You are on page 1of 26

HHS Public Access

Author manuscript
Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Author Manuscript

Published in final edited form as:


Drug Alcohol Depend. 2020 September 01; 214: 108137. doi:10.1016/j.drugalcdep.2020.108137.

The State of the Science in Opioid Policy Research


Megan S. Schuler1, Sara E. Heins2, Rosanna Smart3, Beth Ann Griffin4, David Powell2,
Elizabeth A. Stuart5, Bryce Pardo4, Sierra Smucker3, Stephen W. Patrick6, Rosalie Liccardo
Pacula7, Bradley D. Stein2,8
1.RAND Corporation, 20 Park Plaza #920, Boston MA USA 02216
2.RAND Corporation, 4570 Fifth Ave #600, Pittsburgh PA 15213
Author Manuscript

3.RAND Corporation, 1776 Main Street, Santa Monica CA 90401


4.RAND Corporation, 1200 S Hayes Street, Arlington VA 22202
5.Departmentof Mental Health, Johns Hopkins Bloomberg School of Public Health, 624 N.
Broadway, Baltimore MD 21205
6.Vanderbilt
Center for Child Health Policy, Vanderbilt University Medical Center, 2200 Children's
Way, 11111 Doctors' Office Tower, Nashville TN 37232
7.Schaeffer
Center for Health Policy and Economics, University of Southern California, 635
Downey Way, Verna and Peter Dauterive Hall, Los Angeles CA 90089
8.Department of Psychiatry, University of Pittsburgh School of Medicine, 3811 O'Hara Street,
Pittsburgh PA 15213
Author Manuscript

Abstract
Objective: Characterize the state of the science in opioid policy research based on a literature
review of opioid policy studies.

Methods: We conducted a scoping review of studies evaluating the impact of U.S. state-level and
federal-level policies on opioid-related outcomes published in 2005–2018. We characterized: 1)
state and federal policies evaluated, 2) opioid-related outcomes examined, and 3) study design and
analytic methods (summarized overall and by policy category).

Results: In total, 145 studies were reviewed (79% state-level policies, 21% federal-level policies)
and classified with respect to 8 distinct policy categories and 7 outcome categories. The majority
Author Manuscript

Corresponding Author: Megan Schuler, 20 Park Plaza #920, Boston MA 02216, mschuler@rand.org, (p) 617-338-2059 x8602.
Contributors: MSS led analyses and manuscript writing. SEH conducted analyses and significantly contributed to manuscript writing
and review. BDS and RLP conceptualized the literature review; MSS, SEH, RS, BAG, DP, BP, SS, and RLP conducted literature
review. All authors contributed substantively to the content of the manuscript, as well as contributing to the manuscript writing and
review. All authors have approved the final version of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our
customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of
the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered
which could affect the content, and all legal disclaimers that apply to the journal pertain.
Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi: …
Conflict of Interest: No conflict declared
Financial disclosure: No financial disclosures were reported by the authors of this paper.
Schuler et al. Page 2

of studies evaluated policies related to prescription opioids (prescription drug monitoring


Author Manuscript

programs (PDMPs), opioid prescribing policies, federal regulation of prescription opioids, pain
clinic laws) and considered policy impacts with respect to proximal outcomes (e.g., opioid
prescribing behaviors). In total, only 29 (20% of studies) met each of three key criteria for
rigorous design: analysis of longitudinal data with a comparison group design, adjustment for
difference between policy-enacting and comparison states, and adjustment for potentially
confounding co-occurring policies. These more rigorous studies were predominately published in
2017–2018 and primarily evaluated PDMPs, marijuana laws, treatment-related policies, and
overdose prevention policies.

Conclusions: Our results indicated that study design rigor varied notably across policy
categories, highlighting the need for broader adoption of rigorous methods in the opioid policy
field. More evaluation studies are needed regarding overdose prevention policies and policies
related to treatment access. Greater examination of distal outcomes and potential unintended
Author Manuscript

consequences are also warranted.

Keywords
opioid policy; state policy; federal policy; statistical methodology; review

1. INTRODUCTION
The country is in the midst of an opioid-related public health crisis, characterized by
increased opioid misuse and dependence and accompanying sequalae, including fatal opioid
overdose. National survey estimates indicate that in 2018 nearly 10 million people misused
prescription opioids, approximately 800,000 used heroin, and 2 million people had an opioid
use disorder (OUD) (SAMHSA, 2019), although household surveys likely significantly
Author Manuscript

undercount heroin use (Midgette et al., 2019). Mortality rates from opioid overdoses
increased by nearly 400% between 2000 and 2018 with important spatiotemporal
heterogeneity in opioids involved in fatal overdoses (Hedegaard et al., 2020). As highlighted
by Heins (2019), prescription opioids have persistently been a major contributor to overdose
deaths, while heroin overdose deaths began notably increasing in 2010 (Ciccarone, 2017).
More recently, the rise of synthetic opioids, including illicitly manufactured fentanyl, has
particularly affected Appalachia, New England, and the Midwest (Pardo et al., 2019).
Fundamentally, there is important heterogeneity regarding the nature and impact of the
opioid crisis, with respect to geographic region, rurality, race/ethnicity, gender, and age
(Marsh et al., 2018; SAMHSA, 2020; Schuler et al., 2020; Singh et al., 2019).

In response, states and the federal governments have enacted a broad array of policies,
Author Manuscript

producing a complex and dynamic policy landscape. Given the multi-faceted nature of the
opioid crisis, these policies target multiple stakeholders (including healthcare providers,
patients, substance use treatment providers, individuals using heroin or street-sourced
opioids, and communities affected by opioid misuse) and are intended to impact a wide
range of outcomes. Broadly, opioid-related laws and regulations have focused on deterring
high-risk opioid prescribing behaviors; improving access to treatment, including medications
for OUD treatment (e.g., buprenorphine, methadone); and expanding overdose prevention
efforts (e.g., increasing access to naloxone, an overdose reversal medication). Furthermore,

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 3

numerous evaluation studies have also examined impact of broader policies (e.g., Medicaid
Author Manuscript

expansion, state marijuana laws) on opioid-related outcomes. The growing opioid policy
literature can provide policymakers and other stakeholders with key insights regarding how
various policies may impact different aspects of the opioid crisis, yet published studies vary
regarding methodological rigor.

To characterize the state of the science in opioid policy research, we conducted a scoping
review of studies evaluating the impact of U.S. state-level and federal-level policies on
opioid-related outcomes published in 2005–2018. Our primary focus was to characterize: (1)
which state and federal policies have been evaluated, (2) which opioid-related outcomes
have been examined, and (3) study designs and analytic methods employed. We empirically
define categories of policies and outcomes examined across studies, highlighting those that
have received more extensive, as well as more limited, evaluation to date. For each policy
category, we summarize study design and analytic methods across studies. Overall, our
Author Manuscript

review characterizes the current opioid policy evidence base and highlights important
methodological challenges and analytic considerations facing the field.

2. METHODS
2.1 Data Sources and Search Strategy
We conducted a scoping literature review of U.S. state- and federal-level opioid policy
evaluation studies published between 2005 and 2018. The search included English language
articles within 13 medicine, social science, economic, and legal databases using a defined set
of search terms (full search strategy including list of databases and search terms detailed in
Appendix 11). Search criteria required studies to contain at least one search term from each
of three categories: (1) policy terms (e.g., prescription drug monitoring program, pain clinic,
Author Manuscript

Good Samaritan), (2) opioid terms (e.g., oxycodone heroin), and (3) opioid-related outcome
terms (e.g., opioid overdose, measures of opioid prescribing). The search term list was
compiled iteratively by study team members, with input from Advisory Board members of
the RAND-USC Schaeffer Opioid Policy Tools and Information Center (OPTIC).

2.2 Screening Strategy and Inclusion Criteria


Studies identified in the search were screened and selected for full review if they met both of
the following criteria: (1) quantitatively evaluated at least one policy (i.e., legislation or
regulation) implemented in the U.S. at the state or federal level, and (2) included as a
dependent variable at least one opioid-related outcome. Eligible studies included articles
published in peer-reviewed journal articles or Morbidity and Mortality Weekly Report as
well as economic working papers (e.g., NBER). We excluded purely descriptive studies,
Author Manuscript

qualitative studies, editorials, dissertations, and review articles. Prior to the start of the
screening process, the screening protocol was refined by research team members applying it
to a small random sample of studies; study team members discussed protocol inconsistencies
or ambiguity until consensus was reached and the screening protocol was refined
accordingly.

1Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi…

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 4

Studies flagged as eligible for full text review during screening were cross-checked by a
Author Manuscript

second reviewer; a small random sample of studies flagged as ineligible for full text review
were cross-checked by a second reviewer. Studies flagged as ineligible for inclusion during
full text review were cross-checked by a second reviewer. Study team members discussed
studies for which reviewers disagreed until consensus was reached.

2.3 Data Extraction and Synthesis


For studies eligible for full text review, research team members extracted details regarding
study design and analytic approach from each article using a standardized extraction form
and accompanying codebook. Key study features assessed included: (1) primary opioid-
related policies evaluated, (2) all opioid-related outcomes assessed, and (3) details on study
design and analytic methods. Policies were classified as primary if stated as the exposure of
interest in the study objectives. Policies that were controlled for in the main analyses or only
Author Manuscript

evaluated in sensitivity analyses or supplemental materials were not classified as primary


policies. Study characteristics of interest included: policy level (federal or state); study
design (longitudinal data with a comparison group; longitudinal data with no comparison
group; cross-sectional data with a comparison group; cross-sectional data without a
comparison group); length of study period; adjustment for covariate differences across
policy and comparison groups (if applicable); specification of primary policy (single binary
variable or more complex specification); and adjustment for co-occurring opioid-related
policies.

We identified categories for the opioid policies evaluated in the studies reviewed. A priori
policy categories enumerated by the study team were refined based on literature review
results; categories were iterated among the study team until consensus was reached.
Similarly, we identified broad categories of opioid-related outcomes examined, using
Author Manuscript

empirical results to refine a priori categories (see Appendix 22). We note that neither policy
nor outcome categories are intended to represent an exhaustive description of all opioid
policies and outcomes; instead they provide a classification schema for policies and
outcomes examined in the literature published to date. We present analytic results from our
extraction database including descriptive statistics regarding frequency of policies and
outcomes, as well as study design characteristics (overall and by policy category).

3. RESULTS
The database search yielded 5,113 studies for screening (Figure 1). Preliminary screening of
titles and abstracts eliminated 4,829 studies, most commonly because they did not conduct
an evaluation of a U.S. state- or federal-level policy. The remaining 284 studies underwent
Author Manuscript

full-text assessment; 139 did not meet inclusion criteria, whereas 145 met criteria and were
included in our review (see listing in Appendix 33). All 145 studies were observational (i.e.,
nonexperimental) in nature. The number of opioid evaluation studies published annually has
increased significantly in recent years (Figure 2), with approximately 60% of evaluated

2Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi…
3Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi…

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 5

studies published in 2017 and 2018. Overall, 21% of studies evaluated federal policies while
Author Manuscript

the remaining 79% evaluated state-level policies.

3.1 State and Federal Policies Evaluated


Based on the primary policies evaluated across studies, we defined 8 distinct policy
categories (Figure 3), described below and further detailed in Appendix 24. We report the
percentage of studies that evaluated policies in each category; the sum across categories
exceeds 100% as some studies evaluated more than one primary policy.

Prescription Drug Monitoring Programs—The most commonly evaluated category


was state prescription drug monitoring programs (PDMPs), electronic databases that track
prescribed and dispensed controlled substances, evaluated in 52 (35%) studies. While some
states have had a PDMP for decades, the opioid crisis has fueled the adoption of PDMPs
Author Manuscript

across the country – currently, the District of Columbia and all states with the exception of
Missouri operate a state-wide PDMP (PDAPS, 2019).

Opioid Prescribing Policies—The next most common category comprised policies


targeting opioid prescribing, evaluated in 31 (22%) studies. State-level policies included
Medicaid formulary restrictions (e.g., (Cochran et al., 2017; Keast et al., 2018)), Medicaid
controlled substance lock-in programs (Roberts et al., 2016; Skinner et al., 2016), and state-
level prescribing guidelines (e.g., (Johnson et al., 2011; Penm et al., 2018; Weiner et al.,
2017)). While most prescribing policies were enacted at the state level, federal-level
initiatives included Medicare formulary restrictions (Chen et al., 2017), the Veterans Health
Administration’s Opioid Safety Initiative (Lin et al., 2017), and the 2016 Centers for Disease
Control and Prevention’s (CDC) Guideline for Prescribing Opioids for Chronic Pain
(Bohnert et al., 2018).
Author Manuscript

Federal Regulation of Prescription Opioids—Federal regulations relating to the


approval, formulation, or scheduling of prescription opioids were evaluated in 20 (14%)
studies. Specific policies examined included: up-scheduling hydrocodone combination
products to Schedule II (e.g., (Kuo et al., 2018; Murimi et al., 2019; Raji et al., 2018),
introduction of the abuse-deterrent reformulation of OxyContin (e.g., (Alpert et al., 2018;
Cicero and Ellis, 2015; Larochelle et al., 2015), FDA approval of generic OxyContin (Bailey
et al., 2006), and FDA-requested withdrawal of propoxyphene from the U.S. market
(Delcher et al., 2017; Larochelle et al., 2015).

Treatment Access Policies—Policies relating to substance/OUD treatment access were


evaluated in 17 (12%) studies. State-level policies included state Medicaid expansion,
Author Manuscript

including through the Patient Protection and Affordable Care Act (e.g., (Meinhofer and
Witman, 2018; Sharp et al., 2018; Wen et al., 2017a)); changes to state Medicaid coverage
for methadone or buprenorphine (e.g., (Bachhuber et al., 2017; Saloner et al., 2016)); and
targeted funding for substance use treatment programs (Abraham et al., 2018). Federal-level
policies included the Patient Protection and Affordable Care Act and federal waiver

4Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi…

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 6

requirements for buprenorphine prescribing (e.g., (Dick et al., 2015; Feder et al., 2017; Stein
Author Manuscript

et al., 2015)).

State Marijuana Laws—State marijuana laws (i.e., laws that decriminalize possession of
marijuana, legalize medical marijuana, or legalize recreational marijuana) were evaluated in
14 (10%) studies (e.g., (Shi, 2017; Shi et al., 2019; Wen and Hockenberry, 2018; Wong and
Lin, 2019)). A hypothesized mechanism is that marijuana, particularly medical marijuana,
may have a substitution effect with opioid analgesics.

Pain Management Clinic Laws—Pain management clinic laws (also referred to as “pill
mill laws”) were evaluated in 13 (9%) studies. These laws impose regulations regarding the
licensing, personnel, operation and inspection of medical facilities that primarily diagnosis
and manage chronic pain. The majority of these studies focused exclusively on Florida’s
pain clinic laws (e.g., (Chang et al., 2016; Chang et al., 2018; Kennedy-Hendricks et al.,
Author Manuscript

2016; Rutkow et al., 2015)), with only a single study conducting a multi-state evaluation
(Dowell et al., 2016).

Overdose Prevention Policies—Policies aimed primarily at reducing opioid overdoses


(e.g., naloxone access laws, overdose education/naloxone distribution (OEND) initiatives,
and Good Samaritan laws) were evaluated in 13 (9%) studies. Naloxone, an opioid
antagonist that can be administered by a layperson via nasal spray or intramuscular injection
(using an automatic injection device), can rapidly reverse an opioid overdose (Robinson and
Wermeling, 2014; Sumner et al., 2016). Naloxone access laws evaluated included a broad
range of provisions including: third-party provisions that allow naloxone to be dispensed/
used by someone other than to whom it was prescribed; provisions that permit naloxone
possession without prescription; layperson dispensing provisions that allow naloxone
Author Manuscript

dispensing by individuals who are not otherwise authorized to dispense prescription


medication; immunity provisions that protect clinicians from liability associated with
naloxone administration; and standing order provisions that allow naloxone dispensing
without a prescription to any person in need (e.g., (Gertner et al., 2018; Lambdin et al.,
2018; Rees et al., 2017; Xu et al., 2018)). OEND programs provide overdose prevention
training and naloxone education and distribution (Bounthavong et al., 2017; Walley et al.,
2013). Good Samaritan laws are intended to increase bystander assistance to individuals
experiencing a drug overdose by providing some legal immunity regarding drug
possession/use by the bystander respondent’s and limiting liability in the case of an
unsuccessful intervention (Nguyen and Parker, 2018; Rees et al., 2017).

Punitive Drug Use Policies—Criminal justice laws targeting drug use, including
Author Manuscript

mandatory sentencing minimums for drug distribution/possession (e.g., (Davies, 2010)) and
expansion of child neglect/abuse laws to include use of opioids while pregnant (e.g.,
(Angelotta et al., 2016)) were examined in 6 (4%) studies.

