Although the adverse effects of harmful drinking on the family are well known, scientific studies quantifying the burden have been lacking. The study by Salize et al.  provides scholarly documentation showing that, after alcoholism treatment, the family or care-givers improve their life quality substantially, spend less time on caring and save money, especially because less money is spent on alcohol. Corresponding studies have been published for mental diseases, mainly schizophrenia, but not for alcohol dependence. Thus, this is a pioneering study, contributing to an increased focus on the relatives’ problems.
For every family that agreed to participate it is estimated that two or three declined, and among those included at baseline, 21 of 69 did not participate in the follow-up interview. This means that the data are based on fewer than one in four potential respondents. It is surprising that, when people were invited to take part by the drinking relative's helpers, the response rate was no higher than in many general population surveys, where the invitation comes from distant, unknown individuals. With such a low response rate the calculated results must be considered as approximations, but are still very interesting.
Improvement after treatment, however, does not necessarily mean that all the improvement is caused by the treatment. Measuring treatment effects requires an untreated control group, which the authors acknowledge is lacking. Establishing a control group by not offering treatment to people asking for treatment may be questionable or difficult. Nevertheless, the lack of a control group warrants caution in estimating how much improved life quality and financial savings may be attributed to the treatment.
Without reservations, the authors credit all improvements to the treatment, both in the title and at several places throughout the text. This issue is problematic.
All kinds of treatment for alcohol dependence have a certain proportion of patients showing improvement, probably for two reasons: in most cases, willingness to enter treatment implies contemplating the cessation of harmful drinking. In addition, treatment is most probably entered into after an exceptionally hard drinking period, so that a spontaneous ‘regression to the mean’ may be expected to take place often. Over the years, most alcohol-dependent people recover, although only a minority ever receives treatment .
Salize et al.'s paper does not mention which type of treatment the patients have received. This is surprising, as we must assume that various treatments have different effects. The paper does, however, state that the study was conducted alongside the PREDICT study (PRognostic Evaluation of Diagnostic IGRA ConsorTium), and that many of the patients were recruited from that study. According to another paper , the PREDICT study was a trial comparing acamprosate, naltrexone and placebo. In recent years, Cochrane studies have concluded that, on average, one in nine patients was helped by naltrexone , and that acamprosate gave the same result . This may indicate that only a minor part of the improvement (72.9% remaining abstinent) should be attributed to the treatment itself.
In the absence of an untreated control group, with patients who are willing to enter treatment but remain untreated, it is hardly possible to calculate how much improvement should actually be attributed to the treatment and how much is due to ‘regression to the mean’, or to the patients’ own decisions and motivation when accepting treatment.
Declaration of interest