The paper by Wells et al. raises interesting questions about the phenomenon now called ‘pre-drinking’ and its implications for research and prevention. To call off-premise consumption of alcohol ‘pre-drinking’ is to assume that there is a main event which it precedes. Thus the term and the paper both tend to assign on-premise drinking the status of main event; but, depending on the young drinker's age and context, this may not be the case. On-premise drinking currently accounts for considerably less than half of all drinking in most societies (Britain and Ireland are still exceptions). Those under legal drinking age will do most of their drinking off-premises, so that most drinkers begin their drinking careers as off-premise drinkers. We know more of the phenomenology of on-premise than off-premise drinking because it is easier to observe and to study, but whether and when on-premises drinking is the main event for young drinkers is a matter for investigation rather than assumption.
Wells and her colleagues tend to assume that drinking will be less problematic on-premise than off-premise. There is a logic here around the formal control of the drinking environment: it is usually easier for the state to intervene in drinking occasions in public places than in private. The bartender and bouncer have responsibilities, potentially enforced by economic threats of licence suspension or removal. It is tempting to conclude from this that on-premise drinking will be more controlled than off-premise. This was also the policy assumption some years ago in Finland. However, Partanen  found, comparing the same individual's drinking occasions lasting 2 hours or more at home and on-premises, that Helsinki respondents drank about the same amount in the two contexts, but did it over a longer period of time at home, resulting in less intoxication. A recent British case–control study of young offenders found that the respondent having assaulted someone was associated with going out to a nightclub (not quite significant), associated negatively with going to a pub, and showed no association with other incident contexts . Inspection of the data shows that, combining the results for pubs and nightclubs, there is no significant difference between on-premise and off-premise results.
These studies suggest that whether on-premise drinking is less problematic than off-premise drinking should be—again—a matter for investigation rather than assumption. Clearly, it can go both ways. Furthermore, policies that can reduce harm from on-premise consumption apply regardless of whether pre-drinking takes place. For instance, in today's ‘night-time economy’, there is a high premium on assembling large drunken crowds in a small area, increasing the odds of trouble. Thus Hadfield  reports that British pub rentals are twice as expensive in such ‘entertainment districts’ as elsewhere. Allowing such clustering of outlets is against the interest of public health and order, regardless of whether patrons consume most of their alcohol in pre-drinking sessions, or within the pubs themselves.
Perhaps the most important point made by Wells et al. is that ‘getting drunk appears to be an underlying motivation for drinking (and pre-drinking) among many young people’. This point is supported, for instance, by data from the Victorian Youth Alcohol and Drug Survey of 16–24-year-olds in 2004. To the question, ‘What proportion of the times when you are drinking do you intend to get drunk?’, 20% of drinkers said ‘every time’ or ‘most times’, 53% said ‘some’ or ‘a few times’ and only 27% said ‘never’[5, p. 11]. Getting drunk on a young person's weekend night out in many societies is often not accidental; it is often intended.
To reduce problems from drinking, it may be more important to start from this point rather than from a concern about whether policies favour on-premise or off-premise drinking. Faced with the intention to get drunk, the main policy choices are to dissuade from or frustrate the intent, or to channel the drinking, or organize the drinking environment to reduce the harms. Such measures as high taxes, shorter opening hours (both off-sale and on-sale), banning promotions, server interventions and making low-alcohol beverages differentially available would all potentially serve as efforts to counter the intention to get drunk. Drink-driving counter-measures, provision of late-night public transportation, bans on drinking in public places and training bar staff or re-designing bars to minimize conflict are all potentially efforts to channel the drinking or organize the environment to reduce harms. The crucial issue, in evaluating the public utility of all these measures, is not whether they favour ‘predrinking’ or on-premise drinking, but what they do to the rates of harm to the drinker and to others from a night of drinking, whether the drinking occurs off-premise, on-premise, or in both contexts.