Table 1 Characteristics of studies evaluating the association between MMLs and opioid- related outcomes in the US
Outcome | Author(s), year | Study time period | State | Study subjects | Study design/analysis | Opioid-related outcome measure | Outcome data source | Covariates | Key findings | aQuality score |
|---|---|---|---|---|---|---|---|---|---|---|
Mortality | Powell et al. 2018 | 1992–2013 | All states | Subjects from 24 states with MMLs compared with those from non-MML states | Difference-in-differences | Prescription opioid-related mortality | National Vital Statistics System | Age, % male population, unemployment rate, alcohol taxes, log of population | MMLs were associated with a 4.8% reduction in opioid overdose mortality | 8 |
Phillips and Gazmararian 2017 | 2011–2014 | All states and D.C | US population during the study period | Ecological analysis | Age-adjusted opioid-related mortality | Multiple Cause of Death database, CDC WONDER | State urban population, state disability rates, education, annual unemployment rates | MMLs were associated with a 1.7% increase in opioid-related mortality | 8 | |
Smart 2016 | 1999–2013 | 48 states | Subjects who died from prescription opioid overdose | Poisson regression | Prescription opioid-related mortality | Multiple Cause of Death database, CDC WONDER | Age, % male population, unemployment rate, alcohol taxes, population | MMLs were associated with a 7.2% reduction in opioid overdose mortality. | 7 | |
Bachhuber et al. 2014 | 1999–2010 | All states | Subjects from 13 MML states; 3 states with MML enacted prior to the study and 10 enacted during the study period | Time-series analysis | Age-adjusted prescription opioid overdose death rate | Multiple Cause of Death database, CDC WONDER | PDMP status, laws requiring identification before dispensing, state oversight, unemployment rates | MMLs were associated with a 24.8% reduced state-level prescription opioid overdose mortality rates | 8 | |
Livingston, 2017 | 2000–2015 | Colorado | Subjects who from opioid overdose in Colorado (recreational marijuana law), Nevada (MML), and Utah (no MML) | Time-series analysis | Opioid-related mortality | Multiple Cause of Death database, CDC WONDER | PDMP status, trends in opioid-related deaths in Nevada and Utah | Recreational marijuana was associated with a 6.5% reduction in opioid-related deaths | 6 | |
Prescriptions dispensed | Bradford et al. 2018 | 2010–2015 | All states | All fee-for-service Medicare Part D prescriptions for all opioids | Multi-level regression | Daily opioid doses prescribed (in millions) per state-year | Medicare Part D Prescription Drug Event Standard Analytic File | PDMP status, Physician market competition, % below FPL, total population, % enrolled in Medicare, % in Medicare Advantage plans, state fixed effects | MMLs of any type were associated with a decrease of 8.5% daily opioid doses prescribed per state-year | 7 |
Liang et al. 2018 | 1993–2014 | All states | Patients enrolled in fee-for-service Medicaid programs | Time-series analysis | Opioid prescriptions per quarter year per 100Medicaid enrollees | Medicaid State Drug Utilization Data | PDMP, Medicaid expansion, household income, active physicians per 1000 population, % residents with household income below FPL, unemployment rate | MMLs were not associated with Schedule II opioid prescriptions dispensed. However, MMLs were associated with 15% decrease in Schedule III opioid prescriptions | 8 | |
Powell et al. 2018 | 1992–2013 | All states | Subjects from 24 states with MMLs compared with those from non-MML states | Difference-in-differences | Opioid prescriptions filled | National Vital Statistics System | Age, % male population, unemployment rate, alcohol taxes, log of population | MMLs were associated with a 3.3% increase in opioid prescriptions | 8 | |
Stith et al. 2018 | 2010–2015 | New Mexico | 83 chronic pain patients enrolled in New Mexico medical marijuana program; 42 non-enrolled patients | Retrospective cohort | Schedule II drug prescriptions | Prescription drug monitoring program records | Age, sex | Enrolling in the medical marijuana program was associated with a 4% reduction in Schedule II drug prescriptions filled. | 6 | |
Wen and Hockenberry 2018 | 2011–2016 | All states | All fee-for-service Medicaid and managed care enrollees | Difference-in-differences | Opioid prescriptions filled | Medicaid State Drug Utilization Data | Age, sex, PDMP status, Pain medication laws, poverty rates, household income, unemployment status, number of Medicaid prescriptions | MMLs were associated with a 5.9% reduction in the rate of opioid prescriptions and legalizing creational marijuana was associated with a 6.38% reduction in the rate of opioid prescriptions. | 7 | |
