The absence of data on driving under the influence of alcohol in road traffic studies: a scoping review of non-randomized studies with vote counting based on the direction of effects of alcohol policies

Abstract

Background: Data on driving under the influence of alcohol (DUIA) are not always available, accurate, or reliable, making it difficult to study the effects of alcohol policies on road traffic outcomes. The objectives of our study were twofold: 1) to describe how road traffic outcomes of alcohol policies are assessed when DUIA data are missing, and 2) to explore the effects of alcohol policies when DUIA data are missing.

Methods: We conducted a scoping review of non-randomized studies that assessed the road traffic outcomes of alcohol policies when DUIA data are missing. Until November 2021, we searched studies published between 2000 and 2021, in English or French, via MEDLINE, APA PsycInfo, CINAHL, and SocINDEX. We assessed the risk of bias in the included studies with the Quality Assessment Tool for Before-After (Pre-Post) Studies With No Control Group. The selection process, data extraction, and the risk of bias assessment were conducted independently and in duplicate. We used vote counting based on the direction of the effects of alcohol policies as a synthesis method. The protocol for this review was published in PROSPERO under record number CRD42021266744.

Results: Twenty-four eligible studies were included. Regarding objective 1, most studies used uncontrolled interrupted time series designs to assess road traffic fatalities resulting from night-time crashes. The reasons for missing DUIA data were generally not reported. Regarding objective 2, we found evidence for an association between alcohol policies and decreased road traffic fatalities. Subgroup analyses found no evidence for an association between methodological modifiers and positive effect directions for road traffic fatalities.

Conclusion: Caution is needed when interpreting road traffic outcomes associated with alcohol policies when DUIA data are missing. Greater efforts should be made to improve the reporting of outcomes assessments. Future studies must address several methodological issues (e.g., more granular data, well-defined intervention and implementation, and controlled designs). Our results should be compared to those from others reviews where DUIA data were available to confirm or recalibrate the associations found in studies where DUIA data were missing.

Keywords: Alcohol drinking, Legislation & jurisprudence; Driving under the influence; Non-randomized controlled trials; Systematic review; Traffic accidents.

Fig. 1

PRISMA 2020 flow diagram

Fig. 2

Summary of the risk of bias assessment Notes. Risk of bias domains were as follows: D1, study question; D2, eligibility criteria and study population; D3, representativeness of study participants; D4, participants' enrollment; D5, sample size; D6, description and implementation of the intervention; D7, outcomes measures; D8, statistical analysis; and D9, multiple outcome measures. Studies having a low risk of bias in a specific domain were represented by a green plus sign. Those studies judged as having some concern for bias in a specific domain were represented by a red negative sign

Fig. 3

Effect direction plot summarizing the direction of road traffic outcomes in studies assessing alcohol policies. [–60]. Notes. Effect direction: upward arrow ▲ = negative health impact (e.g., increase in fatalities); downward arrow ▼ = positive health impact (e.g., decrease in injuries); sideways arrow ◄► = no change/mixed effects/conflicting findings. Units of time in the study: large arrow    > 50; medium arrow  25-49; small arrow  < 24. Numbers beside each arrow represent the number of road traffic outcomes summarized, and no number appears in cases where only one outcome was evaluated. In Miller et al. (2004), we classified the actual outcome (severe injuries or fatalities) as fatalities