A Retrospective, Single-institution Review of Loop Ileostomy Reversal Outcomes

Although loop ileostomies are created to mitigate the risk of anastomotic leaks, their reversal is associated with significant morbidity. Complications such as ileostomy site hernia and wound infections are not uncommon. A retrospective study was conducted among 176 consecutive patients who had undergone ileostomy reversal between September 2009 and November 2012 at a tertiary care teaching hospital. The main purpose of the study was to compare differences in short-term outcomes associated with hand-sewn versus stapled anastomotic techniques and purse-string versus stapled/sutured skin closure versus wounds left to heal by secondary intention.Primary outcomes analyzed were length of stay (LOS), wound infection rates, and incisional hernia occurrence. All variables were abstracted from electronic medical records. Surgical techniques used were based on surgeon preference. Categorical variables were compared using Pearson chi-square test or Fisher’s exact test. ANOVA and the Kruskal-Wallis test were used to compare closure technique, age, body mass index (BMI), and hemoglobin; ANOVA was used to compare the remaining continuous/ordinal variables. Anastomotic technique, age, BMI, and hemoglobin were compared using the t-test, and the Wilcoxon rank-sum test was used to compare the remaining continuous/ordinal variables. Age, BMI, and hemoglobin were compared using t-test, and Wilcoxon rank-sum test was used to compare the remaining continuous/ordinal variables. Univariate analysis for preoperative factors associated with postoperative hernia was performed as a secondary analysis due to hernia rates higher than reported in the literature. A P value <0.05 was considered statistically significant. Of the 176 patients (median age 53, median BMI 26), 53% were male. The number of hand-sewn and stapled anastomoses were similar (89 versus 87, respectively) and purse-string skin closures were more common (n = 108) than stapled (n = 45) closures or wounds left to heal by secondary intention (n = 23). The mean follow up time for all patients was 16.9 months (508 days, range 8–1406 days); the overall complication rate was 35.4%. LOS was shorter in the hand-sewn anastomoses than in the stapled anastomoses’ group (3.36 days versus 3.82 days, P = 0.028) due to a greater proportion of patients with shorter stay (44 out of 89 [49%] 1 to 2 days versus 28 out of 87 [32%], hand-sewn versus stapled, respectively). No statistically significant differences were found in short-term outcomes among the 3 skin closure techniques, including wound infections (8 out of 108 [7%] in purse-string versus 2 out of 45 [4%] in stapled/sutured versus 0 out of 23 [0%] in open wounds, P = 0.84). Ileostomy site hernia (16.5%) was associated with a significantly higher BMI (32.2 with hernia versus 26.1 without, P <0.0001) and longer follow-up time (25.1 months with hernia versus 11.5 months without, P = 0.0003). In this study, anastomotic and skin closure technique did not result in clinically important differences in short-term outcomes. The results suggest that, depending on BMI, preoperative loop ileostomy closure counseling of patients should include the risk of developing an incisional hernia.  

A loop ileostomy often is devised at the time of high-risk intestinal or low pelvic colorectal anastomosis creation to mitigate the complications of anastomotic leak.¹ However, according to a prospective, nonrandomized study by Garcia-Botello et al² of 127 consecutive patients, subsequent ileostomy reversal is associated with significant morbidity, including wound infection rates approaching 20%. Other common complications include ileus, intra-abdominal abscess, bowel obstruction, nonreversal, and hernia, with overall complications rates ranging from 11.4% to 43.0%. Wong et al³ identified 1504 patients undergoing ileostomy closure after restorative proctocolectomy and found an overall complication rate of 11.4%; Gessler et al⁴ studied 262 colorectal cancer patients undergoing loop ileostomy reversal and reported an overall complication rate of 43.0%. 

A retrospective chart review was performed of patients who had undergone loop ileostomy closure between September 2009 and November 2012. One hundred, eighty-eight (188) consecutive patients were initially identified via CPT codes 44620 or 44625 using the Department of General Surgery billing database; these codes identified patients who underwent closure of enterostomy with or without resection. Twelve (12) patients were excluded because loop ileostomy closure was not the primary procedure. The Institutional Review Board approved this project.

