eligibility (by month) and service and provider type are also included. In

addition to standard demographic variables such as age and gender, the database includes variables of particular importance for investigating Medicaid populations, such as

aid category (blind/disabled, Medicare eligible) and race. Data are collected from employers, health plans, or state Medicaid agencies. https://truvenhealth.com/your

-

healthcare-focus/analytic-research/marketscan-research-databases

Truven Health

Analytics

(private)

Multistate

(12 states in

2010)

Semiannua

lly

Pharmaceutical claims for filled

prescriptions, outpatient service claims

records, inpatient admissions records

Costs vary. Customized data sets and

licensing agreements available.

Accessing the data requires data

management software. DataProbe® and

MarketScan Online Tools (e.g., Sample

Select, Sample Select Prevalence,

Inpatient View, Outpatient View,

Disease Profiler, Treatment Pathways)

can facilitate access.

Online access available,

but must be purchased

Can be linked with other

MarketScan databases

Optum database: Large database of eligibility-controlled claims information (commercial and Medicare members of affiliated plans, and commercial members of Optum

Employer customers’ and Optum Payer customers’ health plans). Comprises complete inpatient, outpatient, and pharmacy claims.

Optum

(private)

National Not stated Opioid episode duration and dosage;

opioid overdose; enrollment, utilization,

all available clinical data in EMR/EHR

Costs and access restrictions not stated. None identified State (and possibly

county) identifiers

support linkage at

aggregate level

Symphony Health Solutions’ Integrated Dataverse: Comprehensive source providing insight to all the factors that drive pharmaceutical brand success—medical, hospital and

prescription claims, and point-of-sale prescription data, nonretail invoice data, and demographic data; designed more for market research than policy research. It contains

pharmacy retail transactions from more than 80 percent of pharmacies nationwide, including high-volume national chain pharmacies, resulting in information on

approximately 90 percent of prescriptions filled at retail pharmacies in the United States. Missing pharmacies are generally independent or part of small chains. Symphony

obtains pharmacy data directly from prescription drug claim processors and payers, using the same data that get verified against standard reporting information to the U.S.

government. https://symphonyhealth.com/product/idv/

Symphony

Health (private)

National Not stated Medical, hospital and prescription claims

related to opioid prescribing and/or

overdose, point-of-sale prescription

data, nonretail invoice data, and

demographic data

Costs vary by request None identified Can be merged with

other state- or county-

level information

IQVIA (formerly IMS) National Disease and Therapeutic Index (NDTI): The NDTI is a monthly audit of office-based physicians that provides information about patterns and

treatment of disease in the continental United States. For each patient seen during a consecutive two-day period each calendar quarter, participating physicians complete an

encounter form that includes information about diagnoses and drug therapies. Each record of a drug therapy within the NDTI is linked to a specific six-digit taxonomic code

capturing diagnostic information similar to the ICD-9. http://www.imshealth.com/en and https://www.iqvia.com

IQVIA (private) National Monthly;

quarterly

analysis

suggested

Diagnosis codes; underlying and

concomitant conditions; prescription

information; drug appearance or drug

use; patient and physician characteristics

Costs vary depending on request Available (with

payment) via the

customer portal

Can be merged with

other state- or county-

level information

81

https://www.optum.com/solutions/data

analytics/data/real-world-data-analytics-a-cpl/claims-data.html

-

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

IQVIA (formerly IMS) National Prescription Audit: Measures retail dispensing of prescriptions to consumers via formal prescriptions. http://www.imshealth.com/en and

https://www.iqvia.com

IQVIA (private) National Monthly Prescriptions (by National Drug Code),

channel (i.e., where prescription filled),

prescriber specialty

Costs vary depending on request.

Geographic identifiers not available

below three zip-code levels

Available (with

payment) via the

customer portal. The

CDC has online graphs

of aggregate data by

state and county here

Can be merged with

other state- or county-

level information

IQVIA (formerly IMS) National Sales Perspectives: Measures sales volume of dollars and units of pharmaceutical products purchased by retail and nonretail providers. Data

collected from a large sample of manufacturers, wholesalers, outlets, and projected to national estimates. http://www.imshealth.com/en/ and https://www.iqvia.com/

IQVIA (private) National

(projected)

Monthly Prescription sales volume (by product

type), number of units sold

Costs vary depending on request. Flat

files can be delivered through secure

File Transfer Protocol platform

Available (with

payment) via the

customer portal

Projected data intended

for national analyses;

however, state or

county linkages may be

possible

IQVIA (formerly IMS) PayerTrak: PayerTrak is a web-based approach to trends in prescription drug utilization by payer. PayerTrak provides access to payer prescription volume

in all markets and all payers within the retail channel. With the PayerTrak tool, subscribers can quickly assess market share and copay for desired prescription products or

prescription markets in an easy-to-use tool. Data are projected to national estimates. http://www.imshealth.com/en/ and https://www.iqvia.com/

