Bias in retrospective studies of trends in asthma incidence.

There is evidence for an increase in adult asthma prevalence. Several retrospective studies have shown an increase in asthma incidence by year of birth, consistent with an increasing trend in asthma incidence. The validity of this retrospective approach is unknown. Retrospective and prospective asthma incidence by year of birth were compared in the same community, during the same time period, from two independent studies; a cohort study and a cross-sectional study in Western Norway. In the prospective study, subjects without asthma were followed from 1985-1996/1997. In the retrospective study in 1998, subjects reported the age at which the disease started. Analyses of incident asthma in the period 1985-1996 were compared between the studies. The retrospective analysis showed a large increase in asthma incidence by year of birth, with an odds ratio (OR) of 2.9 comparing those born in 1969 with those born in 1927. The prospective study showed the opposite, with an OR of 0.2 comparing those born in 1969 with those born in 1927. There was only a 20% difference in the cumulative incidence of asthma. To conclude, retrospective estimates of trends in asthma incidence are likely to be severely biased by differential recall.