3.2. Opioid-Related Outcomes Evaluated


Based on the opioid-related outcomes evaluated across studies, we defined 7 distinct
outcome categories. We report the percentage of studies that evaluated outcomes in each

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 7

category; the sum across categories exceeds 100% as some studies evaluated outcomes in
Author Manuscript

more than one category. Figure 4 shows the distribution of policy and outcome combinations
evaluated. As this figure highlights, the majority of studies examined proximal outcomes
(e.g., effect of PDMPs on opioid prescribing, effect of treatment access policies on treatment
utilization), yet some studies considered more distal outcomes (e.g., effect of marijuana laws
on opioid-related mortality) or unintended consequences (e.g., effect of PDMPs on heroin
use). We note that not all outcome categories would necessarily be expected to be impacted
by a given policy; thus, an absence of studies examining a specific policy-outcome pairing
does not necessarily imply an evidence “gap” in the literature.

Opioid Prescribing Behavior—Measures of opioid prescribing behavior were examined


in 81 (55%) studies. Specific measures included: type of opioids prescribed, days of opioids
prescribed, total number of opioid prescriptions, morphine milligram equivalents of
prescribed dose, high-dose prescribing, number of unique opioid patients per physician,
Author Manuscript

opioid prescription refill rate, and prescriber use of PDMPs. Measures of prescribing
behavior were generally obtained from state PDMP administrative records, the Drug
Enforcement Administration’s Automation of Reports and Consolidated Orders System
(ARCOS), IQVIA retail pharmacy data, or insurance claims data (e.g., Medicaid, Truven
Marketscan Research Database).

Opioid-Related Morbidity and Mortality—Opioid-related morbidity or mortality were


examined in 55 (38%) studies. Specific outcomes included: fatal opioid-related overdoses,
opioid-related hospitalizations and emergency room visits, opioid-related emergency
medical service responses, and opioid-related poison center calls. Data on fatal overdose
rates were obtained from the CDC’s National Vital Statistics System. Measures of healthcare
utilization for acute opioid-related outcomes were primarily derived from insurance claims
Author Manuscript

or administrative data.

Prescription Opioid Misuse and Opioid Use Disorder—OUD and prescription


opioid misuse (including nonmedical use of prescription opioids and potentially high-risk
prescription opioid behaviors) were examined in 32 (22%) studies. In studies using national
survey data (in particular, the National Survey on Drug Use and Health), prescription opioid
misuse measures included self-reported nonmedical use of prescription opioids, source of
prescription opioids used for nonmedical purposes, and motivation for nonmedical use.
Measures of high-risk prescription opioid use or drug-seeking behavior included: the
number of different pharmacies where an individual filled opioid prescriptions, number of
physicians from whom an individual received an opioid prescription, purchase of opioid
prescriptions with cash, early refills or overlapping prescriptions for opioids, and patient
Author Manuscript

disenrollment following formulary restrictions. These measures were generally


operationalized using self-reported survey data, PDMP data, or insurance claims data. OUD
was operationalized in national survey data as a self-reported diagnosis from a healthcare
provider or based on survey items corresponding to DSM criteria for abuse or dependence
(for either prescription opioids or heroin) and operationalized in administrative/claims data
based on International Classification of Diseases (ICD) codes for opioid abuse or
dependence.

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 8

Heroin and Other Illicit Opioid Use—Measures of heroin and other illicit opioid use
Author Manuscript

were examined in 25 (17%) studies. Note that this category of outcomes does not include
measures that encompass both prescription opioids as well as heroin/illicit opioids (e.g., any
opioid misuse, OUD). Specific outcomes included: heroin or illicit opioid use, heroin use
abuse/dependence, overdose deaths involving heroin or synthetic opioids, heroin-related
poison control calls, sales volume of heroin and illicit opioids through U.S. cryptomarkets,
and street prices of heroin/illicit opioids. These measures were operationalized in survey
data as self-reported heroin use or based on survey items corresponding to DSM criteria for
heroin abuse or dependence and operationalized in claims data based on ICD codes for
heroin abuse or dependence. Overdose data were obtained from the Centers for Disease
Control and Prevention’s National Vital Statistics System and street price data were obtained
from RADARS’ StreetRx Program.

Treatment Access and Utilization—Measures of treatment access and utilization were


Author Manuscript

examined in 25 (17%) studies. Measures included: any OUD treatment, any OUD
medication treatment, type of OUD medication treatment, treatment setting, treatment
referral source, treatment adherence, number of buprenorphine waivered physicians per
capita, and treatment admission rate per capita. Data sources included self-reported
utilization measures from national surveys, insurance claims and administrative data, as well
as data from RADARS’ Opioid Treatment Program and the Survey of Key Informants’
Patients Program.

Opioid-related Crime—Measures of opioid-related crime, primarily diversion of


prescription opioids, were examined in 10 (7%) studies. Specific measures included opioid-
related arrests, rate of prescription opioid diversion cases, listings for prescription opioids on
online cryptomarkets, and street prices of prescription opioids. Data sources include the
Author Manuscript

Federal Bureau of Investigation’s Uniform Crime Reports, the National Incident-Based


Reporting System, the Drug Enforcement Administration’s System to Retrieve Information
From Drug Evidence (STRIDE) data, and the RADARS System.

Naloxone Use—Measures of naloxone use were examined in 7 (5%) studies, primarily in


the context of evaluating naloxone access laws and OEND initiatives. Measures included the
number of naloxone prescriptions, volume of naloxone stocked at pharmacies, volume of
naloxone dispensed, and pharmacist intention to stock and dispense naloxone. Data sources
included administrative and claims data from commercial insurers, Medicaid, and the
Veterans Health Administration as well as survey data from pharmacists.

3.4 Study Design and Analytic Methods used in Evaluation Studies


Author Manuscript

Prescription Drug Monitoring Programs studies—Of the 52 PDMP studies, 31


(60%) analyzed longitudinal data with a comparison group design (i.e., the gold standard for
observational research), 16 (31%) analyzed longitudinal data with no comparison group, and
5 (9%) analyzed cross-sectional data with a comparison group design. Of the 36 studies with
a comparison group design, the majority accounted for differences between policy and
comparison states using regression adjustment (n=27) or propensity scores (n=2), yet 7
studies (19%) did not, which may result in biased effect estimates. Additionally, 32 studies

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 9

(62%) did not control for the potential impact of co-occurring policies, which may also bias
Author Manuscript

effect estimates. Comparing across policy categories, a strength of PDMP studies is their
relatively longer study period lengths: 18 (35%) had a study period of 8 years or longer.
Another strength of PDMP studies was their examination of policy heterogeneity, with 34
(65%) operationalizing the policy with more complexity than a binary variable in order to
assess distinct policy components. Specifically, multiple studies compared mandatory access
PDMP, non-mandatory access PDMP, and no PDMP (e.g., (Ayres and Jalal, 2018; Dhaval et
al., 2018)). Alternatively, several studies differentiated between PDMPs with mandatory
enrollment, with mandatory access, both mandatory access and enrollment, and neither
mandatory access nor enrollment (e.g., (Ali et al., 2017; Wen et al., 2017b)). Other studies
considered even more PDMP characteristics – Pauly et al. (2018) contrasted the
effectiveness the following 5 characteristics: (1) PDMP operational status, (2) controlled
substance schedules monitored by the PDMP, (3) frequency of data reporting to the PDMP
Author Manuscript

central server, (4) provisions for unsolicited reporting of patients’ controlled substance
prescription history, and (5) provisions regarding mandated prescriber access. Finally, some
studies evaluated differences among mandatory access laws – Buchmueller and Carey
(2018) defined three subtypes of these laws: (1) limited laws that apply only to certain drugs
or settings of care, (2) discretionary laws that mandate PDMP access upon provider
suspicion and (3) broad laws that apply to all drugs and settings and do not rely on provider
suspicion.

Opioid Prescribing Policies studies—Opioid prescribing policies have increasingly


been evaluated in recent years, with 25 (81%) of the 31 total studies published during 2016–
2018. Overall, 24 (77%) examined state-level policies and 7 (23%) evaluated federal-level
policies. A relative weakness of these studies is that 22 (71%) did not employ a comparison
group design (in part due to the prevalence of federal policy studies). Overall, only 8 (26%)
Author Manuscript

employed the strongest design (longitudinal data with a comparison group), whereas 21
analyzed longitudinal data with no comparison group, 1 analyzed cross-sectional data with a
comparison group, and 1 analyzed cross-sectional data without a comparison group.
However, all 9 studies with a comparison group design accounted for differences between
policy-enacting and comparison states, using either regression adjustment or propensity
scores. Additional weaknesses include that 27 (87%) did not adjust for co-occurring policies
and only 12 (39%) examined policy components via more complex specifications of the
policy variable. Additionally, these studies had shorter study periods on average: 23 (75%)
had study periods less than 6 years.

