Trends in adolescent psychosomatic complaints: a quantile regression analysis of Swedish HBSC data 1985-2017.

Abstract

According to recent criticism, survey-based measures of adolescent psychosomatic complaints have poor content validity insofar as they conflate trivial with severe complaints. It is argued that this means that estimates of prevalence and trends in complaints may reflect trivial complaints that are not indicators of health problems. In this study, two observable implications of this criticism were investigated: (a) that self-reported psychosomatic complaints should have a bimodal distribution; and (b) that the increase in complaints over time should be of approximately equal size throughout the distribution of complaints. Three decades (1985/1986-2017/2018) of repeated cross-sectional data from the Swedish Health Behaviour in School-aged Children survey were used. Psychosomatic complaints were measured using the screening instrument Health Behaviour in School-aged Children symptom checklist. Histograms, bar charts and quantile regression models were used for the analysis. With regard to the first implication, the results showed that the distribution of complaints was not bimodal and that there were no clusters of respondents. This suggests that binary categorisations of students can be reductive and conceal important variations across students. With regard to the second implication, the results showed that the increase in complaints was greatest among students who report frequent and co-occurring complaints. This suggests that reports of increasing complaints in adolescents cannot be explained as being primarily due to a greater inclination to report trivial complaints. It is concluded that any conflation of trivial and more severe complaints in surveys of psychosomatic complaints is not reflected in population-based estimates.