Sage Journals: Discover world-class research

Abstract

Objectives

In the United States, rising rates of overdose deaths and recent outbreaks of hepatitis C virus and HIV infection are associated with injection drug use. We updated a 2014 review of systems-level opioid policy interventions by focusing on evidence published during 2014-2018 and new and expanded opioid policies.

Methods

We searched the MEDLINE database, consistent with the 2014 review. We included articles that provided original empirical evidence on the effects of systems-level interventions on opioid use, overdose, or death; were from the United States or Canada; had a clear comparison group; and were published from January 1, 2014, through July 19, 2018. Two raters screened articles and extracted full-text data for qualitative synthesis of consistent or contradictory findings across studies. Given the rapidly evolving field, the review was supplemented with a search of additional articles through November 17, 2019, to assess consistency of more recent findings.

Results

The keyword search yielded 535 studies, 66 of which met inclusion criteria. The most studied interventions were prescription drug monitoring programs (PDMPs) (59.1%), and the least studied interventions were clinical guideline changes (7.6%). The most common outcome was opioid use (77.3%). Few articles evaluated combination interventions (18.2%). Study findings included the following: PDMP effectiveness depends on policy design, with robust PDMPs needed for impact; health insurer and pharmacy benefit management strategies, pill-mill laws, pain clinic regulations, and patient/health care provider educational interventions reduced inappropriate prescribing; and marijuana laws led to a decrease in adverse opioid-related outcomes. Naloxone distribution programs were understudied, and evidence of their effectiveness was mixed. In the evidence published after our search’s 4-year window, findings on opioid guidelines and education were consistent and findings for other policies differed.

Conclusions

Although robust PDMPs and marijuana laws are promising, they do not target all outcomes, and multipronged interventions are needed. Future research should address marijuana laws, harm-reduction interventions, health insurer policies, patient/health care provider education, and the effects of simultaneous interventions on opioid-related outcomes.

Keywords

systematic review opioid dependence drug overdose health policy public health systems research

The age-adjusted rate of opioid-related overdose deaths in theUnited States increased from 2.9 per 100 000 population in 1999 to 14.9 per 100 000 population in 2017.¹ Synthetic opioids (non-methadone) were involved in 59.8% of opioid-related overdose deaths in 2017.^1,2 The opioid epidemic and associated increases in injection drug use have increased hepatitis C virus (HCV) incidence, particularly in the West and Appalachia, and contributed to HIV outbreaks among persons who inject drugs.^3
-5

Systems-level policy interventions, defined as state- or health systems–level regulations, laws, and policies, are needed to combat drug-related harms, including overdoses, and increased morbidity, mortality, and economic consequences of HIV infection, HCV infection, and related conditions. Some interventions address prescribing behavior: (1) state-level prescription drug monitoring programs (PDMPs), which require pharmacies to submit prescribing information on controlled substances to health departments or pharmacy boards; (2) health insurer and pharmacy benefit management strategies (eg, lock-in programs that limit opioid access among beneficiaries with a potential for overuse to a single prescriber or pharmacy or prior authorization programs that allow reimbursement only with the health care providers’ confirmation of a patient’s need for medication); (3) clinical guidelines on appropriate prescribing (eg, dosing, duration, and type of medication); and (4) pain clinic regulations (also referred to as “pill-mill” laws) that regulate certification, ownership, and other procedures in pain clinics that prescribe opioids. Other interventions address harm reduction and education: (1) legalizing the medical or recreational use of marijuana as a potential alternative for pain control; (2) naloxone distribution to persons who use drugs, community members, and first responders; (3) Good Samaritan laws, which provide immunity from investigation about possession of drugs for persons who call 911 while having or witnessing an overdose; and (4) educational materials for patients and/or health care providers about opioid use, adverse effects, and abuse, as well as alternative options for pain treatment.

Despite rapidly expanding literature evaluating systems-level opioid policies, consensus is limited. Results are mixed on whether PDMPs and state-level prescribing policies encourage responsible prescribing.^6,7 Naloxone distribution programs and legalization of medical marijuana show positive results, but studies have substantial limitations.^8,9 A comprehensive review of systems-level policies conducted in 2014 concluded that the evidence about which policies are effective was weak; many studies used low-quality designs and examined intermediary outcomes (eg, knowledge or prescribing practices) rather than direct health outcomes (eg, opioid use or opioid-related overdose).¹⁰

Many studies of systems-level policies have been published since the 2014 comprehensive review.¹⁰ Recent literature reviews on opioid policies have focused on 1 or few interventions,^6

-9 making it difficult to summarize current knowledge across interventions. The opioid epidemic has evolved since 2014, and interventions have expanded accordingly. To our knowledge, no similar updated review had been published at the time of our analysis. Our review updates the 2014 review with recent evidence about the effectiveness of systems-level interventions on opioid use, overdose, and mortality to identify promising policy interventions and research priorities.

Methods

We conducted a rapid review of literature published during 2014-2018, using the search strategy from a 2014 systematic review on this topic.¹⁰ Rapid reviews and systematic reviews use similar techniques for finding and synthesizing evidence, but rapid reviews emphasize timeliness rather than exhaustiveness.¹¹ Although systematic reviews are the gold standard, rapid reviews address decision makers’ needs for quick syntheses of broader scopes of knowledge. They expedite the review process by narrowing the search, restricting review of the title/abstract or limiting review of the full-text article, and limiting the amount of data extracted or the quality of data assessed.¹² We used a rapid review because studies on opioid policies are evolving quickly, and timely assessment of the current state of knowledge and evidence gaps of these studies is crucial to inform current policy and research. Because we conducted a rapid review,¹³ we did not perform a systematic assessment of each study’s bias risk. However, we extracted the limitations of each study, which are in an appendix (available from the authors upon request). A summary of the themes of these limitations is provided (Box).

Box

Summary of major study limitations^a among studies that examined the effectiveness of systems-level policies and interventions on opioid use, overdose, and deaths, United States and Canada, 2014-2018

Study Design Limitations
Selection bias possibility: jurisdictions were not randomly assigned to implement interventions (For example, states that implemented prescription drug monitoring programs [PDMPs] may be more affected by the opioid epidemic than states that did not implement PDMPs or may have implemented multiple interventions concurrently.) Uncontrolled secular changes, such as changes in heroin purity, implementation of other opioid-related policies, changes in drug scheduling (eg, from Schedule III to Schedule II), law enforcement activities, cross-state trade of prescription opioids, and commercial health insurer policy changes Limited isolation of specific intervention effects when multiple interventions were implemented during a short period Interstate variation in implementation, including policy design features and the extent to which clinicians use PDMPs Limited analysis of long-term effects, or ability to extrapolate to longer time horizons, because of a rapidly changing environment Limited generalizability to other states with various policy environments and populations and to other populations not included in the analysis, particularly for studies using administrative data from a single payer, such as Medicare Part D Ecologic analyses of aggregate state-level differences fail to account for heterogeneity in states’ populations or determine individual changes and causal pathways Difficulty in identifying appropriate control groups in comparative state designs because of substantial interstate variation in populations, opioid epidemic characteristics, and other environmental conditions Low statistical power because of small sample sizes for experimental studies and ecologic studies with state and time observations

Study Design Limitations

Selection bias possibility: jurisdictions were not randomly assigned to implement interventions (For example, states that implemented prescription drug monitoring programs [PDMPs] may be more affected by the opioid epidemic than states that did not implement PDMPs or may have implemented multiple interventions concurrently.)

Uncontrolled secular changes, such as changes in heroin purity, implementation of other opioid-related policies, changes in drug scheduling (eg, from Schedule III to Schedule II), law enforcement activities, cross-state trade of prescription opioids, and commercial health insurer policy changes

Limited isolation of specific intervention effects when multiple interventions were implemented during a short period

Interstate variation in implementation, including policy design features and the extent to which clinicians use PDMPs

Limited analysis of long-term effects, or ability to extrapolate to longer time horizons, because of a rapidly changing environment

Limited generalizability to other states with various policy environments and populations and to other populations not included in the analysis, particularly for studies using administrative data from a single payer, such as Medicare Part D

Ecologic analyses of aggregate state-level differences fail to account for heterogeneity in states’ populations or determine individual changes and causal pathways

Difficulty in identifying appropriate control groups in comparative state designs because of substantial interstate variation in populations, opioid epidemic characteristics, and other environmental conditions

Low statistical power because of small sample sizes for experimental studies and ecologic studies with state and time observations

Data Limitations
Lags in data availability limited post-intervention periods reviewed Possible misclassification of legitimate opioid users as opioid abusers, including determining whether medications such as methadone were intended for pain or for addiction treatment Data inaccuracies such as misclassification of overdose deaths, particularly among coroner vs medical examiner counties; incorrect coding of overdoses in hospital claims data; inaccuracies in medical records; possible reporting error in surveys from social desirability bias and recall bias; and potential misclassification of primary cause of death when multiple drugs were present Underreporting of overdoses for numerous reasons including underrecognition of opioid-related toxicity by clinicians. (Also, overdoses treated in the field by emergency medical services may not be brought to a health care facility and therefore may not generate a billing claim.) Undercounting of total prescriptions, because medications paid out of pocket or through other sources were not included in prescription claims data Examination of overdose deaths related to all opioids rather than separate analyses for deaths related to prescription opioids vs heroin, which some authors attribute to a lack of specificity in substances reported in toxicology data Pills dispensed may not reflect actual use among patients Data were frequently limited to 1 health care setting or insurer, reducing generalizability and providing potentially incomplete information on patient-level data, such as total number of pills prescribed Low response rates to surveys Opioid prescription dosage was not always available, particularly for publicly available aggregate data

Data Limitations

Lags in data availability limited post-intervention periods reviewed

Possible misclassification of legitimate opioid users as opioid abusers, including determining whether medications such as methadone were intended for pain or for addiction treatment

Data inaccuracies such as misclassification of overdose deaths, particularly among coroner vs medical examiner counties; incorrect coding of overdoses in hospital claims data; inaccuracies in medical records; possible reporting error in surveys from social desirability bias and recall bias; and potential misclassification of primary cause of death when multiple drugs were present

Underreporting of overdoses for numerous reasons including underrecognition of opioid-related toxicity by clinicians. (Also, overdoses treated in the field by emergency medical services may not be brought to a health care facility and therefore may not generate a billing claim.)

Undercounting of total prescriptions, because medications paid out of pocket or through other sources were not included in prescription claims data

Examination of overdose deaths related to all opioids rather than separate analyses for deaths related to prescription opioids vs heroin, which some authors attribute to a lack of specificity in substances reported in toxicology data

Pills dispensed may not reflect actual use among patients

Data were frequently limited to 1 health care setting or insurer, reducing generalizability and providing potentially incomplete information on patient-level data, such as total number of pills prescribed

Low response rates to surveys

Opioid prescription dosage was not always available, particularly for publicly available aggregate data

^aThis list was developed through a summary of limitations reported by authors of studies included in this review and additional synthesis by the authors of this review. The described limitations focus only on the outcomes of opioid use, overdose, and mortality consistent with the rapid review. Other important limitations outside the scope of our review (eg, not having data on quality of life, functional status, or impact on pain, which are common to all study designs) were not included. A full list of limitations for each study is available from the authors upon request.

Search Strategy

We searched MEDLINE for articles published from January 1, 2014, through July 19, 2018. We selected the start date because the previous review went through 2013. The full list of search keywords is available in our study protocol, which is available from the authors upon request. A public health reference librarian reviewed our search strategy. We included all interventions from the 2014 review¹⁰: PDMPs, health insurer benefit management strategies, pain clinic regulations, clinical guidelines, harm-reduction interventions, patient and health care provider education, and safe storage and disposal of prescription drugs. We added marijuana laws. We excluded clinic-focused studies, such as studies of the effectiveness of naloxone in reversing overdose or marijuana in treating certain health conditions because we were interested in systems-level interventions. We supplemented our search procedure with a manual search of articles cited in references.

Review Process

Two coders (B.A., K.M.T.) reviewed titles and abstracts for eligibility and discussed discrepancies with other authors until they reached consensus. If inclusion criteria were unclear during title and abstract review, we assessed eligibility using full-text review. Study eligibility followed the population, intervention, comparator, outcomes, and timing (PICOT) approach from the Institute of Medicine guidance.¹⁴ We limited the population to the United States and Canada. Our outcomes were opioid use, opioid-related overdose, and opioid-related overdose death. These population-level outcomes reflect the health experience of target populations receiving systems-level interventions, unlike individual-level outcomes from clinical studies such as the effects of medical marijuana on acute myocardial infarction,¹⁵ oncology management,¹⁶ and chronic neuropathic pain.¹⁷ Opioid use outcomes included total number or volume of opioid prescriptions filled, per-capita or per-transaction morphine milligram equivalent (MME) dispensing, and number of opioid-abusing patients. Opioid-related overdose outcomes included overdose-related emergency department visits or inpatient admissions and self-reported surviving or witnessing an overdose. Opioid-related overdose death outcomes included age-adjusted opioid-related overdose death rates or death rates related to specific opioids, such as methadone or heroin. Included articles provided primary data analysis with an appropriate comparator for evaluating the policies using cross-sectional data, control groups, or pre–post designs.

Data Extraction and Synthesis

Two coders (B.A., K.M.T.) extracted study data, and other authors (E.S.R., E.G.M.) reviewed the data. We extracted data on location, data source, population covered, study design, treatment and control group characteristics, measures, key findings, and reported limitations. We categorized study locations into 4 US Census regions (Northeast, Midwest, South, and West), plus the New England Census division and Appalachia, which have high overdose rates.¹ Data sources were classified as survey data collected by investigators, administrative data (eg, PDMP and claims data), health surveillance systems (eg, vital statistics and national health monitoring surveys), and medical records.

We classified studies into 4 categories: (1) randomized experimental (investigators manipulated a treatment intervention, with randomized assignment of treatment and control conditions), (2) nonrandomized experimental (nonrandom assignment of treatment and control conditions), (3) quasi-experimental (longitudinal observational study of an intervention, with or without control groups), and (4) cross-sectional. Generally, evidence is strongest for randomized experimental studies, moderate for nonrandomized experimental or quasi-experimental designs, and weakest for cross-sectional studies.¹⁸ Our qualitative synthesis focused on whether studies on a given intervention provided consistent or contradictory results for each outcome and the reasons for contradictory results (eg, intervention design features, data sources, outcomes, or other factors). Highly variable outcome definitions precluded a meta-analysis.

