Introduction

Inflammatory bowel disease (IBD) is a collection of diseases characterized by inflammation of the digestive tract and includes Crohn’s disease (CD), ulcerative colitis (UC), and indeterminate colitis. Surgical intervention varies between IBD types and is often reserved for medically refractory patients and those in extremis [1,2,3,4]. Although more prominent in the white population, the incidence of IBD is rising in minority populations [5,6,7,8], with a doubling of the incidence rate in non-white patients since 1970 [9]. Race, socioeconomic status, and ethnicity often play a significant role in healthcare disparities [10, 11], particularly for postoperative outcomes [12,13,14,15,16,17,18,19]. Several national database studies evaluating postoperative outcomes based on race have demonstrated differences in postoperative morbidity [6, 12,13,14,15, 20, 21], mortality [10], and length of stay [12,13,14,15], often finding increased morbidity and mortality and longer lengths of stay in minority races compared to white patients. A study using the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) data from 2005 to 2017 found higher odds of renal and cardiac complications, death, and serious morbidity in black compared to white patients [12]. The NSQIP IBD data collaborative [22, 23], which was created in 2017, invited collaborators from high volume IBD centers to evaluate outcomes for patients with IBD using IBD-specific data (e.g. medication use within 60 days, ileostomy creation, ileoanal anastomotic technique, and presence of colonic dysplasia) in addition to standard NSQIP variables. These variables provide tailored insight into the pre- and post-operative care of patients with IBD. With the incidence of IBD increasing in non-white populations and previous data suggesting differences in postoperative outcomes based on race, it is imperative to elucidate the contemporary state of disparities in surgical outcomes in patients with IBD. The objective of this study was to assess whether rates of postoperative complications, specifically wound complications (WC) and adverse events (AE), in patients with IBD vary based on race.

Methods

Patient Selection

We conducted a retrospective cohort study using data from the American College of Surgeons NSQIP IBD database collaborative [22, 23], which collected patient information from 15 tertiary care medical centers across the United States from March 2017 to March 2022. Patients over 18 years old with a diagnosis of IBD who underwent surgery were included in this study. Subjects with missing data regarding outcomes (AE and WC), race, and covariates were excluded.

Exposures

The primary exposure variable was race, which was defined as “white”, “black” or African American, and “other minorities” as self-reported in the patient chart. The “other minorities” racial category consisted of Asian Americans, Pacific Islanders, and all other races not otherwise defined as black or white.

Outcomes

The primary outcomes of the study were adverse events (AE) and wound complications (WC). AE was defined as the subject having any one of the following: death, pneumonia, unplanned reintubation, mechanical ventilation for ≥ 48 h, pulmonary embolism (PE), deep venous thrombosis (DVT), progressive renal insufficiency, acute kidney injury (AKI), stroke, urinary tract infections (UTI), cardiac arrest requiring cardiopulmonary resuscitation (CPR), myocardial infarction, postoperative transfusion requirement within 72 h of surgery, sepsis and septic shock, and unplanned reoperation. WC was defined as a subject having any surgical site infection, including superficial, deep, or organ space infection, or disruption of the wound.

Covariates

All subjects had a diagnosis of IBD based on the International Classification of Diseases 10th revision (ICD-10) coding: K50- for Crohn’s disease, K51- for ulcerative colitis, and K53.3 for indeterminate colitis. Perioperative information collected included operative type (open surgery, minimally invasive surgery (MIS) [robotic, laparoscopic, or hybrid], or MIS converted to open), emergency surgery, age, number of comorbidities, medication (i.e. steroids, immunologic, biologic therapy) use within 60 days prior to surgery, sex assigned at birth (male or female), ethnicity (Hispanic or non-Hispanic) and body mass index (BMI). Number of comorbidities was defined as patients having 0, 1–2, or ≥ 3 of any of the following: diabetes, chronic obstructive pulmonary disease, dyspnea, hemodialysis, AKI, disseminated cancer, bleeding disorder, pre-sepsis, smoking history, hypertension, or congestive heart failure. BMI was categorized into underweight (< 18.5 kg/m2), normal to overweight (18.5–30 kg/m2), and obese (> 30 kg/m2). Immunologic therapy was defined as using any one of the following within 60 days of surgery: 6-meraptopurine, Azathioprine, or Methotrexate. Biologic agent use was defined as using any one of the following: Infliximab, Adalimumab, Certolizumab, Golimumab, Vedolizumab, Natalizumab, Ustekinumab, or Tofactinib. Ileostomy was defined as having an ileostomy present at the time of surgery, creating a new one, or revising a previously created one. Surgery type was defined using Current Procedural Terminology (CPT) coding and a user input variable included in the NSQIP IBD collaborative. Data, which is aggregated yearly and disseminated to participating institution, was collected on an iterative basis and entered into the database by a trained member of each institution’s research team.

