Els van Meijel

86 Chapter 5 | Parental posttraumatic stress Data analysis We described parental and child characteristics using counts, percentages, means and standard deviations. Differences between follow-up participants and non-participants were analyzed with Mann-Whitney U tests for the continuous variable posttraumatic stress at the time of the first assessment, and a Fisher’s exact test for the categorical variable sex. In these tests, an alpha level of .05 was considered statistically significant. We described the association between the level of parental PTSS and the level of child PTSS using Spearman’s correlation coefficient. We used univariable linear regression analysis to describe associations between the independent variables prior trauma history, acute parental stress, witnessing the accident, hospitalization of more than 1 day and severe pain, and the dependent variable parental PTSS severity as measured with the IES-R. We also added the independent variable permanent physical impairment of the child to the analyses of T2. We performed multivariable linear regression analysis using the independent variables with p < .10 in the univariable analysis. We then performed a backwards selection procedure until all independent variables had p < .05. Due to the skewed distribution of the PTSS data, we performed the linear regression analysis on log10 transformed data. To avoid taking the log10 of values of zero, we added one point to each parent’s score on the IES-R before performing the log10 transformation. To aid interpretation of the results, we back transformed the regression parameter estimates and corresponding upper and lower limits of confidence intervals. All analyses were performed using SPSS 24 (IBM Statistical Product and Service Solutions, Chicago, Ill). Results Participants In the follow-up study we included 69 families, 58 mothers (84.1%), 11 fathers (15.9%), 28 girls (40.6%) and 41 boys (59.4%). The children had been exposed to various types of accidents: 43 (62.3%) had been involved in a traffic accident, 14 (20.3%) in a sports accident and 12 (17.4%) in other types of accidents, including falls. There was no significant difference between follow-up participants and non- participants with regard to posttraumatic stress 3 months after the accident (U = 2082,