A seemingly simple question has perplexed alcohol researchers over recent decades—does having more retail alcohol outlets in a neighbourhood increase alcohol consumption for people who live and spend time there [1]? The question has firm theoretical and empirical foundations. Individuals’ behaviour is affected by proximal environmental conditions [2], and most ecological studies find aggregate alcohol consumption [3,4] and alcohol-related harms (e.g., injuries [3,4] and liver disease [5–9]) to be greater in neighbourhoods with more outlets. However, researchers continue to report mixed findings for individual-level studies relating alcohol outlets to alcohol consumption. Many authors report overall positive associations between alcohol outlet density and consumption [10–13]; others report no association [14,15]. This paper explores one explanation for these mixed results: individuals might travel to alcohol outlets outside the neighbourhoods where they live and routinely spend time.

Measuring travel to alcohol outlets is methodologically challenging. The proportion of the population who obtain alcohol either for on-premise consumption at a bar or off-premise consumption from a liquor store is relatively small, so sample sizes must be large to achieve stable estimates. However, collecting detailed data about individuals’ trips to specific alcohol outlets is burdensome for both researchers and participants. Commonly used methods for collecting travel data (e.g., travel diaries, interviews, questionnaires) are available, albeit impractical for large samples [16–18]. These obstacles mean that few empirical observations are available regarding the distance that individuals travel to access alcohol at retail establishments. Without this information, studies that examine individuals’ exposure to alcohol outlets rely on educated guesses about how far away alcohol outlets must be to influence an individuals’ alcohol consumption and risks for alcohol-related harm. These can vary, but authors typically assume that individuals access outlets near their place of residence, echoing colloquial notions of the “local bar” [19].

One underutilised solution for measuring travel to alcohol outlets at a population level is household travel surveys. Travel surveys, commonly used by transportation planners to capture representative information about the geographic movements of populations, typically include large samples recruited via careful sample frames and collect detailed trip-level information necessary to describe travel patterns over time (lengths of trips, modes of transportation, etc.) [16]. Household travel surveys can reveal exposures affecting health outcomes, both at home and in various daily visited locations (activity locations).

The aim of this study was to investigate individuals’ travel to alcohol outlets using data from a household travel survey conducted in Victoria, Australia, reporting origin and destination locations with high spatial resolution. We addressed this aim through one descriptive and two analytic objectives. First, we described the distance, time and modes of travel for individuals’ trips to alcohol outlets, accounting for the possibility that these could be part of complex trips involving multiple stops. Second, we examined whether the alcohol outlets to which individuals travelled were in the same neighbourhoods as other locations that they attend routinely (e.g., home, work). Lastly, we assessed statistical associations between individuals’ exposure to alcohol outlets and their trips to alcohol outlets at multiple spatial scales. We interpret results in the context that individual-level studies of exposure to alcohol outlets and alcohol consumption are generally mixed.

There were 23,512 participants in the VISTA survey eligible for inclusion in this analysis. Median age was 46 years (SD = 18 years); 57% of the sample was male (Table 1). Median personal income per week was $600–999. Included participants were from 13,519 households, and the day of survey completion was approximately uniformly distributed. There was one participant aged 116, according to the household travel survey. While this value is implausibly high, no exclusionary criteria were applied to maintain the integrity of the dataset and analysis.

This study of one-day trip paths for 23,512 individuals in Melbourne, Australia, identifies that individuals travel well beyond their immediate residential area to access on-premise alcohol outlets. Travel to distal alcohol outlets could contribute to mixed results for individual-level studies of exposure to alcohol outlets and alcohol consumption.

The first study aim was to describe individuals’ travel to alcohol outlets. We found one-way trips to bars had a mean distance of 10.2 km and took 25.7 minutes, but these trips added only 8.2 km and 18.1 minutes to the cumulative travel distance over the full day. The difference between one-way trip distance and cumulative trip distance was more substantial for liquor stores (0.8 km difference) than bars (2 km difference). We found one-way trips to liquor stores had a mean distance of 5.5 km and took 12.2 minutes, but these trips added only 4.7 km and 8.9 minutes to the cumulative travel distance over the full day. The attenuation occurs because trips to liquor stores are commonly part of complex trips involving multiple stops. While few studies have explored this topic, one smaller study (n=831) found that distance and affordability are key in alcohol purchase decisions [28]. People might travel farther for unique bar attributes [28,29], but proximity influences liquor store purchases. Our results concord with prior findings, in that there are differences in the distance people are willing to travel to an alcohol outlet (i.e., “alcohol access”) [30,31]. The observation that individuals combine trips to liquor stores with trips to other shops is partly a function of the liquor licensing regime in Victoria—which requires that liquor sales occupy separate retail space compared to other items [32]—and partly a reflection that alcohol is generally considered a non-durable good that consumers purchase routinely [33].

