Abstract
BACKGROUND:
Existing studies on the effects of biologic medications on surgical complications among ulcerative colitis patients have mixed results. Because biologics may hinder response to infections and wound healing, preoperative exposure may increase postoperative complications.
OBJECTIVE:
Evaluate associations between biologic exposure within six months preceding colectomy or proctocolectomy and postoperative complications among ulcerative colitis patients.
DESIGN:
Retrospective cohort study with multivariate regression analysis following coarsened exact matching.
SETTING:
A large commercial insurance claims database (2003-2016).
PATIENTS:
A total of 1,794 ulcerative colitis patients underwent total abdominal colectomy with end ileostomy, total proctocolectomy with end ileostomy, or total proctocolectomy with ileal pouch anal anastomosis. 22% were exposed to biologics in the six months preceding surgery.
MAIN OUTCOMES MEASURES:
Healthcare utilization (length of stay, unplanned reoperation/procedure, emergency department visit, or readmission) and complications (infectious, hernia or wound disruption, thromboembolic, or cardiopulmonary) within 30 postoperative days.
RESULTS:
Exposure to biologic medications was associated with shorter surgical hospitalization (7 vs. 8 days, p <0.001), but otherwise was not associated with differences in healthcare utilization or postoperative complications. Patients who underwent total proctocolectomy with ileal pouch anal anastomosis had higher odds of infectious complications compared to those who underwent total abdominal colectomy with end ileostomy (aOR 2.2, 95% CI [1.5-3.0], p<0.001) but had lower odds of cardiopulmonary complications (aOR 0.4, 95% CI [0.3-0.6], p<0.001).
LIMITATIONS:
Analysis of private insurance database claims data may not represent uninsured or government-insured patients and may be limited by coding accuracy. Matched cohorts differed in age and Charlson comorbidity index, which could be influential even after multivariate adjustments.
CONCLUSION:
Biologic exposure among ulcerative colitis patients is not associated with higher odds of postoperative complications or healthcare resource utilization. These data, in combination with clinical judgment and patient preferences, may aid in complex decision-making regarding operative timing, operation type, and perioperative medication management. See Video Abstract.
Keywords: Biologic medications, Colectomy, Inflammatory bowel disease, Postoperative complications, TNF-alpha inhibitor, Ulcerative colitis
Abstract
ANTECEDENTES:
Estudios existentes sobre los efectos de medicamentos biológicos, en complicaciones quirúrgicas, en pacientes con colitis ulcerativa, presentan resultados mixtos. Debido a que los productos biológicos pueden retrasar la respuesta a las infecciones y curación de heridas, su exposición preoperatoria pueden aumentar las complicaciones postoperatorias.
OBJETIVO:
Evaluar las asociaciones entre la exposición biológica dentro de los seis meses anteriores a la colectomía o proctocolectomía y las complicaciones postoperatorias entre los pacientes con colitis ulcerativa.
DISEÑO:
Estudio de cohorte retrospectivo con análisis de regresión multivariante después de una coincidencia exacta aproximada.
AJUSTE:
Una gran base de datos de reclamaciones de seguros comerciales (2003-2016).
PACIENTES:
Un total de 1.794 pacientes con colitis ulcerativa, se sometieron a colectomía abdominal total con ileostomía terminal, proctocolectomía total con ileostomía terminal o proctocolectomía total con anastomosis anal y bolsa ileal. 22% estuvieron expuestos a productos biológicos, seis meses antes de la cirugía.
PRINCIPALES MEDIDAS DE RESULTADOS:
Utilización de la atención médica (duración de la estadía, reoperación o procedimiento no planificado, visita al servicio de urgencias o reingreso) y complicaciones (infecciosas, hernias o dehiscencias de heridas, tromboembólicas o cardiopulmonares) dentro de los 30 días postoperatorios.
