Introduction

Leakage of low colorectal, coloanal or ilealpouch-anal anastomoses is a severe complication leading to considerable morbidity and mortality [1]. Defunctioning ileostomies are constructed routinely in these cases to mitigate the associated clinical consequences such as intraabdominal abscess formation which potentially causes pelvic sepsis, prolonged postoperative stay, impaired quality of life or even higher recurrence of cancer [2, 3]. Loop ileostomies are also constructed whenever considered necessary during emergency surgery including traumatic or inflammatory bowel perforations with severe peritonitis, intestinal haemorrhage or neonatal necrotizing enterocolitis [4]. Ileostomy closure, on the other hand, requires subsequent hospital admission with overall postoperative morbidity rates ranging from 3 to 40% and mortality from 0 to 4% [5, 6].

Today, surgical training is increasingly confronting with rising demands such as providing high quality (i.e. low morbidity), cost-effectiveness (in the era of diagnosis-related group reimbursement) and decreasing operative exposure due to work hour restrictions.

Resident participation in procedures is essential for surgical education [7]. However, the call for high quality (i.e. low morbidity) surgery and cost-effectiveness (in the era of diagnosis-related group reimbursement) on the one hand, but decreasing operative exposure due to work hour restrictions on the other hand result in increasing educational demands. Data on the safety of common general surgical procedures—such as laparoscopic appendectomy (LA), cholecystectomy (LC) or hernia repair (HR)—conducted by residents are contradictory, mainly due to the shortcoming of inconsistently collected data and/or arbitrary defined postoperative complications [8,9,10,11,12,13]. However, Loiero and colleagues conducted a well-designed Swiss trial with 2715 general surgeries on the impact of residency training level on surgical quality by using the comprehensive complication index (CCI) as a measure of postoperative morbidity [14]. By adjusting for important confounders, they could not detect any difference in quality outcome in respect to proficiency level. These results lead the authors to the statement that patient safety in general surgery represented by LA, LC and HR performed by surgical trainees is established [15].

Ileostomy reversal is generally considered a simple procedure and therefore a typical teaching procedure for young residents learning the construction of a proper bowel anastomosis. Our institutional approach of applying hand sewn techniques offers ideal prerequisites for surgical trainees to gain confidence with intestinal wall suturing before being confronted with more complex procedures. So far, there is only little evidence on the impact of surgical training level on postoperative outcome. The aim of this study was therefore to compare postoperative morbidity of ileostomy reversal performed by either consultants or residents (surgical trainees).

Methods

All consecutive adult patients undergoing closure of a defunctioning ileostomy between September 2008 and January 2017 were identified from a prospectively collected surgical administrative database. Patients were included independently of the underlying disease or the indication for constructing a loop ileostomy. Patient and treatment characteristics were retrospectively collected from patient records. All procedures performed were in accordance with the standards of the institutional ethical committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Baseline characteristics included age, sex, body mass index (BMI), American Society of Anaesthesiologists (ASA) classification, health insurance coverage (private vs non-private), co-morbidities such as smoking, diabetes mellitus, immunosuppression or cancer as the primary disease as well as the type of primary surgery and indications for primary (cancer, diverticulitis, inflammatory bowel disease, ileus) or secondary diversion (complications following primary surgery, i.e. anastomotic leakage).

Operative data included operative time, median time to ileostomy closure, proficiency levels (resident vs. consultant), operative technique (end-to-end, end-to-side, side-to-side anastomosis), the need for meticulous adhesiolysis, bowel resection or performing a midline incision as well as median length of hospital stay. Postoperative morbidity and mortality were ranked by the Clavien-Dindo classification [16]. Those with a Clavien-Dindo score ≥ 3 were considered as major. For patients who developed more than one complication, each was counted. The following complications were recorded: surgical site infections (SSI), urinary tract infections (UTI), pneumonia, intraabdominal abscesses (confirmed by abdominal imaging and requiring interventional drainage) as well as anastomotic leakage or bowel obstruction (either temporary obstruction requiring reinsertion of a nasogastric (NG) tube, nil per os and pharmacological treatment or manifest obstruction, i.e. ileus, requiring reoperation).