Results

Study Screening

The search yielded 535 unique articles, 465 of which were removed after screening titles and abstracts and 6 of which were subsequently removed after full-text review, leaving 66 studies for inclusion. The flow diagram of the study selection process is available from the authors upon request.

Overview of Study Designs, Outcomes Assessed, and Study Characteristics

Quasi-experimental designs were most common overall (n = 56, 84.8% of all studies) and among all interventions except patient and/or health care provider education (Table 1). These studies leveraged states’ phased implementations to compare outcomes of intervention and control jurisdictions or evaluate interventions’ effectiveness by measuring changes in outcome trends before and after implementation. Experimental designs were used in 8 (12.1%) articles overall but were the most common design in studies evaluating educational interventions. Opioid use was the most common outcome assessed among all studies (n = 51, 77.3%), followed by opioid overdose (n = 16, 24.2%) and opioid-related overdose mortality (n = 14, 21.2%).

Table 1

Frequency of study designs and outcomes assessed among studies that examined the effectiveness of systems-level policies and interventions on opioid use, overdose, and deaths, United States and Canada (N = 66), 2014-2018^a

Intervention	Study Design				Opioid-Related Study Outcome^b
Intervention	Randomized Experiment, No.^b (n = 3)	Nonrandomized Experiment, No.(n = 5)	Quasi-experiment, No.(n = 56)	Cross-sectional, No.(n = 2)	Use, No.(n = 51)	Overdose, No.(n = 16)	Overdose Mortality, No.(n = 14)
PDMP^c (n = 39)	1	3	33	2	30	8	9
Health insurer and pharmacy benefit manager strategies^d (n = 8)	0	0	8	0	7	3	1
Pain clinic regulation^e (n = 10)	0	0	10	0	7	3	3
Clinical guidelines^f (n = 5)	0	1	4	0	4	1	1
Harm-reduction interventions^g (n = 7)	1	1	5	0	3	4	3
Patient and/or health care provider education^h (n = 6)	1	3	2	0	5	2	1
Medical or recreational marijuanaⁱ (n = 9)	0	0	9	0	5	1	4

Abbreviation: PDMP, prescription drug monitoring program.

^aStudies that include multiple interventions are itemized in multiple places.

^bExamples of opioid use outcomes include total number or volume of opioid prescriptions filled, per-capita or per-transaction dispensing of morphine milligram equivalents, and number of opioid-abusing patients. Examples of opioid-related overdose outcomes include overdose-related emergency department visits or inpatient admissions and self-reported surviving or witnessing an overdose. Examples of opioid-related overdose death outcomes include age-adjusted opioid-related overdose death rates or death rates related to specific opioids, such as methadone or heroin.

^cPDMPs are state programs that require pharmacies to submit information on any controlled substance prescriptions electronically to health departments or pharmacy boards. This information is used to identify health care provider overprescription or patient overuse of controlled substances.

^dHealth insurer and pharmacy benefit strategies include interventions by state Medicaid programs and commercial health insurers to limit some medications including opioids. For example, they use lock-in programs to limit the access of beneficiaries with a potential for overuse of opioids to a single prescriber and/or pharmacy for a period of time; they use prior authorization programs to allow reimbursement only with the health care provider’s confirmation of a patient’s need for medication; or they use drug-use review to identify beneficiaries with unsafe opioid use and contact their health care providers to reassess their medication.

^ePain clinic regulations, also known as “pill-mill” laws, regulate certification, ownership, and other procedures in pain clinics that prescribe opioids.

^fClinical guidelines provide instructions on dosing, duration, and type of medication (short-acting vs long-acting opioids).

^gHarm-reduction interventions include naloxone distribution and Good Samaritan laws. Naloxone distribution programs provide naloxone medication for overdose reversal to persons who use drugs, community members, and first responders. Good Samaritan laws provide immunity from investigation about possession of drugs when persons call 911 while having or witnessing an overdose.

^hPatient education interventions provide materials to patients about opioid use, adverse effects, and abuse. Health care provider education interventions provide materials about overprescription statistics and alternative options for pain treatment.

ⁱMedical marijuana regulations legalize the medical use of marijuana, which has been suggested as an alternative to opioids for pain control.

All outcomes were evaluated in each region, although the Midwest was slightly underrepresented among studies that evaluated overdose, and the South and West had higher coverage than the Northeast and Midwest among studies that reported mortality (Table 2). Half (n = 35, 53.0%) of the studies included states in Appalachia and/or New England. The most commonly used data source was administrative data (n = 45, 68.2% of all studies), followed by surveillance data (n = 17, 25.8%), medical records (n = 6, 9.1%), and survey data (n = 5, 7.6%). Among 14 studies reporting opioid-related overdose mortality, 13 articles used mortality surveillance data. Overall, only 12 (18.2%) studies examined multiple policies.

Table 2

Summary of geography, data source, and other characteristics of studies that examined the effectiveness of systems-level policies and interventions on opioid use (N = 66), United States and Canada, 2014-2018

Study Characteristics	Overall(n = 66)	Opioid-Related Study Outcome, No.^a
Study Characteristics	Overall(n = 66)	Use(n = 51)	Overdose(n = 16)	Overdose Mortality(n = 14)
US Census regions included^b
Northeast	35	27	9	7
Midwest	30	22	4	8
South	39	28	8	11
West	31	22	7	9
Appalachia/New England	35	27	9	7
Multistate census regions	35	27	9	7
Data source
Survey data^c	5	5	2	0
Administrative data^d	45	38	14	5
Surveillance data^e	17	5	4	13
Medical record review^f	6	6	0	0
Other features
Multiple policies examined	12	8	4	4

^aThe total number exceeds the number of studies reviewed because studies may be counted multiple times if they cover multiple geographic areas, use multiple data sources, or have multiple outcomes.

^bFor studies with treatment and control states, only the treatment states are classified here.

^cSurvey data include surveys developed by the investigators and population health monitoring surveys such as the National Survey on Drug Use and Health (NSDUH), National Health and Nutrition Examination Survey (NHANES), and Behavioral Risk Factor Surveillance System (BRFSS); the latter would be coded under “surveillance.”

^dAdministrative data include hospital discharge data, health care claims from public and private health insurers, treatment episode data, and prescription drug monitoring data.

^eSurveillance data include vital records, information from the Centers for Disease Control and Prevention WONDER (Wide-ranging Online Data for Epidemiologic Research) database, and population health monitoring surveys such as NSDUH, NHANES, and BRFSS.

^fMedical record reviews include analyses of electronic medical records. Surveys that are part of surveillance systems are classified as surveillance data.

Impact of Systems-Level Interventions on Opioid Use, Overdose, and Mortality

We report key findings from each study, by intervention type (Tables 3 -5). For each study, we report the population, type of data and data sources used, and key findings per outcome (ie, opioid use, overdose, and mortality outcomes).

Table 3

Findings on the effect of prescribing interventions on opioid use, overdose, and mortality from studies examining the effectiveness of systems-level policies and interventions on opioid use (n = 41), United States and Canada, 2014-2018^a

