Unplanned readmissions after colorectal surgery are a common and costly problem. Studies from have shown 30-day readmission rates after colorectal surgery are increasing in the United States, from an estimated 10.2 % between 1986 and 1990 to 13.7 % between 2001 and 2005 [1]. For colorectal surgery alone, the cost of readmission is estimated $9,000 per episode, accounting for $300 million annually nationwide. With new penalties for readmitted patients, postoperative readmissions are emerging as a quality indicator in surgery [2, 3]. To date, readmissions have not been able to be predicted from the postoperative course [4]. Previous work has attempted to identify predictors of readmission; distal bowel resection, benign diagnosis, young age and old age, more comorbidities, social deprivation, surgical field contamination, longer operative times, hemoglobin levels <12 g/dl, and absence of air testing of a colorectal anastomosis were all found to be related to readmissions [5–7]. These variables have not been consistent or used to enact change and improve outcomes.

In previous work, we found the ileostomy reversal subgroup to be a source of higher readmissions, with a 30-day readmission rate of 12.4 %, versus our overall readmission rate of 5.5 % [8]. Ileus/small-bowel obstruction was the most common cause for readmission. Readmissions had a median length of stay (LOS) of 4 days, and more patients required home care services or temporary skilled nursing after discharge from the readmission episode-all significant sources of healthcare utilization. The next logical step would be to find the causes for the readmission problem in this group.

Our objective was to identify factors related to readmission in ileostomy reversal patients. By identifying these factors, we could develop interventions to identify these high-risk patients prospectively and alter discharge criteria to reduce unnecessary readmissions and improve healthcare utilization.

Materials and methods

After obtaining Institutional Review Board approval, review of a prospective department database was performed from August 2006 to August 2012 to identify patients who underwent an ostomy reversal. The patients were identified by current procedural terminology (CPT) code (44620, 44625, 44626, 44227, and 44320) and considered for evaluation. The ostomy reversal had to be the primary procedure for the episode of care for inclusion. Patients younger than age 18 years and patients with incomplete medical records cases were excluded from the study. Patients were stratified into nonreadmitted and readmitted within 30 days of the ostomy reversal. All cases followed a well-implemented standardized ERP and discharge criteria [9–12]. Preoperative, perioperative, and postoperative factors for the ostomy reversal admission were analyzed. Additional demographic and clinical information was supplemented from the electronic medical records. Data fields evaluated included age, gender, body mass index (BMI), comorbidities, procedure type, procedure indication, LOS, operative time, blood loss, intraoperative complications, postoperative complications, need for intensive care unit (ICU) care, total cost of care for the hospital episode, and discharge disposition. The main outcome measures were predictors of readmission and characteristics of patients readmitted and not readmitted.

Closure techniques

All closures were performed under general anesthesia. For all ileostomy closures, a circumferential peristomal incision was made and deepened down to the fascia. The fascia was mobilized off the small bowel. The small bowel was adequately mobilized to return the ileostomy closure site to the abdominal cavity. The stoma was everted and the edges freshened. Depending on the operating surgeon’s preference, the bowel was then either closed transversely with interrupted sutures or resected with a side-side stapled anastomosis.

For loop colostomy closures, the same peristomal mobilization was performed. For end colostomy closures, a circular stapler anvil was inserted into the distal end of colon. Then, either laparoscopically (through a 12-mm port at the stoma site, a 10-mm port at the umbilicus for the camera, and 5-mm working ports in the right iliac and right upper quadrant) or through a midline laparotomy, the small bowel and omentum were moved towards right upper quadrant. Adhesiolysis was performed as needed to identify the rectal stump. Lateral mobilization of the left colon and splenic flexure, then a medial to lateral mobilization of the sigmoid colon off of Gerota’s fascia, were performed to allow tension-free reach of colon to rectum. The proximal rectum was mobilized until free, then a circular stapled anastomosis completed and leak test performed.

Data analysis was performed using unpaired Student’s t test, χ 2 test, or Fisher exact test, as appropriate, with a significance level defined as p = 0.05. Univariate logistic regression models were built to estimate the odds of readmission for all variables. Multivariate logistic regression was performed to determine which significant factors from the univariate analysis significantly impacted readmission. All analysis was performed using SPSS (v18.0).

Results

During the study period, 351 patients had ostomy reversals and met inclusion criteria. The case classes were 345 elective (98.3 %) and 6 emergent (1.7 %). Total patients were 186 males (53 %) and 165 females (47 %). The main indications for the index case were rectal cancer (33.9 %) and inflammatory bowel disease (21.7 %). Of the 351 cases, 302 (86 %) were ileostomy and 49 (14 %) were colostomy reversals. Fifty-two cases were end ostomy reversals (14.8 %), and 299 were loop ostomy reversals (85.2 %). Twenty-one ostomy closures were performed laparoscopically (6 %); the majority of closures were performed through an open approach (94 %). All laparoscopic cases were end ostomies. There were six total intraoperative complications (1.7 %): two bleedings, two stapler malfunctions, one anastomotic leak test failure, and one rectal serosal tear. Five patients (1.4 %) required an ICU stay after the ostomy reversal. The median LOS was 3 days (range 1–59). At discharge, 339 patients (96.6 %) were discharged home, whereas 12 (3.4 %) required temporary nursing care. After a median follow-up of 24 months (range 2–86), the median time to ostomy reversal was 2.5 months (Table 1).