Surgical techniques used.
Anastomosis. Two (2) approaches were used to re-establish gastrointestinal continuity after completing standard dissection of the loops of ileum of the skin, subcutaneous tissues, and rectus. In the hand-sewn approach, the bowel edges of the ileostomy were freshened by excising the mucocutaneous junction and a single layer closure of the enterotomy was performed. For a stapled side-to-side, functional, end-to-end anastomosis, the ileostomy was resected using gastrointestinal staples and the mesentery controlled with a suture tie. Anastomosis then was performed in a standard fashion with a common channel created via intestinal stapling through enterotomies in each loop of ileum and closure of the common enterotomy with an additional thoracoabdominal (TA) stapler firing.¹¹ 

Skin closure. Fascia defects were closed in a standard fashion with absorbable sutures. The skin closure technique was classified as open, stapled or sutured closed, or purse-string. Open skin wounds were packed with wet-to-dry gauze to heal by secondary intention. Stapled or sutured skin closures were standard closed surgical wounds with either a subcuticular running absorbable suture or with standard skin staples. The purse-string technique as described by Banerjee^10,11 creates a partially open wound using a running purse-string, subcuticular absorbable suture (see Figure 1). 

Data collection. Demographic information and preoperative and operative characteristics were extracted from notes within the individual patients’ electronic medical records by one of the authors. Available data for each patient were entered into a spreadsheet with predetermined data columns. Patients were divided into groups based on anastomotic technique (hand-sewn versus stapled) and skin closure technique (purse-string versus stapled/sutured versus open) as described in the procedural notes. 

Data collected included age, gender, body mass index (BMI), American Society of Anesthesiologists¹² (ASA) physical status, albumin, hemoglobin, primary diagnosis, comorbidities, procedure type, time between primary surgery and ostomy takedown, wound closure technique, operative time, days to flatus, days to bowel movement, postoperative complications, LOS, and total follow-up duration. These data were treated as numerical or categorical entries as appropriate. Obesity was defined as BMI >30; smoking exposure was defined as smoking within 1 year of surgery; warfarin or steroid usage was considered if listed as current medication preoperatively; and radiotherapy or chemotherapy was considered if administered within 6 weeks of surgery. Physical status as defined by the ASA is characterized by the patient’s health at the time of surgery and documented into the electronic medical record by the anesthesiologist as a 1 for a normal health patient, 2 for a patient with mild systemic disease, 3 for a patient with severe systemic disease, 4 for a patient with severe systemic disease that is a constant threat to life, or 5 for a moribund patient who is not expected to survive without surgical intervention. All complications were evaluated within 30 days of operation and included reoperation, surgical site infection, bowel obstruction or ileus, abdominal abscess, hernia, and readmission. Complications were determined from physician notes, usually from documentation of the finding on physical exam or as a diagnosis stated within the assessment/plan section. Reoperations were performed for complications associated with ileostomy reversal and included small bowel obstruction, postoperative ileus, gastrointestinal bleeding, abscess, and anastomotic stricture. Preoperative and operative characteristics were compared to assess for basic differences in groups and differences in the primary outcomes of LOS and wound infection rates were analyzed. 

Statistical analysis. Statistical analysis was completed using SAS Version 9.4 (SAS Institute Inc, Cary, NC). Univariate analysis for preoperative factors associated with postoperative hernia was performed as a secondary analysis due to higher hernia rates than reported in the literature. Categorical variables were compared using Pearson chi-square test or Fisher’s exact test. ANOVA and Kruskal-Wallis test were used with regard to closure technique comparison, age, BMI, and hemoglobin; ANOVA was use to compare the rest of the continuous/ordinal variables. Anastomotic technique, age, BMI, and hemoglobin were compared using t-test, and the Wilcoxon rank-sum test was used for comparing the rest of the continuous/ordinal variables. With regard to hernia development, age, BMI, and hemoglobin were compared using t-test, and Wilcoxon rank-sum test was used for comparing the rest of the continuous/ordinal variables. A P value <0.05 was considered statistically significant. 

One hundred, seventy-six (176) charts of patients who underwent ileostomy closure were identified and reviewed; 93 patients (53%) were male, median age of all patients was 53 (mean 50.3, range 18–83) years, median BMI was 26 (mean 27.1, range 14.7–46), and median ASA physical status class was 2 (65%, n = 115). The mean time between creation of the stoma and closure was 134 (range 17–635) days. The median number of days to discharge was 3 (mean 4, range 1–20). The mean follow-up time for the entire cohort was 16.9 months (508 days, range 8–1406 days). 