IQVIA (private) National

(projected)

Monthly Total prescriptions (by product), pay

type, state, copay

Costs vary depending on request Available (with

payment) via the

customer portal

Projected data intended

for national analyses;

however, state or

county linkages may be

possible

Massachusetts Medicaid Claims and Enrollment Data (MassHealth): Massachusetts state insurance data on claims. MassHealth claims and encounter data provided a

comprehensive history of health care utilization and expenditures, as well as associated diagnoses, in both general medical and behavioral health services sector across a

broad range of health care settings. http://www.mass.gov/eohhs/provider/insurance/masshealth/claims/claims-data/

Massachusetts

Health and

Human

Services (state)

Single state Annual

(may be

possible at

other

levels)

Treatment for addictions, diagnosis of

opioid dependence, expenditures on

treatment, mortality (in the eligibility

file)

Not stated None identified Has been merged with

other state data sets at

the individual level

Massachusetts All-Payer Claims Database (MA APCD): The MA APCD is the most comprehensive source of health claims data from public and private payers in Massachusetts.

With information on the vast majority of Massachusetts residents, the MA APCD promotes transparency and affords a deep understanding of the Massachusetts health care

system. http://www.chiamass.gov/ma-apcd/

Massachusetts

Center for

Health

Information

and Analysis

(state)

Single state Annual

(may be

possible at

other

levels)

Health and pharmacy insurance claims

related to opioids or other prescription

medication, infant diagnosis codes for

neonatal abstinence syndrome,

demographics

Fees may apply. Data must be

requested and approved. See links to

"Steps to Request the Data" for

government and non-government

entities: http://www.chiamass.gov/ma

-

apcd/

None identified Forms the spine of the

Chapter 55 data set,

linked to mortality,

prescription drug

monitoring program

(PDMP), criminal justice,

treatment, and other

data sets

82

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Vermont Health Care Uniform Reporting and Evaluation System (VHCURES): Vermont’s APCD, a comprehensive, longitudinal, multipayer data set that regularly collects

medical and pharmacy claims data and eligibility data from both private and public payers. http://gmcboard.vermont.gov/hit/vhcures

Vermont Green

Mountain Care

Board (state)

Single state Annual

(may be

possible at

other

levels)

Medical expenditures, costs of treatment

for opioid use disorders

Costs apply. Through data use

agreements, de-identified VHCURES

data is being utilized by state agencies,

state contractors, and academic

researchers to support analysis of

health care access, spending, utilization,

and quality.

None identified None identified

83

Agency Coverage Timing Measures Costs and restrictions Available analytics Linking capability

CDC WONDER Multiple Cause of Death Data: The Multiple Cause of Death data available on CDC WONDER provide county-, state-, and national-level mortality and

population data. Data are based on death certificates for U.S. residents. Each death certificate contains a single underlying cause of death, up to 20 additional multiple

causes, and demographic data. https://wonder.cdc.gov/mcd.html

CDC National Released Number of deaths, crude death rates, age- No costs and publicly available. Online data portal here Merged with other

(federal) annually

(but can

obtain

monthly

aggregate)

adjusted death rates (can be analyzed by

drug and alcohol related causes of death,

injury intent and injury mechanism

categories)

Subnational data representing zero to

nine deaths are suppressed

state- or county-level

information

National Death Index (NDI): The NCHS established the NDI as a resource to aid epidemiologists and other health and medical investigators with their mortality-ascertainment

activities. https://www.cdc.gov/nchs/ndi/index.htm

CDC National Annual Study participant death, dates of death, and Fee per study subject with fee schedule None identified Can be linked at the

(federal) the corresponding death certificate numbers.

NDI Plus provides cause of death

here. NDI service is available to

investigators solely for statistical

purposes in medical and health

research. The service is not accessible

to organizations or the general public

for legal, administrative, or genealogy

purposes.

individual level to the

NHIS; National Health

and Nutrition

Examination Survey;

longitudinal study of

aging; and VA health

care data; has been

linked with a variety of

state-specific health

data sets

National Vital Statistics System (NVSS) Multiple-Cause-of-Death files: Mortality data from NVSS are a fundamental source of demographic, geographic, and cause-of-death

information. Comparable for small geographic areas and available for a long time period in the United States. The data are used to present the characteristics of those dying

in the United States to determine life expectancy and to compare mortality trends. https://www.cdc.gov/nchs/nvss/mortality_methods.htm and

http://www.nber.org/data/vital-statistics-mortality-data-multiple-cause-of-death.html

CDC

(federal)

National Annual Mortality with information on drugs involved

in death

No costs. Microdata files must be

requested and approved before being

provided on CD or DVD.

