Introduction

Ulcerative colitis (UC) is a chronic inflammatory condition that is typically treated pharmacologically. However, up to 20% of patients will develop severe colitis requiring hospitalization and up to 25% of patients may require surgery during the course of their disease [1]. Surgery is utilized in cases of disease that is refractory to medical therapy, disease complications (e.g., toxic colitis, perforation, uncontrolled hemorrhage), or the presence of dysplasia or cancer [2, 3].

Prior studies have indicated variable rates of mortality associated with surgery for UC with higher rates in patients undergoing emergent surgery ranging from 2.4 to 13.1% compared to rates of 0.6–2.2% in elective surgery [4,5,6,7,8,9,10]. Some variability in mortality may be related to older studies and differences in surgical technique over time. Additionally, many studies are from tertiary care referral centers which may result in a bias from sicker patients leading to higher reported mortality rate. While it appears that overall mortality is decreasing in patients undergoing colectomy for UC, a large meta-analysis found that mortality was more common following emergent surgery (5.3% 95% CI 3.8–7.4%) compared to elective surgery (0.7% 95% CI 0.6–0.9%) [9]. However, these data were still limited by the included studies, many of which were from tertiary care centers and from national databases lacking details on each individual patient.

Complications are also an important factor in deciding between continuing pharmacologic therapy versus surgery as well as in determining the surgical approach (e.g., anastomosis vs no anastomosis). Similar to mortality, studies have reported variable rates of complications following surgery for UC ranging from 27 to 61% [11,12,13,14]. A large population database study from Calgary reported a complication rate of 61% in patients undergoing colectomy for UC [14]. Surprisingly, these rates were higher in the elective cohort compared to the emergent cohort (64.9 vs 55.4% p = 0.021) [14]. Complications were more frequently seen in elderly patients and those with multiple comorbidities [14]. However, lower complication rates of 30% have been reported in the USA [10]. The exact cause of this difference in reported complications is unclear, but referral bias as well as older studies reflecting open, as opposed to more current minimally invasive, surgical approach may have contributed. Additionally, there were significant variations in how complications were defined from study to study [11, 13, 15].

Based on this variability of the current rates of morbidity and mortality following surgery for UC, we sought to determine the 90-day rates of mortality and complications classified by the Clavien–Dindo postoperative surgical complication system following surgery for UC in a tertiary care referral center.

Methods

This is a single-center retrospective cohort study performed at a tertiary care academic medical center, Beth Israel Deaconess Medical Center (BIDMC). The center has a large gastroenterology and colorectal surgery practice and a specialized center for inflammatory bowel disease (IBD). The hospital utilizes an electronic health record system (EHR) from which all the data were collected.

Inclusion Criteria

All records with a hospital discharge International Classification of Disease-9 (ICD-9) code 556.x between January 1, 2002, and January 1, 2014, were reviewed to determine if the patient underwent surgery for UC during the admission. Only patients who underwent one of the following surgical procedures were included in the study: total proctocolectomy and end ileostomy, abdominal colectomy with end ileostomy, proctocolectomy with ileoanal pouch anastomosis (IPAA) formation and diverting ileostomy, one-stage IPAA (j-pouch), or abdominal colectomy with ileorectal anastomosis. Only the initial surgical procedure was included in this study with follow-up surgeries including proctectomy with IPAA formation and ileostomy take down excluded from the analysis.

Exclusion Criteria

Non-UC-related colitis including Crohn’s disease, ischemic colitis, and infectious colitis was excluded.

Chart Review

Each chart was reviewed manually by two authors (MA and JDF). Patient demographics, age, gender, ethnicity, highest level of education, and type of medical insurance were collected. UC-related history including duration of disease, prior medications and past medical history, and Charlson comorbidity index was noted. Information related to the admission including admission laboratories, body mass index, medications, systemic steroids (intravenous vs oral), ASA class, indication for surgery, type of surgery, clinical data, and operative details and discharge details was all reviewed. Biologics included infliximab, adalimumab, golimumab, and vedolizumab. Current use of biologics was defined as use of infliximab or vedolizumab within 8 weeks of admission or within 4 weeks of admission for adalimumab and golimumab. Disease severity and fulminant colitis were defined based on Truelove and Witts criteria [16]. Fulminant colitis was defined as a combination of signs and symptoms including those with more than ≥ 10 bowel movements per day, ongoing bleeding, toxicity, abdominal tenderness, blood transfusions, or colonic dilation on abdominal films [17]. Surgery was classified as urgent/emergent or elective based on the urgency of the procedure according to the notes in the EHR. Charts were then reviewed for any complications or death within 90 days of the initial surgical procedures. Complications were classified using the Clavien–Dindo classification system (Supplemental Table 1). Time to complication was determined by the first complication reported following surgery. Only complications related to the initial surgery were considered for this manuscript. Readmissions were reviewed to determine the time to readmission and cause for the readmission. All complications and mortality were assessed to determine if they were related to the surgery or not by three authors (MA, JDF, and VP). Any disagreements related to if a complication or mortality was related to surgery were reconciled using a modified Delphi method with senior author (VP) making the final decision [18]. Study data were collected and managed using REDCap electronic data capture tools hosted at BIDMC [19].

