Research paper
The effects of extended public transport operating hours and venue lockout policies on drinking-related harms in Melbourne, Australia: Results from SimDrink, an agent-based simulation model

https://doi.org/10.1016/j.drugpo.2016.02.016Get rights and content

Abstract

Background

The late-night accessibility of entertainment precincts is a contributing factor to acute drinking-related harms. Using computer simulation we test the effects of improved public transport (PT) and venue lockouts on verbal aggression, consumption-related harms and transport-related harms among a population of young adults engaging in heavy drinking in Melbourne.

Methods

Using an agent-based model we implemented: a two-hour PT extension/24-hour PT; 1 am/3 am venue lockouts; and combinations of both. Outcomes determined for outer-urban (OU) and inner-city (IC) residents were: the number of incidents of verbal aggression inside public and private venues; the number of people ejected from public venues for being intoxicated; and the percentage of people experiencing verbal aggression, consumption-related harms and transport-related harms.

Results

All-night PT reduced verbal aggression in the model by 21% but displaced some incidents among OU residents from private to public settings. Comparatively, 1 am lockouts reduced verbal aggression in the model by 19% but led to IC residents spending more time in private rather than public venues where their consumption-related harms increased. Extending PT by 2 h had similar outcomes to 24-hour PT except with fewer incidents of verbal aggression displaced. Although 3 am lockouts were inferior to 1 am lockouts, when modelled in combination with any extension of PT both policies were similar.

Conclusions

A two-hour extension of PT is likely to be more effective in reducing verbal aggression and consumption-related harms than venue lockouts. Modelling a further extension of PT to 24 h had minimal additional benefits but the potential to displace incidents of verbal aggression among OU residents from private to public venues.

Introduction

There is a complex relationship between the late-night accessibility of entertainment precincts and drinking-related harms. In particular, negotiating a journey home late at night is considered a concern for personal safety (Measham & Brain, 2005) and when public transport (PT) is not available regulated closing times for public venues create spikes in taxi demand that can lead to disputes or aggression in taxi queues. This is potentially exacerbated by policies such as mandatory venue lockouts (where venues and services remain open but entrance is not allowed) that leave many people simultaneously requiring transport. In Melbourne, Australia, creating a safe 24-h city has been set as a goal of local planners, and limited late-night transport has been flagged as a current barrier (City of Melbourne, 2010).

Melbourne's CBD is heavily populated with bars and nightclubs that are popular among 18–25 year olds. Travelling from outer-urban (OU) areas for a night out in inner-city (IC) entertainment precincts is common among this population (MacLean & Moore, 2014), yet despite many venue licences and individuals’ nights out exceeding 1 am there is limited PT available after this time. Beyond 1 am, the only PT option is a ‘nightrider’ bus network that operates from the city centre (Public Transport Victoria, 2015). However, due to poor coverage, lack of connectivity and security concerns, this is not considered to be an attractive or safe option for many young adults (Duff and Moore, 2015, MacLean and Moore, 2014). This inadequacy of PT is problematic, as not only can taxi fares be in excess of AUD80—leaving individuals who are unable to afford a taxi home waiting in the street for PT to start in the morning—but Melbourne's liquor licensing means that venues share common closing times, leading to long waits for taxis and the potential for the above-mentioned disputes and aggression. To reduce these harms it has been proposed that 24-hour PT should operate on Friday and Saturday nights and a 12-month trial has been commissioned to start in January 2016. However, the benefits and indirect effects of such a trial are yet to be quantified.

It is hypothesised that the provision of inexpensive PT throughout the night will ease taxi demand and minimise the amount of people who extend their nights out while waiting for the next morning's first train. Thus, extending PT could potentially reduce disputes and lower incidental alcohol consumption. However, it is also plausible that increasing the late-night accessibility of entertainment precincts could increase alcohol consumption, particularly if the flow of city-bound individuals were to increase after 1 am, or if people who would typically catch the last train home were to extend their nights because of extra PT provisions. In both cases, the additional time spent in entertainment precincts would be after 1 am, the time when individuals are most likely to be intoxicated (Miller et al., 2013) and consequently at highest risk of experiencing consumption-related harms (i.e. drinking beyond their physiological limits, in contrast to other drinking-related harms such as experiencing verbal aggression) (Measham & Brain, 2005). These contrasting outlooks highlight the wide range of effects that could emerge from the implementation of PT-related alcohol policies.

