Page 51 - Prevention and Treatment of Incisional Hernia- New Techniques and Materials

scar perceptible or palpable by clinical examination or imaging”(18). Secondary outcomes were short-term postoperative complications (eg, surgical site infection [scored as super cial, deep, or involving organ or space, as speci ed in the protocol(13)]), burst abdomen (fascia dehiscence), cardiac events, length of hospital stay, and health-related quality of life. Main endpoints regarding quality of life were di erences between patients assigned to the small bites technique and those assigned to the large bites technique, and between patients with and without development of incisional hernia during follow-up.
Statistical analysis
We postulated a reduced incidence of incisional hernia in the small bites group. On the basis of the results of the Swedish trial(10), we calculated that 259 patients would be needed in each group to provide 80% power to detect a reduction of 50% (15% vs 7•5%) in the incidence of incisional hernia at a two- sided α level 0•05. We aimed for a total of 576 patients (n=288 per group) to correct for an estimated 10% loss to follow-up(10, 13). We analysed di erences between groups with t tests for continuous variables and χ2tests for categorical variables. For continuous variables, we tested equality of variance with Levene’s test. Normal distribution of data was tested and con rmed by limited skewness and kurtosis. We analysed the primary outcome with cross-tables with χ2testing and logistic regression to adjust for baseline covariates(19). We estimated nal treatment e ects with stratum of randomisation as a random e ect in a generalised linear mixed model. We used a binomial error and logit link function in the glmer function of the lme4 package in R statistical sofware (version 3.1.0).
Considered baseline covariates were prede ned potential predictors of incisional hernia: abdominal aneurysm aorta, body-mass index, diabetes mellitus, corticosteroid usage, preoperative chemotherapy, preoperative radiotherapy, chronic obstructive pulmonary disease (COPD), smoking, age, collagen disorders, non-incisional hernias (including inguinal hernia), and cardiovascular disease(13). For patients with missing covariate data for BMI, we imputed the mean BMI value. We assessed subgroup e ects by tests of interaction to prevent over-interpretation of apparent di erences in e ectiveness for all baseline characteristics. We chose not to do Cox-regression analysis as speci ed in the protocol. Because most patients had available two-time measurements (1 month and 1 year postoperatively), we de ned
3
STITCH trial
49