RESULTADOS:
La exposición a medicamentos biológicos se asoció con una hospitalización quirúrgica más corta (7 frente a 8 días, p <0,001), pero por lo demás, no se asoció con diferencias en la utilización de la atención médica o complicaciones postoperatorias. Los pacientes que se sometieron a proctocolectomía total con anastomosis anal y bolsa ileal, tuvieron mayores probabilidades de complicaciones infecciosas, en comparación con aquellos que se sometieron a colectomía abdominal total con ileostomía final (aOR 2.2, IC 95% [1.5-3.0], p <0.001) pero tuvieron menores probabilidades de complicaciones cardiopulmonares (aOR 0.4, IC 95% [0.3-0.6], p <0.001).
LIMITACIONES:
El análisis de los datos de reclamaciones, de la base de datos de los seguros privados, puede no representar a pacientes no asegurados o asegurados por el gobierno, y puede estar limitado por la precisión de la codificación. Las cohortes emparejadas diferían en la edad y el índice de comorbilidad de Charlson, lo que podría influir incluso después de ajustes multivariados.
CONCLUSIÓN:
La exposición biológica entre los pacientes con colitis ulcerativa, no se asocia con mayores probabilidades de complicaciones postoperatorias, o a la utilización de recursos sanitarios. Estos datos, en combinación con el juicio clínico y las preferencias del paciente, pueden ayudar en la toma de decisiones complejas con respecto al momento quirúrgico, el tipo de operación y el manejo de la medicación perioperatoria. Consulte Video Resumen.
INTRODUCTION
Management of ulcerative colitis (UC) increasingly involves the use of biologic medications to induce and maintain disease remission. The biologic medications most commonly used to treat UC are monoclonal antibodies inhibiting tumor necrosis factor alpha (TNFa) activity, such as adalimumab, infliximab, and golimumab. More recently, antibodies to integrin alpha 4/beta 7 (vedolizumab)—which prevent T-cell homing to the gut—are also being used. Despite the introduction of these new therapies, a significant proportion of patients with UC must undergo colectomy for severe disease refractory to treatment, intolerance of medical therapy, or neoplasia.1 Postoperative complications are frequent among this population, ranging from 10-65%.2
Given the effect of biologic medications on the immune system, there has been concern for increased risk of infectious and wound-healing complications after surgery for patients on biologics. However, results are mixed regarding the effects of biologic medications on postoperative complications. Initial studies showed that UC patients taking TNFa inhibitors are more likely to suffer from complications after surgery.3 Several studies have compared postsurgical outcomes in the “pre-biologic era” to those in the “biologic era” and found an increase in postsurgical complications in the era in which biologics are more commonly prescribed.4,5 However, meta-analyses,6–8 an insurance claims database9 and large retrospective studies10 have shown no effect on infectious complications and mixed results on overall complications. The risk of postoperative complications appear to be similar for patients taking vedolizumab compared to those on TNFa inhibitors.11,12 The discrepant findings on the effects of biologics on postoperative complications have spurred debate on the management of perioperative biologic medications, timing of major operations in relation to the most recent dose of biologic medications, and type of operation to perform in the setting of recent biologic therapy exposure.8,13 Thus, understanding the association between biologics and postoperative complications is critical for surgical decision-making and providing optimal patient care.
To address this question on a broader scale, we used an insurance claims database to evaluate a large cohort of patients with UC who underwent definitive surgical management for their disease (total colectomy or total proctocolectomy). The primary outcome of this study was the risk of postoperative complications in patients exposed to biologic medications in the six months prior to surgery, compared to those who were not exposed to biologics. The secondary outcome was postoperative healthcare resource utilization in surgical patients exposed to biologic medications preoperatively, compared to those who were not.
METHODS
Data Source
We used the Optum Clinformatics® Data Mart administrative health claims database to evaluate a retrospective cohort of patients with UC who underwent surgery. Optum Clinformatics® Data Mart is a national private insurance claims database with deidentified inpatient, outpatient, and pharmaceutical claims previously used to evaluate surgical outcomes14–16 as well as medications in inflammatory diseases.17–19 Each year, data from approximately 15 million members of a national managed care company affiliated with Optum are included. Data include demographic characteristics, encounter data (inpatient and outpatient encounters and associated procedures/surgeries), and pharmaceutical data (medications prescribed, days supply, and dose). The Stanford University Institutional Review Board determined that this project did not meet the definition of human subjects research and exempted it from further review.