Authors	Study Population	Data Type, Data Sources, and Study Design	Opioid Use, Overdose, and Mortality Outcomes
Prescription Drug Monitoring Programs (PDMPs)
Ali et al ¹⁹	Residents in states with PDMPs in operation >1 year compared with residents in states without operational PDMPs; 50 US states, 2004-2014	Design: quasi-experimental Data type: surveillance Data source: Substance Abuse and Mental Health Services Administration (SAMHSA) National Survey on Drug Use and Health	Opioid use: no associations were found between PDMP implementation and past-year use, past-year dependence/abuse, and past-year initiation. Having an operational PDMP was associated with a 10-day reduction of nonmedical use of prescription pain relievers in past year (P < .05) and a 20-day reduction when PDMPs had a provision requiring mandatory health care provider enrollment and access (P < .01).
Bao et al²⁰	Adult patients who reported pain as their reason for visit to office-based physicians and resided in states with active PDMPs compared with residents in these states before PDMPs were implemented; 24 US states, 2001-2010	Design: quasi-experimental Data type: surveillance Data source: Centers for Disease Control and Prevention (CDC) National Ambulatory Medical Care Survey (NAMCS)	Opioid use: reduction in Schedule II opioid prescribing, from 5.5% to 3.7%; >30% reduction relative to rate of prescribing before PDMP implementation.
Brady et al²¹	Patients who filled prescriptions in US states with operational PDMPs compared with patients who filled prescriptions in US states without PDMPs; 50 states and the District of Columbia (DC), 1999-2008	Design: quasi-experimental Data type: administrative Data source: US Drug Enforcement Agency (DEA) Automation of Reports and Consolidated Orders System (ARCOS)	Opioid use: the implementation of PDMPs had no significant relationship with morphine milligram equivalents (MMEs) dispensed per capita (P = .68). The effect of PDMPs on MMEs dispensed per capita varied by state.
Branham²²	Patients admitted for heroin abuse treatment in 22 states that implemented operational PDMPs during 1992-2012 compared with patients in states that had not implemented PDMPs	Design: quasi-experimental Data type: administrative Data source: SAMHSA Treatment Episodes Data Set	Opioid use: post-PDMP, 13 states had a significant change in heroin-related hospital admissions (10 positive and 3 negative); 11 states had significant positive changes in prescription opioid-related hospital admissions.
Brown et al²³	New York State residents who filled prescriptions for opioids and New York State residents who were admitted for prescription opioid and heroin overdose; comparing residents in 2013 (post-PDMP implementation) with residents in 2010 (pre-PDMP implementation)	Design: quasi-experimental Data type: administrative Data sources: New York State’s ARCOS, New York Bureau of Narcotics Enforcement data on prescription fills, Statewide Planning and Research Cooperative System inpatient and emergency department (ED) discharge data	Opioid use: the number of prescriptions filled for opioids declined after PDMP implementation, although the supply chain data (ARCOS) showed a decline in the distribution of MMEs before implementation of a PDMP, with a significant increase after implementation. Opioid overdose: for prescription opioids, slopes before and after PDMP implementation were not significantly different (F = 1.04, P = .37). The rate of heroin overdoses sharply increased after PDMP implementation (F = 14.2, P < .001).
Buchmueller and Carey²⁴	5% random sample of Medicare Part D and fee-for-service beneficiaries, nationwide, comparing data on beneficiaries before and after implementation of PDMPs, 2007-2013	Design: quasi-experimental Data type: administrative Data sources: Centers for Medicare & Medicaid Services (CMS) Medicare Part D and fee-for-service Medicare prescription and medical claims	Opioid use: “must access” PDMPs (ie, PDMPs that require health care providers to access data under certain circumstances) were associated with a 2.4% decline in the share of Part D enrollees taking opioids. Opioid overdose: no significant findings on whether persons had an opioid poisoning incident.
Delcher et al²⁵	Florida residents after PDMP implementation in 2011 compared with residents before PDMP implementation and an external control group of New York City residents, 2003-2012	Design: quasi-experimental Data type: surveillance Data source: Florida Medical Examiners Commission drug-related deaths	Opioid mortality: oxycodone-caused mortality declined 25% the month after PDMP implementation (P = .01). Systemwide increase of 1 PDMP query per health care provider resulted in a significant decline in oxycodone-caused deaths (0.229 persons per month, P = .002).
Deyo et al²⁶	Physicians in Oregon who registered with the PDMP from December 2011 through February 2012 who had ever written an opioid prescription, compared with physicians who had not registered with the PDMP by October 2014	Design: quasi-experimental Data type: administrative, surveillance Data sources: Oregon PDMP, vital records, and statewide hospital discharge registry	Opioid use: the number of opioid units dispensed per capita decreased beginning in the first quarter of the PDMP, continuing through 3 years of operation (from 16.9 to 15.0 units per capita per quarter), and the total number of daily MMEs dispensed per capita decreased (from 2.8 to 2.4 MMEs per capita); declines were not different between clinicians who registered with the PDMP and those who did not. Opioid hospitalization and mortality: the number of prescription opioid-related hospitalizations and deaths was greater among patients of registered physicians than patients of nonregistered physicians at 6- and 30-month follow-up.
Gomes et al²⁷	Ontario residents who were eligible for public drug coverage (persons who were unemployed, persons who had a disability, residents of long-term care facilities, or persons aged ≥65), before and after implementation of a narcotic drug monitoring system in 2012	Design: quasi-experimental Data type: administrative, surveillance Data source: Ontario Public Drug Benefit Database	Opioid use: in the first 6 months after implementation of the drug monitoring system, the prevalence of potentially inappropriate opioid prescriptions decreased by 12.5% (P = .01), but no significant changes were observed thereafter (P = 0.8).
Haffajee et al²⁸	Enrollees of plans offered by a large national health insurer, aged 18-64, enrolled anytime during 2010-2014, and located in Kentucky, New Mexico, Tennessee, and New York compared with similar enrollees in Georgia, Missouri, New Jersey, and Texas	Design: quasi-experimental Data type: administrative Data source: Optum prescription claims for large commercial health insurance plan	Opioid use: in the quarter after PDMP implementation, the percentage of persons filling opioid prescriptions declined in Kentucky, New York, and Tennessee. Declines were sustained only in Tennessee. MME dose declined in Kentucky, New Mexico, and Tennessee.
Li et al²⁹	Residents in states with PDMPs compared with residents in states without PDMPs at that time; 50 US states and DC, 1999-2008	Design: quasi-experimental Data type: surveillance Data source: CDC National Vital Statistics System (NVSS)	Opioid mortality: overdose death rates in states with PDMPs were higher than in states without PDMPs (9.5 vs 8.3 per 100 000 population per year, P < .001). PDMP implementation was associated with a decreased risk of overdose mortality in 3 states, no effect in 11 states, and an increased risk of overdose mortality in 17 states.
Lin et al³⁰	Adult patients with a high risk for developing opioid dependence (eg, diagnosed with non-cancer chronic pain) in states with PDMPs compared with patients in states without PDMPs; 39 US states, 2012	Design: cross-sectional Data type: surveillance Data source: CDC NAMCS	Opioid use: no significant associations were found between states with PDMPs and likelihood of being prescribed opioid analgesics, when any pain medication had to be prescribed.
Maughan et al³¹	Persons visiting EDs before and after PDMP implementation in 11 US metropolitan areas	Design: quasi-experimental Data type: administrative Data source: Drug Abuse Warning Network	Opioid use: no association was found between PDMP implementation and the rate of ED visits involving opioid analgesic misuse, ED visits involving Schedule II opioids, or ED visits involving Schedule II opioid misuse.
McAllister et al³²	Adults who were directly discharged from the ED of an urban hospital in Florida during a 2-week period in February 2014 (hospital intervention period, whereby prescribers received summaries of their patients’ PDMP data), compared with adults who were discharged in December 2013	Design: quasi-experimental Data type: administrative Data source: Florida PDMP	Opioid use: no change was seen in average number of controlled substances prescribed per patient or number of patients prescribed opioids when PDMP data were provided to prescribers compared with control group.
McDonald et al³³	Patients whose controlled substance use history indicated possible misuse assigned to intervention group in which prescribers were sent PDMP history, compared with non-notification control group; Nevada	Design: experimental, randomized Data type: administrative Data source: Nevada PDMP	Opioid use: notified health care providers were 13% less likely than non-notified health care providers to continue prescribing to patients (hazard rate ratio = 0.87; 95% CI, 0.77-0.99). Patients of notified health care providers purchased 42% fewer opioid prescriptions than patients of non-notified health care providers (8.4 vs 14.5).
Moyo et al³⁴	5% national sample of Medicare beneficiaries, 2007-2012, among residents of states with PDMPs (Florida, Louisiana, Nebraska, New Jersey, Vermont) compared with states without PDMPs (Arkansas, Georgia, Maryland, New Hampshire, Wisconsin)	Design: quasi-experimental Data type: administrative Data source: CMS Medicare Part D Prescription Drug Event claims	Opioid use: PDMP implementation was associated with a 2.36 kg per month (95% CI, −3.44 to −1.28) reduction in total opioid volume compared with non-PDMP states. A strong moderation effect was found for daily MME dose or monthly number of opioid prescriptions. MME reductions were observed among persons with a disability and Medicare Advantage drug plan members, but no changes occurred among older adults and fee-for-service beneficiaries. The number of prescriptions increased among persons with a disability and decreased among Medicare Advantage beneficiaries.
Nam et al³⁵	Residents of states with PDMPs compared with residents in states without PDMPs; 35 US states and DC, 1999-2014	Design: quasi-experimental Data type: surveillance Data source: CDC NVSS	Overdose mortality: PDMP implementation had no effect on overall overdose mortality rates.
Norwood and Wright³⁶	Pharmacists in Indiana who held valid licenses to dispense controlled substances who indicated that having a PDMP influenced their dispensation practices, compared with pharmacists in Indiana who did not have valid licenses to dispense controlled substances, 2012	Design: cross-sectional Data type: survey Data source: Indiana Professional Licensing Agency’s web-based pharmacist prescribing pattern survey	Opioid use: 53% of pharmacists who reported a change in dispensation practices attributed the change to the PDMP. Pharmacists who reported using the PDMP were more likely than pharmacists who did not report using the PDMP to have refused dispensation of a controlled substance (odds ratio = 2.2; 95% CI, 1.34-3.69).
Patrick et al³⁷	Residents of states with PDMP legislation compared with residents of states without PDMP legislation; 50 US states and DC, 1999-2013	Design: quasi-experimental Data type: surveillance Data source: CDC WONDER (Wide-ranging Online Data for Epidemiologic Research)	Opioid mortality: PDMP implementation was associated with a decrease of 1.12 opioid-related overdose deaths per 100 000 population annually. Programs that monitored ≥4 drug schedules (from 5 schedules of the DEA’s Controlled Substance Schedule system) and updated their data at least weekly had greater reductions in opioid-related overdose deaths than programs with <4 drug categories included or with less frequent data updates.
Pauly et al³⁸	Nationally representative sample, residents of US states with PDMPs with various features compared with residents of states without PDMPS	Design: quasi-experimental Data type: administrative Data source: Truven Health MarketScan Commercial Claims and Encounters	Opioid overdose: states with PDMPs had lower increases in risk ratio of prescription drug overdoses over time than states without PDMPs. Estimated risk ratio of prescription drug overdoses in states with PDMPs that monitored only Schedule II drugs or Schedule II and III drugs was not different from risk ratios in states without PDMPs.
Ranapurwala et al³⁹	A comparison of privately insured Iowa residents before and after implementation of the PDMP in 2009; 2003-2014	Design: quasi-experimental Data type: administrative Data source: pharmaceutical and medical claims from unspecified large commercial health insurer	Opioid use: in 2014, the post-PDMP opioid pain reliever prescription rate (395 per 1000 person-years) was 28% lower than the predicted rate if the pre-PDMP trend continued (550 prescriptions per 1000 person-years). In 2014, the days’ supply (12.9 days) was 4% lower than the predicted rate under no PDMP implementation (13.4 days).
Rasubala et al⁴⁰	A comparison of patients in an urgent dental care center in Rochester, New York, before and after implementation of the PDMP, 2012-2014	Design: quasi-experimental Data type: medical records Data source: electronic medical records	Opioid use: among patients prescribed pain medications, 30.6% received opioid analgesics before the PDMP was implemented, 14.1% in the first year after implementation, and 9.6% in the second year after implementation. After implementation of the PDMP, the number of opioid prescriptions decreased significantly (P < .05).
Suffoletto et al⁴¹	Adult patients discharged with an opioid prescription from 15 EDs in the University of Pittsburgh Medical Center, July 2015–March 2017, before and after implementation of the PDMP	Design: quasi-experimental Data type: medical records Data source: electronic medical records	Opioid use: the total number of opioid prescriptions decreased by 19.5% from August 2016 (before PDMP) to September 2016 (after PDMP). No immediate decrease in the proportion of opioid prescriptions written for >12 tablets was seen after PDMP implementation.
Sun et al⁴²	Medicaid beneficiaries enrolled during January 2013–September 2015, after implementation of the PDMP in November 2014 compared with before implementation of the PDMP; Washington State	Design: quasi-experimental Data type: administrative Data source: Washington State Medicaid claims and PDMP data	Opioid use: PDMP queries were not associated with changes in the proportion of visits with opioid prescribing (5.8 per 1000 encounters; 95% CI, 0.1 to 11.8) or in the amount of MMEs prescribed (2.66; 95% CI, 0.15 to 5.48).
Wen et al⁴³	Medicaid enrollees in states exposed to mandatory PDMP compared with Medicaid enrollees in states without a mandatory PDMP, 2011-2014; 46 US states	Design: quasi-experimental Data type: administrative Data source: CMS Medicaid state drug use data	Opioid use: the average number of Schedule II opioid prescriptions per quarter per 100 Medicaid enrollees was 15.3 in states without any mandatory PDMP, compared with 13.9 in states with a PDMP that requires the use of registration of the PDMP. The difference in the number of Schedule III opioid prescriptions between states with and without mandates was minimal and not significant.
Winstanley et al⁴⁴	Patients in Ohio who received all medications from Ohio-based health care providers, November 2014–March 2017, before and after PDMP implementation	Design: quasi-experimental Data type: administrative Data source: Ohio PDMP	Opioid use: after implementation of the PDMP, the monthly quantity of opioids dispensed decreased significantly; the mean days’ supply of opioids increased, but there was no change in the MME.
Yarbrough⁴⁵	Medicare Part D enrollees, 2010-2013, before and after PDMP implementation; nationwide	Design: quasi-experimental Data type: administrative Data source: ProPublica and CMS Medicare Part D physician prescribing data	Opioid use: all opioids as well as oxycodone and hydrocodone showed limited decreases in days’ supply prescribed after implementation of a PDMP.
Young et al⁴⁶	Massachusetts residents who obtained Schedule II prescriptions from ≥4 health care providers or pharmacies during a 6-month period, unsolicited PDMP information reported compared with propensity-matched control group whose health care providers were not sent a report	Design: experimental, randomized Data type: administrative Data source: Massachusetts PDMP	Opioid use: intervention had significant effect on the number of prescriptions, number of prescribers, number of pharmacies, dosage unit days’ supply, total MMEs, and average daily MMEs for persons with average daily dosages <100 MME and no effect on these outcomes for persons with average daily dosages ≥100 MME.
Health Insurer and Pharmacy Benefit Manager Strategies
Cochran et al⁴⁷	Pennsylvania Medicaid beneficiaries who were aged 18-64 with opioid prescriptions, had no previous cancer treatment or addiction, were not in long-term care, were not receiving hospice services, and were in a high prior authorization policy (wherein payers require confirmation of the necessity of medication for 17-74 opioids) compared with beneficiaries in a no/low prior authorization policy (wherein payers require confirmation of the necessity of medication for 0 or 1 opioid)	Design: quasi-experimental Data type: administrative Data source: Pennsylvania Medicaid prescription/medical claims	Opioid use: persons in high prior authorization plans were 11% less likely to develop opioid medication abuse than persons in plans with no prior authorization (95% CI, 0.85 to 0.93; P < .001). Low prior authorization plans were associated with a 7% lower rate of developing opioid medication abuse compared with no prior authorization plans (95% CI, 0.87 to 1.00; P = .04). Opioid overdose: low prior authorization plans were associated with 25% lower rates of overdosing (95% CI, 0.59 to 0.95; P = .02) compared with no prior authorization plans. A nonsignificant 12% reduction in overdose was observed for enrollees in high prior authorization plans compared with enrollees not in prior authorization plans.
Daubresse et al⁴⁸	Pharmaceutical benefits plan members who were aged ≥16 and had >12 controlled substance claims during a 3-month period in 2 Midwestern states where a drug use review program was conducted (in which health care providers were contacted about beneficiaries with unsafe opioid use) compared with a similar group in 2 similar states without a drug use review program	Design: quasi-experimental Data type: administrative Data source: pharmaceutical claims from unspecified large pharmacy benefits manager	Opioid use: reductions in the mean controlled substance score were higher among members in the plan with drug use review than among members in the plan without drug use review (5.7 vs 4.3; P = .01). Changes were driven by reductions in the number of controlled substance claims filled (30.5% vs 23.1%; P = .01).
Faul et al⁴⁹	Medicaid enrollees in Florida (which lists methadone as a preferred drug) and North Carolina (which gives methadone a preferred status without listing it) compared with Medicaid enrollees in South Carolina (which did not include methadone as a preferred drug); 2007-2014	Design: quasi-experimental Data type: surveillance, administrative Data source: CDC NVSS, Truven Health MarketScan Commercial Claims and Encounters database, Medicaid multistate databases, Agency for Healthcare Research and Quality State Emergency Department Databases, and State Inpatient Database	Opioid overdose: rates of fatal and nonfatal methadone overdose per 100 000 population were significantly higher in Florida (1.75; 95% CI, 1.57 to 1.94) and North Carolina (1.67; 95% CI, 1.35 to 1.98) than in South Carolina (0.81; 95% CI, 0.65 to 0.96).
Hartung et al⁵⁰	Fee-for-service Medicaid beneficiaries, January 2011 and December 2013, with ≥1 opioid prescription in Oregon, which implemented a prior authorization policy for opioid prescriptions >120 MMEs, compared with Colorado, which did not implement this policy	Design: quasi-experimental Data type: administrative Data sources: Oregon and Colorado Medicaid fee-for-service administrative claims and enrollment data	Opioid use: among Oregon beneficiaries with high-dose opioid use before prior authorization, the probability of having a high-dose fill after implementation declined (−20.3%; 95% CI, −15.3% to −25.3%) compared with changes in Colorado. Opioid overdose: no significant changes in ED encounters or hospitalizations for opioid overdose in Oregon compared with Colorado.
Naumann et al⁵¹	Independently living North Carolina Medicaid beneficiaries aged 18-64 before and after the lock-in program’s implementation, whereby the access for beneficiaries with potential overuse of opioids was limited to a single prescriber and/or pharmacy; October 2010–September 2012	Design: quasi-experimental Data type: administrative Data source: North Carolina PDMP and Medicaid claims	Opioid use: compared with the pre–lock-in program, after the lock-in program, the number of opioids dispensed decreased, but the average daily MMEs of opioids dispensed was higher during and after the lock-in program (daily mean difference per person: 18.7; 95% CI, 13.9 to 23.6 and 11.1; 95% CI, 5.1 to 17.1, respectively).
Qureshi et al⁵²	Health care providers serving commercially insured members of Midwest Blue Cross Blue Shield, post–drug use review, whereby health care providers were contacted about beneficiaries with unsafe opioid use, compared with pre–drug use review	Design: quasi-experimental Data type: administrative Data source: pharmacy claims	Opioid use: a 28.1% reduction in unsafe opioid combination therapy was seen after policy implementation compared with before policy implementation. When the prescribers were surveyed, 70.6% of health care providers said no change to pain management therapy was made, but 8.5% indicated that the opioid dose was decreased (response rate was 23.5%).
Riggs et al⁵³	Adult Medicaid beneficiaries enrolled in Kaiser Permanente Colorado health insurance plan who purchased at least 1 short-acting opioid, comparing beneficiaries pre-implementation of a prior authorization program for short-acting opioids (May 2014–July 2014) with beneficiaries post-implementation (January 2015–March 2015)	Design: quasi-experimental Data type: administrative Data source: electronic medical/pharmacy records	Opioid use: total daily MME doses decreased from 6.8 mg (interquartile range [IQR], 2.2-25.8) in the pre-policy period to 6.6 mg (IQR, 1.7-24.0) in the post-policy period (P = .03). The proportion of patients purchasing >120 mg dose per day and the proportion of patients purchasing long-acting opioids did not change.
Skinner et al⁵⁴	North Carolina Medicaid beneficiaries aged 18-65, comparing those before and after implementation of a lock-in program, whereby the access for beneficiaries with potential overuse of opioids was limited to a single prescriber and/or pharmacy; October 2008–June 2013	Design: quasi-experimental Data type: administrative Data source: North Carolina Medicaid prescription drug claims	Opioid use: odds of having monthly opioid claims were 84% lower during Medicaid lock-in enrollment than before the lock-in program was implemented (adjusted odds ratio = 0.16; 95% CI, 0.15 to 0.16). The program was associated with 1.13 (95% CI, −1.16 to −1.11) fewer opioid prescriptions per member per month.
Clinical Guidelines
del Portal et al⁵⁵	Adult patients visiting 2 large urban ED sites in Philadelphia, Pennsylvania, discharged with certain conditions; 6 and 12-18 months after a guideline for opioid prescription was implemented, compared with 6-12 months before the guideline was implemented. The guideline had 11 instructions, including limiting the amount of prescription to 7 days until follow-up, restricting the prescription of long-acting opioids such as oxycontin, and refusing to prescribe opioids for a condition previously treated by another physician.	Design: quasi-experimental Data type: medical records Data source: electronic medical records	Opioid use: a significant (P < .001) sustained decrease was seen in opioid prescription rates, from 52.7% before the guideline was implemented to 29.8% immediately after implementation and 33.8% at 12-18 months’ follow-up.
Sullivan et al⁵⁶	Patients aged 18-64 with claim for opioid prescription in Washington State Medicaid fee-for-service system, post-guideline compared with pre-guideline; April 2006–December 2010. The guideline included instructions for safe prescription of opioids for chronic non-cancer patients, including limiting the opioid prescription dosing to 120 mg MME per day for patients whose conditions did not improve.	Design: quasi-experimental Data type: administrative Data source: Washington State Medicaid fee-for-service prescription claims	Opioid use: decreases were observed in doses corresponding to the 75th (ß = −0.44; 95% CI, −0.50 to −0.37); 90th (ß = −1.82; 95% CI, −2.14 to −1.50); 95th (B = −4.29; 95% CI, −5.37 to −3.22); and 99th (ß = −25.40; 95% CI, −31.39 to −19.41) percentiles. A significant reduction was seen in the percentage of patients receiving high daily doses (P < .001).
Pain Clinic Regulations
Kennedy-Hendricks et al⁵⁷	Florida residents, before and after implementation of pain clinic regulations, compared with North Carolina residents without pain clinic regulations; 2003-2012. The Florida pain clinic regulations prohibited the distribution of most narcotic medications and increased the penalty for physicians involved in such activities.	Design: quasi-experimental Data type: surveillance Data sources: Florida and North Carolina vital statistics mortality data	Opioid mortality: implementation of pain clinic regulations was estimated to be associated with 1034 fewer opioid-related overdose deaths, with minimal change in 2010 (−0.3% difference, not significant), 1.5 (95% CI, −1.9 to −1.1) fewer overdose deaths per 100 000 population in 2011, and 3.9 (95% CI, −5.1 to −2.8) fewer deaths per 100 000 population in 2012.
Lyapustina et al⁵⁸	Texas patients with any prescription claims before and after implementation of pain clinic regulations in September 2010; September 2009–September 2011. Legislation requires pain management clinics to be certified by the state medical board and owned by a licensed physician who must review at least one-third of patient records and verify all employees’ licensures and qualifications.	Design: quasi-experimental Data type: administrative Data source: IMS LifeLink Longitudinal Prescription (LRx) multi-payer prescription claims data	Opioid use: laws were associated with declines in average MME dose per transaction (−0.57 mg/month; 95% CI, −1.09 to −0.06), monthly opioid volume (−9.99 kg/month decline from 181 kg; 95% CI, −12.86 to −7.11), monthly number of opioid prescriptions (−12 200 prescriptions/month, from 233 000; 95% CI, −15 300 to −9150), and monthly quantity of opioid pills dispensed (−714 000 pills/month, from 15 million; 95% CI, −877 000 to −550,000).
Raji et al⁵⁹	Texas Medicare beneficiaries aged ≥66 who were not in a health maintenance organization and did not have cancer, before and after implementation of pain clinic regulations; 2007-2012. Legislation requires pain management clinics to be certified by the state medical board and owned by a licensed physician who must review at least one-third of patient records and verify all employees’ licensures and qualifications.	Design: quasi-experimental Data type: administrative Data source: CMS Medicare Part D Prescription Drug Event claims and outpatient claims	Opioid use: laws were associated with a temporary decline in the monthly percentage of patients who filled ≥1 Schedule II or Schedule III opioid prescription. Opioid overdose: no association was found between acute hospitalization rate for opioid toxicity and laws regulating clinics.