Statistical Analysis

Continuous variables were analyzed with univariate analysis (means and standard deviations) and then compared between each exposure group using a student’s t-test and ANOVA. For nominal and categorical variables, frequencies were calculated for each group and the groups were compared using Mantel–Haenszel chi-squared tests.

We computed odds ratios (OR) and 95% confidence intervals (95% CI) for the association between race and AE and WC using logistic regression. A crude model was created including all possible covariates of the relationship between race and postoperative outcomes, and the p-value was determined for each covariate. A final adjusted model was then created, which included all statistically significant covariates (p < 0.10) from the crude analysis. To assess for potential confounding, effect measure modification (EMM) analyses were performed for variables with a greater than 10% change in OR from crude analysis to the final model. For variables with statistically significant findings on EMM, an adjusted logistic regression was performed using the variables from the final model. A second logistic regression analysis investigated race’s relationship with each component AE and WC. A p-value less than 0.05 was considered statistically significant.

The data analysis for this paper was generated using SAS, Version 8.3 of the SAS System for Academics. Copyright © 2021 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA.

Results

Cohort Characteristics

The NSQIP IBD database collaborative included 5,603 patients. After applying inclusion and exclusion criteria, 4,284 subjects were included in the study (Table 1). Of these, 3,899 (91%) were white, 276 (6.4%) were black, and 113 (2.6%) were other minority subjects. Notably, there were baseline cohort differences in IBD diagnoses, BMI, age, biologic agent and steroid use, and ileostomy presence based on race. While not statistically significant, 55% of white subjects had CD and 37% had UC, while 46% of other minority and 63% of black subjects had CD, and 43% of other minority and 26% of black subjects had UC respectively (p = 0.36). BMI differed between groups with 20% of white subjects having a BMI > 30 and 7.7% having a BMI < 18.5 compared to black subjects with 24% having a BMI > 30 and 13.2% having a BMI < 18.5 and other minority subjects having 8% with a BMI > 30 and 24% having a BMI < 18.5 (p < 0.05). The average age of white subjects was higher than that of black and other minority subjects with an average age of 44.3 years old compared to 40.4 and 41.6 years old respectively (p < 0.05). Use of biologic agents also differed with 45% of white subjects being on a biologic agent pre-operatively compared to 34% and 36% of black and other minority subjects (p < 0.05). Other minority and white subjects had higher pre-operative rates of steroid use as well (44% and 45% respectively) compared to black subjects (34.9%) (p < 0.05). 50% of white subjects had an ileostomy compared to 37% of black and 42% of other minority subjects (p < 0.05). There were no statistically significant differences between number of comorbidities, operative type, need for emergency surgery, and immunologic medication use when stratified by race.

Table 1 Cohort characteristics and post-operative outcomes
Full size table

No difference in AE or WC was found between racial groups on univariate analysis (AE: 20% white, 25% black, 22% other minority, p = 0.13 and WC: 11.3% white, 9.9% black, 12.4% other minority, p = 0.899; Table 1). Differences in rates of AE were present in open surgery and MIS converted to open compared to MIS (27% vs 29% vs 16.2% respectively, p < 0.01); emergency surgery compared to non-emergency surgery (52% vs 19.6%, p < 0.01); 1–2 and ≥ 3 comorbidities versus no comorbidities (26% vs 44% vs 16.6%, respectively, p < 0.01); BMI < 18.5 (27%) compared to BMI > 18.5 (< 20%); pre-operative steroid use (23%) compared to not using steroids (18.2%, p < 0.01); and in subjects with an ileostomy as compared to without (22% vs 18.4%, p < 0.01). Rate of WC was higher in open surgery compared to MIS and MIS converted to open surgery (16% vs 8.7% vs 9.8%, respectively, p < 0.01); ≥ 3 comorbidities (22%) compared to 1–2 (13.4%) and no comorbidities (22% vs 13.4% vs 9.7%, p < 0.01); and in patients with as compared to without an ileostomy (13.1% vs 9.5%, p < 0.01).