Another policy option central to much debate within Australia is venue lockouts. In February 2014, following a rise in drinking-related violence in Sydney, a two-year trial of 1.30 am lockouts and 3 am closing times were introduced in Kings Cross and the Sydney CBD. Preliminary findings suggest that this led to a 26–32% decrease in assaults (Menéndez, Weatherburn, Kypri, & Fitzgerald, 2015) with limited displacement effects (a simultaneous 9% decrease across the rest of NSW was also reported), although it is unclear whether these decreases were simply the result of reduced pedestrian traffic at night. In 2008, Melbourne's city council introduced a three-month trial (March, April and May) of 2 am lockouts for public venues across four local government areas (LGAs) (Department of Transport, 2015)—Melbourne's CBD (including Docklands), Port Phillip, Yarra and Stonington. However, the implementation of the policy was flawed, as although there were 487 venues within this area, 120 (25%) were granted exemptions by the Victorian Civil and Administrative Tribunal, many of which were nightclubs either in close proximity to venues with lockouts or located centrally within entertainment precincts (46 out of the 85 nightclubs obtained exemptions). The large number of exemptions created community confusion about the policy that limited its effectiveness, and an evaluation found mixed results (Department of Justice, 2008). Despite this, the Sydney data would indicate the potential for success in Melbourne, in particular in the context of improved PT, and modelling these benefits would be useful for informing policy discussion.

Changes to transport and venue lockout policies are likely to affect different people in different ways, depending on where they live, where they normally drink and other personal characteristics (Callinan et al., 2015, Hart, 2015, MacLean et al., 2013, Meier et al., 2010). Often models used to test alcohol policy options inadequately capture these differences, and are therefore prone to error if results are extrapolated. Agent-based models (ABMs) are types of models that address this issue by using a set of autonomous ‘agents’ to represent a population (Gilbert, 2008). Each agent is given unique characteristics and follows simple behavioural rules to interact with others and their environment. When many agents are combined and simulated together, their individualised characteristics provide a representation of a real-world population, and large scale behavioural patterns can emerge from a multitude of local, stochastic interactions. This offers a powerful and complex method for describing human behaviour, which has been successfully applied to alcohol policy research previously (Giabbanelli and Crutzen, 2013, Gorman et al., 2006, Lamy et al., 2011).

In this paper we use an ABM SimDrink, developed in Scott et al. (2015), to virtually implement combinations of 24-hour PT and venue lockout policies in Melbourne. The model is designed to capture the net effects of alcohol policies on a population of 18–25 year old heavy drinkers, measuring the resulting prevalence of experiencing verbal aggression, consumption-related harms and difficulty getting home. The approach taken is novel because it involves simulating and tracking a population on an hourly time scale throughout the course of a night. This is consistent with the shift in contemporary alcohol and other drug research towards considering the consumption event as the unit of analysis (Bøhling, 2014, Callinan et al., 2014, Dilkes-Frayne, 2014, Kuntsche et al., 2014); researchers are attempting to understand individuals’ decisions and their consequences within a single drinking event (a ‘big night out’). By using this type of simulation model to compare hypothetical time-specific (i.e. hour of day specific) policies, this study is an example of how modelling can provide insight into the mechanisms by which interventions can affect outcomes. Further, although the model has been applied to Melbourne, these results are applicable to other settings that have similar characteristics—namely locations with a central entertainment precinct that attracts both local residents and residents from surrounding suburbs.