Patient characteristics. The patient and operative characteristics of 89 patients who underwent hand-sewn anastomosis and 87 patients who underwent stapled anastomosis were compared (see Table 1A and 1B). Patient groups were similar in age (mean age 49.5 and 51.1 years, respectively; P = 0.48), BMI (for both groups mean 27.1 kg/m², P = 0.98) and gender (number of men = 47 out of 89 and 46 out of 87, respectively [53% male], P = 0.99), and the majority of patients were ASA class 2. Significant differences were seen between groups in exposure to radiotherapy (hand-sewn = 1 out of 89 [1%] and stapled 7 out of 87 [8%], P = 0.032) and chemotherapy (hand-sewn 4 out of 89 [4%] and stapled 20 out of 87 [24%], P = 0.0003). The 2 groups were significantly different with regard to primary diagnosis (P = 0.009): inflammatory bowel disease (IBD, hand-sewn 43 out of 89 [49%] versus stapled 25 out of 87 [29%]), diverticulitis (hand-sewn 19 out of 89 [22%] versus stapled 15 out of 87 [18%]), cancer (hand-sewn 20 out of 89 [23%] versus stapled 32 out of 87 [38%]), and “other” (hand-sewn 5 out of 89 [6%] versus stapled 13 out of 87 [15%]). A significant difference also was noted in skin closure method (P <0.0001): purse-string method (71 out of 89 [80%] hand-sewn versus 37 out of 87 [43%] stapled), closed (7 out of 89 [8%] hand-sewn versus 38 out of 87 [44%] stapled), and open (11 out of 89 [12%] hand-sewn versus 12 out of 87 [14%] stapled). Finally, LOS was lower in the hand-sewn group than in the stapled (mean 3.36 days versus 3.82 days, respectively; P = 0.028). The median LOS was clinically similar in both groups (3 versus 3), but due to the greater proportion of patients with short LOS (1 to 2 days) in the hand-sewn group as compared to the stapled group (44 out of 89 [49%] versus 28 out of 87 [32%]), a significant difference was noted.

Outcomes by technique. Comparisons of outcomes by anastomotic technique were compared (see Table 2). The most frequent complications were hernia (14 out of 89 [16%] in hand-sewn versus 15 out of 87 [17%] in stapled, P = 0.79) and bowel obstruction (9 out of 89 [10%] in hand-sewn versus 12 out of 87 [14%] in stapled, P = 0.45). No significant differences in complication rates were found between techniques, including wound infections. Follow-up time was similar in both groups (median 13.9 months in hand-sewn versus 12.7 months in stapled, P = 0.30). 

Patient and operative characteristics as well as outcomes were compared for the 108 patients who underwent purse-string closure, 45 patients who underwent stapled/sutured closure, and 23 patients for whom wounds were left open (see Table 3A and 3B, and Table 4). Demographic information for the groups demonstrated similar age (49.4 years versus 52.1 years versus 50.9 years, P = 0.60), BMI (26.8 versus 26.9 versus 28.6, P = 0.48), and gender (54 out of 108 [50%] versus 27 out of 45 [60%] versus 12 out of 23 [52%] male, P = 0.53). Comorbidities and exposures for the groups also were similar, but a significantly greater number of patients in the stapled/sutured group had undergone preoperative chemotherapy (12 out of 108 [11%] versus 11 out of 45 [24%] versus 1 out of 23 [4%], P = 0.038). A significant difference between the 3 groups was noted with respect to primary diagnosis (P = 0.019): irritable bowel disease (46 out of 108 [43%] for purse-string method versus 14 out of 45 [32%] for the closed method versus 8 out of 23 [36%] for the open method), diverticulitis (23 out of 108 [22%] versus 3 out of 45 [7%] versus 8 out of 23 [36%]), cancer (28 out of 108 [26%] versus 20 out of 45 [45%] versus 4 out of 23 [18%]), and “other” (9 out of 108 [8%] versus 7 out of 45 [16%] versus 2 out of 23 [9%]). A significant difference also was noted in anastomosis method (71 out of 108 [66%] purse-string versus 7 out of 45 [16%] stapled versus 11 out of 23 [48%] left open, P <0.0001). No differences in complications rates were noted among techniques, including wound infections. 

