In several previous studies in CS patients, we assessed post-traumatic stress using the PTSS-10 instrument that will also be the primary outcome measure in this study. Our preliminary work indicated an 8-point difference on the PTSS-10 scale between homozygous Bc
l l *G carriers as opposed to non-carriers [
12]. This study also showed that the PTSS-10 score did not significantly decrease until six months after surgery. We may assume that the non-trauma-focused control intervention does not change post-traumatic stress levels in homozygous patients and that the difference of 8 points in post-traumatic stress scores between non-carriers and homozygous individuals is clinically meaningful. This assumption is supported by the recent study discussed in the section
Primary endpoint[
26] that showed that even a relatively low PTSS-10 score (>20) in patients one week after ICU therapy and weaning from mechanical ventilation was highly predictive for the later development of PTSD. Based on the results of our pilot study [
12], we make the following assumptions: first, the mean PTSS-10 score one year after ICU discharge is 18±8 (mean ±SD) for non-Bc
l l *G carriers and heterozygous Bc
l l *G carriers and 25±8 for homozygous Bc
l l *G carriers each under control intervention. Second, the mean PTSS-10 score one year after ICU discharge is 18 ±8 for all patients under experimental intervention. Both assumptions show a quantitative intervention-SNP interaction. An overall sample size of 252 patients (84 per SNP group) is needed assuming a type I error rate of 5% and a type II error rate of 80% to detect a statistically significant differential treatment effect between the genotype groups. The sample size calculation was performed by Nquery (Statistical Solutions Ltd.,4500 Airport Business Park, Cork, Ireland). Further, experience from former studies shows a perioperative mortality of 8% and a loss to follow-up of approximately 20% of patients at the six-month time point [
2,
12]. Adjustment for a drop-out rate of 30% leads to a total of 120 required patients (120–0.3 × 120 = 84) per SNP group. In our previous study in 126 Western European CS patients from our institution, 43% were non-carriers of the Bc
l l *G, 43% were heterozygous Bc
l l *G carriers and 14% were homozygous [
12]. This means that we will have to genetically screen approximately 872 patients to detect 120 homozygous individuals eligible for randomization. Genetic screening will also result in the identification of approximately 376 non-carriers and 376 heterozygous individuals. These patients will also be randomly allocated to the control and the targeted intervention group but, to reduce costs, only 120 individuals with each genotype will be randomly selected by choosing every third consecutive heterozygous or non-carrier patient included in the study. Thus, we will have to screen and genotype 872 patients to include 360 patients into the final analyses (Figure
4).