Study Population
We identified patients with UC, aged 12+ years, who underwent total colectomy or proctocolectomy from 2003-2016. A diagnosis of IBD was defined as ≥ 2 distinct outpatient or inpatient claims for either Crohn’s disease or ulcerative colitis. Patients with IBD were considered to have UC if 80% or more of their distinct IBD-related encounter codes were for UC20 (Supplementary Table 1). We then selected patients with an inpatient claim for colectomy. Colectomy was grouped by total abdominal colectomy (TAC) with end ileostomy, total proctocolectomy (TPC) with end ileostomy, or TPC with ileal pouch anal anastomosis (IPAA) based on ICD or CPT codes, as shown in Supplementary Table 2. Patients <12 years of age at time of surgery were excluded due to the low likelihood of surgical treatment for UC in this population. We included patients with continuous enrollment for at least six months prior to surgery to provide sufficient enrollment time to ascertain preoperative risk factors and 30 days following surgery to evaluate postoperative healthcare utilization and 30-day complications.
Outcomes
Our primary outcomes of interest were 30-day complications categorized in four domains: infectious, hernia or wound disruption, thromboembolic, or cardiopulmonary (Supplementary Table 3). We specifically modeled infectious and hernia/wound disruption complications because inhibition of T-cell action by biologic medications has previously been implicated in causing these complications.21–23 UC patients are at higher risk for thromboembolic events due to their chronic inflammatory state, so we additionally modeled the effect of biologics on thromboembolic complications. Lastly, we evaluated cardiopulmonary complications, hypothesized to be unrelated to biologic exposure. Secondary outcomes were healthcare utilization within 30 days after surgery including: length of hospital stay, unplanned procedure or reoperation, emergency department visit, and readmission (Supplementary Table 3).
To determine postoperative complications, we extracted all diagnosis and procedure codes for each patient in the 30 days after surgery. Two authors (KR and CK) classified each code as either not a complication or as an infectious, hernia/wound disruption, thromboembolic, or cardiopulmonary complication. Complications not falling in those categories, such as bleeding, were grouped as “any complication.” The goal of this approach was to capture the actual complication burden experienced by patients rather than limiting the analysis to a predetermined set of complication codes. If a patient had a previously documented complication code within the six months preceding surgery, that code was not classified as a postoperative complication for that patient (e.g. postoperative ventral hernia was not counted as a complication if ventral hernia was present on claims within six months before surgery). Unplanned procedures and operations occurring within 30 days after the index operation were identified using procedure and anesthesia codes.
Variable Classification
The primary independent variable of interest was biologic prescription in the six months preceding surgery. Biologic medications have long half-lives, and typical dosing intervals range from two to eight weeks. To determine the study definition of biologic exposure timing, we conducted analyses comparing multiple definitions: 1) exposure to biologics within six months prior to surgery, 2) exposure within 30 days prior to surgery, and 3) exposure within three half-lives of the given biologic therapy prescribed. We found minimal differences in the demographic and clinical characteristics of resulting patient cohorts from these definitions. Thus, we chose to include patients exposed to biologic medications within six months prior to surgery to fully capture exposure to biologics with longer half-lives. By looking only at 30-day exposure, the effects of longer half-life biologics would be inappropriately assigned to the non-exposed group. Patients were classified as having received a biologic medication based on presence of Healthcare Common Procedure Coding System J-codes (procedure codes) or pharmaceutical claims for infliximab, adalimumab, golimumab, vedolizumab, certolizumab, ustekinumab, and natalizumab. Steroid prescriptions for prednisone, prednisolone, methylprednisolone or hydrocortisone in the six months preceding surgery were also evaluated. The Charlson Comorbidity Index (CCI)24 was calculated over a six-month preoperative period.