According to our standard protocol, ileostomy reversal is usually performed 8–12 weeks after construction, except for cases with severe stoma-related problems (recurrent acute renal failure, dehydration, electrolyte disturbances and severe skin problems). The reversal is performed either by a consultant or a resident supervised by a board-certified senior surgeon or consultant (depending on the insurance type of the patient and the severity of the case). In detail, patients with a complicated course after primary surgery, patients needing laparotomy or difficult cases with complex parastomal hernias and/or dense adhesions are predominantly operated by a senior or consultant surgeon. We usually perform a peristomal incision. The stoma is dissected from the mucocutaneous junction and delivered from the rectus sheath and peritoneal cavity by sharp dissection. Ileostomies are usually closed extraperitoneally in an end-to-end fashion by a single layer of interrupted 4-0 Vicryl sutures (Gambee technique [17]), either as hemi-anastomosis or in case of segmental bowel resection as full-anastomosis. Rectus sheaths are closed. After inserting a subcutaneous drain, the skin is closed by interrupted sutures. NG-tubes are removed after surgery and patients initiated a liquid diet until first bowel movements, followed by a normal diet.

According to distribution, descriptive data were reported as median with interquartile range or median +/− standard deviation. Continuous variables were analysed using the Mann-Whitney U test or the independent T test according to the distribution. Categorical variables were analysed with the chi-square-test or Fisher’s exact test. A p value ≤ 0.05 was considered to be statistically significant for all the analyses performed with IBM SPSS Statistics, version 24.0 (IBM Corp., NY, USA).

Results

Between September 2008 and January 2017, 558 adult patients underwent ileostomy reversal. Approximately two thirds were male (61.8%), median age was 63 years (20–93), median BMI was 24.9 kg/m2 (15.5–51.1) and the most common ASA scores were 2 or 3 (97.7%). One hundred one patients (18.1%) reported a history of smoking habitually, 69 patients (12.4%) were diagnosed with type 2 diabetes mellitus and 32 patients (5.8%) were under immunosuppression. Two hundred seventy-five patients (49.3%) had a history of colorectal cancer. One hundred seventy-six patients (31.5%) received neoadjuvant and 171 received adjuvant treatment according to the individual tumour stage (30.6%). One hunderd twenty-one patients (21.7%) had private health insurance coverage (Table 1).

Table 1 Clinical patient data
Full size table

The initial creation of a loop ileostomy during major surgery was planned in 380 (68.1%) patients, whereas loop ileostomy was performed in 178 patients (31.9%) in an emergency setting. Most of these latter cases were due to colon perforation or anastomotic leakage. We performed 482 (86.4%) primary open surgeries, 382 (68.5%) colorectal resections, 157 (28.1%) primary or secondary loop ileostomies and 19 (3.4%) small bowel resections. Indications for the construction of a defunctioning ileostomy were low anterior rectum resections due to cancer (43.7%), anastomotic leakage (14.9%), diverticulitis (10.4%, most cases in terms of perforated diverticulitis with peritonitis), inflammatory bowel disease (9.1%) and ileus (8.4%) (Table 2).

Table 2 Primary operation data
Full size table

Median time to ileostomy closure was 225 days (3–1693) with a median operation time of 93 min (23–614). A peristomal incision was adequate in all but 49 cases, in which conversion to a midline laparotomy was necessary due to massive adhesions (8.8%). Five hundred thirty-two (95.5%) stoma reversals were constructed by an end-to-end, 3 (0.6%) by an end-to-side and 22 (3.9%) by a side-to-side anastomosis. (Table 3).

