Comparison of Stayers, Dropouts, and Newcomers in a Longitudinal Population Study of Asthma and Bronchial Hyperresponsiveness: Introduction of Bias?

Abstract

A random sample of children and adolescents from the general population in Copenhagen, Denmark, has been examined twice (6 years apart) with respect to asthma, allergy, and nonspecific bronchial hyperresponsiveness. To investigate potential bias resulting from loss of baseline subjects at follow-up (dropout bias) and bias resulting from subjects entering the study at the follow-up stage (newcomer bias), stayers (subjects who participated in both examinations) were compared to dropouts and newcomers, respectively. The sample consisted of 983 subjects (aged 7–17 years at the time of the baseline study), of whom 408 (199 boys) participated in both examinations (stayers), 119 (62 boys) in the baseline study only (dropouts), and 257 (124 boys in the follow-up study only (newcomers). Thus, a total of 784 subjects (80% of the sample) were examined either once or twice. At baseline dropouts did not differ from stayers with respect to anthropometric data, smoking habits, pulmonary function, or prevalence of positive skin prick tests, bronchial hyperresponsiveness, asthma, and allergic diseases. Likewise, these variables for newcomers were not different from those of the stayers apart from a significantly higher smoking rate in newcomers (45% vs. 32%, p = 0.003). Based on these findings and the high overall response rate, it seems reasonable to assume that the group of stayers is representative of the whole sample, apart from an underestimation of the number of smokers and, therefore, an underestimation of the risks associated with smoking. The findings in the present study suggest that the possible effect of dropout and newcomer bias needs to be assessed, especially concerning self-selected risk factors, before final evaluation of data in longitudinal population studies.