^aPolicy types are explained in detail in Table 1.

Table 4

Findings on the effect of harm-reduction and educational interventions on opioid use, overdose, and mortality from studies examining the effectiveness of systems-level policies and interventions on opioid use (n = 13), United States and Canada, 2014-2018^a

Authors	Study Population	Data Sources and Study Design	Opioid Use, Overdose, and Mortality Outcomes
Medical and Recreational Marijuana Legislation
Bradford et al⁶⁰	Medicaid enrollees in stand-alone Part D plans or with prescription coverage under Medicare Advantage Prescription Drug plan in states with medical cannabis laws, 2010-2015, compared with enrollees in states without these laws	Design: quasi-experimental Data type: administrative Data source: Centers for Medicare & Medicaid Services (CMS) Medicare Part D Prescription Drug Event claims	Opioid use: states with active dispensaries saw 3.74 million (95% CI: −6.29 million to −1.19 million) fewer daily doses filled, whereas states with home cultivation only saw 1.79 million (95% CI, −3.53 million to −0.05 million) fewer daily doses filled. Secondary analyses of the association between marijuana legislations separated by class of opioid show that significant associations of active dispensaries were observed only for hydrocodone and morphine, and a significant association of cultivation-only laws was observed only for hydrocodone.
Liang et al⁶¹	Medicaid enrollees in fee-for-service programs from 1993 to 2014 who were not enrolled in managed care in states with medical cannabis legislation compared with Medicaid enrollees in states without such legislation; 50 US states and the District of Columbia (DC)	Design: quasi-experimental Data type: administrative Data source: CMS Medicaid State Drug Utilization Data	Opioid use: medical cannabis legalization was associated with a 29.6% reduction in the number of opioid prescriptions, a 29.9% reduction in dosage, and a 28.8% reduction in Medicaid spending for Schedule III opioids. No effect was observed on Schedule II opioid prescriptions.
Livingston et al⁶²	A comparison of Colorado residents before and after implementation of laws legalizing recreational marijuana, 2000-2015	Design: quasi-experimental Data type: administrative Data source: Centers for Disease Control and Prevention (CDC) WONDER (Wide-ranging Online Data for Epidemiologic Research)	Opioid mortality: opioid-related deaths decreased by 0.7 deaths per month relative to the baseline period (ß = −0.68; 95% CI, −1.34 to −0.03), an estimated 6.5% reduction overall in opioid-related deaths.
Vigil et al⁶³	Patients with chronic back pain enrolled in a medical cannabis program in a single-physician rehabilitation clinic in Albuquerque, New Mexico, compared with patients with chronic back pain who were not enrolled, 2010-2015	Design: quasi-experimental Data type: administrative Data source: New Mexico Prescription Drug Monitoring Program (PDMP)	Opioid use: medical cannabis program participants were more likely than nonparticipants to reduce their daily opioid prescription doses (odds ratio = 5.12; 95% CI, 1.56 to 16.88; P < .01).
Wen and Hockenberry⁶⁴	Medicaid fee-for-service and managed care enrollees residing in states with cannabis legislation compared with Medicaid fee-for-service and managed care enrollees residing in states without such legislation; 50 US states and DC, 2011-2016	Design: quasi-experimental Data type: administrative Data source: CMS Medicaid State Drug Utilization Data	Opioid use: compared with states without any medical or recreational marijuana law, states with medical marijuana laws had a 5.88% (95% CI, −11.55% to −0.21%) lower rate of opioid prescribing, which was concentrated in Schedule III opioids. Recreational laws (in states with existing medical marijuana laws) were associated with a −6.38% (95% CI, −12.20% to −0.56%) prescribing rate, which was concentrated in Schedule II opioids.
Harm-Reduction Interventions
Banta-Green et al⁶⁵	Seattle, Washington, emergency department (ED) patients at elevated risk of opioid overdose, aged 18-70, randomized to receive overdose education, behavioral change counseling, and naloxone kit, compared with ED patients not receiving the intervention	Design: experimental, randomized Data type: surveillance, administrative Data sources: hospital encounter and billing data; Washington State Department of Health Comprehensive Hospital Abstract Reporting System; Washington State death certificate data	Opioid overdose: during the follow-up period, 24% of all participants had ≥1 overdose event. The study found no significant difference in the number of overdose incidents or in the time to first overdose event between ED patients who received the naloxone kit and overdose education and ED patients who did not receive the intervention.
Doe-Simkins et al⁶⁶	Massachusetts participants trained in overdose education and naloxone distribution compared with untrained participants who rescued an overdose event, 2006-2010	Design: quasi-experimental Data type: administrative Data source: participant questionnaires (refill-rescue questionnaires after overdose rescues and enrollment questionnaires)	Opioid use: no significant overall change in the number of days using heroin in past 30 days between trained and untrained participants who reported using naloxone kits (P = .52). Opioid overdose: no significant differences in overdose event characteristics or actions taken during overdose in which rescuer was previously trained compared with overdose events in which rescuer was not trained.
Dwyer et al⁶⁷	Massachusetts ED patients receiving overdose education and naloxone distribution compared with ED patients receiving	Data source: telephone survey on knowledge, experiences, and behaviors	Opioid use: no significant differences were seen in reported opioid use between patients who received overdose education and naloxone and patients who received overdose education only. Opioid overdose: no significant differences in rate of experiencing or witnessing an overdose were seen among patients who received overdose education and naloxone and patients who received overdose education only.
McClellan et al⁶⁸	Residents of states with naloxone distribution laws or Good Samaritan laws compared with residents of states without such	Data sources: CDC National Vital Statistics System, Substance Abuse and Mental Health Services Administration National Survey on Drug Use and Health	Opioid use: no relationships were observed between examined laws and nonmedical opioid use. Opioid mortality: states with naloxone access laws or Good Samaritan laws had a 14% (P = .03) and 15% (P = .05) lower incidence of opioid-related overdose mortality, respectively.
Nguyen and Parker⁶⁹	Patients visiting EDs in New York State for opioid overdoses before and after implementation of the 2011 Good Samaritan laws compared with patients visiting EDs in New Jersey, where no policy was in place, 2010-2012	Design: quasi-experimental Data type: administrative Data sources: Agency for Healthcare Research and Quality State Emergency Department Databases and State Inpatient Databases	Opioid overdose: rates of ED visits and inpatient admissions for heroin overdoses in New York State increased compared with changes in New Jersey (incidence rate ratio [IRR] = 1.34; 95% CI, 1.00 to 1.86) after law implementation. Results were inconclusive for non-heroin opioid-related overdoses (IRR = 0.98; 95% CI, 0.86 to 1.13).
Rando et al⁷⁰	Deceased persons in Lorain County, Ohio, after (October 2013–October 2014) and before (January 2011–September 2013) police naloxone training and distribution	Design: quasi-experimental Data type: surveillance, administrative Data sources: Lorain County, Ohio, coroner office’s opioid-related deaths; Lorain County, Ohio, Alcohol and Drug Addiction Services Board’s naloxone administration events	Opioid mortality: after the intervention was delivered, the number of opioid-related overdose deaths decreased, with an overall average of 13.4 fewer deaths per quarter year.
Health Care Provider and Patient Education
Hill et al⁷¹	New Hampshire patients who underwent 5 common outpatient surgeries at Dartmouth-Hitchcock Medical Center during an educational intervention (June–September 2016) compared with the pre-intervention period (2015). The intervention included presentations, meetings, and emails for prescribers on opioid prescription topics such as dosing recommendations for procedures and encouraging the use of non-opioid pain medications.	Design: experimental, nonrandomized Data type: survey data, medical records Data sources: patient surveys on pill use, electronic medical records	Opioid use: the mean number of opioids prescribed per operation decreased significantly during the intervention period compared with the pre-intervention period (P < .001). The total number of pills prescribed decreased by 53% during the intervention period compared with the number of pills that would have been prescribed in the absence of the intervention.
Syed et al⁷²	Eligible patients undergoing primary arthroscopic rotator cuff repair at Rothman Institute in Manhattan who received opioid preoperative education and standard care compared with patients who received standard care only	Design: experimental, randomized Data type: survey Data source: repeated patient surveys	Opioid use: compared with 3-month changes among patients who received standard care only, patients who received preoperative education and standard care consumed 42% fewer narcotics (P = .01) and were 2.2 times more likely to discontinue narcotic use (odds ratio [OR] = 2.19; P = .03). Patients with a history of preoperative narcotic use who received preoperative education and standard care were 6.8 times more likely to discontinue narcotics by the end of follow-up (OR = 6.8; P = .01) than patients who received standard care only.

^aSee Table 1 footnotes for a detailed explanation of each policy type.

Table 5

Findings on the effect of implementing multiple concurrent interventions on opioid use, overdose, and mortality from studies that examined the effectiveness of systems-level policies and interventions on opioid use (n = 12), United States and Canada, 2014-2018^a