Multivariable Analysis

The odds of AE or WC did not differ based on race on unadjusted logistic regression analysis (Table 2). However, in the final multivariable analysis, black patients were found to have a 40% increase in odds of AE compared to white patients (OR 1.40 [95% CI 1.04–1.88] p = 0.03). We observed a 22% increase in odds of AE in UC subjects compared to CD (OR 1.22 [1.02–1.46], p = 0.03). Subjects who underwent open or MIS converted to open surgery had a > 85% increase in odds of AE compared to MIS. A BMI < 18.5 was associated with a nearly 1.5 times increase in AE compared to BMIs 18.5–30 (OR 1.48 [1.14–1.92], p < 0.01). Pre-operative steroid use increased the odds of AE by 1.36 compared to not using steroids (OR 1.36 [1.16–1.60], p < 0.01). In the multivariable analysis for WC, no difference in WC was observed based on race. An increasing number of comorbidities was associated with a 28% (1–2 comorbidities) and 96% (≥ 3 comorbidities) increase in odds of WC compared to having no comorbidities (p = 0.02). There was a nearly doubling in the odds of WC for subjects who underwent open surgery compared to MIS (OR 1.93 [1.58–2.35], p < 0.01).

Table 2 Unadjusted and multivariable logistic regression analysis evaluating the odds of a post-operative adverse event (AE) and wound complications (WC). IBD—Inflammatory bowel disease; BMI—Body mass index; MIS—Minimally invasive surgery
Full size table

Effect Measure Modification Analysis

To assess for confounding, EMM was performed using the variables included in the final multivariable model for AE in which statistically significant changes were noted in the odds of AE when stratified by IBD diagnosis, gender, and steroid use (Table 3). Black subjects with CD (OR 1.88 [1.31–2.72], p < 0.01), black males (OR 1.79 [1.21–2.64], p < 0.01), and black subjects on steroids (OR 2.14 [1.35–3.39], p < 0.01) had higher odds of AE compared to their white counterparts. No EMM was seen in the WC cohort.

Table 3 Effect measure modification analysis. AE – Adverse Event; *Adjusted OR controls for age, race, gender, comorbidities, BMI, operative intervention type, emergency surgery, steroid use, and ileostomy creation or formation
Full size table

Postoperative Complication Analysis

When we evaluated the relationship between race and each component of the composite postoperative outcomes independently, we saw few significant differences between racial groups (Table 4). Black subjects had a higher incidence of blood transfusions within 72 h of surgery (14%) compared to white (8.7%) and other subjects (8.0%) (p = 0.13). Other minority and black subjects also had higher rates of sepsis compared to white subjects (6.2%, 6.2% vs 3.4% p = 0.01). None of the non-white subjects experienced myocardial infarctions, cardiac arrest requiring CPR, or AKI, and no other minority subjects experienced PE, progressive renal insufficiency, reintubation or pneumonia. In the entire subject pool, no one suffered from a stroke. On logistic regression, there were increased odds of black subjects requiring a blood transfusion compared to white subjects (OR 1.71 [1.19–2.45], p < 0.01; Table 4), as well as increased odds of sepsis for black (OR 1.90 [1.12–3.18], p = 0.02) and other minority subjects (1.87 [0.85–4.10], p = 0.12).

Table 4 Secondary analysis of individual post-operative outcomes. CPR—cardiopulmonary resuscitation, SSI—surgical site infection
Full size table

Discussion

Over the last several decades, race has been shown to be a determinant of health [24, 25]. In this study, we demonstrated that racial differences in postoperative complications for patients undergoing IBD surgery have decreased [12,13,14,15,16]. We found no difference in WC based on race. However, we did demonstrate increased odds of AE in black subjects. In our EMM analysis, only black patients who were male, had CD, or received preoperative steroids had increased odds of AE. The rate of post-operative blood transfusions was higher for black subjects compared to white subjects as were the rates of post-operative sepsis for black and other minority subjects. Similar to previous reports, a diagnosis of UC [26, 27], presence of comorbidities [27, 28], decreased BMI [29], increased age [28], open versus minimally invasive surgery [30], and emergency surgery [31] were associated with increased odds of postoperative complications. In particular, an increased number of comorbidities, open surgery, and emergency surgery had the highest odds of AE and WC for all groups suggesting that medical optimization and pursuing elective surgery are the best measures to reduce postoperative complications.