Comparison of patient and operative characteristics associated with postoperative development of a hernia (see Table 5A and 5B) and associated complications (see Table 6) demonstrated the group that developed postoperative hernia had a higher mean BMI (32.2 versus 26.1, P <0.0001), higher incidence of obesity (20 out of 29 [69%] versus 31 out of 147 [21%], P <0.0001), and longer follow-up time (25.1 months versus 11.5 months, P = 0.0003). Twenty-one (21) out of 29 patients with hernia occurrence had documented surgical repair of their hernia. Of the 8 patients without repair, the mean follow-up was 28.0 months compared to 24.1 months in patients who underwent repair (P = 0.37). 

A diverting loop ileostomy often is deemed necessary during the creation of a high-risk colon or rectal anastomosis. The procedure is known for technical ease of creation and reversal, but a review¹³ and retrospective study¹⁴ showed the ileostomy itself can lead to complications, mainly dehydration and readmission. The ileostomy reversal procedure also can be associated with significant complications; the literature includes prospective and retrospective reviews^15-17 focused on comparing skin closure methods and anastomotic technique as an area for outcomes modification. The results of relatively substantial literature review have largely been equivocal. At the authors’ institution, a balanced proportion of techniques were employed for both ileal anastomosis and skin closure secondary to surgeon preference and individualization of patient cases.

Results of this study demonstrated no difference between hand-sewn or stapled ileal anastomosis technique in terms of operative and postoperative complications and short-term outcomes outside of a statistically significant but clinically similar difference in LOS favoring the hand-sewn anastomotic technique. No differences were noted in wound infection rates between standard skin closure and purse-string technique or healing by secondary intention. Unique to this study was the long-term follow-up and documentation of ileostomy site hernia, with an overall rate of 16.5% (29 out of 176). 

The literature examining outcomes in ileostomy closure by anastomotic technique has largely been retrospective case series reviews,^4,18 with results showing no or minimal advantage to either technique. In addition, several small, prospective, randomized, controlled trials^15,19 have been performed but small sample size has limited strong conclusions: Hull et al¹⁹ included 61 consecutive patients in their study; the HASTA trial¹⁵ of 27 centers included 337 patients. Several meta-analyses showed advantages to stapled techniques in operative time and potentially decreased ileus or small bowel obstruction rate, but these reports have not been unanimous in their conclusions. The results of meta-analyses conducted by Markides et al⁶ (14 studies) and Leung et al¹⁶ (6 studies) suggested a trend toward lower postoperative small bowel obstruction rates and decreased operative times in the stapled group but also many without significant difference between anastomotic technique. Of note: no difference in operative time was identified in this study population, and both groups had a median LOS of 3 days. However, the LOS in the stapled group was lower when comparing median length of stay using the Wilcoxon-rank sum test. This test allows for shifts in distribution of LOS to be interpreted as statistically significant. The proportion of patients with short LOS was greater in the hand-sewn patients, while the proportion of patients with longer LOS was greater in the stapled anastomosis group. Despite the statistically significant difference in LOS, the results of the present study were in agreement with the literature finding no clinical differences in patient outcomes by anastomotic technique. 

Wound infection rates comparing skin closure technique also have been studied extensively to determine whether significant reductions in wound infection rates occurred with the purse-string technique. Klink et al²⁰ conducted a 2-center, retrospective cohort study of 140 patients with surgical site infection rates of 17% in conventional primary closure versus 5% in purse-string closure (P = 0.047); Li et al¹⁷ conducted a meta-analysis including 15 retrospective studies or randomized, controlled studies involving a total of 2921 patients where  results of surgical site infection in purse-string closure were calculated at an odds ratio of 0.12 (95% confidence interval 0.02–0.40). In these authors’ anecdotal experience, lower infection rates in patients closed using the purse-string technique was not observed. Therefore, the original hypothesis that patients would not show demonstrable differences in wound infection rates based on skin closure technique, while confirmed in this study, is disparate from the previously mentioned literature suggesting lower infection rates in purse-string closure.^8,9 One potential explanation for why patients closed using the purse-string method in this study did not exhibit a decreased wound infection rate may be due to the selective practice of leaving some wounds open to heal by secondary intention. The criteria by which these wounds were not closed were not identified by chart review, but presumably this was a decision made by the surgeon during the procedure due to concern for a high risk of infection. The impact of this decision on overall infection rates is likely uncontrolled in this review.