Table 3 Ileostomy closure—procedure related data
Full size table

Overall morbidity according to Clavien-Dindo classification was 45.2% (Table 4). For patients suffering from more than one complication, each event was calculated separately which resulted in a higher number of total events per classification category. Major complications, i.e. a Clavien-Dindo score ≥ 3, occurred in 61 patients (10.9%). Seventeen patients (3.0%) developed an anastomotic leakage requiring reoperation. Twelve patients (2.2%) developed an ileus, all but one (treated by endoscopic decompression) needed a reoperation. Two patients (0.4%) died, one of septic multiorgan failure due to an anastomotic leakage with faecal peritonitis, the other one of severe postoperative pneumonia. With regard to infectious complications, the rate for both superficial and deep surgical site infections was 13.6%. 2.9% of patients developed a urinary tract infection, 0.9% pneumonia. Five patients (0.9%) required a percutaneous drainage due to intraabdominal abscess formation (Table 4).

Table 4 Ileostomy closure—morbidity and mortality
Full size table

Two hundred three (36.4%) patients were operated by residents and 355 (63.6%) ileostomies were closed by senior surgeons (board certified) or consultants (Table 5). One hundred twenty-one patients had private health insurance coverage on the condition that the operation was performed by the Chief of Department or one of his consultants. There was no significant difference in baseline characteristics (age, ASA score, BMI, smoking habit, diabetes, immunosuppression) between both groups. Operation time (94 vs. 76 min, p < 0.001) and hospital stay (6 vs. 7 days, p < 0.001) differed significantly. Residents needed longer for the operation, whereas patients treated by consultants stayed longer in hospital. Forty-four (89.8%) laparotomies were done by consultants, 5 (10.2%) by residents (p < 0.001) All ileostomy closures performed by residents were constructed by an end-to-end anastomosis with interrupted Vicryl sutures using the Gambee technique, either by direct closure or following a segmental resection. Consultants decided to perform an end-to-side anastomosis in 3 cases and a side-to-side reconstruction in 22 cases (p = 0.001). No differences were detected in terms of adhesiolysis or segmental bowel resection. With regard to surgical site infections, leakage rate or bowel obstruction showed equal results in both groups. When focusing on postoperative complications according to the Clavien-Dindo classification, there was a significant difference with more grade 1 and 2 complications (37.4 vs. 28.6%), more grade 3a and 3b complications (31.8 vs. 18.2%), three out of four grade 4a complications, all grade 4b complications and both deaths occurring in the consultant cohort (p = 0.032, Table 5).

Table 5 Comparative analysis—resident vs. consultant
Full size table

In a further analysis, we decided to exclude patients with non-private insurance status primarily operated by a consultant and/or patients requiring laparotomy (Table 6). These parameters were considered surrogate markers for difficult cases/more complex surgeries and thus possible confounders beyond patients’ baseline characteristics. There were no differences detected regarding age, ASA score, the percentage of smokers or patients with diabetes and the need for immunosuppression. Operation time and length of hospital stay remained significantly different. We observed a trend of more bowel resections performed by residents (46 vs 35.2%, p = 0.071). Higher numbers of minor postoperative complications (Clavien-Dindo grade 1 and 2) were recorded in the consultant cohort and thereby maintained a statistically significant difference in overall morbidity (p = 0.001, data not shown). However, when focusing on major complications (Clavien-Dindo grade ≥ 3), this difference dissolves (p = 0.948) (Table 6).

Table 6 Comparative analysis w/o laparotomy—resident vs. consultant (w/o non-private cases)
Full size table

Discussion

Faecal diversion drastically reduces the number of symptomatic dehiscences of colorectal, coloanal or ilealpouch-anal anastomoses [2]. Besides the risk of stoma-related complications (dehydration, acute renal failure, parastomal hernia, stoma prolapse, severe parastomal skin alterations), as well as the fact that up to 25% of defunctioning stomata become permanent [18, 19], ileostomy closure is associated with a high rate of morbidity ranging from 3 to 40% and mortality from 0 to 4% [5, 6]. Up to 10% of major complications were reported in a large series of more than 5000 ileostomy reversals [20]. Differences in study design, heterogeneity in study populations as well as inconsistency and incompleteness in outcome reporting account for broad data variability [16, 21, 22]. Consequently, the interpretation (direct comparison) of outcome data after ileostomy reversal remains demanding. Through consistent data acquisition of postoperative complications, our series of 558 ileostomy reversals came up with an overall morbidity of 45.2%. For patients suffering from any postoperative complication according to the Clavien-Dindo system, each event was calculated separately resulting in a higher number of total events per classification category (especially grade 1 or 2 events). When focusing on major complications, i.e. postoperative events with a Clavien-Dindo score ≥ 3, we found 61 cases (11%) including 2 deaths (0.4%). These results are in line with previous reports [5, 20].