Bradford and Bradford 2017 | 2007–2014 | All states | All fee-for-service Medicaid prescriptions covering 9 clinical areas of prescription drugs for which MM could be an alternative | Difference-in-differences | Daily doses of prescriptions for pain medications per quarter year per Medicaid enrollee | Medicaid State Drug Utilization Data | Physicians per capita, poverty rate, unemployment rate, state total population, median income, PDMP status | MMLs were associated with an 11% reduction in daily doses of prescriptions for pain medications | 7 | |
Bradford and Bradford 2016 | 2010–2013 | All states | All fee-for service Medicare Part D prescriptions covering 9 clinical areas of prescription drugs for which MM could be an alternative | Difference-in-differences | Daily doses of prescriptions for pain medications filled per physician per year | Medicare Part D Prescription Drug Event Standard Analytic File | Physicians per capita, county unemployment rate, county total population, racial composition, SES, county mortality rate, physician sex | MMLs were associated with a 14.3% reduction in daily doses of prescriptions for pain medications filled per physician per year | 6 | |
Hospitalizations | Powell et al. 2018 | 1992–2013 | All states | Subjects from 24 states with MMLs compared with those from non-MML states | Difference-in-differences | Prescription opioid-related hospitalizations | National Vital Statistics System | Age, % male population, unemployment rate, alcohol taxes, log of population | MMLs were not associated with prescription opioid-related hospitalizations | 8 |
Shi 2017 | 1997–2014 | 27 states | Subjects who were hospitalized in states that participated in the State Inpatient Databases | Time-series analysis | Opioid pain reliever overdose hospitalizations per state per year | State Inpatient Databases, Healthcare Cost Utilization Project | State population size, unemployment rate, median household income, beer tax per gallon, health uninsured rate | MMLs was associated with a 13% reduction related to opioid pain reliever overdose hospitalizations | 8 | |
Non-medical use | Cerda et al. 2018 | 1991–2015 | 48 states | 8th, 10th, and 12th graders | Difference-in-differences | Self-reported nonmedical use of prescription opioids | National Monitoring the Future annual survey | Grade, age, sex, race/ethnicity, SES, students per grade, type pf school, urbanicity, percent of state population that was male, White and aged 10–24 years or 25 years and older, alcohol and cigarette taxes | MML was associated with a 0.3% reduction, and a 0.3% increase in nonmedical use of prescription opioids among 10th and 12th graders respectively. The was no change among 11th graders | 8 |
Shi 2017 | 1997–2014 | 27 states | Subjects who were hospitalized in states that participated in the State Inpatient Databases | Time-series analysis | Opioid pain reliever abuse or dependence –related hospital discharges per state per year | State Inpatient Databases, Healthcare Cost Utilization Project | State population size, unemployment rate, median household income, beer tax per gallon, health uninsured rate | MMLs was associated with a 23% reduction in opioid pain reliever abuse or dependence-related hospitalization | 7 | |
Wen et al. 2015 | 2004–2012 | 10 states | Civilian, non-instutionalized subjects aged 12 years and older | Probit regression | Non medically used prescription pain medications | National Survey on Drug Use and Health | Age, gender, race/ethnicity, health status, smoking status, health insurance status, family income, urbanicity, marital status, education attainment, college enrollment, employment status, state unemployment rate, average personal income, median household income, beer tax per gallon | MML was not associated with any significant change in the rate of nonmedical prescription pain medications use | 8 | |
Opioid positivity among fatally injured drivers | Kim et al. 2016 | 1999–2013 | 18 states that tested for alcohol and drugs in at least 80% of all fatally injured drivers | Fatally injured drivers who died within 1 h of crash | Multi-level logistic regression | Opioid positivity | Fatality Analysis Reporting System | Age, sex, PDMP status, blood alcohol concentration | MMLs were associated with a reduction on opioid positivity among 21–40 year old fatally injured drivers (OR = 0.50 95%ci = 0.37–0.67) | 7 |