Federal Regulation of Prescription Opioids studies—Of the 20 studies, 12 (60%)


were published in 2016–2018 and 7 (35%) were published in 2011–2015. Identifying an
Author Manuscript

appropriate comparison group for federal policies is challenging – as such, 19 studies (95%)
did not use a comparison group design. Instead, many performed an interrupted time series
analysis, in which the pre-policy trend is used to extrapolate the counterfactual trend of how
the outcome would have evolved in the absence of policy implementation. While this is one
of the most rigorous quasi-experimental study designs, studies without a comparison group
are subject to potential bias arising from exogenous factors that also impact the outcome. As
such, it is particularly notable that 17 studies (85%) did not control for any co-occurring

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 10

policies. Additionally, these studies had shorter study periods: 17 (85%) had study periods
Author Manuscript

less than 6 years. Given that the federal policies evaluated did not differ across states, none
of these studies utilized more complex specifications of the policy variable to examine
policy heterogeneity.

Treatment Access Policies studies—Of the 17 studies, 15 (88%) examined state-level


policies and 2 (12%) evaluated federal-level policies. Overall, 9 (53%) analyzed longitudinal
data with a comparison group design, 4 (24%) used longitudinal data with no comparison
group, and 4 (24%) used cross-sectional data with a comparison group. Of the 13
comparison group design studies, 12 accounted for differences between policy and
comparison states using covariate regression. A relative strength of these studies is that the
majority (9; 53%) accounted for co-occurring policies, with 8 studies adjusting for 2 or more
co-occurring policies. Additionally, 9 studies (53%) used more complex specifications of the
policy variable. These studies also had relatively longer study periods -- 6 (36%) had a study
Author Manuscript

period of 8 years or longer.

State Marijuana Laws studies—Studies examining the impacts of state marijuana laws
on opioid-related outcomes were more likely to be published in recent years, with 12 (86%)
of the 14 studies published during 2016–2018. Compared to other policy categories, this
category had the highest percentage of studies employing a comparison group design with
longitudinal data (n=12; 86%), all of which accounted for differences between policy and
comparison states. Half of all studies (n=7) used more complex specifications of the policy
variable (e.g., to differentiate states that permitted marijuana dispensaries versus only home
cultivation). Another relative strength was that these studies were the most likely to account
for co-occurring policies (n=11; 79%), with 8 studies controlling for 2 or more co-occurring
policies. It is possible that addressing potential confounding of co-occurring policies (e.g.,
Author Manuscript

PDMPs) may be perceived as more salient for this policy category, given that marijuana laws
do not directly target opioid-related outcomes. A relative weakness of these studies is their
shorter study period length (9 studies (64%) were less than 4 years), due to the recency of
state marijuana laws.

Pain Management Clinic Laws studies—Of the 13 pain clinic studies, 8 (62%)
analyzed longitudinal data with a comparison group design, and the remaining 5 (38%) used
longitudinal data with no comparison group. Of the 8 studies with a comparison group
design, 3 (38%) did not account for differences between policy and comparison states,
which may bias policy effect estimates. Furthermore, 9 studies (69%) did not adjust for co-
occurring policies. The majority (n= 7; 54%) did not examine policy heterogeneity through
more complex specifications of the policy variable, largely due to a preponderance of studies
Author Manuscript

that only considered a single policy state (e.g., Florida, Texas). Notably, due to the near
simultaneous enactment of Florida’s PDMP and pain clinic laws, two studies strictly
estimated the joint effect of both policies (Chang et al., 2016; Chang et al., 2018). A relative
strength of pain clinic studies is their relatively longer study periods -- 39% (n=5) had a
study period of 8 years or longer.

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 11

Overdose Prevention Policies studies—Given the recency of policies targeting opioid


Author Manuscript

overdose prevention, 12 (92%) of the 13 studies were published in 2016–2018. Compared to


other policy categories, this category had the second highest percentage of studies
employing a comparison group design with longitudinal data (n=11; 85% of studies). Of the
11 studies with a comparison group, 2 (18%) did not account for differences between policy
and comparison states, which may bias policy effect estimates. Furthermore, 8 (62%) of the
13 studies did not adjust for the potential effect of co-occurring policies. Overall, 7 (54%)
considered policy heterogeneity through more complex specifications of the policy variable.
Naloxone law studies varied in their classification scheme, with studies considering four
categories (i.e., third party, standing order, prescriber immunity, and layperson dispensing)
(Gertner et al., 2018), five categories (i.e., third party, standing order, possession, prescriber
immunity, and dispenser immunity) (McClellan et al., 2018), and six categories (i.e., third
party, standing order, possession, prescriber immunity, dispenser immunity, and layperson
Author Manuscript

dispensing) (Lambdin et al., 2018). A relative weakness of these studies is their relatively
shorter study periods: 31% of studies (n=4) had a study period less than 2 years.

Punitive Drug Use Policies studies—Overall, only 6 studies examined the impact of
punitive drug use policies on opioid-related outcomes; as such, a primary weakness is the
limited number of published evaluation studies. Overall, 3 (50%) analyzed longitudinal data
with a comparison group design, 2 (33%) used longitudinal data with no comparison group,
and 1 (17%) used cross-sectional data with a comparison group. Relative strengths included
that all 4 studies with a comparison group design adjusted for differences between policy
and comparison states and 4 of the 6 studies accounted for co-occurring policies in the
primary analyses. Only 1 study considered policy heterogeneity through more complex
specifications of the policy variable. A relative weakness of these studies is their relatively
shorter study periods, as 33% (n=2) of studies had a study period less than 2 years.
Author Manuscript

4. DISCUSSION
This study provides a novel overview of the state of the opioid policy literature between
2005 to 2018, focusing on evaluation studies of U.S. state and federal policies on opioid-
related outcomes. In contrast to prior systematic reviews regarding specific opioid policies
(e.g., (Fink et al., 2018; Finley et al., 2017)), we provide a more expansive perspective on
the state of the field by considering the breadth of policies and outcomes evaluated to date.
Furthermore, our study is distinct from recent opioid policy scoping reviews (Beaudoin et
al., 2016; Haegerich et al., 2019; Mauri et al., 2020) in that: (1) we provide a detailed
characterization of study design and methods by policy category and (2) we include policy
categories that are not specifically targeting opioids but have been evaluated in terms of their
Author Manuscript

impacts on opioid-related outcomes, including state marijuana laws and Medicaid expansion
through the Affordable Care Act.

We identified 8 categories of policies evaluated across studies, listed from most to less
frequent: (1) prescription drug monitoring programs, (2) opioid prescribing policies, (3)
federal regulation of prescription opioids, (4) treatment access policies, (5) pain
management clinic laws, (6) overdose prevention policies, (7) state marijuana laws, and (8)
punitive drug use policies. Similarly, we identified 7 categories of opioid-related outcomes

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 12

examined, listed from most to least frequent: (1) opioid prescribing behavior, (2) opioid-
Author Manuscript

related mortality and morbidity, (3) prescription opioid misuse/OUD, (4) heroin and other
illicit opioid use, (5) treatment utilization, (6) naloxone use, and (7) opioid-related crime.
Currently, the majority of existing studies examined policies aimed strictly at prescription
opioids (PDMPs, opioid prescribing policies, federal regulation of prescription opioids, pain
clinic laws). Furthermore, the majority of studies evaluated policy effectiveness with respect
to proximal outcomes (e.g., effect of PDMPs on opioid prescribing, effect of treatment
access policies on treatment utilization). We note that the research objectives and study
design of published studies, in part, have been driven by data availability. The preponderance
of studies of prescription opioid-related policies likely reflects the fact that many of the
earliest policies enacted targeted opioid prescribing, resulting in richer data availability (e.g.,
greater number of implementation states, longer window of post-policy data). Furthermore,
data on prescription opioids is accessible via administrative data, whereas reliable data
Author Manuscript

regarding use of heroin and other illicit opioids is relatively scarce.

Our results indicated that study design rigor varied notably across policy categories,
highlighting the need for broader adoption of rigorous methods in the opioid policy field. Of
the 145 studies, only 29 (20%) met each of three key criteria for rigorous design: analysis of
longitudinal data with a comparison group design, adjustment for difference between policy
and comparison states, and adjustment for potentially confounding co-occurring policies.
These 29 studies were predominately published in 2017–2018 and primarily evaluated
PDMPs (n=11), marijuana laws (n=9), treatment-related policies (n=6), and overdose
prevention policies (n=5). In addition to the greatest number of rigorous studies, key
strengths of PDMP studies were that (1) the studies included more years of data and (2) the
majority (65%) examined policy heterogeneity (i.e., compared distinct PDMP components).
While opioid prescribing policies was the second-most common category evaluated, a
Author Manuscript

notable weakness of these studies was that the majority (71%) did not employ a comparison
group design. While a limited number of studies evaluated marijuana laws and overdose
prevention policies, these studies were more likely to have rigorous designs (with respect to
the three aforementioned criteria) relative to studies in other policy categories.