NVSS is the underlying

data for CDC WONDER

Merged at the county-

level with other data

sets

Fatal Accident Reporting System (FARS): Data derived from a census of fatal traffic crashes within the 50 states, the District of Columbia, and Puerto Rico primarily from the

police accident report in those states, but also from death certificates, state coroners and medical examiners, state driver and vehicle registration records, and emergency

medical services records. https://www.nhtsa.gov/research-data/fatality-analysis-reporting-system-fars

National National Annual 143 different coded data elements (as of No costs and publicly available See online query system Geocode identifiers

Highway 2013) that characterize the fatal crash, here. support linkage at the

Traffic including toxicology reports city, county, and state

Safety levels.

Administ

ration

(federal)

84

Table A.3. Mortality Records

Agency Coverage Timing Measures Costs and restrictions Available analytics Linking capability

Examples of State Death Certificate Data Provided Below

Florida Department of Health mortality data: http://www.floridahealth.gov/statistics-and-data/

Death certificate data for the state of Florida containing information on cause of death

Data access: application form and information provided here

Prior studies using this data: Kennedy-Hendricks et al. (2016)

Past linkages: Merged at the state level with Florida PDMP information

North Carolina State Center for Health Statistics: http://www.schs.state.nc.us/aboutus.htm

Death certificate data for the state of North Carolina containing information on cause of death

Data access: Requests requiring extensive analysis or computer programming may be subject to a charge and completed as available staff time permits.

Prior studies using these data: Albert et al. (2011); Hirsch et al. (2014); Dasgupta et al. (2016); Kennedy-Hendricks et al. (2016)

Past linkages: Linked at the individual level, matching decedent names to controlled substance–prescription histories through PDMP data

North Carolina Office of the Chief Medical Examiner: http://www.ocme.dhhs.nc.gov/

Detailed data on all deaths in North Carolina caused by injury or violence, as well as natural deaths that are suspicious, unusual, or unattended by a medical professional;

contains postmortem serum toxicological analyses

Data access: Autopsy, investigation, and toxicology reports are also public records and once finalized, may be obtained from the Office of the Chief Medical Examiner. To

request any of these documents, use the Document Request web form.

Prior studies using these data: Albert et al. (2011); Hirsch et al. (2014); Dasgupta et al. (2016)

Past linkages: Linked at the individual level to state death certificate data and state PDMP data

Massachusetts Registry of Vital Records and Statistics: http://www.mass.gov/eohhs/gov/departments/dph/programs/admin/dmoa/vitals/

Vital records and deaths for Massachusetts

Data access: Information provided here

Prior studies using these data: Walley et al. (2013); The Commonwealth of Massachusetts, Executive Office of Health and Human Services, Department of Public Health

(2016, 2017)

Past linkages: Linked at the individual level to multiple other state databases under Chapter 55 (see The Commonwealth of Massachusetts, Executive Office of Health and

Human Services, Department of Public Health [2016, 2017])

Tennessee Department of Health, Division of Health Statistics, Death Statistical System: https://www.tn.gov/health/health-program-areas/statistics/health-data/

death-statisticshtml

Cause of death statistics for Tennessee (note: Tennessee Department of Health has many public health statistics publicly available)

Data access: Individual-level data not publicly available. Contact department for further information.

Prior studies using this data: Baumblatt et al. (2014)

Past linkages: Linked at the individual level to state PDMP data

85

Agency Coverage Timing Measures Costs and Restrictions Available Analytics Linking Capability

Automation of Reports and Consolidated Orders System (ARCOS): Measure of prescription drug supply based on mandatory reporting for Schedule I and II controlled

substances and selected Schedule III and IV substances from manufacture to sale. Data for each substance reported by quantity (e.g., mg, dosage unit) and three-digit zip

code. https://www.deadiversion.usdoj.gov/arcos/

Drug National Annual Amount of manufactured controlled Costs not stated. Available to all DEA Summary reports Merged with other data

Enforcem substance circulating through legal means, manufacturers and distributers; must publicly available sources at the county or

ent by compound procure data through Freedom of state level

Administr Information Act (FOIA) request; public

ation data are usually released only at the

(DEA) state level, but three-digit zip-level data

(federal) have been used under special agreement

Prescription Behavior Surveillance System: Epidemiological surveillance and evaluation tool based on de-identified longitudinal data from state PDMPs to measure trends in

controlled substance prescribing and dispensing and indicators of medical use and possible nonmedical prescription drug abuse and diversion.