Statistical Analysis
Patient characteristics were compared between surgery groups using χ2 tests for categorical variables and independent sample t-tests for continuous variables.
We applied a many-to-one coarsened exact matching (CEM) approach to minimize treatment selection bias for patients receiving biologics in this retrospective observational study. CEM is a monotonic imbalance bounding method that prunes observations, creating balanced treatment and control groups within the matched observations.25 While the more common approach of propensity score matching can still produce large imbalance through complete randomization, CEM eradicates this imbalance through full blocking.26 Matched covariates included: age, sex, malnutrition (263.x, E44.x), diabetes (250.x, E08.x-E11.x, E13.x), obesity (278.x, E66.x), and anemia (280.x-285.x, D50.x-D53.x, D55.x-D63.x). Selected matching characteristics were thought likely to be related to disease severity and also to affect postoperative complications in the IBD population.
We calculated multivariate logistic regression models using CEM-derived weights to compare matched biologic and non-biologic cohorts. Covariates included age group (12-18 years, 19-45 years, 46-65 years, and >65 years), sex, CCI (0, 1, 2+), region (Northeast, Midwest, South, West), race (White, African-American, Hispanic/Latino, Asian), steroids, net worth range (<$25,000, $25,000-$149,000, $150,000-$249,000, $250,000-$499,000, and ≥$500,000), and surgery type. Respective referent groups include 12-18 years, female, CCI of 0, Northeast region, White race, no steroid exposure, <$25,000, and total abdominal colectomy with end ileostomy. Missing data were considered a separate category within each variable. Patients with missing region were excluded from the analysis due to the small population size within the matched cohort (n=4) and lack of outcome occurrences in this small group. Statistical significance was defined as α=0.05. All statistical analyses were completed using Stata v14.2 (College Station, TX).
RESULTS
We identified 175,191 patients with a diagnosis of UC between 2003 and 2016. 3,328 of those patients (2%) underwent total colectomy or proctocolectomy. After many-to-one coarsened exact matching, 1,794 patients were included in regression analysis (Figure 1). Most patients exposed to biologic medications in the six months before surgery were prescribed TNFa inhibitors [infliximab (n=248, 62%) adalimumab (n=122, 30%), golimumab (n=20, 5%), and certolizumab (n=20, 5%)]. Few were prescribed an integrin α4β7 inhibitor (vedolizumab (n=9, 2%)]. No patients in this cohort were taking ustekinumab (interleukin-12/23 inhibitor) or natalizumab (integrin α4β7 inhibitor). Sixteen patients (4%) were prescribed two biologics in the six months before surgery, and one (0.3%) was prescribed three biologics within six months before surgery. We found minimal differences in the cohorts when we looked at biologic exposure in the six months, 30 days, or three half-lives (for each given medications) before surgery; we used the cohort with biologic exposure within six months of surgery so as not to miss exposure to biologics with longer half-lives. Although medication adjustments are often made within the 30 days before surgery, the 8-25 day half-life of these biologics means that they could still be exerting effects even months after the last dose is administered.
Figure 1:
Cohort flow diagram.
In the pre-matching cohort, the biologic-exposed group was considerably younger than the non-exposed group (43±16 years vs. 54±18 years, p<0.001, Supplementary Table 4). After matching, the imbalance was minimized with consideration for maintaining adequate sample size. The group of patients excluded after coarsened exact matching were generally an older, sicker group than the matched group (CEM matched age 48.1 +/− 16.5, CCI 1.5 +/− 2.4 vs. unmatched age 66.8 +/− 17.4, CCI 3.3 +/− 3/0, p< 0.001 for both). However, even after matching, the biologic-exposed cohort was still younger than the non-exposed cohort (44±15 vs. 48±15 years, p=0.03) but differences had been minimized without overly sacrificing sample size. The matched, biologic-exposed group had a lower CCI (0.8±1.5 vs. 1.7±2.5, p<0.001) and had higher level of educational attainment (p<0.001) than patients not exposed to biologics. Patients in the biologic-exposed group were more likely to have also been exposed to steroids in the six months preceding surgery (82% vs. 51%, p<0.001). The biologic-exposed group was more likely to have undergone IPAA (44% vs. 29% in the non-exposed group), whereas the non-exposed group was more likely to have undergone TAC with end ileostomy (54% vs. 41% in the biologic-exposed group, p<0.001) (Table 1).