In a previous study, Dos Santos Marques et al. used NSQIP data from 2005 to 2017 to demonstrate that black subjects with IBD had higher odds of postoperative AE, primarily due to higher rates of renal insufficiency, blood transfusions, and sepsis compared to white subjects [12, 14, 16]. In another study using NSQIP data from prior to 2008, which evaluated postoperative complications following emergency surgery, Esnaola et al. found that odds of postoperative cardiac arrest and renal insufficiency and/or failure were higher in black subjects compared to white [16]. Compared to the previous literature, our study had similar distributions of races (white patients 91 vs 89%, black 6.4 vs 7.6, other 2.6 vs 3.8%) and demonstrated lower rates of AE and WC for all populations [1]. Our secondary analysis of AE showed a decrease in cardiac and renal complications compared to the existing literature. Yet, we showed the continued increased need for post-operative blood transfusions for black subjects, and higher rates of sepsis in both black and other minority subjects.

Since the odds of AE increased for black subjects on multivariable analysis, we assessed for confounding using EMM. When stratified by IBD diagnosis, gender, and steroid use, the analyses showed that black subjects who were male, had CD, and who used pre-operative steroids had increased odds of AE. Separately, we found that subjects with UC had higher odds of AE compared to those with CD [32]. For subjects with UC, race did not influence odds of AE in contrast to a NSQIP study from 2016 to 2019 [33].

Of the medications recorded in the dataset, only preoperative steroid use was associated with increased odds of AE, while both biologic and immunologic agents were not associated with either of AE or WC, consistent with previous literature [34,35,36,37]. In particular, the EMM analysis revealed that black subjects using pre-operative steroids had increased odds of AE compared to white subjects. We also found that black subjects used biologics the least. Considering that biologic medications cost significantly more than steroids and immunologic agents, financial status and access to care may also influence racial differences in surgical morbidity [38, 39].Other factors, such as socioeconomic status and cultural differences can contribute to disparities in surgical outcomes as shown by Akinyemiju et al. who demonstrated that patients with Medicare or Medicaid experience more post-surgical complications than privately insured patients [11].

This study has several strengths. First, this data includes typical NSQIP variables as well as IBD specific ones (e.g. medication use, ileostomy formation) allowing us to have a more detailed assessment of the effect these variables have on the IBD patient population. Next, the dataset is a large national dataset utilizing data from high volume tertiary IBD centers, which are at often the forefront of surgical and medical care. Lastly, the effect of medication usage and ileostomy creation and revision as predictors of postoperative outcomes based on race has rarely been incorporated into larger national studies until now.

This study also has several limitations. First, our definition of AE included multiple postoperative complications of varying severity. To account for this heterogeneity and increased sensitivity, we analyzed each outcome independently and its association with race. Next, this study is retrospective in nature and is therefore limited by the data quality entered into the database. Third, given that the data collected was exclusively from tertiary care centers, this study’s results are less generalizable to non-academic or academic-affiliated centers. For this same reason, our study population likely underrepresents black and other minority races with UC as well as other minority races with CD [1, 5]. Finally, this study follows subjects for 30-days post-operatively. However, IBD is a chronic disease and 30-day follow up may not suffice in determining the long-term outcomes for these patients.

Despite improvements in care for patients with IBD, race was associated with increased odds of AE for black patients with IBD who underwent surgery. On further analysis, black subjects who were male, had CD, or who were using steroids had increased odds of AE compared to their white counterparts. Blood transfusions continue to be the primary AE associated with black race, while other AE have decreased in frequency compared to the previous literature. Importantly when using this IBD specific database, typical predictors of postoperative complications, including comorbidities, BMI, open surgery, and emergency surgery played a stronger role than race alone.