The literature^3,4 on ileostomy closure reports overall complication rates in the range of 11.4% to 43%. This study demonstrated an overall complication rate of 35.4%, adding to the literature by providing additional data confirming the significant complication rate but with a longer follow-up time (median follow-up time of 16.9 months compared to follow-up times ranging from 1 to 33 months in the literature^3,21). In addition, significant follow-up likely allowed for inclusion of more complications — in particular, the development of ileostomy site hernia, which provides further validation of reported complication rates. This is consistent with several studies with longer follow-up times that also reported higher hernia rates: Guzman-Valdivia²² reported a hernia rate of 31.4% (22 out of 70) at a median detection time of 6 months, El-Hussuna et al²³ reported a hernia rate of 5% (5 out of 159) at a median follow-up of 95 weeks, and Saha et al²⁴ reported a hernia rate of 5.5% (18 out of 325) for a median interval from closure of 63 months. Hernia rates in loop ileostomy closure reported in prospective²⁵ and retrospective studies²⁶ range from 0% to 50%; Cingi et al²⁵ reported a 50% hernia rate in 2 of 4 patients. A range of ileostomy site hernia occurrence of up to 12% has been more realistically reported²² where length of follow-up has been correlated with hernia detection. Incisional hernias usually are considered a “late complication” that often is documented several months post-procedure, long after the traditional 30-day window for “short-term” outcomes. In this review, follow-up time was collected as most recent documented contact with the patient rather than correlation with complications. 

Results from this study indicated a strong correlation between obesity and hernia occurrence (mean BMI 32.2 versus 26.1, P <0.0001). Few studies have reported possible clinical associations with stoma site hernia occurrence. A meta-analysis²¹ and a randomized comparison²⁷ suggest hernia development may be associated with concomitant illnesses (diabetes, chronic obstructive pulmonary disease, or hypertension) or postoperative complications such as wound infection; however, these studies include colostomy closure operation as well as ileostomy closure sites. Subsequent hernia repair occurred in 21 out of 29 hernia patients (72%), which is comparable to the literature^2,8 in which hernia repair following loop ileostomy reversal ranges from 61.5% to 100% (8 out of 13 and 11 out of 11, respectively). Given the strong correlation between obesity and postoperative hernia development, it would be appropriate to counsel patients with a BMI >30 on increased risk for development of postoperative hernia and subsequent hernia repair. 

The results of this study are subject to several limitations commonly associated with retrospective cases series review and include the inability to control patient- and surgeon-related factors. The indication for ileostomy creation at the index operation, which could theoretically affect outcomes, was not known. In addition, uncontrolled intraoperative decision-making may have impacted surgical technique and outcomes, specifically with regard to some wounds being left open. Additionally, the differences in groups observed may be related to surgeon factors and their preferred anastomotic and skin closure technique. With respect to patient factors, follow-up information about complications also is limited by the retrospective study design and the observed complication estimates may be lower than the actual rates. 

The sample size also limits the external validity of the results, even though the outcomes are very similar to those reported in the literature. The sample size of loop ileostomy reversal studies ranges from 4 to 1504^24,4; the sample size of most studies^21,23,26,28 is within the range of 56 to 159 patients. 

A single institution, retrospective case series study of 176 patients undergoing loop ileostomy closure did not find clinically important differences in LOS or wound infection rates between hand-sewn or stapled ileal anastomosis technique or between purse-string, stapled/sutured skin closure, or healing by secondary intention. During a median follow-up of 16 months, 16.5% of patients developed an ileostomy site hernia. Ileostomy site hernias were significantly more common in patients who were obese and among patients with longer length of follow-up. The majority (72.4%) of ileostomy hernia site patients within this study underwent hernia repair surgery. An area of future interest and study may be the quantification of risk factors for ileostomy closure site hernia. The results of this study suggest patients, especially patients who are obese, should be counseled about the risks of hernia after ileostomy closure. 

Please address correspondence to: John Byrn, MD, University of Michigan Health System, Department of Surgery, Division of Colorectal Surgery, Department of Surgery, 2214 Taubman Center, 1500 E. Medical Center Drive, Ann Arbor, MI 48109; email: jcbyrn@med.umich.edu.