The study protocol was approved by the ethics committee of the University of Heidelberg (S-206/2007), the trial was internationally registered (NCT 00544583, October 16, 2007), and the study protocol was published to ensure transparency of the design and analysis procedures[27]. CONTINT was initiated as a single center RCT in the Department of General, Visceral and Transplantation Surgery of the University of Heidelberg. The number of trial institutions was extended, and the trial protocol was amended to a total of 12 surgical sites on August 31, 2009. The multi-center, pragmatic, intra-operatively randomized, controlled, two-group trial, was managed and monitored by the Study Centre of the German Surgical Society (SDGC) and analyzed by the Institute of Medical Biometry (IMBI), University of Heidelberg. Reporting of the trial adheres to the recommendations of the updated and extended Consolidated Standards of Reporting Trials (CONSORT) Statement.

The inclusion/exclusion criteria, trial interventions, randomization process, definitions of endpoints and follow-up have been previously described [27]. In short, patients ≥ 18 years of age in need of an emergency midline laparotomy because of a septic focus (e.g., perforated stomach ulcer, perforated diverticulitis) with written informed consent and a life expectancy of at least 12 months were eligible for participation. While patients with previous laparotomy and a planned second look operation were excluded, patients with previous minor laparoscopic surgery (apart from colon surgery) were included.

Randomization was performed in permuted blocks using sealed opaque envelopes prepared by the IMBI. Intraoperatively, randomization took place after successful source control and abdominal lavage and before the closing of the abdominal wall.

The outcome assessors of the CONTINT trial were blinded to the patients’ trial intervention. The primary endpoint had to be assessed by a board-certified surgeon familiar with the examination of the abdominal wall and at least six months training in ultrasound.

For both groups, the distance between the stitches had to be no more than 1.5 cm and the distance from the edge of the fascia had to be at least 2 cm. For patients in the continuous suture group the abdomen was closed by a continuous, all-layer suture using two Monoplusâ USP 1 (0.4 mm diameter), 150 cm loops, which are made of a slowly absorbable monofilament material. Two sutures started at the wound edges, had to be anchored cranial and caudal of the incision and had to overlap in the middle for at least 2 cm. In the interrupted suture group Vicryl© USP 2 (0.5mm diameter), 45 cm absorbable sutures were used starting from the cranial end to the middle of the incision and then from the caudal pole also with anchoring of stitches cranial and caudal of the incision. The sutures were tied only after all the stitches had been performed. The subcutaneous tissue was not sutured, and no subcutaneous drainage was applied while the skin closure was performed with clips. Antibiotic prophylaxis and therapy were carried out according to local standards. Electric cautery was used to cut skin, the subcutaneous tissue, the abdominal fascia, and the peritoneum, carefully avoiding damage to the umbilicus. Opening of the peritoneum was performed with scissors. Abdominal drains were placed at the end of surgery.

The composite primary endpoint was the presence of burst abdomen after 30 days or incisional hernia after 12 months. Burst abdomen was defined as postoperatively missing continuity of the abdominal fascia in combination with a wound dehiscence and/or a consecutive redo surgery due to fascial dehiscence occurring up to day 30 after surgery. Incisional hernia was assessed by physical examination and abdominal ultrasound 12 months postoperatively and was defined as a fascia gap and a protruding hernia sac on ultrasound or with a clinical examination consistent with a hernia. In cases of hernia confirmed by a surgical intervention within 12 months after the index operation no ultrasound examination was mandatory.

Secondary outcome measures included length of skin and fascia incision, time needed for fascial closure, frequency of re-operation due to burst abdomen and due to any cause, frequency of abdominal re-interventions, postoperative pulmonary infection, duration of artificial respiration and postoperative hemodialysis. Furthermore, frequency of wound infection, duration of vacuum therapy and wound healing, time to first bowel movement, duration of abdominal drainage via intraoperatively placed drains and duration of closed abdominal lavage were evaluated. Finally, we assessed LOS, duration of intensive care unit stay, quality of life (by using the standardized form (SF 36)) and overall mortality.

As empirical data for the primary endpoint were not available in the planning stage, the overall rate and treatment effect regarding the primary endpoint were uncertain. Consequently, a sample size calculation was highly uncertain, and thus, the study was performed with an adaptive interim analysis. This design allowed for early stopping of the trial or, if continued, modification of design characteristics—such as recalculation of the sample size—under control of the global type I error rate. The adaptive interim analysis was planned beforehand to take place after the completion of the 12 months follow-up for 80 evaluable patients [28]. The null hypothesis was assessed by testing the effect of the wall closure procedure in a logistic regression model that takes into account the covariates “wall closure procedure” (continuous / interrupted), BMI (values as measured on the original scale), and age (values as measured on the original scale). The global one-sided type I error rate was set at \(\alpha = 0.025\) and the boundary for the one-sided p value for accepting the null-hypothesis within the interim analysis was \(\alpha_ = 0.40\). This approach is equivalent to two-sided testing of \(H_\) and assures control of the global two-sided type I error rate of 0.05 within the chosen adaptive two-stage design.

Data were described using appropriate measures of location. Due to missing documentation of endpoints and as per initial protocol, patients with missing endpoint documentation on visit 5 (12 months) or more than one occurrence of missing endpoint documentation were documented as missing while patients with at least one endpoint confirmation during the follow-up period (i.e., burst abdomen or incisional hernia) were categorized as positive cases; all others were treated as negative cases. A detailed depiction of this ruling is presented in Additional file 1: Figure S1. If a patient discontinued from the study prematurely, missing data with respect to the primary outcome variable were replaced by the Imputed Case Analysis- reasons (ICA-r) method described by Higgins et al.

The primary endpoint was investigated in a logistic regression model taking into account the group (continuous / interrupted), as well as the covariates BMI (values as measured on the original scale), and age (values as measured on the original scale). For the evaluation of the serious adverse events all available data in the database were considered. Calculations were performed using SAS software (version 9.4; SAS Institute, Inc., Cary, NC, USA). In the case of missing data in secondary endpoints, patients were excluded from statistical analysis of the outcome measure concerned. Due to the explorative nature of the trial all reported p values have to be treated as descriptive statistics without confirmatory value.