While challenging in the context of a federal-level policy, the use of comparison group
design significantly strengthens inference when evaluating state-level policies. Confounding
due to systematic differences between policy and comparison groups may bias effect
estimates if not rigorously controlled for. While the commonly-used DID design can account
for unobserved confounders if assumptions are upheld (Wing et al., 2018), these may not
always be plausible. In particular, policy and comparison groups may have differential
outcome trends over time or state population composition may change significantly across
Author Manuscript

time (Stuart et al., 2014). While the majority of comparison group design studies reviewed
adjusted for some set of observed covariates, the most common method was regression
adjustment. Policy researchers should consider more robust approaches such as propensity
score methods (Austin and Stuart, 2015; Linden and Adams, 2011; Stuart et al., 2014) and
synthetic control designs (Abadie et al., 2010; Kreif et al., 2016; Xu, 2017), or, in the
context of time-varying confounding, marginal structural models or a principal stratification
framework (Cole and Hernan, 2008; Robins et al., 2000; Santacatterina et al., 2019;
Shinohara et al., 2013).

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 13

Another key methodological consideration our review identified was the importance of
Author Manuscript

controlling for co-occurring policies. While addressing potential confounding of co-


occurring opioid policies may be viewed as more necessary for face validity when
examining the impact of policies that don’t directly target opioid-related outcomes (e.g.,
marijuana laws), we underscore that all evaluation studies are subject to potential bias when
co-occurring policies are not accounted for. Furthermore, co-occurring policies may yield
synergistic policy effects, although such additive or interactive effects have rarely been
examined to date. Going forward, it is imperative that evaluation studies more rigorously
account for concurrent policies when estimating the effects of a given policy as well as
estimate more expansive policy effects, such as potential additive effects of concurrent
policies. Recent methodological work has focused on causal inference methods in the
context of concurrent treatments (Ellis and Brookhart, 2013; Jackson, 2016; Lusivika-
Nzinga et al., 2017).
Author Manuscript

As the state-level policy landscape has become more complex, increasing attention has been
paid to heterogeneity in state policies and resultant variation in policy effectiveness. Indeed,
nearly half of all studies and 65% of PDMP studies used more complex specifications (e.g.,
categorical or composite variables) to examine effectiveness across distinct policy
components or characteristics. We note that early studies of a given policy are often unable
to examine policy heterogeneity due to the limited number of enactment states; as more
states implement a given policy, it is increasingly possible to rigorously examine differences
across policy components or characteristics. Yet, as highlighted in our findings, there is
currently tremendous variation in how studies classify and operationalize heterogeneity
across policies, suggesting the need for a more structured and systematic approach (Grant et
al., 2020). Going forward, an essential challenge for the field is defining and
operationalizing meaningful policy categories in evaluation studies.
Author Manuscript

Finally, our findings highlight several policies and outcomes for which more evidence is
needed. More studies are needed regarding overdose prevention policies (e.g., naloxone
access laws, overdose education/naloxone distribution initiatives, Good Samaritan laws),
particularly as the prevalence of overdose deaths due to synthetic opioids is rising in some
regions (Hedegaard et al., 2020). Additional research regarding treatment-related policies is
needed, as the policy landscape continues to evolve regarding state Medicaid coverage of (or
alternatively, restrictions relating to) medication treatment (Kaiser Family Foundation, 2019;
Leslie et al., 2019) as well as federal policies regarding buprenorphine waiver requirements
(McBain et al., 2020). Also, few studies to date have examined punitive drug use policies,
such as laws that stipulate that substance use during pregnancy may be grounds for
terminating parental rights – despite enactment in 23 states, it is unclear whether these laws
Author Manuscript

promote maternal and child well-being. Furthermore, examination of more distal outcomes
of policies is needed, including the impact of PDMPs and opioid prescribing policies on the
use of non-opioid pain management techniques and patient functional ability. Similarly,
existing studies of overdose prevention policies have primarily assessed the impacts on
overdoses and naloxone use, yet the potential effects on opioid misuse or treatment
engagement post-overdose has not been well-characterized. Unintended consequences of
policies also deserve more attention; for example, a recent study found that punitive state

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 14

policies related to substance use during pregnancy were linked with increased neonatal
Author Manuscript

abstinence syndrome rates (Faherty et al., 2019).

4.1 Limitations of the current study


Several limitations of the current study deserve mention. The scope of our literature review
was U.S. state and federal level policies – many opioid-related policies and initiatives have
also been adopted at the community or health system level. We did not include gray
literature publications, unpublished evaluation studies, dissertations, or abstracts from
conference proceedings. Given that our study objective was a scoping review of the state of
the field, we did not systematically review evidence regarding effectiveness of specific
policies nor did we conduct a meta-analysis to obtain pooled estimates of policy
effectiveness.
Author Manuscript

5. CONCLUSION
The current opioid policy literature is rapidly evolving, but still has limitations. More
evaluation studies are needed regarding overdose prevention policies and policies related to
treatment access. Greater examination of distal outcomes and potential unintended
consequences are also warranted. There are still important methodological challenges and
considerations facing the field, including standardized classification of opioid policies and
heterogeneous policy components, identifying an optimal comparison group and rigorously
controlling for differences across policy and comparison groups, and disentangling effects of
co-occurring policies. Increasing the methodological rigor of opioid evaluation studies is
imperative to identifying and implementing opioid policies that are most effective at
reducing opioid-related harms.
Author Manuscript

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

Acknowledgments
Role of funding source: This work was funded by awards P50DA046351. The content is solely the responsibility
of the authors and does not necessarily represent the official views of NIDA, the NIH or the US Government.

Funding: P50DA046351

REFERENCES
Abadie A, Diamond A, Hainmueller J, 2010 Synthetic control methods for comparative case studies:
Estimating the effect of California’s tobacco control program. J. Am. Stat. Assoc 105, 493–505.
Author Manuscript

Abraham AJ, Andrews CM, Grogan CM, Pollack HA, D’Aunno T, Humphreys K, Friedmann PD,
2018 State-targeted funding and technical assistance to increase access to medication treatment for
opioid use disorder. Psychiatr. Serv 69, 448–455. [PubMed: 29241428]
Ali MM, Dowd WN, Classen T, Mutter R, Novak SP, 2017 Prescription drug monitoring programs,
nonmedical use of prescription drugs, and heroin use: Evidence from the National Survey of Drug
Use and Health. Addict. Behav 69, 65–77. [PubMed: 28152391]
Alpert A, Powell D, Pacula RL, 2018 Supply-side drug policy in the presence of substitutes: Evidence
from the introduction of abuse-deterrent opioids. Am. Econ. J.-Econ. Polic 10, 1–35.

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 15

Angelotta C, Weiss CJ, Angelotta JW, Friedman RA, 2016 A moral or medical problem? The
relationship between legal penalties and treatment practices for opioid use disorders in pregnant
Author Manuscript

women. Womens Health Issues 26, 595–601. [PubMed: 27773527]


Austin PC, Stuart EA, 2015 Moving towards best practice when using inverse probability of treatment
weighting (IPTW) using the propensity score to estimate causal treatment effects in observational
studies. Stat. Med 34, 3661–3679. [PubMed: 26238958]
Ayres I, Jalal A, 2018 The impact of prescription drug monitoring programs on U.S. Opioid
prescriptions. J. Law. Med. Ethics 46, 387–403. [PubMed: 30146997]
Bachhuber MA, Mehta PK, Faherty LJ, Saloner B, 2017 Medicaid coverage of methadone
maintenance and the use of opioid agonist therapy among pregnant women in specialty treatment.
Med. Care 55, 985–990. [PubMed: 29135769]
Bailey JE, Barton PL, Lezotte D, Lowenstein SR, Dart RC, 2006 The effect of FDA approval of a
generic competitor to OxyContin (oxycodone HCl controlled-release) tablets on the abuse of
oxycodone. Drug Alcohol Depend 84, 182–187. [PubMed: 16510252]
Beaudoin FL, Banerjee GN, Mello MJ, 2016 State-level and system-level opioid prescribing policies:
The impact on provider practices and overdose deaths, a systematic review. J. Opioid Manag 12,
Author Manuscript

109–118. [PubMed: 27194195]