http://www.pdmpassist.org/content/prescription-behavior-surveillance-system

TTAC @ 12 states Quarterly Forty-three prescription behavior Costs not stated. Data-sharing Online access for Compiles PDMP

Brandeis submitting; measures: overall usage within drug classes agreement specifies how Brandeis will authorized federal information across states

(federally more being and for selected individual drugs; daily manage, secure, and protect the PDMP researchers

funded) reviewed to

join

dosage; overlapping prescriptions within

each drug class or across classes;

questionable activity; payment source;

indicators of possible pill mills;

inappropriate prescribing measures; and

pharmacy-based measures of possible

inappropriate dispensing

data; data are maintained securely at

Brandeis, and access by Brandeis

research staff is limited in accordance

with the IRB-approved protocol.

Procedures are in process to provide

access by authorized federal researchers.

Examples of State Prescription Drug–Monitoring Program (PDMP) Data

Maine Prescription Monitoring Program: http://www.maine.gov/dhhs/samhs/osa/data/pmp/index.htm

Maine’s PDMP data, hosted by the Maine Substance Abuse and Mental Health Services

Data access: Agency has demonstrated willingness to provide data sets needed for research to address the problem of opioid misuse and abuse. De-identified data have been

made available to researchers.

Prior studies using this data: Piper et al. (2016), Kreiner et al. (2017)

Past linkages: Linked at the individual level to other prescriber information; merged at the county level with Maine Diversion Alert Program data

Maryland Prescription Drug Monitoring Program: https://bha.health.maryland.gov/pdmp/Pages/Home.aspx

Maryland’s PDMP data, hosted by the Maryland Department of Health and Mental Hygiene, Behavioral Health Administration

Data access: Individuals requesting data must complete training prior to submitting any data requests.

Prior studies using this data: Lin et al. (2016)

Past linkages: Linked at the prescriber level to a different survey on physician attitudes and use of PDMP

86

Table A.4. Prescription Monitoring Secondary Data Sources

Massachusetts Prescription Drug Monitoring Program: http://www.mass.gov/eohhs/gov/departments/dph/programs/hcq/drug-control/pmp/reports-and-data.html

Massachusetts’s PDMP data, hosted by the Massachusetts Department of Public Health

Data access: Data request form available here

Prior studies using this data: Katz et al. (2010), The Commonwealth of Massachusetts, Executive Office of Health and Human Services, Department of Public Health (2016,

2017)

Past linkages: Linked at the individual level to multiple other state databases under Chapter 55 (see The Commonwealth of Massachusetts, Executive Office of Health and

Human Services, Department of Public Health [2016, 2017]); also allows interstate data sharing

Tennessee Controlled Substances Monitoring Program/Database:

Tennessee’s PDMP data, hosted by the Tennessee Department of Health

Data access: The law allows a number of other state and federal officials to register with the database, including certain law enforcement officers, medical examiners, drug

court judges, and others.

Prior studies using this data: Baumblatt et al. (2014)

Past linkages: Linked at the individual level to state death certificate data

Ohio Automated Rx Reporting System: https://www.ohiopmp.gov/

Ohio’s PDMP data, hosted by the State of Ohio Board of Pharmacy

Data access: Not stated

Prior studies using this data: Baehren et al. (2010), Weiner et al. (2017)

Past linkages: Linked at the individual level with patient emergency department data

Kentucky All Schedule Prescription Electronic Reporting System:

Kentucky’s PDMP data, hosted by the Kentucky Cabinet for Health and Family Services

Data access: Not stated

Prior studies using this data: Blondell et al. (2004), Brady et al. (2014), Becker et al. (2017), Slavova et al. (2017)

Past linkages: Merged with zip-, county-, or state-level social and economic variables. The Kentucky Department of Public Health, Cabinet for Health and Family Services, has

established a multisource drug-overdose surveillance system, including the PDMP and various other state data sources (e.g., emergency department discharges, overdose

death and postmortem toxicology, and heroin/fentanyl submissions to Kentucky State Police crime labs).

Florida’s Prescription Drug Monitoring Program: http://www.floridahealth.gov/statistics-and-data/e-forcse/

Florida’s PDMP data, hosted by the Florida Department of Health

Data access: Not stated

Prior studies using this data: Delcher et al. (2015)

Past linkages: Merged with other state-level data sources (e.g., mortality)

87

https://www.tn.gov/health/health-program-areas/health-professional-boards/csmd-board.html

https://chfs.ky.gov/agencies/os/oig/dai/deppb/Pages/kasper.aspx

North Carolina Controlled Substances Reporting System: https://nccsrsph.hidinc.com/nclogappl/bdncpdmqlog/pmqhome

https://www.ncdhhs.gov/divisions/mhddsas/ncdcu/csrs

North Carolina’s PDMP data, hosted by the North Carolina Department of Health and Human Services, Division of Mental Health, Developmental Disabilities, and Substance

Abuse Services

Data access: Permission to query the system must be obtained from system administrators.

