Are cannabis policy changes associated with alcohol use patterns? Evidence for age-group differences based on primary care screening data

Abstract

Background and Aims

In California, USA, adult cannabis legalization (ACL) was passed in 2016, with implementation of legal sales commencing in 2018. This study examined whether these policy changes were associated with changes in alcohol use patterns and explored age-group differences.

Design

We examined monthly rates of positive alcohol screens in primary care patients in California from 2015 to 2019. Using an interrupted time series design and autoregressive integrated moving average (ARIMA) models, changes were evaluated in the overall sample and stratified by age groups (21–34, 35–64 and 65+) across two key intervention points for ACL: 9 November 2016 (ACL passage) and 1 January 2018 (ACL implementation). Time-varying confounders of seasonality and autocorrelation were controlled for.

Setting

California, United States; Kaiser Permanente Northern California Healthcare System.

Participants

Across the study period, n = 8 028 627 screenings were completed for  n= 3 525 493 unique patients aged 21+ .

Measurements

Aggregated monthly rates of patients exceeding daily limits, weekly limits, exceeding both daily and weekly limits and frequent heavy episodic drinking (HED), based on current US guidelines set by the National Institute on Alcohol Abuse and Alcoholism.

Findings

Statistically significant, gradual declines in rates of both exceeding weekly alcohol limits [estimate (95% confidence interval [CI]) of slope change = −0.013(−0.019 to −0.008), P < 0.001] and frequent HED followed ACL passage [estimate (95% CI) of slope change = −0.015(−0.024 to −0.005), P = 0.003]. ACL implementation was generally not associated with further changes, suggesting changes after ACL passage were sustained over time. Age-group patterns varied: young adults (21–34) demonstrated an immediate increase, followed by a statistically significant gradual decline in rates of HED post-ACL passage [estimate (95% CI) of slope change = −0.096(−0.160, −0.033), P = 0.003], as well as an immediate decrease post ACL implementation; adults 35–49 showed consistent declines in step and slope in rates of frequent HED post ACL passage; adults 50–64 showed an immediate increase in rates of exceeding weekly limits [estimate (95% CI) of step change = 0.167(0.036–0.297), P = 0.012], with gradual declines in the other drinking limits; adults 65 + had small, gradual decreases. In addition, exploratory analyses in 18–20-year-olds showed gradual increases in rates of frequent HED post ACL passage, but the P value was not significant after Bonferroni correction.

Conclusions

Cannabis policy changes in California, USA, appear to be linked to age-specific changes in alcohol use, with moderate reductions, particularly among middle-aged adults. Findings were associative in nature and causality could not be determined.