Section snippets

The model

SimDrink is an existing ABM (Scott et al., 2015), which simulates a population of young (18–25 year olds) people from Melbourne (either residing in IC or OU areas) meeting up with friends, who then move between private, public-niche (e.g. pubs, bars) and public-commercial (e.g. nightclubs) venues over the course of a night (Barton & Husk, 2012). The model tracks individuals’ alcohol consumption, spending and whether or not they experience verbal aggression, drink more than their physiological

Results

Using baseline parameters, IC residents in the model were more likely to be involved in verbal aggression or ejected from public venues than OU residents, and less likely to be involved in verbal aggression in private venues (Table 1 and Fig. 1); however, this is a reflection of IC residents spending more time in public venues (Scott et al., 2015) rather than IC residents having a higher propensity for harm overall. When weighted by the numbers attending each venue type, the overall modelled

Discussion

This study takes the novel approach of using a simulation model of a single drinking occasion to implement and compare time-specific alcohol policies. Using an existing ABM we simulate a population of young heavy drinkers in Melbourne in order to virtually implement and compare the effects of PT extensions and venue lockout policies on verbal aggression, consumption-related harms and transport-related harms. Further, although the setting was specified as Melbourne, the results apply more

Conclusion

Our model suggests that a two-hour extension of PT is likely to be more effective in reducing verbal aggression and consumption-related harms than venue lockouts. Modelling a further extension of PT to 24 h had minimal additional benefits and the potential to displace incidents of verbal aggression among OU residents from private to public venues. When implemented in conjunction with any extension of PT, 3 am lockouts were equally as effective as 1 am lockouts in reducing verbal aggression.

Conflict of interest statement

The authors declare no conflicts of interest.

Acknowledgements

The research reported here was funded by an Australian Research Council Discovery Project (DP110101720). The authors gratefully acknowledge the contribution to this work of the Victorian Operational Infrastructure Support Program. The National Drug Research Institute is supported by funding from the Australian Government under the Substance Misuse Prevention and Service Improvement Grants Fund. NS is the recipient of a Burnet Institute Jim and Margaret Beever Fellowship, PD is the recipient of

References (29)

  • P.M. Dietze et al.

    The big night out: What happens on the most recent heavy drinking occasion among young Victorian risky drinkers?

    Drug and Alcohol Review

    (2014)
  • E. Dilkes-Frayne

    Tracing the ‘event’ of drug use: Examining ‘context’ and the co-production of a night out on MDMA

    Contemporary Drug Problems

    (2014)
  • C. Duff et al.

    Going out, getting about: Atmospheres of mobility in Melbourne's night-time economy

    Social & Cultural Geography

    (2015)
  • J. Foster et al.

    Why do people drink at home?

    Journal of Public Health

    (2010)
  • Cited by (17)

    • Evaluation of public health interventions from a complex systems perspective: A research methods review

      2021, Social Science and Medicine
      Citation Excerpt :

      Models cannot truly capture the complexities and unpredictability of the real world, but they may be of use to decision-makers in anticipating likely impacts of interventions. Agent-based models (ABMs) were typically used to hypothesise and simulate how agents within a system might react and interact in response to an intervention (White and Levin, 2016; Adams and Schaefer, 2016; Atkinson et al., 2018; Beheshti et al., 2017; Combs et al., 2019; Luke et al., 2017; Jin and White, 2012; Kasman et al., 2019; Keyes et al., 2019a, 2019b; Koh et al., 2019; Lee et al., 2018; Li et al., 2018; Scott et al., 2016; Spicer et al., 2012; Yang et al., 2014; Yonas et al., 2013; Zhang et al., 2014). System dynamics (SD) modelling was used to hypothesise and simulate how an intervention may impact on and interact within a wider complex system (Allender et al., 2019; Araz et al., 2018; Biroscak et al., 2014; Caroleo et al., 2017; Chen et al., 2018; Eker et al., 2018; Guo et al., 2016; Guzman et al., 2013; Haghshenas et al., 2015; Hirsch et al., 2010; Honeycutt et al., 2019; Jalali et al., 2019; Kuo et al., 2016; Lich et al., 2017; Loyo et al., 2013; Lyon et al., 2016; Manohar et al., 2014; Nyabadza and Coetzee, 2017; Soler et al., 2016; Tengs et al., 2001; Tobias et al., 2010; Wakeland et al., 2013; York et al., 2017).

    View all citing articles on Scopus
    View full text