Prior studies using this data: Albert et al. (2011), Hirsch et al. (2014), Ringwalt et al. (2015), Dasgupta et al. (2016), Roberts et al. (2016)

Past linkages: Linked at the individual level with Medicaid claims data and mortality data

California’s Controlled Substance Utilization Review and Evaluation System: https://oag.ca.gov/cures

California’s PDMP data, hosted by the California Department of Justice

Data access: For access, researchers must obtain a background check from the California Department of Justice. An SQL server is used within the Department of Justice to de-

identify the database using a record-linkage methodology to permit identification of sequential prescriptions for each patient. Unique computer-generated identifiers are

devised for each provider and pharmacy to remove identifying information at the patient, provider, or pharmacy level.

Prior studies using this data: Wilsey et al. (2011), Gilson et al. (2012), Han et al. (2014)

Past linkages: No individual-level linkages identified.

Oregon’s Prescription Drug Monitoring Program: http://www.orpdmp.com/researchers.html

Oregon’s PDMP data, hosted by the Oregon Health Authority

Data access: The Oregon Health Authority may provide de-identified PDMP data for research purposes. The Oregon Health Authority is accepting research requests.

Prior studies using this data: Hartung et al. (2012); O'Kane et al. (2016); Deyo et al. (2017)

Past linkages: Linked at the patient level to state vital records, hospital discharge registry, and Medicaid administrative pharmacy claims

88

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

Area Health Resource Files (AHRF): The AHRF data include county, state, and national-level files in eight broad areas: health care professions, health facilities, population

characteristics, economics, health professions training, hospital utilization, hospital expenditures, and environment. The AHRF data are obtained from more than 50 sources.

https://www.hrsa.gov/about/contact/ehbhelp.aspx

Health Resources National Annual Information on health care cost and utilization, No costs and publicly See tools and Merged county-level

and Services (some demographics, health care facilities and services, available data portal here. contextual factors with

Administration measures are vital events, and other health information based other data on opioid

Data Warehouse available daily, on geographic region outcomes

(federal) monthly, and

quarterly)

Current Population Survey (CPS): Primary source of labor force statistics for the U.S. population. Supplemental questions are added to the basic CPS questions;

supplemental inquiries vary month to month and cover a wide variety of topics such as child support, volunteerism, health insurance coverage, and school enrollment.

Supplements are usually conducted annually or biannually. https://www.census.gov/programs-surveys/cps.html

U.S. Census National Monthly Information on educational status, health No costs and publicly See interactive Merged state- or county-

Bureau and the insurance, work and labor market outcomes, available. Not all counties are data tools here. level contextual factors

U.S. Bureau of income, disability, household characteristics (e.g., included, and data are not with other data on opioid

Labor Statistics household size), demographics (e.g., age, race, available for most sampled outcomes

(federal) gender), labor force participation, and poverty

rates

counties due to

confidentiality laws.

National Alliance Model for State Drug Laws (NAMSDL) policy data: Provides information on current state statutes and policies related to controlled substances and

prescription drugs. http://www.namsdl.org/index.cfm

NAMSDL National Updated Statutes related to naloxone access; pain No costs and publicly See maps of state Merged with state-level

(federally funded) semiannually management, pain clinics, and prescribing

practices; Good Samaritan Laws; PDMPs; doctor

shopping laws; prescription trafficking statutes;

regulation of internet pharmacies

available. Historical data are

not available or readily

downloadable for all policies.

policies here. data on opioid-related

outcomes

National Conference of State Legislatures (NCSL) policy data: NCSL maintains legislative tracking databases about public health issues such as criminal justice, education,

employment policy, immigrant policy, transportation, health care access, and public health. Users can search tracking databases for relevant legislation by year, topic, and

keyword. Users can download state legislation as a PDF file. http://www.ncsl.org/research/health/ncsl-prescription-drug-policy-resources-center.aspx

NCSL

(nongovernment

al organization)

National Annual State legislation related to Medicaid prescription

drug policies; PDMPs; prescribing guidelines;

naloxone; pain clinics

No costs and publicly

available. Historical data are

not available or readily

downloadable for all policies.

Online database

search here

.