The second study aim was to examine whether individuals’ chosen alcohol outlets were in the same neighbourhoods as their homes, workplaces and other destinations. We used LGAs to represent neighbourhoods, which are much larger geographic areas than are used in most published studies that relate individuals’ exposure to alcohol outlets and alcohol consumption. Despite the potential for bias in detecting co-location, many trips were to unique neighbourhoods. Almost half the observed trips to bars (47%) and one-fifth of liquor store trips (23%) were outside individuals’ LGAs. This corroborates other studies that have examined associations of alcohol outlet density with alcohol consumption using measures of activity spaces (i.e., convex hull polygons, standard deviational ellipses, etc.) [34,35], which are the “set of spatial locations visited by an individual over a given period.” [36] These studies, by Freisthler et al. [34] and Morrison et al. [37], find that measures incorporating activity space information, like activity location measures (places individuals frequent) and activity path-based measures (places individuals frequent and paths between each of those places) do not necessarily relate to residence-based measures, like outlet density in a census tract. Implications of this study are important for research on the impact of the alcohol environment on consumption. If a significant percentage of trips to bars are located outside of an individual’s “neighbourhood”, limiting studies to only the alcohol environment of where they live or work may miss their true exposure and ultimately their risk of alcohol consumption.

The third study aim was to examine associations between alcohol outlet density and trips to alcohol outlets within and outside of the home LGA. One extrinsic and mutable environmental feature that could theoretically affect individuals’ alcohol consumption is the availability of alcohol through retail outlets. Alcohol availability can be separated into four dimensions —physical (prevalence of retail establishments that sell alcohol like bars and liquor stores), social (prevalence of alcohol in social environments), economic (cost relative to income) and subjective (perceived access) [38]. Retail alcohol outlet density is generally used as the measurement technique for the four availability dimensions listed above. Much of the evidence examining the relationship between alcohol outlet density and alcohol consumption does not explicitly isolate the type of alcohol availability under study, potentially pointing to another reason for mixed findings. Epidemiologically, these four dimensions of alcohol availability act as multiple correlated exposures, each with separate paths to alcohol access (mediator) and to alcohol consumption (outcome). Given the mixed results from individual-level studies, further research into the specific pathways linking outlet density and consumption is needed to clarify the overall impact. For example, distance to alcohol outlets may mediate the relationship between outlet density and trips to alcohol outlets; that is, high outlet density could shorten the distance to alcohol outlets, consequently increasing trips and opportunities for consumption. Future research examining distance to alcohol outlets may elucidate yet another component pathway in relationship between outlet density and consumption. Our primary focus was on exploring the broader relationship between outlet density and trips to alcohol outlets. Thus, the third aim found significant associations between individuals’ exposure to alcohol outlets and their physical access to alcohol outlets within their home, isolating one of the multiple theoretical pathways by which exposure to alcohol outlets affects alcohol consumption for individuals.

This study did not find the same significant associations between alcohol outlet density in individuals’ home LGAs or postcodes and visits to outlets outside their home LGA. However, it is crucial to note that over half of bar visits are outside the home LGA. While local alcohol outlet densities affect local trips to alcohol outlets, additional research is necessary to understand the factors influencing trips to outlets outside of the home LGA if density is not a significant predictor.

There are several limitations to this study. First, VISTA contains no alcohol purchase or consumption information for individuals when traveling to retail alcohol outlets. Second, we cannot confirm that the destination outlets were licensed for on-premise or off-premise alcohol sales. Third, the VISTA data capture individuals’ travel over just one day. The alcohol outlets to which they travelled may be located within LGAs that they frequented on other days, and studies using activity location-based or activity path-based approaches may capture these neighbourhoods as being within the individuals’ activity space. However, the trip-level data allows identification of dedicated trips to specific locations, so these data are arguably a better fit for testing direct access to alcohol outlets than data that measure exposure to alcohol outlets over longer periods (e.g., weeks, months).

Studies examining associations between individuals’ exposure to retail alcohol outlets and the volume and pattern of their alcohol consumption often have mixed results. This study identifies one explanation for this puzzling finding—trips to the “local bar” may be the exception, not the norm. Further, in sync with other literature, we find alcohol outlet density to be associated with trips to alcohol outlets, thus emphasising one possible pathway toward alcohol consumption. Refining the scientific methods for measuring exposure to alcohol outlets and alcohol consumption is a research priority because the retail alcohol environment is a modifiable exposure, and alcohol consumption takes a considerable toll on public health.