Authors	Study Population	Data Sources and Study Design	Opioid Use, Overdose, and Mortality Outcomes
Prescription Drug Monitoring Program (PDMP) Combined With Pain Clinic Regulations (Pill-Mill Laws)
Chang et al⁷³	Opioid prescribers in Florida, with pill-mill and PDMP laws, compared with opioid prescribers in Georgia, without such legislation, 2010-2012. High-risk prescribers were prescribers in the top fifth percentile of prescribed opioid volume during 4 consecutive quarters. Low-risk prescribers were those who did not qualify for the high-risk category.	Design: quasi-experimental Data type: administrative Data source: IMS LifeLink Longitudinal Prescription (LRx) multi-payer prescription claims data	Opioid use: compared with high-risk prescribers in Georgia, high-risk prescribers in Florida had large relative reductions in monthly numbers of opioid patients (−536 patients/month; 95% CI, −829 to −243) and monthly opioid prescriptions (−847 prescriptions/month; 95% CI, −1498 to −197); monthly morphine milligram equivalent (MME) dose (0.88 mg/month; 95% CI, −1.13 to −0.62); and total monthly prescribed opioid volume (−3.88 kg/month; 95% CI, −5.14 to −2.62). Low-risk prescribers did not have significant reductions.
Chang et al⁷⁴	A comparison of high-risk patients in Florida, which has pill-mill and PDMP laws, and high-risk patients in Georgia, which did not have such legislation, 2010-2012. High-risk patients included concomitant users (filling >30 days of concomitant opioids and benzodiazepines per year), chronic users (consuming >100 MMEs per day for >90 consecutive days), and opioid shoppers (visiting >3 pharmacies and >3 prescribers to acquire opioids). Low-risk patients are those who did not qualify for the high-risk category.	Design: quasi-experimental Data type: administrative Data source: IMS LRx multi-payer prescription claims data	Opioid use: compared with high-risk patients in Georgia, high-risk patients in Florida had relative reductions in monthly MME dose (opioid shoppers: −1.08 mg/month; 95% CI, −1.62 to −0.54), total monthly prescribed opioid volume (chronic users: −4.58 kg/month; 95% CI, −5.41 to 3.76), and monthly number of dispensed prescriptions (concomitant users: −640 prescriptions/month; 95% CI, −950 to −340). Low-risk patients did not have significant reductions.
Rutkow et al⁷⁵	A comparison of a closed cohort of patients, prescribers, and pharmacies in Florida, which had pill-mill and PDMP laws, and Georgia, which did not have such legislation, 2010-2012.	Design: quasi-experimental Data type: administrative Data source: IMS LRx multi-payer prescription claims data	Opioid use: Florida laws were associated with significant relative declines in opioid volume (2.5 kg/month, P < .05) and MMEs per transaction (0.45 mg/month, P < .05). After 1 year, policies were associated with a 1.4% relative decrease in prescriptions, a 2.5% relative decrease in volume, and a 5.6% relative decrease in MMEs per transaction. Reductions were limited to prescribers and patients with highest baseline prescribing and use.
Health Care Provider Education Combined With Patient Histories or Clinical Guidelines
Pasquale et al⁷⁶	Kentucky physicians linked to Medicare opioid patients at high risk of opioid abuse in 4 intervention arms: (1) physicians receive only patients’ opioid prescription history from any prescriber, (2) physicians receive only educational materials about opioids and safe prescribing behaviors, (3) physicians receive both educational materials and patients’ history, and (4) physicians receive neither the educational materials nor the patients history; 2012-2014.	Design: experimental, nonrandomized Data type: administrative Data source: Humana Research Database of multi-payer medical and prescription claims	Opioid use: none of the intervention arms had a significant effect on the number of opioid or pain medication prescriptions filled, chronic high-dose opioid use, uncoordinated opioid use, emergency department (ED) visits, or rates of diagnosed opioid abuse.
Stanek et al⁷⁷	Prescribing habits of hand surgeons in an urban Midwest hospital who received clinical dosing guidelines for opioid prescribing after surgery procedures and a “memory guide reminder” (including a multimodal pain management plan for before, during, and after operations and opioid recommendations) educational assistance device compared with their prescribing habits pre-intervention.	Design: quasi-experimental Data type: medical records Data source: electronic health records	Opioid use: the average postoperative opioid prescription size decreased for all 4 case types (release of the first dorsal compartment, primary excision of wrist ganglion, open reduction internal fixation of a metacarpal fracture, and trigger finger release) by 15% to 48%; prescription size decreased significantly for 2 procedures (primary excision of a wrist ganglion [P = .02] and open reduction internal fixation of a metacarpal fracture [P = .04]).
Legalizing Medical Marijuana Combined With PDMP and Pain Clinic Regulations
Bachhuber et al⁷⁸	Residents in states with medical cannabis laws compared with residents in states without cannabis law; 50 US states, 1999-2010.	Design: quasi-experimental Data type: surveillance Data source: Centers for Disease Control and Prevention (CDC) Wide-ranging Online Data for Epidemiologic Research (WONDER)	Opioid mortality: states with medical cannabis laws had a 24.8% lower mean annual opioid-related overdose mortality rate (95% CI, −37.5% to −9.5%; P = .003) compared with states that had no medical cannabis laws. The relationship between medical cannabis laws and opioid-related overdose mortality strengthened yearly. No detected association was found between the other policies and opioid-related overdose mortality rates.
Pardo⁷⁹	Residents of states with robust PDMPs, during 1999-2014, scored on an ordinal “robustness” scale based on multiple dimensions: monitoring more than Schedule II drugs; permitting or requiring prescribers to use PDMPs to identify suspicious activity, dispensing, or purchasing; access for law enforcement, prosecutors, and a larger range of health care professionals; requiring prescribers’ use of PDMPs before prescribing; permitting PDMP to share data with other states; requiring program evaluation of PDMP; having an oversight board; data retention; and the funding mechanism.	Design: quasi-experimental Data type: surveillance Data source: CDC WONDER	Opioid mortality: a 1-point increase in the PDMP robustness score was associated with a 1% reduction in the overdose death rate and about 200 additional lives saved in 2014. Good Samaritan laws, naloxone distribution programs, and pain clinic regulations did not have a significant effect on opioid-related mortality.
Phillips and Gazmararian⁸⁰	Residents of states with PDMPs and/or medical cannabis laws compared with residents of states that did not have such laws; 50 US states, 2011-2014.	Design: quasi-experimental Data type: administrative Data source: CDC WONDER	Opioid-related mortality: medical cannabis laws and PDMPs were associated with a 21.7% and 11.4% increase in mean opioid-related mortality, respectively (P < .001). Interaction between both types of legislation produced a non-significant decrease of 10.1% (P = .06). These effects persisted after accounting for the time since enactment.
Shi⁸¹	Residents of states with medical marijuana laws during 1997-2014 compared with the period before medical marijuana laws were implemented; 27 US states.	Design: quasi-experimental Data type: administrative Data source: Agency for Healthcare Research and Quality State Inpatient Databases	Opioid use: medical marijuana policies were associated with a 23% reduction in hospitalizations related to opioid dependence or abuse (P = .01). Reductions were most salient after 1 year of policy implementation (9.4%, P = .03). Opioid overdose: medical marijuana policies were associated with a 13% reduction in hospitalizations related to overdose (P = .03), most notably after the third year of policy implementation (12%, P = .01). No association was found between the PDMP and pain management clinic regulations and hospitalization outcomes.
Other Concurrently Implemented Interventions
Alexandridis et al⁸²	Residents of 74 North Carolina counties implementing a 7-strategy intervention to prevent opioid-related overdose (diversion control, community education, health care provider education, naloxone policies, support for patients with pain, hospital ED policies, and addiction treatment), compared with North Carolina residents outside those counties.	Design: experimental, nonrandomized Data type: surveillance, administrative Data sources: North Carolina vital statistics data and Disease Event Tracking and Epidemiologic Collection Tool ED discharge data	Opioid-related overdose: compared with residents in counties with the intervention, health care provider education was associated with limited changes in overdose-related ED visits, and policies to limit ED opioid dispensing were associated with more ED visits (incidence rate ratio [IRR] = 1.06; 95% CI, 1.01 to 1.12). Medication-assisted therapy expansion was associated with fewer ED visits in counties with the intervention vs those without the intervention. Opioid-related mortality: compared with residents in counties without the intervention, health care provider education was associated with lower overdose mortality (IRR = 0.91; 95% CI, 0.8 to 1.02), and policies to limit ED opioid dispensing were associated with lower mortality (IRR = 0.97; 95% CI, 0.87 to 1.07). Expansions of medication-assisted therapy were associated with increased mortality (IRR = 1.22; 95% CI, 1.08 to 1.37) in counties with the intervention compared with counties that did not have the intervention.
Losby et al⁸³	Large sample (N = 3.2 million) of adult Kaiser Permanente Southern California members, excluding cancer, hospice, and palliative care patients, after implementation of a multipronged intervention (including prescribing and dispensing limits, health care provider education, and providing prescriber-specific performance feedback to health care providers), compared with before implementation of the intervention, 2010-2015.	Design: quasi-experimental Data type: administrative, medical records Data sources: pharmacy claims and electronic health records	Opioid use: a 30% reduction in prescribing opioids at high doses, a 98% reduction in the number of prescriptions with quantities >200 pills, a 90% reduction in the combined prescribing of opioids with benzodiazepines and carisoprodol, a 72% reduction in the prescribing of long-acting/extended-release opioids, and a 95% reduction in the prescribing of brand-name opioid-acetaminophen products occurred after the intervention.
Meara et al⁸⁴	Medicare beneficiaries aged <65 with disabilities in states with controlled substance laws (including PDMP, pain clinic regulations, prescription limits, tamper-resistant prescriptions, patient identification, doctor-shopping restrictions, and/or pharmacist verification), compared with Medicare beneficiaries aged <65 with disabilities in states without these laws, 2006-2012.	Design: quasi-experimental Data type: administrative Data source: Centers for Medicare & Medicaid Services Medicare fee-for-service prescription and medical claims	Opioid use: no significant associations were found between opioid outcomes and types of laws or the number or types of laws enacted. Opioid-related overdose: associations between types of laws and rates of nonfatal prescription opioid-related overdose were not significant for any types of laws.

^aSee Table 1 for detailed explanations of each policy type.

PDMPs

We found little consensus on the effectiveness of PDMPs. Although several studies found no effect on opioid prescribing and use after PDMP implementation^{21,26,30–32,32} opioid prescription rates declined in Iowa³⁹ and pharmacists’ self-reported dispensing decreased in Indiana.³⁶ One study reported a 12.5% reduction in potentially inappropriate opioid prescriptions to Ontario residents with publicly funded drug insurance after implementation of a narcotic drug monitoring program.²⁷ Several multistate studies showed reductions in opioid prescriptions after PDMP implementation, including a 30% reduction in the self-reported rate of Schedule II prescriptions (opioids with high potential for abuse, such as hydromorphone) among patients reporting pain as a reason for visit²⁰ a limited decline in days’ supply of opioids prescribed among Medicare Part D enrollees,⁴⁵ and a reduction in the amount of opioid pills prescribed among persons with disabilities and older adult Medicare enrollees.³⁴

PDMP effects were stronger in regions where health care provider participation was mandatory, with varying effect sizes. Compared with pre-PDMP implementation, mandatory PDMP registration and use in Ohio were associated with fewer pills dispensed but not with reduced MME per prescription.⁴⁴ In Pittsburgh, Pennsylvania, mandatory PDMP access was associated with fewer patients prescribed opioids in the emergency department than pre-PDMP implementation⁴¹ and mandatory PDMP access in New York State was associated with reduced prescription rates and quantity of pills prescribed by dentists.⁴⁰ States with mandatory PDMPs (vs states without mandatory PDMPs) reported reduced volume of opioid prescriptions filled among Medicaid enrollees⁴³ and reduced opioid misuse among Medicare Part D beneficiaries.²⁴

Unsolicited or automated reporting of PDMP data led to a significant reduction in the number of prescriptions and total MME prescribed in Massachusetts⁴⁶ yet it had no effect on the same measures in Nevada³³ or on the proportion of emergency department visits resulting in opioid prescriptions or prescribed MME in Washington State.⁴² These differences may have been the result of state PDMP characteristics, such as different methods of automated reporting and/or analytic designs (eg, the study of unsolicited PDMP reporting in Nevada used propensity score matching³³). PDMP effectiveness may change over time: a 4-state study found an immediate decline in the percentage of persons filling opioid prescriptions after implementation of robust PDMPs that mandate access, monitor multiple drug schedules, and update data regularly, but only 1 state saw a sustained effect.²⁸

Some evidence suggests that PDMPs result in increased heroin use, presumably because opioid users will substitute heroin for opioids. Compared with residents of states without active PDMPs, residents of states with active PDMPs self-reported fewer days of nonmedical prescription pain medication use but more days of past-year heroin use.¹⁹ Another study found higher treatment admissions for heroin and prescription opioids in 22 states after PDMP implementation (vs pre-PDMP implementation).²²

Findings on PDMPs’ influences on overdose and mortality rates were mixed. PDMPs were not associated with changes in nonfatal overdose rates among Medicare Part D enrollees nationwide,²⁴ but opioid- and heroin-related mortality in New York State increased after PDMP implementation.²³ In Florida, PDMP implementation was associated with decreased oxycodone-related overdose mortality²⁵; in Oregon, a higher number of overdoses occurred among patients of PDMP-registered physicians than among patients of clinicians who did not register for PDMPs.²⁶ Although 1 study found no effect of PDMPs on the rate of fatal drug overdoses in the US general population³⁵ PDMPs were associated with an increased risk of fatal overdose in 17 states.²⁹ Robust PDMP policies were associated with greater overdose mortality protection among states with PDMPs (vs states without PDMPs).^37,38

Health insurer benefit management strategies

State Medicaid programs and commercial health insurers commonly use preferred drug lists of certain brand-name and generic medications and require physician authorization for nonpreferred products. Compared with no policy, preferred drug lists and authorization policies were associated with fewer high-dose opioid prescription fills and a reduced likelihood of abusing opioid medication in several studies.^47,50 Rates of methadone overdose mortality were higher in Florida and North Carolina, which included methadone as a preferred drug, than in South Carolina, which did not include methadone as a preferred drug.⁴⁹ Health insurer–initiated drug-use review programs, which provide patients’ histories with controlled substances to health care providers, were associated with fewer controlled substance prescriptions filled compared with no drug-use review programs.⁴⁸ A drug-use review intervention in a Midwest Blue Cross Blue Shield plan led to a 28% decrease in the number of patients who received unsafe opioid combination prescriptions (eg, opioid and benzodiazepine).⁵² Medicaid lock-in programs require patients to use a single designated prescriber and pharmacy. In North Carolina, the number of Medicaid-funded prescriptions and the average daily dose declined after implementation of lock-in programs⁵⁴ but this effect may have been offset by patients seeking non–Medicaid-reimbursed health care providers and prescriptions.⁵¹ Similarly, a policy that limited the number of Medicaid-reimbursed short-acting opioids in Colorado was associated with a 3% decrease in the number of overall daily prescriptions reimbursed.⁵³

Clinical guidelines

Limited evidence exists on the effect of clinical guidelines for prescribing opioids^55,56 and no studies on clinical guidelines in our review examined overdose or mortality outcomes. A hospital system in Philadelphia found sustained decreases in the rate of opioid prescriptions after physician guidelines were distributed.⁵⁵ In Washington State, the median number of opioid prescription doses did not change among all users, but the median dose was reduced among patients who received high daily opioid doses.⁵⁶

Pain clinic regulations

Pill-mill laws in Texas were associated with reduced opioid use (number of prescriptions, average daily MME doses, and number of pills dispensed)⁵⁸; however, they were not associated with changes in the monthly percentage of Medicare beneficiaries who filled Schedule II or Schedule III prescriptions or with rates of opioid-related hospitalizations.⁵⁹ Pain clinic regulations and increased penalties for overprescribing were associated with fewer prescription opioid and heroin overdose deaths in Florida compared with North Carolina, which did not have pain clinic regulations.⁵⁷

Medical and recreational marijuana policy

One study with a small sample size found that patients with chronic pain who received medical cannabis had lower doses of opioid prescription than patients with chronic back pain who did not receive medical cannabis.⁶³ However, studies of Medicaid beneficiaries show that states with medical cannabis policies had fewer Schedule III opioid prescriptions, but not fewer Schedule II prescriptions, than states without medical cannabis policies.^61,64 States with medical cannabis laws with active dispensaries had fewer morphine and hydrocodone prescriptions among Medicaid Part D enrollees than states without these regulations.⁶⁰