Table 1 Patient demographic data
Full size table

Forty-four were readmitted within 30 days of ostomy reversal, a readmission rate of 12.5 %. Comparing readmitted and nonreadmitted patients, there were no significant differences across demographic factors of age, BMI, gender, or comorbidities, as demonstrated by comparable American Society of Anesthesiologists class, the Modified Frailty Index, and Charlson Comorbidity Indexes. Readmitted and nonreadmitted patients had similar indications for the index operation and time to ostomy reversal (Table 1).

Readmitted had significantly longer operative times (p = 0.002), intraoperative blood loss (p = 0.003), and intraoperative complications (p = 0.005). More readmitted patients required ICU care after the ostomy reversal (p ≤ 0.0001). LOS for the ostomy reversal was significantly longer for readmitted patients (p = 0.001); 61.4 % of readmitted patients had delayed discharge (≥3 days). At discharge, significantly more nonreadmitted patients were discharged home without need for home care or skilled nursing services (p < 0.001). Total hospital costs for the ostomy reversal episode for significantly higher for readmitted than nonreadmitted patients ($22,048 vs. $12,526, respectively; p = 0.0162). All readmitted patients had postoperative complications, compared with only 5.5 % of nonreadmitted patients (p < 0.0001). The most common complication in both groups was small-bowel obstruction/postoperative ileus (36.4 % readmitted vs. 5.5 % nonreadmitted; p < 0.0001). Eleven readmitted patients required an unplanned intervention during readmission (25 %): five interventional radiology drainage procedures, three exploratory laparotomies, with lysis of adhesion, two exploratory laparotomies with bowel resection and recreation of the ostomy, and one therapeutic endoscopy. Perioperative and short-term patient outcomes are seen in Table 2.

Table 2 Stoma reversal perioperative and short-term outcome data
Full size table

With the logistic regression model, we found that longer operative time [odds ratio (OR) 1.006, 95 % confidence interval (CI) 1.001–1.012], intraoperative complications (OR 7.334, 95 % CI 1.23–43.761), ICU stay (OR 1.291, 95 % CI 1.18–1.893), delayed discharge (OR 1.085, 95 % CI 1.003–1.173), and discharge to a skilled nursing facility (OR 6.936, 95 % CI 1.531–31.332) were independent predictors of readmissions (Table 3). In subgroup analysis, there were no significant differences between ileostomy and colostomy outcomes, and ostomy type was not a predictor of readmission after ostomy reversal.

Table 3 Multivariate logistic regression model for predictors of readmission
Full size table

Discussion

In this study, we found that readmitted and nonreadmitted ostomy reversal patients have similar demographic profiles and comorbidities. However, the two groups have significantly different perioperative and short-term outcome variables, including greater operative time, intraoperative complications, need for ICU care postoperatively, delayed discharge, and the need for home care or skilled nursing services.

Extensive prior research has focused on the complications and subsequent readmissions after ileostomy creation [13–18]. Few studies have described readmission rates after ostomy closure. Published rates are between 10 and 12 %; however, these studies have all focused on ileostomy reversals [19–24]. Our study evaluated ileostomy and colostomy reversals, a strength over previous work. We found the type of stoma had no effect on readmission (OR 1.059, 95 % CI 0.307–3.595; p = 0.938). Previous studies have demonstrated that early discharge was feasible [20–22]. Our results show that early discharge is not only feasible but associated with fewer readmissions; delayed discharge was independently predictive of a postoperative readmission (OR 1.085, 95 % CI 1.003–1.173).

Even fewer studies have attempted to determine factors related to complications and readmissions after ostomy reversal. Increased length of time between ileostomy formation and closure was suggested as a factor responsible for postoperative small-bowel obstruction [23]. Conversely, a longer wait from loop ileostomy creation to closure was shown significantly related to complications requiring readmission and surgical intervention (140 vs. 210 days, p < 0.0001) [25]. When attempting to identify predictors of postoperative complications colostomy (Hartmann) reversal patients during 7-year period, Lin et al. found that BMI was an independent predictor of morbidity (p < 0.04); the authors recommended weight loss to improve outcomes [26]. We identified independent contributors to unnecessary readmissions. Patients with longer operative times, intraoperative complications, and ICU stays can be prospectively identified at high risk for readmission, allowing changes to our ERP. In additional, in patients who require LOS longer than 3 days or skilled nursing at discharge, the discharge criteria and postoperative follow-up can be amended to potentially reduce the 12.5 % readmission rate in this cohort.

With the ability to identify patients at risk for readmission, we can prospectively modify our discharge and follow-up protocols. For example, adding a telephone call to this patient population before the first scheduled postoperative follow-up visit to assess bowel function, abdominal pain, and hydration may be beneficial; future prospective studies will be needed to assess the impact of this change.

We do recognize limitations of this study. The design is a retrospective database review and subject to the biases associated with retrospective studies, as well as limitations in the data available for analysis. It is a single-institution study, so our results may not be generalizable. There also was a relatively small sample size of readmitted patients, so any further subgroup analysis could be underpowered and subject to a Type II error. Regardless of any limitations, the study highlights an issue that affects patient outcomes and healthcare resources and deserves further attention in the current literature.

In conclusion, readmitted and nonreadmitted patients have similar demographics and comorbidities. However, the perioperative and short-term outcomes are significantly different. Longer operative procedure times, intraoperative complications, ICU admissions, need skilled nursing at discharge, and delayed discharge were all predictive of readmission after ostomy reversal. These factors can be used to prospectively identify patients more likely to be readmitted, and modify discharge and follow-up protocols to reduce unnecessary readmissions.