Table 1:
Patient characteristics of matched biologic and non-biologic-exposed groups.
| Patient Characteristics | Total n=1,794 |
Preoperative Biologic Exposure n=401 (22.4%) |
No Biologic Exposure n=1,393 (77.6%) |
p-value |
|---|---|---|---|---|
| Age, Years (mean ± SD)* | 48.1 + 16.5 | 44.0 + 15.1 | 48.2 + 15.4 | 0.034 |
| Female Sex, n (%)* | 811 (45.2) | 186 (46.3) | 625 (44.9) | 0.659 |
| Malnutrition, n (%)* | 306 (17.1) | 76 (18.9) | 230 (16.5) | 0.252 |
| Diabetes, n (%) * | 238 (13.3) | 47 (11.7) | 191 (13.7) | 0.171 |
| Obesity, n (%) * | 151 (8.4) | 38 (9.5) | 109 (7.8) | 0.208 |
| Anemia, n (%)* | 1,461 (81.4) | 322 (80.3) | 1,139 (81.7) | 0.506 |
| Race, n (%) | 0.356 | |||
|
White Black/ African American Hispanic or Latino Asian Unknown |
1,406 (78.4) 112 (6.2) 103 (5.7) 36 (2.0) 137 (7.6) |
324 (80.8) 24 (6.0) 24 (6.0) 8 (2.0) 21 (5.2) |
1,082 (77.7) 88 (6.3) 79 (5.7) 28 (2.0) 116 (8.3) |
|
| Geographic Region, n (%) | 0.330 | |||
|
Northeast Midwest South West Unknown |
201(11.2) 534 (29.8) 761 (42.4) 294 (16.4) 4 (0.2) |
42 (10.5) 107 (26.7) 184 (45.9) 68 (17.0) 1 (0.2) |
159 (11.4) 427 (30.7) 577 (41.4) 226 (16.2) 4 (0.3) |
|
| Charlson Comorbidity Index (mean ± SD) | 1.5 + 2.4 | 0.8 + 1.5 | 1.7 + 2.5 | <0.001 |
| Net Worth, n (%) | 0.213 | |||
|
<$25K $25K-$149K $150K-$249K $250K-$499K >$500K Unknown |
170 (9.5) 358 (20.0) 247 (13.8) 420 (23.4) 360 (20.1) 239 (13.3) |
48 (12.0) 80 (20.0) 52 (13.0) 94 (23.4) 85 (21.2) 42 (10.5) |
122 (8.8) 278 (20.0) 195 (14.0) 326 (23.4) 275 (19.7) 197 (14.1) |
|
| Education, n (%) | <0.001 | |||
|
Less than 12th Grade High School Diploma Less than Bachelor’s At least Bachelor’s Degree Unknown |
12 (0.7) 407 (22.7) 958 (53.4) 358 (20.0) 59 (3.3) |
1 (0.3) 75 (18.7) 226 (56.4) 95 (23.7) 4 (1.0) |
11 (0.8) 332 (23.8) 732 (52.6) 263 (18.9) 55 (4.0) |
|
| Steroids, n (%) | 1,033 (57.6) | 329 (82.0) | 704 (50.5) | <0.001 |
| Surgery Type, n (%)** | <0.001 | |||
|
TAC with end ileostomy TPC with end ileostomy TPC with IPAA |
918 (51.2) 291 (16.2) 585 (32.6) |
164 (40.9) 62 (15.5) 175 (43.6) |
754 (54.1) 229 (16.4) 410 (29.4) |
|
Groups matched based on age, sex, anemia, malnutrition, obesity, and diabetes
TAC: total abdominal colectomy, TPC: total proctocolectomy, IPAA: ileal pouch anal anastomosis
Healthcare utilization was evaluated by length of hospital stay, unplanned reoperation or procedure, emergency department visit, or readmission within 30 days after surgery. The biologic-exposed group had a shorter median length of hospital stay (7 vs. 8 days, p<0.001). There were no statistically significant differences in other healthcare utilization parameters between patients with and without preoperative biologic exposure (Table 2). There were no statistically significant differences in utilization outcomes by biologic exposure or steroid exposure after adjusting for covariates in multivariate regression (Table 3). However, operations involving total proctocolectomy were associated with higher odds of emergency department visits compared to TAC: TPC with end ileostomy had an aOR of 1.9 [1.1-3.1] (p=0.02), and TPC with IPAA had an aOR of 3.0 [2.1-4.3] (p<0.001). IPAA was also associated with higher odds of readmission within 30 days postoperatively (aOR 1.3 [1.1-1.6], p=0.02) compared to TAC.