Bohnert ASB, Guy GP Jr., Losby JL, 2018 Opioid prescribing in the United States before and after the
Centers for Disease Control and Prevention’s 2016 opioid guideline. Ann. Intern. Med 169, 367–
375. [PubMed: 30167651]
Bounthavong M, Harvey MA, Wells DL, Popish SJ, Himstreet J, Oliva EM, Kay CL, Lau MK,
Randeria-Noor PP, Phillips AG, Christopher MLD, 2017 Trends in naloxone prescriptions
prescribed after implementation of a national academic detailing service in the Veterans Health
Administration: A preliminary analysis. J. Am. Pharm. Assoc 57, S68–S72.
Chang HY, Lyapustina T, Rutkow L, Daubresse M, Richey M, Faul M, Stuart EA, Alexander GC, 2016
Impact of prescription drug monitoring programs and pill mill laws on high-risk opioid
prescribers: A comparative interrupted time series analysis. Drug Alcohol Depend 165, 1–8.
[PubMed: 27264166]
Chang HY, Murimi I, Faul M, Rutkow L, Alexander GC, 2018 Impact of Florida’s prescription drug
monitoring program and pill mill law on high-risk patients: A comparative interrupted time series
analysis. Pharmacoepidemiol. Drug Saf 27, 422–429. [PubMed: 29488663]
Author Manuscript

Chen CC, De AP, Sweet B, Wade RL, 2017 Evaluation of patient migration patterns and related health
care costs within a national Medicare advantage prescription drug plan after implementation of an
oxycodone HCl extended-release access restriction. J. Manag. Care Spec. Pharm 23, 902–912.
[PubMed: 28737985]
Ciccarone D, 2017 Fentanyl in the US heroin supply: A rapidly changing risk environment. Int. J. Drug
Policy 46, 107–111. [PubMed: 28735776]
Cicero TJ, Ellis MS, 2015 Abuse-deterrent formulations and the prescription opioid abuse epidemic in
the United States: Lessons learned from OxyContin. JAMA Psychiatry 72, 424–430. [PubMed:
25760692]
Cochran G, Gordon AJ, Gellad WF, Chang CH, Lo-Ciganic WH, Lobo C, Cole E, Frazier W, Zheng P,
Kelley D, Donohue JM, 2017 Medicaid prior authorization and opioid medication abuse and
overdose. Am. J. Manag. Care 23, e164–e171. [PubMed: 28810127]
Cole SR, Hernan MA, 2008 Constructing inverse probability weights for marginal structural models.
Am. J. Epidemiol 168, 656–664. [PubMed: 18682488]
Author Manuscript

Davies RB, 2010 Mandatory minimum sentencing, drug purity and overdose rates. Econ. Soc. Rev 41,
429–457.
Delcher C, Chen G, Wang Y, Slavova S, Goldberger BA, 2017 Fatal poisonings involving
propoxyphene before and after voluntary withdrawal from the United States’ market: An analysis
from the state of Florida. Forensic Sci. Int 280, 228–232. [PubMed: 29080523]
Dhaval D, Deza M, Horn BP, 2018 Prescription drug monitoring programs, opioid abuse, and crime
National Bureau of Economic Research, Cambridge, MA.

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 16

Dick AW, Pacula RL, Gordon AJ, Sorbero M, Burns RM, Leslie D, Stein BD, 2015 Growth in
buprenorphine waivers for physicians increased potential access to opioid agonist treatment, 2002–
Author Manuscript

11. Health Aff. (Millwood) 34, 1028–1034. [PubMed: 26056209]


Dowell D, Zhang K, Noonan RK, Hockenberry JM, 2016 Mandatory provider review and pain clinic
laws reduce the amounts of opioids prescribed and overdose death rates. Health Aff. (Millwood)
35, 1876–1883. [PubMed: 27702962]
Ellis AR, Brookhart MA, 2013 Approaches to inverse-probability-of-treatment--weighted estimation
with concurrent treatments. J. Clin. Epidemiol 66, S51–56. [PubMed: 23849154]
Faherty LJ, Kranz AM, Russell-Fritch J, Patrick SW, Cantor J, Stein BD, 2019 Association of punitive
and reporting state policies related to substance use in pregnancy with rates of neonatal abstinence
syndrome. JAMA Network Open 2, e1914078. [PubMed: 31722022]
Feder KA, Mojtabai R, Krawczyk N, Young AS, Kealhofer M, Tormohlen KN, Crum RM, 2017
Trends in insurance coverage and treatment among persons with opioid use disorders following the
Affordable Care Act. Drug Alcohol Depend 179, 271–274. [PubMed: 28823834]
Fink DS, Schleimer JP, Sarvet A, Grover KK, Delcher C, Castillo-Carniglia A, Kim JH, Rivera-
Aguirre AE, Henry SG, Martins SS, Cerda M, 2018 Association between prescription drug
Author Manuscript

monitoring programs and nonfatal and fatal drug overdoses: A systematic review. Ann. Intern.
Med 168, 783–790. [PubMed: 29801093]
Finley EP, Garcia A, Rosen K, McGeary D, Pugh MJ, Potter JS, 2017 Evaluating the impact of
prescription drug monitoring program implementation: a scoping review. BMC Health Serv. Res
17, 420. [PubMed: 28633638]
Gertner AK, Domino ME, Davis CS, 2018 Do naloxone access laws increase outpatient naloxone
prescriptions? Evidence from Medicaid. Drug Alcohol Depend 190, 37–41. [PubMed: 29966851]
Grant S, Smart R, Stein BD, 2020 We need a taxonomy of state-level opioid policies, JAMA Health
Forum Insights
Haegerich TM, Jones CM, Cote PO, Robinson A, Ross L, 2019 Evidence for state, community and
systems-level prevention strategies to address the opioid crisis. Drug Alcohol Depend 204,
107563. [PubMed: 31585357]
Hedegaard H, Miniño AM, Warner M, 2020 Drug overdose deaths in the United States, 1999–2018
National Center for Health Statistics, Hyattsville, MD.
Author Manuscript

Heins SE, 2019 Prescription opioids: A continuing contributor to the epidemic. Am. J. Public Health
109, 1166–1167. [PubMed: 31390249]
Jackson JW, 2016 Diagnostics for confounding of time-varying and other joint exposures.
Epidemiology 27, 859–869. [PubMed: 27479649]
Johnson EM, Porucznik CA, Anderson JW, Rolfs RT, 2011 State-level strategies for reducing
prescription drug overdose deaths: Utah’s prescription safety program. Pain Med 12 Suppl 2, S66–
72. [PubMed: 21668759]
Kaiser Family Foundation, 2019 The opioid epidemic and medicaid’s role in facilitating access to
treatment https://bit.ly/2TdsxE6. <accessed on Jan 10, 2020>
Keast SL, Kim H, Deyo RA, Middleton L, McConnell KJ, Zhang K, Ahmed SM, Nesser N, Hartung
DM, 2018 Effects of a prior authorization policy for extended-release/long-acting opioids on
utilization and outcomes in a state Medicaid program. Addiction 113, 1651–1660.
Kennedy-Hendricks A, Richey M, McGinty EE, Stuart EA, Barry CL, Webster DW, 2016 Opioid
overdose deaths and Florida’s crackdown on pill mills. Am. J. Public Health 106, 291–297.
[PubMed: 26691121]
Author Manuscript

Kreif N, Grieve R, Hangartner D, Turner AJ, Nikolova S, Sutton M, 2016 Examination of the synthetic
control method for evaluating health policies with multiple treated units. Health Econ 25, 1514–
1528. [PubMed: 26443693]
Kuo YF, Raji MA, Liaw V, Baillargeon J, Goodwin JS, 2018 Opioid prescriptions in older Medicare
beneficiaries after the 2014 federal rescheduling of hydrocodone products. J. Am. Geriatr. Soc 66,
945–953. [PubMed: 29656382]
Lambdin BH, Davis CS, Wheeler E, Tueller S, Kral AH, 2018 Naloxone laws facilitate the
establishment of overdose education and naloxone distribution programs in the United States.
Drug Alcohol Depend 188, 370–376. [PubMed: 29776688]

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 17

Larochelle MR, Zhang F, Ross-Degnan D, Wharam JF, 2015 Rates of opioid dispensing and overdose
after introduction of abuse-deterrent extended-release oxycodone and withdrawal of
Author Manuscript

propoxyphene. JAMA Intern. Med 175, 978–987. [PubMed: 25895077]


