Evidence of harm from late night alcohol sales continues to strengthen.

Abstract

In their evaluation of a 1-hour increase in opening hours of alcohol outlets (from 4 a.m. to 5 a.m. on Friday and Saturday nights, and from 3 a.m. to 4 a.m. on weeknights) in central Amsterdam, de Goeij and colleagues find a 34% increase in ambulance attendances for alcohol-related injury compared with other areas within the city 1. This is the first evaluation of such a law change to utilize ambulance data, where previous studies relied on police or hospital data 2, 3. A significant limitation of police data is the confounding of deterrence and detection. How many police are on the beat and how they respond to crime can vary temporally and spatially in ways that bias inferences about the effects of law changes 4. Hospital admissions or emergency department data are afflicted by a different set of service delivery variables, e.g. policies on what constitutes an admission or episode of care 5. Such data sets typically lack systematically collected information on patient alcohol impairment 6 or even the time or location of the injury, let alone details of the event that caused injury 7. In principle, ambulance data are less susceptible to service delivery variables; being unrelated to the causes of injury, geocoding the location of the event (at least where the patient is picked up), they include diagnostic information and have the potential for linkage with hospital records. It is surprising that more use has not been made of ambulance data in epidemiological research on alcohol-related injury, and in the evaluation of interventions and policy changes. de Goeij and colleagues provide a useful example of what is possible. Authorities seeking to balance commercial interests with the problems arising from the sale and consumption of alcohol typically consider incremental rather than dramatic changes to availability. Given the low sensitivity of harm indicators and high potential for confounding in these natural experiments, evaluating small changes runs a high risk of failing to detect real effects. This is perhaps evident in the review by Hahn and colleagues 3, who distinguished studies of ≥2 -hour versus <2-hour changes in trading, finding clear evidence of effect in the former category but not in the latter, where the quality of evidence was considered inadequate to support a conclusion one way or the other. Since Hahn and colleagues’ review, studies of a 1.5-hour reduction in hours in Newcastle, Australia 8 and of 0.5–2-hour changes in both directions in 18 Norwegian cities 9 have estimated 16–25% effects on assault incidence per hour of trading. As an evaluation of a modest change in trading hours applied within a large city, the findings of de Goeij and colleagues are consistent with these recent studies, but the effect estimated is somewhat larger. Complicating matters for policy makers and scientists is the diversity of options sometimes concealed within the rubric of trading hours. Four types of restriction have been documented: (1) closing time, (2) hours of service (‘last drinks’), (3) hours in which alcohol can be consumed on the premises (‘last drinking’) and (4) lockouts. In Newcastle, outlets were required to close by 3.30 a.m. with a 1.30 a.m. lockout. In central Sydney most outlets are now required to stop selling alcohol at 3 a.m. with a 1.30 a.m. lockout, while California law prohibits drinking in licensed premises from 2–6 a.m. (http://www.abc.ca.gov/FORMS/ABC608.pdf). In addition, some countries (e.g. Norway) require the service of spirits to cease before beer and wine sales have to stop. Whether closing, ‘last drinks’, ‘last drinking’ or lockout laws differ in their effects on violence and other harm is worthy of study. There is now widespread use of lockouts in Australia and New Zealand where, after a specified time, patrons may leave but not enter premises. These are intended to prevent a mass exodus of patrons upon closing, and thereby to help police manage entertainment precincts; however, the evidence base for lockouts is small and the better studies find no effect on assault incidence (see 10 for a summary). In Australia there is widespread confusion among governments, the media, patrons and the wider public about distinctions between closing time, ‘last drinks’ and lockouts, and the times they come into effect, with lockouts or ‘curfews’ often used as blanket terms in public debate. It should be noted that all but the first type of law permit outlets to remain open, and this may be important for businesses that generate revenue from products other than alcohol, e.g. live music. It may be wise for public health advocates to avoid the terms ‘closure’, ‘lockout’ and ‘curfews’, which invoke the discourse of martial law, and to emphasize the cessation of alcohol consumption as the critical element of policy. None.