End Points

The primary end point was mortality at 90 days following initial colectomy for UC. Secondary analysis included the rate and predictors of complications at 90 days following surgery for UC.

Statistical Analysis

Patients experiencing death or complication were compared to those that did not. Two-tailed p values were calculated by Chi-square test for categorical variables or Student’s t test for continuous variables. Independent predictors of complication were identified using logistic regression. All variables with a p value less than 0.1 on bivariate comparison of those with complications versus those that did not were included in the model. Backward stepwise elimination was used to determine final independent predictors with variables eliminated for p value greater than 0.05. All analyses were performed using IBM SPSS Statistics, version 23.0 (IBM Corp., Armonk, New York).

The study was approved by the Institutional Review Board at BIDMC.

Results

Two hundred and fifty-eight patients underwent an initial surgery for UC. Of these, 69% (177) of cases were elective and 31% (81) were urgent/emergent (Fig. 1). There were several variables that were different between the elective and the urgent/emergent cohorts (Table 1). Compared to non-elective cases, patients undergoing elective surgery were older (median 45.4 vs 41.4 p = 0.048), had higher BMI (26.4 vs 23.1 p 0.003), and had higher Charlson comorbidity index (0.4 vs 0.1 p = 0.008), but were less likely to have been on outpatient prednisone (59 vs 84% p < 0.001). During the admission, patients who underwent an elective operation were less likely to have a flare of UC on admission (65 vs 95% p < 0.001) and less likely to receive intravenous steroids (53 vs 85% p < 0.001). Surgical indications (p < 0.001), specific surgical procedure (p < 0.001), number of stages planned (p < 0.001), and ASA class (p < 0.001) were all different between the elective and urgent/emergent cohorts. For additional demographic variables and other significant variables see Table 1.

Fig. 1
figure 1

Patients stratified by type of surgery

Full size image
Table 1 Demographics
Full size table

There were no deaths reported in our cohort at 30 days. At 90 days, there were 3 deaths: 2 in the elective group and 1 in the urgent/emergent group. The cause of death in the urgent/emergent group patient was fatal immune-mediated enteritis 47 days after surgery. While not directly related to the surgery, this was considered to be likely related to her recent ulcerative colitis [20]. In the elective cohort, one patient died from unknown causes at home. A second patient died secondary to sepsis after developing a small perforation at her ileostomy site 56 days after surgery. None of the deaths were determined to be directly related to the initial surgery for UC. See supplemental Table 2 for details of the deaths.

Complications developed in 47% (n = 122) of the entire cohort with 44% in the elective cohort and 54% in the urgent/emergent cohort (Table 2). There was no difference in time to complication based on elective vs urgent/emergent surgery (Fig. 2). Gastrointestinal issues were the most frequently reported complication followed by infectious complications, but there was no difference in the surgical cohorts in any of the complications. 62% of the complications were Clavien–Dindo grade 1 or 2 with no difference between the cohorts (Table 2). There was no significant difference in any Clavien–Dindo grade complications in the elective or urgent/emergent cohorts (Table 2). Readmissions were seen in 24% of the cohort, with no significant difference seen in the two cohorts (p = 0.636) (Table 2). Time to readmission between the groups was different with patients in the elective group readmitted within a mean of 19 days compared to 30.5 days in the urgent/emergent cohort (p = 0.003) (Fig. 3). The median readmission hospitalization was 3.5 days (range 1–23).

Table 2 Complications within 90 days
Full size table
Fig. 2
figure 2

Time to complication: elective versus non-elective surgery

Full size image
Fig. 3
figure 3

Time to complication by surgical approach

Full size image

When comparing predictors of complications Clavien–Dindo grade 3 or higher, univariate predictors with p value < 0.1 include age, female gender, any anastomosis, ASA class, history of myocardial infarction, peptic ulcer disease history, thiopurine use on admission, and adalimumab use on admission. See supplemental Table 3 for a full list of variables. On multivariate analysis, predictors of complications Clavien–Dindo grade 3 or higher included prior myocardial infarction (OR 6.093 95% CI 1.479–25.094), thiopurine use on admission (OR 5.200 95% CI 2.346–11.525), adalimumab use on admission (OR 5.516 95% CI 1.622–18.761), and ASA class 3 or greater (OR 3.014 95% CI 1.387–6.551) (Table 3).

