Our study shows that the scope of societal costs included in five economic evaluations with a societal perspective was smaller than the scope of societal costs as incorporated in four cost-of-illness studies. We found only 5 full-economic evaluation studies on preventive interventions of alcohol abuse. None of the economic evaluations included all types of societal costs that were represented in the cost-of-illness studies. This means that, although it is claimed that the societal perspective is taken, the societal costs as reported in economic evaluations are incomplete in general. In relation to alcohol abuse it is remarkable that none of the studies included drink-driving costs. Only two studies included work-related costs due to productivity losses. These results corresponds with the findings of Barbosa et al. that a societal perspective has never been taken into full account in economic evaluations [19]. The four cost-of-illness studies used to identify the healthcare and societal costs of alcohol abuse present those costs clearly in general. The only exception is that Fenoglio et al. did not include criminal damage [13].
Averaged over all economic evaluations, the proportion of healthcare costs (57%) and societal cost (43%) does not correspond with the proportion as reported in cost-of-illness studies (17% and 83%, respectively). So, the incremental cost-effectiveness ratios (ICER) that are reported in economic evaluations might be underestimations of the true societal costs and effects of preventive interventions directed at alcohol abuse.
It is known that economic evaluations of alcohol abuse interventions performed from a healthcare perspective underestimate true costs, because only healthcare costs are involved in this kind of studies. It will be useful to policy makers if studies make clear how a life-style intervention directed towards alcohol abuse affects both the health care costs and the societal costs [20]. Studies that claim to be done from a societal perspective show an underestimation too. The real part of societal costs will be much higher than presented in those economic evaluations.
Comparing productivity costs between studies is hampered by the existence of two 'schools' that address the issue of valuation of productivity losses in an entirely different manner. Advocates of the friction cost approach come-up with relatively modest estimates of productivity losses, whereas those using the human capital approach may easily reach estimates that are ten to hundred times higher than those generated by frictionists. There is no easy solution to this problem, and both approaches may hold best arguments, depending on the local circumstances of the analysis. In the meantime, absolute clarity about methodology used to derive productivity costs estimates is indispensable.
Two studies presented the QALY as an outcome. In studies done from a healthcare perspective the QALY as an outcome measure is almost standard practice. Although economic evaluations presenting the QALY as outcome measure are more or less comparable with each other, the way the QALY is measured in a public-health intervention performed from a societal perspective is debatable. The questionnaires to estimate the QALY are directed on health status, while in a societal perspective also other benefits besides healthcare, like social welfare, should be involved; these are not included in those questionnaires [21]. In general, it will be useful to think about the method used to make an economic evaluation of public health programs. Recent literature gives arguments to change from the cost effectiveness/cost-utility analyses to cost-benefit analyses (CBA). CBA does assess how social welfare is affected by an intervention, by identifying and measuring all costs and benefits. All gains, also health gains, are expressed in monetary terms [22]. Clearly this would represent a more holistic approach to evaluation of different alcohol policy measures. However, for obvious reasons, CBA is rare in all economic evaluations, not only for those of public health programmes.
Some limitations of this study have to be taken into account. We calculated per-study averaged total healthcare costs and societal costs, but it is quite difficult to compare costs as reported in different economic evaluations. This was due to differences in study design, outcome measures, discount rates, et cetera. For example, three studies used 'preventing the decrease of drinking' and 'abstinence' as a surrogate outcome measure. Another limitation is that the cost-of-illness studies showed also differences in the types of costs involved in the studies. That means that a full comparison of the studies is not really possible, although these studies are more complete than the economic evaluations. Furthermore, in the original studies, it is not always clear which costs are placed under the types of costs. It will be possible that 'property damage' costs for instance are placed under the topic 'criminal damage'.