Bibian van der Voorn

184 CHAPTER 11 Our review has several strengths and limitations. Our strengths lie in the systematic and extensive search performed, which has resulted in the inclusion of 81 studies. Our review is limited by the previously mentioned concerns, but also by the broad range in ages as well as the lack of (reliable) establishment of pubertal stage in the majority of the included articles. Although several studies mention an effect of age or pubertal status on cortisol reactivity, 13,22,24,30,33,34,36,37,39,45,48,50-52,54,57,58,64,65,67,84,95 findings are conflicting between the articles. Moreover, we ourselves were unable to draw any conclusions with regard to age or pubertal status, due to the heterogeneous ways of analyzing these effects as well as limited power within studies. Moreover, pubertal status was often assessed through self-report, which has poor reliability. 112 However, it is possible that the effect of age and/or pubertal status can partly explain our unequivocal conclusions regarding sex differences, as was previously suggested by Jessop and Turner-Cobb. 10 Aside from standardizing the collection and presentation of data, we therefore urge to also always take age and pubertal status into account. This is in line with a recent study in adults, which showed that adjusting for sex hormones significantly alters sex-specific cortisol profiles 113 . CONCLUSIONS In conclusion, we found that gender differences in HPA axis reactivity appear to be present in childhood, suggestive of priming of the HPA axis during early development, although the evidence is not unequivocal. Overall, girls appear to have a more variable diurnal rhythm, a higher CAR, and a higher cortisol response to social stress tests. These differences are not in line with studies in adults, which might be due to changes in gonadal hormones during puberty impacting on HPA axis reactivity. We found various gender differences depending on the type of stressor applied, which stresses the importance of taking the nature of the stressor into account when designing a new study. Moreover, standardization of protocols and reports of results is warranted.