Resident participation in surgical procedures is essential for training, and it remains of major interest to examine quality of surgery and safety of patients undergoing surgical procedures in a teaching environment [14]. The demand for high surgical quality (i.e. low morbidity) and cost-effectiveness in combination with working hour restrictions and a change in young residents’ attitude to fully commit to the strenuous efforts of surgical training (i.e. generation Y) impede the generation of an appropriate organisational framework for surgical education [23]. Nonetheless, large-scale resident involvement in surgical procedures proved to be safe in the present setting [14, 24].

Ileostomy reversal is considered a good learning model to create intestinal anastomoses. As a commonly performed procedure, it allows appropriate learning curves for young trainees to become familiar with intestinal wall suturing. It thereby represents an ideal bridge to more complex bowel surgery. These considerations explain our institutional approach to apply hand-sewn techniques. Data on ileostomy closure remain controversial with regard to proficiency levels in terms of postoperative outcome. The HASTA trial, a randomized controlled multi-centre study, investigated the best surgical technique for closure of loop ileostomy. In the multivariate analysis of potential confounders, surgeons’ training levels (40% of closures were performed as teaching operation) did not correlate with postoperative bowel obstruction rates, the study’s primary endpoint [25]. Another study was conducted in order to identify risk factors for reoperation after ileostomy reversal. Anaemia remained the only risk factor in multivariate analysis, whereas training levels of participating surgeons did not have a relevant prognostic value [26]. Luglio and colleagues analysed 298 ileostomy closures which were performed by junior residents supervised by senior residents [27]. Postoperative complications were compared to the results of the HASTA trial, which may be considered as a limitation as a matched cohort for direct comparison of the cohort was missing. Morbidity was found to be very low leading the authors to argue, that a senior resident’s supervision might be sufficient to avoid higher complication rates [27]. These results become partly contradicted by two consecutive Dutch studies. The initial trial reported major morbidity rates of 11% [28] that could have been significantly reduced to 4% by changing surgical proficiency levels [29]. Stoma reversal being performed or supervised by a specialized colorectal surgeon was found to be associated with a lower risk of 30-day postoperative morbidity. These results, however, should be interpreted with caution as there is a relevant heterogeneity with regard to the type of construction as well the anastomotic technique when the two cohorts are compared [29].

In our comparative analysis on training levels, we detected a significantly longer operation time in the resident group (94 vs. 76 min, p < 0.001) representing the young surgeons’ learning curve. Postoperative complications according to the Clavien-Dindo classification did show statistical significance between both groups in overall morbidity (p = 0.032). All but one Clavien-Dindo ≥ 4a (n = 10) complication was recorded in the consultant cohort, emphasizing the difficulty of those cases that were primarily operated by consultants (including all non-private patients). Additionally, 89.9% of all laparotomies were performed by consultants, a further surrogate marker for complex procedures. Therefore, when adjusting for these possible confounders, baseline characteristics remained similar, but major postoperative morbidity was not found to be still statistically different between both groups (p = 0.948). Moreover, operation time was drastically reduced in the consultant group from initially 76 min down to 64 min. We detected a trend towards more bowel resections performed in the resident group (p = 0.071). This might reflect the different performance levels of our trainees’ preparation techniques leading to serosal or transmural tears which require subsequent segmental resection.

In conclusion, our results extend previous observations that ileostomy reversal in low-risk patients represents an optimal teaching procedure for trainees to become confident with the construction of bowel anastomoses. The application of one standard technique allows fast and sustained learning curves. Supervision by an experienced surgeon remains mandatory to obtain acceptable postoperative morbidity rates.