We found limited but mixed results on recreational cannabis. One study found a decrease in the rate of opioid-related overdose deaths after Colorado legalized recreational marijuana.⁶² Another study found no significant association between legalization of recreational marijuana and any opioid prescription (Schedule II or III) among Medicaid fee-for-service enrollees.⁶¹ Yet another study found that, compared with states without any medical or recreational marijuana law, states with recreational cannabis laws had fewer Schedule II opioid prescriptions.⁶⁴

Harm-reduction interventions

Naloxone distribution programs were understudied in the articles we reviewed. In Lorain County, Ohio, the rate of overdose deaths decreased after a police naloxone training and distribution intervention.⁷⁰ The Good Samaritan law in New York State was associated with increased hospital admissions for heroin overdoses but not for non-heroin opioid-related overdoses.⁶⁹ Consistent with arguments that harm-reduction programs do not increase drug use⁸⁵ states with naloxone distribution and Good Samaritan laws had lower rates of opioid-related overdose mortality but no difference in opioid use compared with states that did not have such laws.⁶⁸ No benefits were observed in studies examining combined overdose education and naloxone distribution. A randomized controlled trial in Seattle, Washington, found that offering naloxone kits and education did not affect the time to first overdose event among patients at risk of opioid abuse.⁶⁵ Similarly, a study of overdose education and a naloxone distribution program in Massachusetts found no significant differences in opioid use, overdose occurrence, and actions taken during overdose among trainees who received a kit and education, trainees who received education only, and nonparticipants.^66,67 Of 6 studies on naloxone, deaths were evaluated in only 2 studies.^68,70

Patient and health care provider education

Several recent studies indicated that education interventions lead to a reduction in inappropriate prescribing and opioid use. A health care provider education intervention at a New Hampshire medical center was associated with fewer opioids prescribed than before the intervention⁷¹ and a patient education intervention in a New York City institution led to lower opioid consumption, particularly among persons at high risk of opioid abuse or with past narcotic abuse, compared with persons who did not receive the intervention.⁷²

PDMP and pain clinic regulations

After Florida’s concurrent implementation of a PDMP and pill-mill laws, the number of opioid prescriptions, total opioid volume, and MME per transaction decreased among prescribers and patients with high rates of opioid prescriptions but not among prescribers and patients with low rates of opioid prescription.^73–75,75

Health care provider education and patient histories or clinical guidelines

Providing hand surgeons with prescription guidelines for various operations and an educational assistance tool reduced the amount of pain medication they prescribed.⁷⁷ However, an experimental study found that sending prescribers educational material and/or information about patients at risk for opioid abuse did not affect the number of opioid prescriptions.⁷⁶

Medical marijuana with PDMP and pain clinic regulations

After controlling for PDMP and pain clinic regulations, 2 studies reported negative associations between medical cannabis laws and opioid-related hospitalization and mortality.^78,81 A study of PDMP and medical marijuana policies found an interaction effect between the 2 policies: mortality was lowered when PDMPs and medical marijuana policies were implemented simultaneously.⁸⁰ A national study of PDMPs, pain clinic regulations, medical marijuana, and harm-reduction strategies found that only robust PDMPs and medical marijuana policies had significant effects on opioid-related mortality.⁷⁹

Other concurrently implemented interventions

One multi-intervention study (including prescribing and dispensing limits, health care provider education, and prescriber-specific performance feedback to health care providers) in a large health care system in Southern California reported 30% fewer high-dose prescriptions and 98% fewer prescriptions with ≥200 pills after implementation compared with before implementation.⁸³ A study of 7 interventions (diversion control, community education, health care provider education, naloxone policies, support for patients with pain, hospital emergency department policies, and addiction treatment) in North Carolina found little to no effect of each strategy on the rate of opioid-related hospital visits and death after controlling for the presence of other policies.⁸² A national evaluation of 8 policies—PDMPs, pain clinic regulations, prescription limits, tamper-resistant prescriptions, patient identification, doctor-shopping (ie, visiting multiple physicians to obtain multiple prescriptions) restrictions, physician examinations, and pharmacist verification—found little effect on opioid use or overdose among Medicare beneficiaries.⁸⁴

Impact of Systems-Level Interventions on HIV, HCV, Tuberculosis, and Sexually Transmitted Infections

Only 1 study reported HIV/HCV outcomes. The study examined implementation of an automated PDMP among Medicaid members in Washington State and reported no change from pre- to post-implementation of the automated PDMP in the percentage of patients with a reported history of HIV/AIDS.⁴²

Study Limitations

We found common study limitations (eg, selection bias, secular changes, isolating intervention effects, implementation variability, short time horizons, generalizability, ecological studies, and appropriate control groups) and data limitations (eg, time lags, misclassification of abuse and deaths, undercounting overdoses and prescriptions, aggregate mortality measures, pills dispensed vs use of pills, and generalizability) among most studies we examined (Box). However, although selection bias and interstate variation in implementation were commonly stated in PDMP studies, they are broadly applicable to any state-level intervention.

Discussion

Considerable developments have taken place in opioid-related policies since the 2014 review, which concluded that the evidence base was weak overall.¹⁰ Robust evidence in the current review supports the effectiveness of PDMPs that mandate physicians’ access to patients’ prescription history for controlled substances. Evidence was limited on new interventions, such as naloxone distribution and marijuana legalization. Furthermore, evidence on clinical guidelines, health insurer policies, and educational interventions continues to be limited. Future research should focus on these areas.

Interventions to lower the prevalence of opioid use disorder may improve downstream control of HIV/HCV infection.⁸⁶ Modeling evidence shows that these interventions increase the efficiency and cost effectiveness of HIV/HCV epidemic strategies, supplementing interventions focused on risk behaviors, such as syringe services programs.^87,88 Systems-level opioid use interventions will likely be a necessary part of achieving national goals of ending the HIV and HCV epidemics.^89,90

Naloxone distribution and Good Samaritan laws are understudied. These interventions are crucial for mitigating negative consequences of the opioid epidemic, and evidence suggests that they may reduce opioid-related overdose mortality.⁶⁸ A 2019 study (outside our review period) found that states with laws on naloxone access had fewer fatal overdoses but more nonfatal overdoses presenting to emergency departments than states without laws on naloxone access.⁹¹ Further research is needed on potential unintended consequences. We focused on the effects of harm-reduction interventions on opioid use, overdose, and mortality outcomes. However, other interventions (eg, syringe services programs) may alleviate the incidence of infectious disease associated with injection drug use.

Medical cannabis laws are associated with reduced opioid-related hospitalizations and mortality.^78,79,81 Laws to legalize recreational marijuana are new, and few studies evaluated their effect on opioid-related outcomes. Because early adopters of medical marijuana legalization may differ, it is speculative to extrapolate from the limited evidence to determine how similar changes in law will affect population outcomes in other states. Further research using longer time horizons and individual-level data is needed.

PDMPs are the most studied and widely implemented opioid policy. Consistent with the previous review,¹⁰ we found inconsistent results for the effect of PDMPs on opioid use, overdose, and deaths, which is partly attributable to heterogeneous policy designs. However, converging evidence suggests that mandatory PDMPs reduce inappropriate prescribing behavior. Our rapid review focused on opioid use, overdose, and mortality outcomes. However, a potential unintended consequence of systems-level policies such as PDMPs and changing health care providers’ prescribing behavior is that individual or population health might worsen if persons with addiction are pushed to illicit markets. This concern is supported by 3 studies we reviewed that found higher heroin poisoning rates among states that implemented PDMPs than among states that did not implement PDMPs.^19,22,23 Most studies of PDMPs use administrative PDMP or claims data on the number of pills or MME dispensed per capita. These studies may underreport total number of prescriptions because other sources, including the illicit market, are excluded and pills dispensed may not reflect actual use. More importantly, the administrative data include behaviors and not meaningful health outcomes such as addiction, quality-of-life indicators (eg, poor health days, self-rated health), and sequelae of heroin use (eg, HCV and HIV infection). Future research should prioritize long-term health consequences of these policies beyond opioid use, overdose, and mortality.

Since the 2014 review,¹⁰ few studies have investigated the effects of clinical guidelines, health insurer policies, and patient or health care provider interventions on opioid mortality. We identified only 1 study that examined the effect of health insurer policies on opioid mortality.⁴⁹ This study was limited in scope and had no causal identification; yet, its positive results suggest that clinical guidelines are promising.

Heterogeneity in implementation, and the concurrent implementation of multiple interventions, precludes meta-analyses. Some studies examined concurrent interventions, using particular study designs to isolate the effects of individual interventions in combination studies.^82
-84 Only 12 included studies examined multiple interventions. Studying simultaneous interventions is a priority research area, because interventions are rarely implemented in isolation.

Given the rapid evolution of studies on opioid policies, additional literature has been published since our review: 23 studies published from July 19, 2018, through November 17, 2019, met our inclusion criteria. PDMPs are still the most commonly studied intervention, and recent research has focused on the program characteristics that improve the effect of PDMPs on opioid-related outcomes. Consistent with the findings from our rapid review, newer studies have mixed conclusions. Two studies investigated interstate PDMP data sharing (not investigated in the studies we reviewed) and found no effect on the number of opioid prescriptions.^92,93 Six studies found that mandatory use of PDMPs, including unsolicited and proactive reporting of patient prescription histories, improved opioid-related outcomes.^94

-99 These positive findings on strong PDMPs were not uniform: 1 study found no significant association between mandatory use requirements and the number of opioid prescriptions among patients undergoing general surgical procedures¹⁰⁰ and another study found that strong and weak PDMPs were associated with similar changes in opioid prescription rates, the number of cumulative doses, and prescription length (in days) among Medicare beneficiaries.¹⁰¹

Studies published after our review period investigated Good Samaritan laws, pain clinic and marijuana regulations, prescribing guidelines, and educational interventions. Evidence in more recent studies refutes positive findings from the studies in our review. We reviewed a national study that found reduced mortality after implementation of Good Samaritan laws and naloxone distribution.⁶⁸ However, a more recent study on Good Samaritan laws found no significant effect on opioid-related overdose mortality.¹⁰² Similarly, studies on pill-mill laws that we reviewed found an increase in the number of opioid prescriptions in Texas^58,59 and opioid-related mortality in Florida⁵⁷; however, a 2019 study found null effects of pill-mill laws on opioid prescriptions in Ohio and Tennessee, despite robust enforcement of these laws.¹⁰³ Two additional studies in 2019 on medical marijuana policies found lower rates of opioid use in areas with legalized medical marijuana policy than in areas without such policy, consistent with our results.^104,105 However, a replication of the article by Bachhuber et al⁷⁸ that used more recent data concluded that associations between opioid-related overdose mortality and cannabis laws may be “spurious”: the association between opioid-related overdose mortality and medical marijuana laws reversed direction, and mortality increased during the longer time horizon, but no association was found between recreational marijuana laws and the number of overdoses.¹⁰⁶

Studies of prescribing guidelines and educational interventions published after our review were consistent with studies we reviewed. The 5 newer studies on institutional, state, or national guidelines also found significant associations between guideline implementation and a decrease in the number of opioid prescriptions.^107

-111 The 3 recent studies on educational interventions found a decrease in opioid prescribing after implementation of an educational intervention for patients¹¹² and a combined guidelines/educational intervention for health care providers.^113,114

Public Health Implications

Tackling the opioid epidemic and its effect on HCV/HIV infection, morbidity, and mortality requires a comprehensive approach with multiple interventions. States that mandate PDMP access have seen declines in inappropriate prescribing practices^24,43 and states that have legalized medical or recreational marijuana have seen reductions in the number of opioid-related overdose deaths.^78,79,81 An integrated approach that combines multiple intervention strategies is needed, but scaling up complex policies to state or regional scales can be difficult to implement and monitor.⁸² Future research should focus on the effects of these combined interventions.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors declared no financial support with respect to the research, authorship, and/or publication of this article.

ORCID iDs

Bahareh Ansari

Erika G. Martin

References

Scholl

Seth

Kariisa

Wilson

Baldwin

. Drug and opioid-involved overdose deaths—United States, 2013-2017. MMWR Morb Mortal Wkly Rep. 2018;67(5152):1419-1427.doi:10.15585/mmwr.mm675152e1

Hedegaard

Miniño

AM.

Warner

. Drug overdose deaths in the United States, 1999-2017. NCHS Data Brief. 2018;329:1-8.

Rosenberg

ES.

Rosenthal

EM.

Hall

et al. Prevalence of hepatitis C virus infection in US states and the District of Columbia, 2013 to 2016. JAMA Netw Open. 2018;1(8):e186371.doi:10.1001/jamanetworkopen.2018.6371

Peters

PJ.

Pontones

Hoover

et al. HIV infection linked to injection use of oxymorphone in Indiana, 2014-2015. N Engl J Med. 2016;375(3):229-239.doi:10.1056/NEJMoa1515195

National Academies of Sciences, Engineering, and Medicine . Integrating Responses at the Intersection of Opioid Use Disorder and Infectious Disease Epidemics: Proceedings of a Workshop . National Academies Press; 2018.

Fink

DS.

Schleimer

JP.

Sarvet

et al. Association between prescription drug monitoring programs and nonfatal and fatal drug overdoses: a systematic review. Ann Intern Med. 2018;168(11):783-790.doi:10.7326/M17-3074

Beaudoin

FL.

Banerjee

GN.

Mello

. State-level and system-level opioid prescribing policies: the impact on provider practices and overdose deaths, a systematic review. J Opioid Manag. 2016;12(2):109-118.doi:10.5055/jom.2016.0322

Mueller

SR.

Walley

AY.

Calcaterra

SL.

Glanz

JM.

Binswanger

. A review of opioid overdose prevention and naloxone prescribing: implications for translating community programming into clinical practice. Subst Abus. 2015;36(2):240-253.doi:10.1080/08897077.2015.1010032

Vyas

MB.

LeBaron

VT.

Gilson

. The use of cannabis in response to the opioid crisis: a review of the literature. Nurs Outlook. 2018;66(1):56-65.doi:10.1016/j.outlook.2017.08.012

10.

Haegerich

TM.

Paulozzi

LJ.

Manns

BJ.

Jones

. What we know, and don’t know, about the impact of state policy and systems-level interventions on prescription drug overdose. Drug Alcohol Depend. 2014;145:34-47.doi:10.1016/j.drugalcdep.2014.10.001

11.

