How robust is the evidence of an emerging or increasing female excess in physical morbidity between childhood and adolescence? Results of a systematic literature review and meta-analyses.

For asthma and psychological morbidity, it is well established that higher prevalence among males in childhood is replaced by higher prevalence among females by adolescence. This review investigates whether there is evidence for a similar emerging female 'excess' in relation to a broad range of physical morbidity measures. Establishing whether this pattern is generalised or health outcome-specific will further understandings of the aetiology of gender differences in health. Databases (Medline; Embase; CINAHL; PsycINFO; ERIC) were searched for English language studies (published 1992-2010) presenting physical morbidity prevalence data for males and females, for at least two age-bands within the age-range 4-17 years. A three-stage screening process (initial sifting; detailed inspection; extraction of full papers), was followed by study quality appraisals. Of 11 245 identified studies, 41 met the inclusion criteria. Most (n = 31) presented self-report survey data (five longitudinal, 26 cross-sectional); 10 presented routinely collected data (GP/hospital statistics). Extracted data, supplemented by additional data obtained from authors of the included studies, were used to calculate odds ratios of a female excess, or female:male incident rate ratios as appropriate. To test whether these changed with age, the values were logged and regressed on age in random effects meta-regressions. These showed strongest evidence of an emerging/increasing female excess for self-reported measures of headache, abdominal pain, tiredness, migraine and self-assessed health. Type 1 diabetes and epilepsy, based on routinely collected data, did not show a significant emerging/increasing female excess. For most physical morbidity measures reviewed, the evidence broadly points towards an emerging/increasing female excess during the transition to adolescence, although results varied by morbidity measure and study design, and suggest that this may occur at a younger age than previously thought.