Merged with state-level

data on opioid-related

outcomes

89

Table A.5. Contextual and Policy Data Sources

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

Prescription Drug Abuse Policy System (PDAPS): Tracks key state laws related to prescription drug abuse. PDAPS provides accurate, detailed information about important

policies designed to promote the safe use of controlled medicines and reduce overdoses. PDAPS users interact with and download legal data through the MonQcle software

platform. http://www.pdaps.org/

Legal Science, LLC

(federally funded)

National Updated

semiannually

Notably, state laws regarding: access to naloxone,

Good Samaritan 911 immunity, PDMPs

administration, and regulation and reporting

Data download is a paid

feature.

See MonQcle

data maps

example here.

Merged at the state-level

with opioid-related

outcomes

Kaiser Family Foundation (KFF) data: Polling data on a variety of public health issues and opinions. Also compiles information from other secondary sources (e.g., CPS) to

provide state-level data on health indicators. http://www.kff.org/

KFF (private) National Varies Public opinion on opioid use; polling data from

public and medical officials; health insurance

coverage

No costs stated. Publicly

available

Access state

profiles here.

Merged with state-level

data on opioid-related

outcomes

Policy Surveillance Program (PSP): Program aiming to increase the use of policy surveillance and legal mapping as tools for improving the nation's health. Data from legal

mapping to understand the laws on a given topic and how those laws differ over time and across jurisdictions. http://lawatlas.org/

Temple National Updated Opioid policies and regulations across states No costs stated. Publicly Maps are Merged at the state level

University semiannually available. Historical data are available online with information on

LawAtlas Project not available or readily here. opioid-related outcomes

(private) downloadable for all policies

PDMP Training and Technical Assistance Center (TTAC) at Brandeis: The PDMP Training and Technical Assistance Center (PDMP TTAC) at Brandeis University provides a

wide range of services and resources to PDMP agencies, researchers, and other stakeholders in an effort to advance the effectiveness of PDMPs to combat misuse and abuse

of prescription drugs. http://www.pdmpassist.org/

TTAC @ Brandeis National Updated fairly Information on timing of state PDMP laws and No costs stated. Publicly See maps and Merged at the state level

(federally funded) regularly PDMP law components available. Historical data are

not available or readily

downloadable for all policies

tables of PDMPs

available here.

with information on

opioid-related outcomes

CDC Public Health Law Program (PHLP): Laws summarizing legal strategies used by states to address the misuse, abuse, and health impacts of prescription drugs.

https://www.cdc.gov/phlp/index.html

CDC (federal) National Not stated Time and dosage limit laws; physical exam

requirements; doctor shopping laws; patient

identification laws; pain management clinic

regulations; Good Samaritan laws

No costs stated. Publicly

available. Historical data are

not available or readily

downloadable for all policies

See state laws on

prescription drug

misuse and abuse

here.

Merged at the state level

with information on

opioid-related outcomes

90

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

SAMSHA buprenorphine physician treatment locator: SAMHSA tracks the number of DATA-Certified Physicians waivered to prescribe buprenorphine in each state and

territory. https://www.samhsa.gov/medication-assisted-treatment/physician-program-data/treatment-physician-locator

SAMHSA National Daily Number and location of DATA-Certified physicians; waiver No costs. Public use files are See counts by Merged with other zip

(federal) limits not a complete census of

providers, but a complete

census is available as

restricted-use files

state here. code–, county- or state-

level information

DEA Active Controlled Substances Act Registrants Database (ACSA): Contains a full list of addresses for physicians with DATA waivers, as well as a full list of practitioners

registered to handle controlled substances. https://classic.ntis.gov/products/dea-csa/

DEA;

distributed

by the

National

Technical

Information

Services of

the U.S.

Department

of

Commerce

(federal)

National Daily Number and location of DATA-Certified physicians See fee schedule here. Online access

available with fee

Merged with other

county- or state-level

information

Drug Abuse Warning Network (DAWN): DAWN is a public health surveillance system that monitors drug-related emergency department visits in the United States and for

select metropolitan areas. DAWN relies on a nationally representative sample of general, non-federal hospitals operating 24-hour emergency departments, with

oversampling of hospitals in selected metropolitan areas. In each participating hospital, emergency department medical records are reviewed retrospectively to find the

emergency department visits that involved recent drug use.