Table 2:
Healthcare utilization of matched biologic and non-biologic-exposed groups.
| n (%) | Total n= 1,794 |
Preoperative Biologic Exposure n= 401 (22.4%) |
No Biologic Exposure n= 1,393 (77.6%) |
p-value |
|---|---|---|---|---|
| Hospital Length of Stay, Days, median (IQR) | 8 (5, 14) | 7 (5, 12) | 8 (6, 15) | <0.001 |
| Unplanned Procedure or Reoperation, n (%) | 188 (10.5) | 39 (9.7) | 149 (10.7) | 0.576 |
| Emergency Department Visit, n (%) | 163 (9.1) | 44 (11.0) | 119 (8.5) | 0.136 |
| Readmission, n (%) | 461 (25.7) | 100 (24.9) | 361 (25.9) | 0.693 |
IQR: Interquartile range
Table 3:
Multivariate regression results of healthcare utilization, by pre-operative biologic exposure, pre-operative steroid exposure, and operation type.
| Independent Variables | Unplanned Procedure or Reoperation | Emergency Department Visits | Readmission | |||
|---|---|---|---|---|---|---|
| aOR [95% CI] | p-value | aOR [95% CI] | p-value | aOR [95% CI] | p-value | |
| No Biologic Exposure Biologic Exposure |
Referent 1.2 [0.8, 1.7] |
0.47 |
1.0 [0.7, 1.5] |
0.83 |
1.0 [0.7, 1.3] |
0.92 |
| No Steroid Exposure Steroid Exposure |
Referent 0.8 [0.6, 1.2] |
0.27 |
1.1 [0.8, 1.6] |
0.63 |
1.1 [0.8, 1.3] |
0.63 |
| Surgery Type | ||||||
|
TAC with end ileostomy TPC with end ileostomy TPC with IPAA |
Referent 0.6 [0.4, 1.1] 0.9 [0.6, 1.2] |
0.09 0.38 |
1.9 [1.1, 3.1] 3.0 [2.1, 4.3] |
0.02 <0.001 |
0.8 [0.6, 1.2] 1.3 [1.1, 1.6] |
0.34 0.02 |
aOR: Odds Ratio adjusted for age, sex, Charlson Comorbidity Index, region, race, and net worth. CI: confidence interval.