Leslie DL, Ba DM, Agbese E, Xing X, Liu G, 2019 The economic burden of the opioid epidemic on
states: the case of Medicaid. Am. J. Manag. Care 25, S243–S249. [PubMed: 31361426]
Lin LA, Bohnert ASB, Kerns RD, Clay MA, Ganoczy D, Ilgen MA, 2017 Impact of the opioid safety
initiative on opioid-related prescribing in veterans. Pain 158, 833–839. [PubMed: 28240996]
Linden A, Adams JL, 2011 Applying a propensity score-based weighting model to interrupted time
series data: Improving causal inference in programme evaluation. J. Eval. Clin. Pract 17, 1231–
1238. [PubMed: 20973870]
Lusivika-Nzinga C, Selinger-Leneman H, Grabar S, Costagliola D, Carrat F, 2017 Performance of the
marginal structural cox model for estimating individual and joined effects of treatments given in
combination. BMC Med. Res. Methodol 17, 160. [PubMed: 29202691]
Marsh JC, Park K, Lin YA, Bersamira C, 2018 Gender differences in trends for heroin use and
nonmedical prescription opioid use, 2007–2014. J. Subst. Abuse Treat 87, 79–85. [PubMed:
29433788]
Author Manuscript

Mauri AI, Townsend TN, Haffajee RL, 2020 The association of state opioid misuse prevention policies
with patient- and provider-related outcomes: A scoping review. Milbank Q 98, 57–105. [PubMed:
31800142]
McBain RK, Dick A, Sorbero M, Stein BD, 2020 Growth and distribution of buprenorphine-waivered
providers in the United States, 2007–2017. Ann. Intern. Med 172, 504–506. [PubMed: 31905379]
Meinhofer A, Witman AE, 2018 The role of health insurance on treatment for opioid use disorders:
Evidence from the Affordable Care Act Medicaid expansion. J. Health Econ 60, 177–197.
[PubMed: 29990675]
Midgette G, Davenport S, Caulkins JP, Kilmer B, 2019 What America’s users spend on illegal drugs,
2006–2016 RAND Corporation, Santa Monica, CA https://www.rand.org/pubs/research_reports/
RR3140.html <accessed on May 10, 2020>
Murimi IB, Chang HY, Bicket M, Jones CM, Alexander GC, 2019 Using trajectory models to assess
the effect of hydrocodone upscheduling among chronic hydrocodone users. Pharmacoepidemiol.
Drug Saf 28, 70–79. [PubMed: 30187574]
Author Manuscript

Nguyen H, Parker BR, 2018 Assessing the effectiveness of New York’s 911 Good Samaritan law-
evidence from a natural experiment. Int. J. Drug Policy 58, 149–156. [PubMed: 29966919]
Pardo B, Taylor J, Caulkins JP, Kilmer B, Reuter P, Stein BD, 2019 The future of fentanyl and other
synthetic opioids RAND Corporation, Santa Monica, CA https://www.rand.org/pubs/
research_reports/RR3117.html <accessed on Jan 10, 2020>
PDAPS, 2019 http://pdaps.org/datasets/pdmp-implementation-dates. <accessed on Jan 10, 2020>
Penm J, MacKinnon NJ, Mashni R, Lyons MS, Hooker EA, Winstanley EL, Carlton-Ford S, Connelly
C, Tolle E, Boone J, Koechlin K, Defiore-Hyrmer J, 2018 Statewide cross-sectional survey of
emergency departments’ adoption and implementation of the Ohio opioid prescribing guidelines
and opioid prescribing practices. BMJ Open 8, e020477.
Raji MA, Kuo YF, Adhikari D, Baillargeon J, Goodwin JS, 2018 Decline in opioid prescribing after
federal rescheduling of hydrocodone products. Pharmacoepidemiol. Drug Saf 27, 513–519.
[PubMed: 29271049]
Rees DI, Sabia JJ, Argys LM, Latshaw J, Dhaval D, 2017 With a little help from my friends: The
effects of naloxone access and Good Samaritan laws on opioid-related deaths National Bureau of
Author Manuscript

Economic Research, Cambridge, MA.


Roberts AW, Farley JF, Holmes GM, Oramasionwu CU, Ringwalt C, Sleath B, Skinner AC, 2016
Controlled substance lock-in programs: Examining an unintended consequence of a prescription
drug abuse policy. Health Aff. (Millwood) 35, 1884–1892. [PubMed: 27702963]
Robins JM, Hernan MA, Brumback B, 2000 Marginal structural models and causal inference in
epidemiology. Epidemiology 11, 550–560. [PubMed: 10955408]
Robinson A, Wermeling DP, 2014 Intranasal naloxone administration for treatment of opioid overdose.
Am. J. Health Syst. Pharm 71, 2129–2135. [PubMed: 25465584]

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 18

Rutkow L, Chang HY, Daubresse M, Webster DW, Stuart EA, Alexander GC, 2015 Effect of Florida’s
prescription drug monitoring program and pill mill laws on opioid prescribing and use. JAMA
Author Manuscript

Intern. Med 175, 1642–1649. [PubMed: 26280092]


Saloner B, Stoller KB, Barry CL, 2016 Medicaid coverage for methadone maintenance and use of
opioid agonist therapy in specialty addiction treatment. Psychiatr. Serv 67, 676–679. [PubMed:
26927578]
SAMHSA, 2019 Key substance use and mental health indicators in the United States: Results from the
2018 National Survey on Drug Use and Health Center for Behavioral Health Statistics and Quality,
Substance Abuse and Mental Health Services Administration, Rockville, MD.
SAMHSA, 2020 The opioid crisis and the black/African American population: An urgent issue Office
of Behavioral Health Equity, Substance Abuse and Mental Health Services Administration.
Santacatterina M, Garcia-Pareja C, Bellocco R, Sonnerborg A, Ekstrom AM, Bottai M, 2019 Optimal
probability weights for estimating causal effects of time-varying treatments with marginal
structural cox models. Stat. Med 38, 1891–1902. [PubMed: 30592073]
Schuler MS, Dick AW, Stein BD, 2020 Heterogeneity in prescription opioid pain reliever misuse
across age groups: 2015–2017 National Survey on Drug Use and Health. J. Gen. Intern. Med 35,
Author Manuscript

792–799. [PubMed: 31792871]


Sharp A, Jones A, Sherwood J, Kutsa O, Honermann B, Millett G, 2018 Impact of Medicaid expansion
on access to opioid analgesic medications and medication-assisted treatment. Am. J. Public Health
108, 642–648. [PubMed: 29565661]
Shi Y, 2017 Medical marijuana policies and hospitalizations related to marijuana and opioid pain
reliever. Drug Alcohol Depend 173, 144–150. [PubMed: 28259087]
Shi Y, Liang D, Bao Y, An R, Wallace MS, Grant I, 2019 Recreational marijuana legalization and
prescription opioids received by Medicaid enrollees. Drug Alcohol Depend 194, 13–19. [PubMed:
30390550]
Shinohara RT, Narayan AK, Hong K, Kim HS, Coresh J, Streiff MB, Frangakis CE, 2013 Estimating
parsimonious models of longitudinal causal effects using regressions on propensity scores. Stat.
Med 32, 3829–3837. [PubMed: 23533091]
Singh GK, Kim IE, Girmay M, Perry C, Daus GP, Vedamuthu IP, De Los Reyes AA, Ramey CT,
Martin EK, Allender M, 2019 Opioid epidemic in the United States: Empirical trends, and a
literature review of social determinants and epidemiological, pain management, and treatment
Author Manuscript

patterns. Int. J. MCH AIDS 8, 89–100. [PubMed: 31723479]


Skinner AC, Ringwalt C, Naumann RB, Roberts AW, Moss LA, Sachdeva N, Weaver MA, Farley J,
2016 Reducing opioid misuse: Evaluation of a Medicaid controlled substance lock-in program. J.
Pain 17, 1150–1155. [PubMed: 27497767]
Stein BD, Gordon AJ, Dick AW, Burns RM, Pacula RL, Farmer CM, Leslie DL, Sorbero M, 2015
Supply of buprenorphine waivered physicians: The influence of state policies. J. Subst. Abuse
Treat 48, 104–111. [PubMed: 25218919]
Stuart EA, Huskamp HA, Duckworth K, Simmons J, Song Z, Chernew M, Barry CL, 2014 Using
propensity scores in difference-in-differences models to estimate the effects of a policy change.
Health Serv. Outcomes Res. Methodol 14, 166–182. [PubMed: 25530705]
Sumner SA, Mercado-Crespo MC, Spelke MB, Paulozzi L, Sugerman DE, Hillis SD, Stanley C, 2016
Use of naloxone by emergency medical services during opioid drug overdose resuscitation efforts.
Prehosp. Emerg. Care 20, 220–225. [PubMed: 26383533]
Walley AY, Xuan Z, Hackman HH, Quinn E, Doe-Simkins M, Sorensen-Alawad A, Ruiz S, Ozonoff
Author Manuscript

