Els van Meijel

53 Acute pain and posttraumatic stress symptoms | Chapter 3 between these variables for boys and girls using interaction effects. However, the number of children with significant PTSS ( n = 20) or PTSD (van Meijel et al., 2015) in our sample was too low to perform logistic regression analysis, including a correction for gender and other potentially relevant factors (Peduzzi, Concato, Kemper, Holford, & Feinstein, 1996). Because of skewness of the data (many children without symptoms and few children with low pain scores), a transformation of the data offered no solution. The use of bivariate correlation precludes corrections for multiple factors. However, the use of Spearman’s rho correlation coefficients was the best alternative to obtain a reliable answer to the research question, certainly because there is a definite ordering of most of the variables in time. Although correlation coefficients should be interpreted as associations, the succession in time of the variables creates possibilities for additional interpretations. However, only an association, and not causation, can be inferred from cross-sectional data (Sedgwick, 2014). An advantage of the method we used was that pain and severity of posttraumatic stress were reported directly by the children and therefore were not biased by the interpretation of parents or professionals. Conclusions and clinical implications This study contributes to the knowledge of factors related to the risk of posttraumatic stress following accidental injury, specifically regarding acute pain. The experience of pain may be subjective, but severe acute pain is associated with the severity of later PTSS. We therefore recommend timely measurement and management of pain according to acute pain protocols in all phases and disciplines after accidental injury. Further research is needed to investigate the role of gender, to clarify the interaction between pain, injury and injury severity, and to examine the usefulness of acute pain in screening tools. Acknowledgements We gratefully acknowledge and thank Joost Daams, clinical librarian, Academic Medical Center, University of Amsterdam, the Netherlands, for his help with the literature search; Rebecca Holman, clinical statistician at Clinical Research Unit,