Twenty-one percent of patients experienced at least one complication within 30 days of surgery. In bivariate analysis, biologic exposure was not associated with a change in the rate of complications (Table 4). In multivariate analysis, the odds of infectious, hernia or wound disruption, thromboembolic, and cardiopulmonary complications did not differ significantly between biologic-exposed and non-exposed groups (Table 5). Patients with preoperative steroid exposure had higher odds of cardiopulmonary complications (aOR 1.4 [1.0-2.0], p=0.05) compared to those without steroid exposure. Patients who underwent TPC with IPAA had twice the odds of developing postoperative infectious complications (aOR 2.2 [1.5-3.0], p<0.001) compared to those who underwent TAC with end ileostomy. Infectious complications were twice as common after proctocolectomy with IPAA (14%) compared to TAC with end ileostomy (7%); 10% of patients who underwent TPC with end ileostomy experienced an infectious complication. Proctocolectomy patients had lower odds of cardiopulmonary complications: those undergoing TPC with end ileostomy had an aOR of 0.6 [0.4-0.9] (p=0.03), and IPAA had an aOR of 0.4 [0.3-0.6] (p<0.001) compared to those undergoing TAC. Cardiopulmonary complications were nearly twice as common following total abdominal colectomy (14%) than after proctocolectomy (TPC with end ileostomy: 9%; TPC with IPAA: 6%).
Table 4:
Postoperative complications of matched biologic and non-biologic-exposed groups.
| n (%) | Total n= 1,794 |
Preoperative Biologic Exposure n= 401 (22.4%) |
No Biologic Exposure n= 1,393 (77.6%) |
p-value |
|---|---|---|---|---|
| Infectious | 176 (9.8) | 39 (9.7) | 137 (9.8) | 0.095 |
| Hernia or Wound Disruption | 124 (6.9) | 29 (7.2) | 95 (6.8) | 0.774 |
| Thromboembolic | 84 (4.7) | 18 (4.5) | 66 (4.7) | 0.835 |
| Cardiopulmonary | 190 (10.6) | 36 (9.0) | 154 (11.1) | 0.233 |
| Any Complication* | 383 (21.4) | 77 (19.2) | 306 (22.0) | 0.234 |
Categories are not mutually exclusive; patients could have more than complication.
Table 5:
Multivariate regression results of postoperative complications in the matched cohort, by preoperative biologic exposure, preoperative steroid exposure, and operation type.
| Pre-operative and Operative Characteristics | Infectious Complication | Hernia or Wound Disruption | Thromboembolic Complication | Cardiopulmonary Complication | ||||
|---|---|---|---|---|---|---|---|---|
| aOR [95% CI] | p-value | aOR [95% CI] | p-value | aOR [95% CI] | p-value | aOR [95% CI] | p-value | |
| No Biologic Exposure Biologic Exposure |
Referent 0.9 [0.6, 1.3] |
0.437 |
1.2 [0.7, 1.8] |
0.534 |
0.9 [0.5, 1.5] |
0.619 |
0.9 [0.6, 1.3] |
0.482 |
| No Steroid Exposure Steroid Exposure |
Referent 0.9 [0.7, 1.3] |
0.657 |
1.1 [0.7, 1.6] |
0.708 |
1.4 [0.8, 2.3] |
0.253 |
1.4 [1.0, 2.0] |
0.05 |
| Surgery Type | ||||||||
|
TAC with end ileostomy TPC with end ileostomy TPC with IPAA |
Referent 1.3 [0.8, 2.1] 2.2 [1.5, 3.0] |
0.356 <0.001 |
0.7 [0.4, 1.3] 0.7 [0.5, 1.1] |
0.247 0.169 |
1.0 [0.5, 2.1] 1.2 [0.7, 2.0] |
0.912 0.485 |
0.6 [0.4, 0.9] 0.4 [0.3, 0.6] |
0.03 <0.001 |
| TAC with end ileostomy | ||||||||
|
No Biologic Exposure (n=754) Biologic Exposure (n= 164) |
Referent 1.4 [0.7, 2.6] |
0.34 |
1.0 [0.5, 2.0] |
0.99 |
0.9 [0.3, 2.2] |
0.76 |
0.6 [0.4, 1.1] |
0.13 |
|
No Steroid Exposure (n=418) Steroid Exposure (n=500) |
Referent 0.9 [0.5, 1.6] |
0.71 |
0.8 [0.5, 1.5] |
0.51 |
0.7 [0.4, 1.6] |
0.43 |
1.1 [0.7, 1.7] |
0.71 |
| TPC with end ileostomy | ||||||||
|
No Biologic Exposure (n=229) Biologic Exposure (n= 62) |
Referent 1.1 [0.4, 3.3] |
0.88 |
3.0 [0.9, 10.0] |
0.08 |
1.7 [0.4, 7.4] |
0.47 |
2.1 [0.7, 6.2] |
0.18 |
|
No Steroid Exposure (n=129) Steroid Exposure (n=162) |