Khangura

Konnyu

Cushman

Grimshaw

Moher

. Evidence summaries: the evolution of a rapid review approach. Syst Rev. 2012;1(1):10.doi:10.1186/2046-4053-1-10

12.

Ganann

Ciliska

Thomas

. Expediting systematic reviews: methods and implications of rapid reviews. Implement Sci. 2010;5(1):56.doi:10.1186/1748-5908-5-56

13.

Grant

MJ.

Booth

. A typology of reviews: an analysis of 14 review types and associated methodologies. Health Info Libr J. 2009;26(2):91-108.doi:10.1111/j.1471-1842.2009.00848.x

14.

Institute of Medicine . Finding What Works in Health Care: Standards for Systematic Reviews. National Academies Press; 2011.

15.

Patel

RS.

Kamil

SH.

Bachu

et al. Marijuana use and acute myocardial infarction: a systematic review of published cases in the literature. Trends Cardiovasc Med. 2019;pii:S1050-1738(19):30112-4.doi:10.1016/j.tcm.2019.08.003

http://www.ncbi.nlm.nih.gov/pubmed/31439383

16.

Brown

Watson

Schloss

. Pharmacological evidence of medicinal cannabis in oncology: a systematic review. Support Care Cancer. 2019;27(9):3195-3207.doi:10.1007/s00520-019-04774-5

17.

Mücke

Phillips

Radbruch

Petzke

Häuser

. Cannabis-based medicines for chronic neuropathic pain in adults. Cochrane Database Syst Rev. 2018;3:CD012182.

18.

Trochim

WMK.

Donnelly

. The Research Methods Knowledge Base . 3rd ed. Atomic Dog Publishing; 2006.

19.

Ali

MM.

Dowd

WN.

Classen

Mutter

Novak

. Prescription drug monitoring programs, nonmedical use of prescription drugs, and heroin use: evidence from the National Survey of Drug Use and Health. Addict Behav. 2017;69:65-77. doi:doi:10.1016/j.addbeh.2017.01.011

28152391

20.

Bao

Pan

Taylor

et al. Prescription drug monitoring programs are associated with sustained reductions in opioid prescribing by physicians. Health Aff (Millwood). 2016;35(6):1045-1051. doi:doi:10.1377/hlthaff.2015.1673

21.

Brady

JE.

Wunsch

DiMaggio

Lang

BH.

Giglio

. Prescription drug monitoring and dispensing of prescription opioids. Public Health Rep. 2014;129(2):139-147. doi:doi:10.1177/003335491412900207

22.

Branham

. Time-series analysis of the impact of prescription drug monitoring programs on heroin treatment admissions. Subst Use Misuse. 2018;53(4):694-701. doi:doi:10.1080/10826084.2017.1363232

23.

Brown

Riley

MR.

Ulrich

et al. Impact of New York prescription drug monitoring program, I-STOP, on statewide overdose morbidity. Drug Alcohol Depend. 2017;178:348-354. doi:doi:10.1016/j.drugalcdep.2017.05.023

http://www.ncbi.nlm.nih.gov/pubmed/28692945

24.

Buchmueller

Carey

. The effect of prescription drug monitoring programs on opioid utilization in Medicare. NBER Working Paper No. 23148. Accessed February 13, 2020. https://www.nber.org/papers/w23148; 2018.

25.

Delcher

Wagenaar

AC.

Goldberger

BA.

Cook

RL.

Maldonado-Molina

. Abrupt decline in oxycodone-caused mortality after implementation of Florida’s prescription drug monitoring program. Drug Alcohol Depend. 2015;150:63-68. doi:doi:10.1016/j.drugalcdep.2015.02.010

26.

Deyo

RA.

Hallvik

SE.

Hildebran

et al. Association of prescription drug monitoring program use with opioid prescribing and health outcomes: a comparison of program users and nonusers. J Pain. 2018;19(2):166-177. doi:doi:10.1016/j.jpain.2017.10.001

27.

Gomes

Juurlink

Yao

et al. Impact of legislation and a prescription monitoring program on the prevalence of potentially inappropriate prescriptions for monitored drugs in Ontario: a time series analysis. CMAJ Open. 2014;2(4):E256-E261. doi:doi:10.9778/cmajo.20140027

28.

Haffajee

RL.

Mello

MM.

Zhang

Zaslavsky

AM.

Larochelle

MR.

Wharam

. Four states with robust prescription drug monitoring programs reduced opioid dosages. Health Aff (Millwood). 2018;37(6):964-974. doi:doi:10.1377/hlthaff.2017.1321

29.

Brady

JE.

Lang

BH.

Giglio

Wunsch

DiMaggio

. Prescription drug monitoring and drug overdose mortality. Inj Epidemiol. 2014;1(1):9. doi:doi:10.1186/2197-1714-1-9

30.

Lin

H-C.

Wang

Boyd

Simoni-Wastila

Buu

. Associations between statewide prescription drug monitoring program (PDMP) requirement and physician patterns of prescribing opioid analgesics for patients with non-cancer chronic pain. Addict Behav. 2018;76:348-354. doi:doi:10.1016/j.addbeh.2017.08.032

31.

Maughan

BC.

Bachhuber

MA.

Mitra

Starrels

. Prescription monitoring programs and emergency department visits involving opioids, 2004-2011. Drug Alcohol Depend. 2015;156:282-288. doi:doi:10.1016/j.drugalcdep.2015.09.024

32.

McAllister

MW.

Aaronson

Spillane

et al. Impact of prescription drug-monitoring program on controlled substance prescribing in the ED. Am J Emerg Med. 2015;33(6):781-785. doi:doi:10.1016/j.ajem.2015.03.036

33.

McDonald

DC.

Carlson

KE.

Jalbert

. An experimental test of the effectiveness of unsolicited reporting by a prescription drug monitoring program in reducing inappropriate acquisition of opioids. Pain Med. 2019;20(5):944-954. doi:doi:10.1093/pm/pny095

34.

Moyo

Simoni-Wastila

Griffin

et al. Impact of prescription drug monitoring programs (PDMPs) on opioid utilization among Medicare beneficiaries in 10 US states. Addiction. 2017;112(10):1784-1796. doi:doi:10.1111/add.13860

35.

Nam

YH.

Shea

DG.

Shi

Moran

. State prescription drug monitoring programs and fatal drug overdoses. Am J Manag Care. 2017;23(5):297-303.

36.

Norwood

CW.

Wright

. Integration of prescription drug monitoring programs (PDMP) in pharmacy practice: improving clinical decision-making and supporting a pharmacist's professional judgment. Res Social Adm Pharm. 2016;12(2):257-266. doi:doi:10.1016/j.sapharm.2015.05.008

37.

Patrick

SW.

Fry

CE.

Jones

TF.

Buntin

. Implementation of prescription drug monitoring programs associated with reductions in opioid-related death rates. Health Aff (Millwood). 2016;35(7):1324-1332. doi:doi:10.1377/hlthaff.2015.1496

38.

Pauly

NJ.

Slavova

Delcher

Freeman

PR.

Talbert

. Features of prescription drug monitoring programs associated with reduced rates of prescription opioid-related poisonings. Drug Alcohol Depend. 2018;184:26-32. doi:doi:10.1016/j.drugalcdep.2017.12.002

39.

Ranapurwala

SI.

Carnahan

RM.

Brown

Hinman

Casteel

. Impact of Iowa’s prescription monitoring program on opioid pain reliever prescribing patterns: an interrupted time series study 2003-2014. Pain Med. 2019;20(2):290-300. doi:doi:10.1093/pm/pny029

40.

Rasubala

Pernapati

Velasquez

Burk

Ren

. Impact of a mandatory prescription drug monitoring program on prescription of opioid analgesics by dentists. PLoS One. 2015;10(8):e0135957. doi:doi:10.1371/journal.pone.0135957

41.

Suffoletto

Lynch

Pacella

CB.

Yealy

DM.

Callaway

. The effect of a statewide mandatory prescription drug monitoring program on opioid prescribing by emergency medicine providers across 15 hospitals in a single health system. J Pain. 2018;19(4):430-438. doi:doi:10.1016/j.jpain.2017.11.010

42.

Sun

BC.

Charlesworth

CJ.

Lupulescu-Mann

et al. Effect of automated prescription drug monitoring program queries on emergency department opioid prescribing. Ann Emerg Med. 2018;71(3):337-347. doi:doi:10.1016/j.annemergmed.2017.10.023

43.

Wen

Schackman

BR.

Aden

Bao

. States with prescription drug monitoring mandates saw a reduction in opioids prescribed to Medicaid enrollees. Health Aff (Millwood). 2017;36(4):733-741. doi:doi:10.1377/hlthaff.2016.1141

44.

Winstanley

EL.

Zhang

Mashni

et al. Mandatory review of a prescription drug monitoring program and impact on opioid and benzodiazepine dispensing. Drug Alcohol Depend. 2018;188:169-174. doi:doi:10.1016/j.drugalcdep.2018.03.036

45.

Yarbrough

. Prescription drug monitoring programs produce a limited impact on painkiller prescribing in Medicare Part D. Health Serv Res. 2018;53(2):671-689. doi:doi:10.1111/1475-6773.12652

46.

Young

LD.

Kreiner

PW.

Panas

. Unsolicited reporting to prescribers of opioid analgesics by a state prescription drug monitoring program: an observational study with matched comparison group. Pain Med. 2018;19(7):1396-1407. doi:doi:10.1093/pm/pnx044

47.

Cochran

Gordon

AJ.

Gellad

et al. Medicaid prior authorization and opioid medication abuse and overdose. Am J Manag Care. 2017;23(5):e164-e171.

48.

Daubresse

Gleason

PP.

Peng

Shah

ND.

Ritter

ST.

Alexander

. Impact of a drug utilization review program on high-risk use of prescription controlled substances. Pharmacoepidemiol Drug Saf. 2014;23(4):419-427. doi:doi:10.1002/pds.3487

49.

Faul

Bohm

Alexander

. Methadone prescribing and overdose and the association with Medicaid preferred drug list policies—United States, 2007-2014. MMWR Morb Mortal Wkly Rep. 2017;66(12):320-323. doi:doi:10.15585/mmwr.mm6612a2

50.

Hartung

DM.

Kim

Ahmed

et al. Effect of a high dosage opioid prior authorization policy on prescription opioid use, misuse, and overdose outcomes. Subst Abus. 2018;39(2):239-246. doi:doi:10.1080/08897077.2017.1389798

51.

Naumann

RB.

Marshall

SW.

Lund

JL.

Gottfredson

NC.

Ringwalt

CL.

Skinner

. Evaluating short- and long-term impacts of a Medicaid “lock-in” program on opioid and benzodiazepine prescriptions dispensed to beneficiaries. Drug Alcohol Depend. 2018;182:112-119. doi:doi:10.1016/j.drugalcdep.2017.10.001

52.

Qureshi

Wesolowicz

LA.

Liu

C-M.

Tungol Lin

. Effectiveness of a retrospective drug utilization review on potentially unsafe opioid and central nervous system combination therapy. J Manag Care Spec Pharm. 2015;21(10):938-944. doi:doi:10.18553/jmcp.2015.21.10.938

53.

Riggs

CS.

Billups

SJ.

Flores

Patel

RJ.

Heilmann

RM.

Milchak

. Opioid use for pain management after implementation of a Medicaid short-acting opioid quantity limit. J Manag Care Spec Pharm. 2017;23(3):346-354. doi:doi:10.18553/jmcp.2017.23.3.346

54.

Skinner

AC.

Ringwalt

Naumann

et al. Reducing opioid misuse: evaluation of a Medicaid controlled substance lock-in program. J Pain. 2016;17(11):1150-1155. doi:doi:10.1016/j.jpain.2016.07.003

55.

del Portal

DA.

Healy

ME.

Satz

WA.

McNamara

. Impact of an opioid prescribing guideline in the acute care setting. J Emerg Med. 2016;50(1):21-27. doi:doi:10.1016/j.jemermed.2015.06.014

56.

Sullivan

MD.

Bauer

AM.

Fulton-Kehoe

et al. Trends in opioid dosing among Washington State Medicaid patients before and after opioid dosing guideline implementation. J Pain. 2016;17(5):561-568. doi:doi:10.1016/j.jpain.2015.12.018

57.

Kennedy-Hendricks

Richey

McGinty

EE.

Stuart

EA.

Barry

CL.

Webster

. Opioid overdose deaths and Florida’s crackdown on pill mills. Am J Public Health. 2016;106(2):291-297. doi:doi:10.2105/AJPH.2015.302953

58.

Lyapustina

Rutkow

Chang

H-Y

et al. Effect of a “pill mill” law on opioid prescribing and utilization: the case of Texas. Drug Alcohol Depend. 2016;159:190-197. doi:doi:10.1016/j.drugalcdep.2015.12.025

59.

Raji

MA.

Kuo

Y-F.

Chen

N-W.

Hasan

Wilkes

DM.

Goodwin

. Impact of laws regulating pain clinics on opioid prescribing and opioid-related toxicity among Texas Medicare Part D beneficiaries. J Pharm Technol. 2017;33(2):60-65. doi:doi:10.1177/8755122516686226

60.

Bradford

AC.

Bradford

WD.

Abraham

Bagwell Adams

. Association between US state medical cannabis laws and opioid prescribing in the Medicare Part D population. JAMA Intern Med. 2018;178(5):667-672. doi:doi:10.1001/jamainternmed.2018.0266

61.

Liang

Bao

Wallace

Grant

Shi

. Medical cannabis legalization and opioid prescriptions: evidence on US Medicaid enrollees during 1993-2014. Addiction. 2018;113(11):2060-2070. doi:doi:10.1111/add.14382

62.

Livingston

MD.

Barnett

TE.

Delcher

Wagenaar

. Recreational cannabis legalization and opioid-related deaths in Colorado, 2000-2015. Am J Public Health. 2017;107(11):1827-1829. doi:doi:10.2105/AJPH.2017.304059

63.

Vigil

JM.

Stith

SS.

Adams

IM.

Reeve

. Associations between medical cannabis and prescription opioid use in chronic pain patients: a preliminary cohort study. PLoS One. 2017;12(11):e0187795. doi:doi:10.1371/journal.pone.0187795

64.

Wen

Hockenberry

. Association of medical and adult-use marijuana laws with opioid prescribing for Medicaid enrollees. JAMA Intern Med. 2018;178(5):673-679. doi:doi:10.1001/jamainternmed.2018.1007

65.

Banta-Green

CJ.

Coffin

PO.

