National Retail Sales of Alcohol and Cannabis During the COVID-19 Pandemic in Canada
James MacKillop, PhD; Alysha Cooper, MSc; Jean Costello, PhD
JAMA Network Open, 2021, 4, (11), e2133076, 1-4.
Doi : 10.1001/jamanetworkopen.2021.33076
There is concern that the societal consequences of the COVID-19 pandemic will be associated with increased substance use.1 Data to date have primarily been self-reported changes, but objective sales data may inform this question. Here, we examined national retail sales of alcohol and cannabis prior to and during the pandemic in Canada.
Methods : Where applicable, the report for this economic evaluation is consistent with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guideline. The Hamilton Integrated Research Ethics Board determined that ethical review board approval and informed consent were not needed because the data were publicly available salesmetrics. The data were seasonally adjusted national monthly retail sales (ie, North American Industry Classification System codes 4453, for beer, wine, and liquor stores, and 453993, for cannabis stores) from November 2018 to June 2021 in Canadian dollars.2 The period was selected to provide a sizable prepandemic window and because of the timing of cannabis legalization (ie, mid-October 2018). Principal analyses were contrasts between intrapandemic sales and a counterfactual intrapandemic linear trend based on prepandemic sales. A subanalysis quantified stockpiling, operationalized as the proportionate change in March 2020, when states of emergency were declared, compared with the counterfactual estimate. Because the data were population level, null hypothesis significance testing was a secondary priority, but overall differences and intrapandemic trends in poststockpiling data were examined statistically using analysis of variance (ANOVA) and segmented regression, respectively.
Significance tests used P < .05 and were 2-sided, and analyses were conducted from May to August 2021 using Excel version 2019 (Microsoft), SPSS statistical software version 26.0 (IBM), and R statistical software version 4.1.1 (R Project for Statistical Computing).
Results : Mean (SD) monthly prepandemic alcohol sales were $2.02 billion ($26.24 million), exhibiting a positive slope of $4.45 million/month (Figure 1). Putative stockpiling in March 2020 was an increase of $330 million (15.97%) from the counterfactual estimate of $2.06 billion. Intrapandemic poststockpiling alcohol national monthly sales were $2.09 billion in April 2020 and $2.29 billion in June 2021. Mean (SD) monthly intrapandemic alcohol sales during the intra–COVID-19 period were $2.214 billion ($87 million), which was an increase of $116 million (5.54%) vs the mean (SD) counterfactual estimate of $2.098 billion ($21 million), reflecting an increase of $1.86 billion over the 16-month period. Based on ANOVA, mean monthly prepandemic saleswere statistically significantly lower than intrapandemic sales ($2.02 billion; 95%CI, $1.99-$2.06 billion vs $2.21 billion; 95%CI, $2.18-$2.25 billion; P < .001). In segmented regression, there was an increasing trend in the time × COVID-19 interaction (B = 7.78; standard error [SE] = 3.07; P = .02) (Figure 1).
Monthly prepandemic cannabis sales exhibited an increase (Figure 2), from $55.40 million in November 2018 to $150.75 million in February 2020 (slope, $6.36 million/month), associated with the expanding legal marketplace. Stockpiling was estimated at an increase of $23.59 million (15.02%) greater than the counterfactual estimate of $157.10 million. Mean (SD) monthly intrapandemic cannabis sales were $255.51 million ($47.71 million), which was 24.78%greater than the mean (SD) counterfactual estimate of $204.78 million ($30.26 million) and reflected an increase of $811.74 million over the 16-month period. The difference in mean monthly prepandemic sales ($100.58 million [95%CI, $78.56-$122.61 million]) vs intrapandemic sales ($255.51 million [95%CI, $233.48- $277.54 million]) was significant (P < .001) but uninformative because of the steep postlegalization slope. In segmented regression, there was no statistically significant increasing trend (B = 1.82; SE = 1.43; P = .22) Figure 2).