A, 2013 Opioid overdose rates and implementation of overdose education and nasal naloxone
distribution in Massachusetts: Interrupted time series analysis. BMJ 346, f174. [PubMed:
23372174]
Weiner SG, Baker O, Poon SJ, Rodgers AF, Garner C, Nelson LS, Schuur JD, 2017 The effect of
opioid prescribing guidelines on prescriptions by emergency physicians in Ohio. Ann. Emerg. Med
70, 799–808 e791. [PubMed: 28549620]
Wen H, Hockenberry JM, 2018 Association of medical and adult-use marijuana laws with opioid
prescribing for Medicaid enrollees. JAMA Intern. Med 178, 673–679. [PubMed: 29610827]

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 19

Wen H, Hockenberry JM, Borders TF, Druss BG, 2017a Impact of Medicaid expansion on Medicaid-
covered utilization of buprenorphine for opioid use disorder treatment. Med. Care 55, 336–341.
Author Manuscript

[PubMed: 28296674]
Wen H, Schackman BR, Aden B, Bao Y, 2017b States with prescription drug monitoring mandates saw
a reduction in opioids prescribed to Medicaid enrollees. Health Aff. (Millwood) 36, 733–741.
[PubMed: 28373340]
Wing C, Simon K, Bello-Gomez RA, 2018 Designing difference in difference studies: Best practices
for public health policy research. Annu. Rev. Public Health 39, 453–469. [PubMed: 29328877]
Wong SW, Lin HC, 2019 Medical marijuana legalization and associated illicit drug use and
prescription medication misuse among adolescents in the U.S. Addict. Behav 90, 48–54. [PubMed:
30359847]
Xu J, Davis CS, Cruz M, Lurie P, 2018 State naloxone access laws are associated with an increase in
the number of naloxone prescriptions dispensed in retail pharmacies. Drug Alcohol Depend 189,
37–41. [PubMed: 29860058]
Xu YQ, 2017 Generalized synthetic control method: Causal inference with interactive fixed effects
models. Polit. Anal 25, 57–76.
Author Manuscript
Author Manuscript
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 20

Highlights
Author Manuscript

• Conducted a scoping review of U.S. opioid policy evaluation studies from


2005–2018

• 145 studies were reviewed: 21% examined federal policies, 79% state policies

• Majority of studies evaluated policies related to prescription opioids

• Our results indicated that study design rigor varied notably across policy
categories

• Greater adoption of more rigorous study designs and statistical methods is


warranted
Author Manuscript
Author Manuscript
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 21
Author Manuscript
Author Manuscript
Author Manuscript

Figure 1.
Flow chart of literature search and study sample selection
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 22
Author Manuscript
Author Manuscript

Figure 2.
Publication dates of opioid policy evaluation studies reviewed (n=145)
Author Manuscript
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 23
Author Manuscript
Author Manuscript

Figure 3.
Categories of opioid-related policies evaluated across the 145 studies reviewed
Note: The sum across categories exceeds 145, as some studies examined multiple policies.
Author Manuscript
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Schuler et al. Page 24
Author Manuscript
Author Manuscript

Figure 4.
Categorization of studies by both policies evaluated and outcomes examined.
Author Manuscript
Author Manuscript

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Table 1.

Study characteristics for 145 studies reviewed

Opioid Fed. Reg. of Overdose


Treatment State Marijuana Pain Clinic Punitive Drug
Overall PDMP Prescribing Prescription Prevention
Access Policies Laws Laws Use Policies
Schuler et al.

Policies Opioids Policies


N=145 N=52 N=31 N=20 N=17 N=14 N=13 N=13 N=6
Publication Year
2005–2010 7 (5%) 4 (8%) 0 (0%) 1 (5%) 1 (6%) 0 (0%) 0 (0%) 0 (0%) 1 (17%)
2011–2015 33 (23%) 13 (25%) 6 (19%) 7 (35%) 3 (18%) 2 (14%) 3 (23%) 1 (8%) 0 (0%)
2016–2018 105 (72%) 35 (67%) 25 (81%) 12 (60%) 13 (76%) 12 (86%) 10 (77%) 12 (92%) 5 (83%)
Policy level
State-level policy 115 (79%) 52 (100%) 24 (77%) 0 (0%) 15 (88%) 14 (100%) 13 (100%) 12 (92%) 5 (83%)
Federal policy 30 (21%) 0 (0%) 7 (23%) 20 (100%) 2 (12%) 0 (0%) 0 (0%) 1 (8%) 1 (17%)
Study design
Longitudinal data, 68 (47%) 31 (60%) 8 (26%) 1 (5%) 9 (53%) 12 (86%) 8 (62%) 11 (85%) 3 (50%)
comparison group
Longitudinal data, no 64 (44%) 16 (31%) 21 (68%) 19 (95%) 4 (24%) 2 (14%) 5 (38%) 2 (15%) 2 (33%)
comparison group
Cross-sectional data, 11 (8%) 5 (10%) 1 (3%) 0 (0%) 4 (24%) 0 (0%) 0 (0%) 0 (0%) 1 (17%)
comparison group
Cross-sectional data, 2 (1%) 0 (0%) 1 (3%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
no comparison group
Adjusted for differences between policy and comparison groups
Covariate regression 62 (43%) 27 (52%) 6 (19%) 0 (0%) 12 (71%) 11 (79%) 5 (38%) 9 (69%) 4 (67%)
adjustment
Propensity score 6 (4%) 2 (4%) 3 (10%) 0 (0%) 0 (0%) 1 (7%) 0 (0%) 0 (0%) 0 (0%)
method

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
No adjustment 11 (8%) 7 (13%) 0 (0%) 1 (5%) 1 (6%) 0 (0%) 3 (23%) 2 (15%) 0 (0%)
Not applicable - no 66 (46%) 16 (31%) 22 (71%) 19 (95%) 4 (24%) 2 (14%) 5 (38%) 2 (15%) 2 (33%)
comparison group
Policy coding other than single binary indicator
Yes 69 (48%) 34 (65%) 12 (39%) 0 (0%) 9 (53%) 7 (50%) 6 (46%) 7 (54%) 1 (17%)
No 76 (52%) 18 (35%) 19 (61%) 20 (100%) 8 (47%) 7 (50%) 7 (54%) 6 (46%) 5 (83%)
Accounted for any co-occurring policies
Primary analyses: 14 (10%) 7 (13%) 1 (3%) 2 (10%) 1 (6%) 2 (14%) 2 (15%) 2 (15%) 2 (33%)
Single co-occurring
Page 25
Author Manuscript Author Manuscript Author Manuscript Author Manuscript

Opioid Fed. Reg. of Overdose


Treatment State Marijuana Pain Clinic Punitive Drug
Overall PDMP Prescribing Prescription Prevention
Access Policies Laws Laws Use Policies
Policies Opioids Policies
N=145 N=52 N=31 N=20 N=17 N=14 N=13 N=13 N=6
Primary analyses: 2+ 29 (20%) 8 (15%) 2 (6%) 1 (5%) 8 (47%) 8 (57%) 1 (8%) 3 (23%) 2 (33%)
Schuler et al.

co-occurring
Secondary / 8 (6%) 5 (10%) 1 (3%) 0 (0%) 0 (0%) 1 (7%) 1 (8%) 0 (0%) 0 (0%)
sensitivity analyses only
No 94 (65%) 32 (62%) 27 (87%) 17 (85%) 8 (47%) 3 (21%) 9 (69%) 8 (62%) 2 (33%)
Length of study period
< 2 yrs 33 (23%) 10 (19%) 8 (26%) 5 (25%) 5 (29%) 2 (14%) 3 (23%) 4 (31%) 2 (33%)
2 yrs - < 4 yrs 32 (22%) 11 (21%) 7 (23%) 9 (45%) 2 (12%) 7 (50%) 3 (23%) 2 (15%) 1 (17%)
4 yrs < 6 yrs 20 (14%) 5 (10%) 8 (26%) 3 (15%) 1 (6%) 1 (7%) 2 (15%) 2 (15%) 2 (33%)
6 yrs < 8 yrs 12 (8%) 8 (15%) 3 (10%) 0 (0%) 3 (18%) 1 (7%) 0 (0%) 0 (0%) 0 (0%)
8 yrs < 10 yrs 14 (10%) 6 (12%) 2 (6%) 1 (5%) 3 (18%) 1 (7%) 1 (8%) 1 (8%) 0 (0%)
10 yrs 34 (23%) 12 (23%) 3 (10%) 2 (10%) 3 (18%) 2 (14%) 4 (31%) 4 (31%) 1 (17%)

Drug Alcohol Depend. Author manuscript; available in PMC 2021 September 01.
Page 26

You might also like