Using simulation modelling to examine the impact of venue lockout and last‐drink policies on drinking‐related harms and costs to licensees

Abstract

Objective: Many variations of venue lockout and last-drink policies have been introduced in attempts to reduce drinking-related harms. We estimate the public health gains and licensee costs of these policies using a computer simulated population of young adults engaging in heavy drinking. Method:Using an agent-based model we implemented 1 am/2 am/3 am venue lockouts in conjunction with last drinks zero/one/two hours later, or at current closing times. Outcomes included: the number of incidents of verbal aggression in public drinking venues, private venues or on the street; and changed revenue to public venues. Results:The most effective policy in reducing verbal aggression among agents was 1 am lockouts with current closing times. All policies produced substantial reductions in street-based incidents of verbal aggression among agents (33-81%) due to the smoothing of transport demand. Direct revenue losses were 1-9% for simulated licensees, with later lockout times and longer periods between lockout and last drinks producing smaller revenue losses. Conclusion:Simulation models are useful for exploring consequences of policy change. Our simulation suggests that additional hours between lockout and last drinks could reduce aggression by easing transport demand, while minimising revenue loss to venue owners.Implications for public health: Direct policies to reduce late-night transport-related disputes should be considered.

Applying a complex systems perspective to alcohol consumption and the prevention of alcohol‐related harms in the 21st century: a scoping review

“…The studies we identified conceptualized, described or modelled systems of interest to alcohol-harm prevention research primarily in the United States (n = 38) [22,23,29,, Australia (n = 17) [28,31,[77][78][79][80][81][82][83][84][85][86][87][88][89][90][91] and the United Kingdom (n = 7) [26,[92][93][94][95][96][97]. We also found examples of generic alcohol systems (n = 16) [24,93,[98][99][100][101][102][103][104][105][106][107][108][109][110][111], as well as examples from Sweden (n = 2) [112,113], South Africa (n = 2) [114,115], Canada (n = 1) [116], Denmark (n = 1)…”

“…These refer to computational models which model non-linear causal relationships between system elements, which may operate at varying temporal and spatial scales in order to understand emergent patterns of system behaviour [22,31]. The majority of the dynamic simulations we identified described the process of developing, or developed, ABMs (n = 29) [28,31,[42][43][44][45][46]58,60,62,64,[74][75][76][77][78]81,83,84,[87][88][89][90]92,[96][97][98]107,119] or advocated for the use of, or developed, system dynamics (SD) models for alcohol-harm prevention (n = 9) [22,23,26,59,60,73,91,99,100]. ABMs model individual agents with different personal characteristics who interact with other agents according to 'rules' that govern their behaviour within a specific environment [42,83].…”

“…The majority of the dynamic simulations we identified described the process of developing, or developed, ABMs (n = 29) [28,31,[42][43][44][45][46]58,60,62,64,[74][75][76][77][78]81,83,84,[87][88][89][90]92,[96][97][98]107,119] or advocated for the use of, or developed, system dynamics (SD) models for alcohol-harm prevention (n = 9) [22,23,26,59,60,73,91,99,100]. ABMs model individual agents with different personal characteristics who interact with other agents according to 'rules' that govern their behaviour within a specific environment [42,83]. SD models represent the inter-relationships between system elements and how behaviour is governed through feedback loops.…”

“…In contrast to ABMs, SDs focus less on individual agents and more on 'population-level influences and whole-system dynamics' ( [31], p. 2). ABMs and SD models have been utilized to understand the dynamics of individual and social drinking behaviour [44,64,[74][75][76]81,92,96,99,107,119], to explore how individuals and their networks interact with their broader environment [22,23,43,59,97,98] and to predict outcomes stemming from the introduction of an intervention or range of policy options [26,28,31,42,45,46,58,62,73,77,78,83,84,[87][88][89][90][91].…”

“…However, some studies theorized that drinking environments or societal norms and roles may also influence consumption and alcohol-related harms [42,43,[74][75][76]. A large number of the models drew on empirical data, both primary [65,81,96] and secondary data from a range of sources, including academic literature [28,31,44,46,48,61-63, 73,78,81,83,84,87,91,97,99,104,116], censuses [42,45,75,76], cohort studies, surveys and local and national data from public agencies [26,28,31,42,43,45,46,48,[60][61][62][63][66][67][68][74][75][76]78,81,83,84,87,91,92,97,104,116,119]. Apostolopoulos et al advocated for participatory model building [23], and we identified one model that used stakeholder engagement in the model-building process [28,31,…”