Methodological Aspects

Letters to the Editor regarding the article Point Prevalence and Risk Factors for Insomnia in Children and Adolescents Findings of a Population-Based Survey by Magdalena Wieder, Rainer Thomasius, and Kerstin Paschke in issue 17/2025

In their informative study, Wieder et al. provide important data regarding the point prevalence and risk factors of insomnia in children and adolescents (1). Some methodological aspects are, however, in need of critical consideration. Table 3 in the article shows the results of a multinomial logistic regression with a 4-categorical target variable. All risk factors are included into a single model as variables of influence, which means that the “table 2 fallacy” is obviously present (2). If, for example, sex is interpreted as a risk factor and it is assumed that sex affects the occurrence of anxiety or depression (3), then one adjusts with anxiety/depression for a potential mediator of the association between sex and insomnia and underestimates the total effect of female sex. The authors mention on several occasions and rightly the bidirectionality between individual risk factors and insomnia. For example, they write that insomnia in children may be a cause of parental insomnia. However, they interpret—in the abstract too—the observed association in such a way that parental insomnia becomes a risk factor for insomnia in children. This is conceivable but the direction of the association cannot be proved by using cross-sectional data. The surprising result for obesity—protective for moderate insomnia (odds ratio [OR]=0.48), increasing the risk for severe insomnia (OR=3.20)—can possibly be explained with the very small number of cases in subjects with obesity (“sparse data bias” [4]). It can be calculated that in obese persons (3% of the total sample) the OR of 3.20 is based on a maximum of two persons with severe insomnia. It would be desirable to make this more transparent by also including in Table 3 data on the number of persons in the exposure categories and on case numbers.

DOI: 10.3238/arztebl.m2025.0178

Prof. Dr. rer. nat. Dr. rer. san. Bernd Kowall

Prof. Dr. med. Andreas Stang

Institut für Medizinische Informatik, Biometrie und Epidemiologie (IMIBE)

Universitätsklinikum Essen

References

1. Wieder M, Thomasius R, Paschke K: Point prevalence and risk factors for insomnia in children and adolescents: Findings of a population-based survey. Dtsch Arztebl Int 2025; 122: 461–6.

2. Westreich D, Greenland S: The table 2 fallacy: Presenting and interpreting confounder and modifier coefficients. Am J Epidemiol 2013; 177: 292–8.

3. Resch F, Parzer P: [Anxiety and depression in adolescents]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2024; 67: 374–82

4. Greenland S, Mansournia MA, Altman DG: Sparse data bias: A problem hiding in plain sight. BMJ 2016; 353: i1981