Table 3 Multivariate model predictors of complications
Full size table

Discussion

The risk of death from surgery for ulcerative colitis is an extremely rare event at a tertiary care academic referral center. None of the deaths were believed to be directly related to the initial surgery (Table 2). In contrast, complications following surgery are not uncommon, occurring in up to 47% of cases. However, the majority of these complications are classified as Clavien–Dindo grade 1 or 2, which by definition are managed without the need for any major interventions. Readmissions, however, remain a concern following surgery for UC in nearly 25% of patients. In general, these readmissions are limited hospital stays of 3.5 days. The exact predictors of complications and readmissions remain elusive, but in our study, prior use of thiopurine, anti-tumor necrosis factor, a personal history of a prior myocardial infarction, and ASA class 3 or higher all increased the risk of a Clavien–Dindo complication grade 3 or greater. Finally, even with potential delays in surgery, urgent/emergent surgery does not carry a higher risk of complications compared to elective surgery.

Prior studies have indicated variable rates of mortality following surgery for UC [4,5,6,7,8,9]. Some studies have reported higher mortality following urgent/emergent surgery when patients are sicker and likely at greater risk of having complications, but this finding has not been substantiated in all studies [4,5,6,7,8,9]. Theoretically, patients seen in tertiary care academic referral centers may be sicker and as more patients are treated with infliximab even in community hospitals, the patients being seen in an academic referral center may be at higher risk of death from surgery [14]. However, our study indicates that even with different colorectal surgeons and a study period spanning 12 years, still not a single patient died within 30 days of discharge from their surgical procedure. Additionally, in the three deaths reported within 90 days, none were considered to be directly related to the initial surgery. Our study indicates that even with potentially sicker patients being referred to an academic medical center, still death related to surgery for UC should be seen in < 1% of cases.

While surgical complications can increase the overall cost of care, some complications may be unavoidable [21]. Interestingly, in our study, there was no significant difference in the likelihood of complications in patients undergoing elective surgery or urgent/emergent surgery. Previous studies have shown higher rates of complications following urgent/emergent surgery [4, 5, 8, 22].

However, our cohort did not show significant difference in mortality in the emergent group. Similar to mortality, this did not differ between surgeons or the 12 years included in this study period. In our study, a personal history of myocardial infarction as well as the use of a thiopurine or adalimumab prior to surgery increased the risk of Clavien–Dindo grade 3 or higher complications. These findings are similar to prior studies indicating medications, and myocardial infarction may increase the risk of postsurgical complications [15, 23,24,25]. Although overall complications rate remained high in both groups, majority of complications were mild and Clavien–Dindo grade 1 or 2 [22, 26, 27]. The cost associated with these complications is thought to be significantly less than those resulting from hospital readmissions and more severe Clavien–Dindo grades [22, 26, 27]. It is likely that these more minor complications are often unavoidable and focus should be on evaluating ways to reduce the risk of the grade 3 and higher complications.

With an increasing focus on reimbursements based on quality of care provided, readmission rates are being carefully evaluated [28]. In our study, readmissions were seen in 24% of cases, but they did not differ based on surgical urgency. Some readmissions are unlikely to be avoidable [29, 30]. Nagle et al. [31], however, showed that in cases of dehydration following new ostomy formation a standardized postoperative care pathway reduces the readmission rates from dehydration to 0%. However, not every risk factor like disease severity or urgency of surgery predicts readmission. Nevertheless, similar to other conditions, readmissions should ideally be reduced to under 13% at 30 days [32]. While the readmission may not be indicative of the quality of the surgical care, it may have more to do with IBD and the medications being used to treat IBD [13]. Prior studies have shown an increase in readmission rates in IBD even without surgery as well as an increase in surgical complications associated with anti-TNF medications prior to surgery [14, 33]. Additionally, similar to our study finding of myocardial infarction increasing the risk of complications, De Silva et al. [14] also noted that patients with IBD with cardiac disease are also at increased risk of surgical complications. While our study did not assess for when medications should be stopped prior to surgery or how to mitigate cardiovascular risk, further studies are warranted to determine if stoppage of IBD-specific medications prior to surgery and optimization of cardiovascular status results in reduced complications and readmissions.

Our study does have some limitations. It is a single-center study performed in a referral-based tertiary care medical center. The center has a robust inflammatory bowel disease program and colorectal surgery program. The ability to generalize these findings to the community and other practice settings is limited. Additionally, the rates of death, complications, and readmissions were all dependent on accurate clinical documentation. It is possible that patients were readmitted to other centers or died without our group being informed of this. However, following a surgical procedure, patients are more likely to return to the hospital at which the surgery was performed.

Conclusion

Surgery for ulcerative colitis is not associated with any mortality at 30 days and very low mortality at 90 days. However, patients undergoing elective and urgent/emergent surgery are associated with increased rates of complications, but the majority of these complications are minor in nature. Readmissions, however, occur in nearly 25% of patients, and interventions are needed to reduce this rate. Further studies are necessary to determine how to further mitigate these complications and reduce the 90-day readmission rates following surgery for ulcerative colitis.