[Statistical methodologic comments on epidemiologic work disability trends].

The term "epidemiological absenteeism trend"-especially as used in Germany-was defined by Wagner in 1977 and has since entered the sickness-absence discussion as an outcome indicator of medical rehabilitation. In its key concepts, the term refers to the fact that although there is a constant trend in absenteeism over time, individuals show changes in absenteeism from one measurement to the next. Evaluations of those individuals who have very high (or low) absence rates change to a higher degree than do those of individuals with mean scores on the absenteeism index; nevertheless, the summation of all deviations will be zero. In evaluating absence reduction due to rehabilitation, this fact must be considered. Because people who take part in medical rehabilitation show high absenteeism before they start its intervention, there will be--according to Wagner's absenteeism trend--a higher rate of reduction in sickness absence in the time following rehabilitation than there will be for people with medium scores. This paper investigates the hypothesis that the epidemiological absenteeism trend is identical with Galton's discovery--roughly 100 years before Wagner's definition--generally referred to as "reversion". Galton developed this term as the fundamental concept for regression and correlation. It can be shown that the absenteeism trend does not give a special description of sickness-absence but is rather the results of applying regression analysis. With the help of the statistical method of log linear modelling, a re-analysis of Wagner's data is performed in which the term epidemiological absenteeism trend is not necessary. In consequence, it is argued that this term should be abandoned or given a new meaning.