Merrill

et al. Impacts of an opioid overdose prevention intervention delivered subsequent to acute care. Inj Prev. 2019;25(3):191-198. doi:doi:10.1136/injuryprev-2017-042676

66.

Doe-Simkins

Quinn

Xuan

et al. Overdose rescues by trained and untrained participants and change in opioid use among substance-using participants in overdose education and naloxone distribution programs: a retrospective cohort study. BMC Public Health. 2014;14:297. doi:doi:10.1186/1471-2458-14-297

67.

Dwyer

Walley

AY.

Langlois

et al. Opioid education and nasal naloxone rescue kits in the emergency department. West J Emerg Med. 2015;16(3):381-384. doi:doi:10.5811/westjem.2015.2.24909

68.

McClellan

Lambdin

BH.

Ali

et al. Opioid-overdose laws association with opioid use and overdose mortality. Addict Behav. 2018;86:90-95. doi:doi:10.1016/j.addbeh.2018.03.014

69.

Nguyen

Parker

. Assessing the effectiveness of New York’s 911 Good Samaritan Law—evidence from a natural experiment. Int J Drug Policy. 2018;58:149-156. doi:doi:10.1016/j.drugpo.2018.05.013

70.

Rando

Broering

Olson

JE.

Marco

Evans

. Intranasal naloxone administration by police first responders is associated with decreased opioid overdose deaths. Am J Emerg Med. 2015;33(9):1201-1204. doi:doi:10.1016/j.ajem.2015.05.022

71.

Hill

MV.

Stucke

RS.

McMahon

ML.

Beeman

JL.

Barth

. An educational intervention decreases opioid prescribing after general surgical operations. Ann Surg. 2018;267(3):468-472. doi:doi:10.1097/SLA.0000000000002198

72.

Syed

UAM.

Aleem

AW.

Wowkanech

et al. Neer Award 2018: the effect of preoperative education on opioid consumption in patients undergoing arthroscopic rotator cuff repair: a prospective, randomized clinical trial. J Shoulder Elbow Surg. 2018;27(6):962-967. doi:doi:10.1016/j.jse.2018.02.039

73.

Chang

H-Y.

Lyapustina

Rutkow

et al. Impact of prescription drug monitoring programs and pill mill laws on high-risk opioid prescribers: a comparative interrupted time series analysis. Drug Alcohol Depend. 2016;165:1-8. doi:doi:10.1016/j.drugalcdep.2016.04.033

74.

Chang

H-Y.

Murimi

Faul

Rutkow

Alexander

. 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. 2018;27(4):422-429. doi:doi:10.1002/pds.4404

75.

Rutkow

Chang

H-Y.

Daubresse

Webster

DW.

Stuart

EA.

Alexander

. Effect of Florida’s prescription drug monitoring program and pill mill laws on opioid prescribing and use. JAMA Intern Med. 2015;175(10):1642-1651. doi:doi:10.1001/jamainternmed.2015.3931

76.

Pasquale

MK.

Sheer

RL.

Mardekian

et al. Educational intervention for physicians to address the risk of opioid abuse. J Opioid Manag. 2017;13(5):303-313. doi:doi:10.5055/jom.2017.0399

77.

Stanek

JJ.

Renslow

MA.

Kalliainen

. The effect of an educational program on opioid prescription patterns in hand surgery: a quality improvement program. J Hand Surg Am. 2015;40(2):341-346. doi:doi:10.1016/j.jhsa.2014.10.054

78.

Bachhuber

MA.

Saloner

Cunningham

CO.

Barry

. Medical cannabis laws and opioid analgesic overdose mortality in the United States, 1999-2010. JAMA Intern Med. 2014;174(10):1668-1741. doi:doi:10.1001/jamainternmed.2014.4005

79.

Pardo

. Do more robust prescription drug monitoring programs reduce prescription opioid overdose? Addiction. 2017;112(10):1773-1783. doi:doi:10.1111/add.13741

80.

Phillips

Gazmararian

. Implications of prescription drug monitoring and medical cannabis legislation on opioid overdose mortality. J Opioid Manag. 2017;13(4):229-239. doi:doi:10.5055/jom.2017.0391

81.

Shi

. Medical marijuana policies and hospitalizations related to marijuana and opioid pain reliever. Drug Alcohol Depend. 2017;173:144-150. doi:doi:10.1016/j.drugalcdep.2017.01.006

82.

Alexandridis

AA.

McCort

Ringwalt

et al. A statewide evaluation of seven strategies to reduce opioid overdose in North Carolina. Inj Prev. 2018;24(1):48-54. doi:doi:10.1136/injuryprev-2017-042396

83.

Losby

JL.

Hyatt

JD.

Kanter

MH.

Baldwin

Matsuoka

. Safer and more appropriate opioid prescribing: a large healthcare system’s comprehensive approach. J Eval Clin Pract. 2017;23(6):1173-1179. doi:doi:10.1111/jep.12756

84.

Meara

Horwitz

JR.

Powell

et al. State legal restrictions and prescription-opioid use among disabled adults. N Engl J Med. 2016;375(1):44-53. doi:doi:10.1056/NEJMsa1514387

85.

Kerr

. The impact of harm reduction on HIV and illicit drug use. Harm Reduct J. 2014;11:7. doi:doi:10.1186/1477-7517-11-7

86.

Institute of Medicine .. Integrating Responses at the Intersection of Opioid Use Disorder and Infectious Disease Epidemics. National Academies Press; 2018.

87.

Barocas

JA.

Morgan

JR.

Fiellin

et al. Cost-effectiveness of integrating buprenorphine-naloxone treatment for opioid use disorder into clinical care for persons with HIV/hepatitis C co-infection who inject opioids. Int J Drug Policy. 2019;72:160-168. doi:doi:10.1016/j.drugpo.2019.05.010

88.

Fraser

Vellozzi

Hoerger

et al. Scaling up hepatitis C prevention and treatment interventions for achieving elimination in the United States: a rural and urban comparison. Am J Epidemiol. 2019;188(8):1539-1551. doi:doi:10.1093/aje/kwz097

89.

US Department of Health and Human Services . What is “Ending the HIV Epidemic: A Plan for America”? Accessed July 12, 2019. https://www.hiv.gov/federal-response/ending-the-hiv-epidemic/overview

90.

US Department of Health and Human Services . National Viral Hepatitis Action Plan. Accessed July 12, 2019. https://www.hhs.gov/hepatitis/viral-hepatitis-action-plan/index.html

91.

Abouk

Pacula

RL.

Powell

. Association between state laws facilitating pharmacy distribution of naloxone and risk of fatal overdose. JAMA Intern Med. 2019;179(6):805-811. doi:doi:10.1001/jamainternmed.2019.0272

92.

Lin

H-C.

Wang

Simoni-Wastila

Boyd

Buu

. Interstate data sharing of prescription drug monitoring programs and associated opioid prescriptions among patients with non-cancer chronic pain. Prev Med. 2019;118:59-65. doi:doi:10.1016/j.ypmed.2018.10.011

93.

Bao

Wen

Johnson

Jeng

PJ.

Meisel

ZF.

Schackman

. Assessing the impact of state policies for prescription drug monitoring programs on high-risk opioid prescriptions. Health Aff (Millwood). 2018;37(10):1596-1604. doi:doi:10.1377/hlthaff.2018.0512

94.

Strickler

GK.

Zhang

Halpin

JF.

Bohnert

ASB.

Baldwin

GT.

Kreiner

. Effects of mandatory prescription drug monitoring program (PDMP) use laws on prescriber registration and use and on risky prescribing. Drug Alcohol Depend. 2019;199:1-9. doi:doi:10.1016/j.drugalcdep.2019.02.010

95.

Wen

Hockenberry

JM.

Jeng

PJ.

Bao

. Prescription drug monitoring program mandates: impact on opioid prescribing and related hospital use. Health Aff (Millwood). 2019;38(9):1550-1556. doi:doi:10.1377/hlthaff.2019.00103

96.

Bachhuber

MA.

Tuazon

Nolan

ML.

Kunins

HV.

Paone

. Impact of a prescription drug monitoring program use mandate on potentially problematic patterns of opioid analgesic prescriptions in New York City. Pharmacoepidemiol Drug Saf. 2019;28(5):734-739. doi:doi:10.1002/pds.4766

97.

Cerdá

Ponicki

WR.

Smith

et al. Measuring relationships between proactive reporting state-level prescription drug monitoring programs and county-level fatal prescription opioid overdoses. Epidemiology. 2020;31(1):32-42. doi:doi:10.1097/EDE.0000000000001123

98.

McDonald

DC.

Carlson

KE.

Jalbert

99.

Castillo-Carniglia

Ponicki

WR.

Gaidus

et al. Prescription drug monitoring programs and opioid overdoses: exploring sources of heterogeneity. Epidemiology. 2019;30(2):212-220. doi:doi:10.1097/EDE.0000000000000950

100.

Stucke

RS.

Kelly

JL.

Mathis

KA.

Hill

MV.

Barth

. Association of the use of a mandatory prescription drug monitoring program with prescribing practices for patients undergoing elective surgery. JAMA Surg. 2018;153(12):1105-1110. doi:doi:10.1001/jamasurg.2018.2666

101.

Moyo

Simoni-Wastila

Griffin

et al. Prescription drug monitoring programs: assessing the association between “best practices” and opioid use in Medicare. Health Serv Res. 2019;54(5):1045-1054. doi:doi:10.1111/1475-6773.13197

102.

Atkins

DN.

Durrance

CP.

Kim

. Good Samaritan harm reduction policy and drug overdose deaths. Health Serv Res. 2019;54(2):407-416. doi:doi:10.1111/1475-6773.13119

103.

Brighthaupt

SC.

Stone

EM.

Rutkow

McGinty

. Effect of pill mill laws on opioid overdose deaths in Ohio & Tennessee: a mixed-methods case study. Prev Med. 2019;126:105736. doi:doi:10.1016/j.ypmed.2019.05.024

104.

Segura

LE.

Mauro

CM.

Levy

et al. Association of US medical marijuana laws with nonmedical prescription opioid use and prescription opioid use disorder. JAMA Netw Open. 2019;2(7):e197216. doi:doi:10.1001/jamanetworkopen.2019.7216

105.

Shah

Hayes

CJ.

Lakkad

Martin

. Impact of medical marijuana legalization on opioid use, chronic opioid use, and high-risk opioid use. J Gen Intern Med. 2019;34(8):1419-1426. doi:doi:10.1007/s11606-018-4782-2

106.

Shover

CL.

Davis

CS.

Gordon

SC.

Humphreys

. Association between medical cannabis laws and opioid overdose mortality has reversed over time. Proc Natl Acad Sci USA. 2019;116(26):12624-12626. doi:doi:10.1073/pnas.1903434116

107.

Bohnert

ASB.

Guy

GP.

Losby

. Opioid prescribing in the United States before and after the Centers for Disease Control and Prevention's 2016 opioid guideline. Ann Intern Med. 2018;169(6):367-375. doi:doi:10.7326/M18-1243

108.

Phillips

AL.

Thiese

MS.

Freeman

Kartchner

Hegmann

. Implementation of an opioid guideline impacts on opioid prescriptions, adverse outcomes, and an association with a state opioid-related fatalities. J Occup Environ Med. 2019;61(8):653-658. doi:doi:10.1097/JOM.0000000000001640

109.

Gillette

Bush

MA.

Rogers

KML

et al. Association between the North Carolina Medical Board opioid guideline update and opioid prescriptions in Medicare Part D beneficiaries. J Opioid Manag. 2018;14(4):239-243. doi:doi:10.5055/jom.2018.0455

110.

Wyles

CC.

Hevesi

Trousdale

et al. The 2018 Chitranjan S. Ranawat, MD Award: developing and implementing a novel institutional guideline strategy reduced postoperative opioid prescribing after TKA and THA. Clin Orthop Relat Res. 2019;477(1):104-113. doi:doi:10.1007/s11999.0000000000000292

111.

Earp

BE.

Silver

JA.

Mora

AN.

Blazar

. Implementing a postoperative opioid-prescribing protocol significantly reduces the total morphine milligram equivalents prescribed. J Bone Joint Surg Am. 2018;100(19):1698-1703. doi:doi:10.2106/JBJS.17.01307

112.

Patel

HD.

Faisal

FA.

Patel

et al. Effect of a prospective opioid reduction intervention on opioid prescribing and use after radical prostatectomy: results of the Opioid Reduction Intervention for Open, Laparoscopic, and Endoscopic Surgery (ORIOLES) Initiative. BJU Int. 2020;125(3):426-432. doi:doi:10.1111/bju.14932

113.

Stanley

Jackson

Norman

Collins

Bonomo

. Opioid prescribing improvement in orthopaedic specialty unit in a tertiary hospital: a retrospective audit of hospital discharge data pre- and post-intervention for better opioid prescribing practice. ANZ J Surg. 2019;89(10):1302-1307. doi:doi:10.1111/ans.15305

114.

Stepan

JG.

Sacks

HA.

Lovecchio

et al. Opioid prescriber education and guidelines for ambulatory upper-extremity surgery: evaluation of an institutional protocol. J Hand Surg Am. 2019;44(2):129-136. doi:doi:10.1016/j.jhsa.2018.06.014

A Rapid Review of the Impact of Systems-Level Policies and Interventions on Population-Level Outcomes Related to the Opioid Epidemic,United States and Canada,2014-2018

Abstract

Objectives

Methods

Results

Conclusions

Keywords

Methods

Summary of major study limitations a among studies that examined the effectiveness of systems-level policies and interventions on opioid use, overdose, and deaths, United States and Canada, 2014-2018

Search Strategy

Review Process

Data Extraction and Synthesis

Results

Study Screening

Overview of Study Designs, Outcomes Assessed, and Study Characteristics

Impact of Systems-Level Interventions on Opioid Use, Overdose, and Mortality

PDMPs

Health insurer benefit management strategies

Clinical guidelines

Pain clinic regulations

Medical and recreational marijuana policy

Harm-reduction interventions

Patient and health care provider education

PDMP and pain clinic regulations

Health care provider education and patient histories or clinical guidelines

Medical marijuana with PDMP and pain clinic regulations

Other concurrently implemented interventions

Impact of Systems-Level Interventions on HIV, HCV, Tuberculosis, and Sexually Transmitted Infections

Study Limitations

Discussion

Public Health Implications

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iDs

References

Summary of major study limitations^a among studies that examined the effectiveness of systems-level policies and interventions on opioid use, overdose, and deaths, United States and Canada, 2014-2018