More Detailed Data does not Always Raise the Costs: Experience from a Swedish Cost of Alcohol Study

Abstract

The aim is to investigate the effect of availability and level of detail of data on cost estimates of alcohol consumption. Using the recent Swedish cost of alcohol study as baseline, limitations on data are applied, forming two models. The costs of alcohol in Sweden are re-estimated in the two models and compared to the baseline, to establish the magnitude and direction of the bias resulting from limited data, and which results are the most sensitive to variations in data availability and level of detail. Almost all differences between the baseline and the two limited models stem from reduced availability and not the level of detail of data. However, the level of detail plays an important role for the prevalence of consumption and the alcohol-attributable fractions (AAFs) for injuries. The conclusion is that it is more important to estimate more cost components than to improve existing estimates, as differences between cost estimates are mainly driven by the availability rather than the level of detail of data. However, the level of detail in the prevalence rates of consumption and in the AAFs for injuries is likely to considerably affect the cost estimates and obtaining the best possible data in these areas should therefore be prioritized.