SAMHSA National Annual Opioid misuse and abuse–related emergency department No stated costs. DAWN was Online analysis Compared with other

(federal) visits; mortality data (only for subset of states). All types of

drugs are included. Alcohol is considered an illicit drug

when consumed by patients aged 20 or younger. For

patients over 21 years old, alcohol is reported only when it

is used in conjunction with other drugs.

discontinued in 2011, but

SAMHSA is developing other

sources of data on drug-

related emergency visits

provided through

ICPSR with

account

surveillance data sources

U.S. Food and Drug Administration (FDA) Adverse Event Reporting System: The FDA Adverse Event Reporting System (FAERS) is a database that contains information on

adverse event and medication error reports submitted to FDA. The database is designed to support the FDA's post-marketing safety surveillance program for drug and

therapeutic biologic products. https://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Surveillance/AdverseDrugEffects/ucm070093.htm

FDA National Quarterly Reports of abuse-related adverse events. Contains detail on No stated costs. Public files See FAERS public None identified

(federal) product and substance with formulation- and composition-

specific differentiation

are available and individual

case safety reports can be

obtained by sending a FOIA

request to the FDA

dashboard here

.

91

Table A.6. Other National, State, and Local Secondary Data Sources

https://www.samhsa.gov/data/data-we-collect/dawn-drug-abuse-warning-network

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

National Emergency Medical Services Information System (NEMSIS): Provides data on EMS events for nearly all states. The consolidated data, while not a random sample

or census, is considered representative of national EMS activity. https://nemsis.org/

NHTSA

(Federal)

National

(49 states

as of 2016)

Annual Basic 911 call information about the scene of injury or

illness, medications administered (including naloxone), and

Emergency Medical Service (EMS) provider level, dispatch

call indicated overdose event, recorded overdose as injury

cause

No stated costs. Public-use

files must be requested;

certain variables are

restricted use and must go

through separate approval

process.

See data explorer

available here.

Has been merged with

mortality data based on

urbanicity

National Forensic Laboratory Information System (NFLIS): Drug cases investigated by the DEA. The data set provides information about chemistry of drugs seized by law

enforcement and analyzed by state, county, and volunteer forensic labs. Available for states, participating localities, and nationally.

https://www.deadiversion.usdoj.gov/nflis/

DEA

(federal)

National Monthly Drug identification results from drug cases submitted to

forensic laboratories

No stated costs. The private

site requires user accounts,

and security roles are

assigned to manage access to

its features, including the

Map Library, NFLIS Data Entry

Application, and Data Query

System. Only participating

laboratories and other DEA-

approved entities are granted

access to the Data Query

System

See information

about Data Query

System here.

None identified; linkages

at the state and

jurisdiction level likely

possible with access to

geocode identifiers

System to Retrieve Information from Drug Evidence (STRIDE): Data on drug exhibits remitted to DEA laboratories. The data set provides nationwide information on purity

and weight of each drug sample by month of seizure and total annual seizure weights by drug. Depending on the method of acquisition, information may be provided on

price of illicit drugs. https://www.dea.gov/resource-center/stride-data.shtml

DEA

(federal)

National Annual Street drug price by geographic area; street drug purity by

geographic area; volume of drug acquisitions (through

seizures, stings, purchases by undercover agents); product-

specific information

Some state-level annual

statistics available for

download online. More

detailed data can typically

only be obtained through a

FOIA request.

See state-level

annual statistics

here for heroin,

cocaine, and

methamphetamin

es.

Linkages at the state,

city, and metropolitan

statistical area level

possible with access to

geocode identifiers

92

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

National Poison Data System (NPDS): Data reported by the American Association of Poison Control Centers members. Provides information on poison call conditions across

the United States, including number of exposure calls by drug/substance at state and national levels. http://www.aapcc.org/data-system/uses-npds-data/

American National Monthly Poison control calls related to opioids or other drugs by Fees vary depending on NPDS offers a Can be merged with

Association “intentional exposures” (includes abuse, misuse, and request and requesting variety of other state or county

of Poison suspected suicidal) or “intentional abuse exposures.” organization. AAPCC NPDS analytical data level information

Control Contains detail on product type/composition Data Request Policy requires products,

Centers certain levels of internal although costs

(federal) approval prior to agreement

execution

apply

Nationwide Inpatient Sample (NIS) and State Inpatient Databases (SID) from HCUP: The NIS is the largest publicly available all-payer inpatient health care database in the

US, providing national estimates of hospital inpatient stays. Weighted, it estimates more than 35 million hospitalizations nationally. The NIS is sampled from the SID, which

are the state inpatient databases that contribute to HCUP (currently 48 states participate in the SID). https://www.hcup-us.ahrq.gov/nisoverview.jsp and

https://www.hcup-us.ahrq.gov/sidoverview.jsp

HCUP, National or Annual Opioid-related inpatient stays for specific diagnosis; patient See database catalog for Online query Previously linked at the

AHRQ state- demographic characteristics; expected payment source; costs. All users, including system through metropolitan statistical

(federal) specific total charges purchasers and collaborators,

must complete the online

training and must read/sign

the DUA for state databases

HCUPnet

Opioid-specific

analytics

level to other data sets.

