Reducing alcohol use through alcohol control policies in the general population and population subgroups: a systematic review and meta-analysis

Abstract

We estimate the effects of alcohol taxation, minimum unit pricing (MUP), and restricted temporal availability on overall alcohol consumption and review their differential impact across sociodemographic groups. Web of Science, Medline, PsycInfo, Embase, and EconLit were searched on 08/12/2022 and 09/26/2022 for studies on newly introduced or changed alcohol policies published between 2000 and 2022 (Prospero registration: CRD42022339791). We combined data using random-effects meta-analyses. Risk of bias was assessed using the Newcastle-Ottawa Scale. Of 1887 reports, 36 were eligible. Doubling alcohol taxes or introducing MUP (Int$ 0.90/10 g of pure alcohol) reduced consumption by 10% (for taxation: 95% prediction intervals [PI]: -18.5%, -1.2%; for MUP: 95% PI: -28.2%, 5.8%), restricting alcohol sales by one day a week reduced consumption by 3.6% (95% PI: -7.2%, -0.1%). Substantial between-study heterogeneity contributes to high levels of uncertainty and must be considered in interpretation. Pricing policies resulted in greater consumption changes among low-income alcohol users, while results were inconclusive for other socioeconomic indicators, gender, and racial and ethnic groups. Research is needed on the differential impact of alcohol policies, particularly for groups bearing a disproportionate alcohol-attributable health burden.

Funding: Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award Number R01AA028009.

Keywords: Alcohol consumption; Alcohol policy; Effectiveness; Ethnicity; Race; Socioeconomic status.

Fig. 1

Flowchart of study selection. Note: List of excluded full-text articles is available in the appendix p. 12. ^∗Numbers do not add up, as three studies reported results on both the impact of alcohol taxation and temporal availability policies.

Fig. 2

Effect of alcohol excise taxes on the level of alcohol consumption based on the identified literature (n = 10). Notes: 1. One study may have provided multiple point estimates, for example, for different alcoholic beverages. 2. Solid shapes indicate the focus of the interventional study: tax reform specific to certain alcoholic beverages (diamond) and a more complex tax reform affecting different alcoholic beverages to varying degrees (triangle). 3. Empty shapes indicate studies reporting tax elasticities. 4. Grey colors indicate studies with a critical risk of bias. 5. The size of the shape indicates the inverse variance weight of each point estimate. 6. The solid regression line depicts the association between alcohol excise taxes and consumption based on the random-effects meta-regression model. The dashed lines illustrate the 95% confidence interval and the dotted grey lines the 95% prediction interval. 7. Where multiple point estimates (k ≥ 2) on the consumption change were available, point estimates were pooled using fixed-effects meta-analysis (applied to one study²⁸).

Fig. 3

Effect of introducing a minimum unit price on the level of alcohol consumption within one year of policy implementation. 95% prediction interval of weighted average: 95% PI: −28.2%, 5.8%. Notes: 1. Solid and empty shapes indicate individual-level and aggregated alcohol consumption data, respectively. 2. Grey colors indicate studies with a critical risk of bias. 3. DPA = Darwin/Palmerston area is a subregion of the Northern Territory. The point estimates from the Darwin/Palmerston area were preferred over that of the Northern Territory, as another policy were implemented shortly before the introduction of MUPs in different areas throughout the Northern Territory (Police Adjunct Licensing Inspectors, PALI: stationing of police officers in front of off-trade alcohol stores). The introduction of PALIs may have amplified the MUP effect in the Northern Territory. 4. “Alcohol” in Taylor et al., 2021 refers to the consumption of alcoholic beverages other than wine. 5. The effect estimate for ready-to-drink (RTD) beverages in Wales (Anderson et al. 2021⁵⁰) could not be included in the model as the effect and standard error was zero. 6. Where multiple point estimates (k ≥ 2) on the consumption change were available, point estimates were pooled using fixed-effects meta-analysis (applied to one study⁴⁷). 7. Inverse variance weighting: larger confidence interval signifies smaller weighting of a point estimate.

Fig. 4

Effect of banning alcohol sales on one additional day on alcohol consumption. 95% prediction interval of weighted average: 95% PI: −7.2%, −0.1%. 1. Solid and empty shapes indicate individual-level and aggregated alcohol consumption data, respectively. 2. Grey colors indicate studies with a critical risk of bias. 3. Asterisks (∗) indicate studies that examined the impact of permitting alcohol sales on one additional day a week; observed consumption changes were inverted. 4. USA = United States of America. 5. Inverse variance weighting; larger confidence interval signifies smaller weighting of a point estimate.

Fig. 5

Summary of existing evidence on the impact of alcohol control policies conditional on gender, socioeconomic status, and race and ethnicity. ‘+’ evidence for conditional effectiveness, ‘〇’ mixed or inconclusive evidence for conditional effectiveness, ‘–’ no evidence available.