Hospital identifier

unavailable for all states

beginning with 2012 NIS

Nationwide Emergency Department Sample (NEDS) and State Emergency Department Databases (SEDD) through HCUP: NEDS is the largest all-payer emergency

department database in the United States, providing national estimates of hospital-based emergency department visits. Weighted, it estimates roughly 143 million

emergency department visits. NEDS is sampled from the SID and SEDD—the SEDD capture emergency visits at hospital-affiliated emergency departments not resulting in

hospitalization (currently 36 states participate in the SEDD). https://www.hcup-us.ahrq.gov/nedsoverview.jsp and https://www.hcup

-

us.ahrq.gov/db/state/sedddbdocumentation.jsp

HCUP, National or Annual Opioid-related emergency department stays for specific See database catalog for Online query Linked at the state-level

AHRQ state- diagnosis; patient demographic characteristics; expected costs. All users, including system through with other data. Hospital

(federal) specific payment source; total charges purchasers and collaborators,

must complete the online

training and must read/sign

the DUA for state databases

HCUPnet

Opioid-specific

analytics

identifiers permit linkage

to hospital inpatient

databases

National Addictions Vigilance Intervention and Prevention Program (NAVIPPRO): NAVIPPRO is a comprehensive risk-management system for prescription opioids and

other Schedule II or III therapeutic agents. Continuous and ‘‘real-time’’ data streams are subjected to temporal and spatiotemporal signal detection strategies, followed up

with signal verification. NAVIPPRO monitors two proprietary data sources (ASI-MV Connect and web-onformed services survey on prescription misuse) and several publicly

available data sources (FDA-AERS, DAWN Live!, AAPCC New Core System database). http://www.inflexxion.com/asi-mv

Inflexxion Most of Near-real Lifetime nonmedical opioid, heroin use; first-time Costs and access restrictions None identified Geographically detailed

(private) the United

States

time nonmedical opioid use, heroin initiates; past-year and -

month heroin use; nonmedical opioid use by product; route

of administration; lifetime and past-year nonfatal opioid

overdose; source of opioids

apply. Costs vary by request.

Propriety data set

information may support

linkages at aggregate

level

93

Agency Coverage Timing Measures Costs and Restrictions

Available

Analytics Linking Capability

Researched Abuse, Diversion and Addiction-Related Surveillance System (RADARS): RADARS consists of several programs: drug diversion, poison center, opioid treatment,

impaired health care worker, Survey of Key Informants, college survey, StreetRx (streetrx.com for street drug price) programs. https://www.radars.org/radars-system-

programs.html

Rocky

Mountain

Poison and

Drug

Center,

Denver

Health and

Hospital

Authority

(private)

Most of

the United

States

Near-real

time

Nonmedical opioid, heroin use; first-time nonmedical

opioid use, heroin initiates; past-year and -month heroin

use, nonmedical opioid use by product; measures of

diversion; street price of opioid products

Costs vary by request. Each

program in RADARS is

approved by the institutional

review board of the principal

investigator's institution

None identified—

will provide

customized

reports for a fee

Can be linked at the zip

code level to other

information

Harm Reduction Coalition (HRC) data on organizations providing naloxone to laypersons: In October 2010 and July 2014, the Harm Reduction Coalition emailed a survey to

staff in a sample of U.S. organizations known to distribute naloxone to laypersons. Surveys asked about year of program implementation and total amount of naloxone kits

distribution and number of individuals receiving training, as well as reported number of overdose reversals because of naloxone administration by program participants.

http://harmreduction.org/ also Link to recent report using data

Harm National Less than When the organization began operating; numbers of sites Costs not stated. Data not None identified Merged with state-level

Reduction annually or local programs providing naloxone kits; number of available publicly rates of overdose

Coalition (2010 and persons trained in overdose prevention and provided mortality

(private) 2014) naloxone kits; and number of reports of overdose reversals

(administration of naloxone by a trained layperson in the

event of an overdose)

Overdose education and naloxone distribution (OEND) program data: OEND programs serve as a source for naloxone distribution as well as training and education for

overdose response in communities throughout the United States. Several studies have used data from state- or site-specific programs to study research questions related to

overdose and overdose reversing drugs.

Varies, but State or Varies Number of trainings, overdose rescue behaviors, naloxone Data generally not available Example of OEND Linked with state- or

generally site- administrations, naloxone kits distributed publicly, although some state information for community-level

state specific agencies provide aggregate Rhode Island information on overdoses

agency statistics. available here and/or hospital

utilization rates

94