Referent 0.5 [0.2, 1.3] |
0.16 |
4.1 [0.9, 18.5] |
0.06 |
1.9 [0.4, 8.7] |
0.39 |
1.4 [0.4, 4.1] |
0.60 |
| TPC with IPAA | ||||||||
|
No Biologic Exposure (n=410) Biologic Exposure (n= 175) |
Referent 0.5 [0.3, 0.9] |
0.03 |
1.2 [0.5, 2.7] |
0.66 |
0.7 [0.3, 1.6] |
0.39 |
0.9 [0.4, 2.1] |
0.89 |
|
No Steroid Exposure (n=214) Steroid Exposure (n=371) |
Referent 1.1 [0.7, 1.8] |
0.72 |
1.1 [0.5, 2.3] |
0.88 |
3.2 [1.1, 9.3] |
0.03 |
4.6 [1.6, 13.0] |
0.004 |
aOR: Odds Ratio adjusted for age, sex, Charlson Comorbidity Index, region, race and net worth. CI: confidence interval
Given the difference in infectious complications based on type of surgery, we wanted to look further at the effect of biologics on complication by surgery type. Neither biologics nor steroids affected odds of complications for patients undergoing TAC or TPC with end ileostomy. Among patients who had TPC with IPAA, biologic exposed patients had lower odds of infectious complications (aOR 0.5 [0.3-0.9], p=0.03). The higher odds of infectious complications in this surgery group were attributable to the patients who had not been exposed to biologic medications before surgery. However, the TPC with IPAA group that had been exposed to steroids had higher odds of both thromboembolic (aOR 3.2 [1.1-9.3], p=0.03) and cardiopulmonary complications (aOR 4.6 [1.6-13.0], p=0.004). Taken together, these data show that the higher odds of cardiopulmonary complications seen among all patients exposed to steroids is largely due to those who had TPC with IPAA. While the odds of cardiopulmonary complications are highest among patients undergoing TAC with end ileostomy, there was no difference in odds whether or not this group was exposed to biologic medications or steroids.
DISCUSSION
This is one of the largest studies to assess postoperative complications among UC patients exposed or not exposed to biologic medications. We found no difference in postoperative complications after total colectomy or proctocolectomy whether or not patients were prescribed biologic medications if the surgery included end ileostomy. Patients who had TPC with IPAA had double the odds of an infectious complication compared to patients who got and end ileostomy. However, for patients who underwent TPC with IPAA, biologic exposure was associated with lower odds of infectious complications. Exposure to biologic medications also did not result in more unplanned reoperations, emergency department visits, or readmissions.
This study reinforces prior work showing that TPC with IPAA is more prone to infectious complications.10 Of the three operations performed for UC, IPAA is the only one involving the formation of anastomoses, which may be the reason behind the highest risk of infection. The higher odds of emergency department visits and readmissions in the IPAA group is likely due to the higher risk of infectious complications, in addition to the higher risk for dehydration from a more proximal loop ileostomy. Interestingly, among the IPAA group, having received preoperative biologic medications was associated with lower odds of infectious complications in our study. This could be due to better disease control in the biologic-exposed group. Kulaylat et al similarly looked at complications after IPAA among UC patients using a different insurance claims database. In their cohort, anti-TNF therapy did not affect the rates of fistula, abscess, sepsis/bacteremia/pneumonia, or wound debridement/dehiscence. The differences with our findings are potentially due to their study having evaluated codes for specific complications as opposed to our approach of evaluating all postoperative claims codes to determine complication rates. Our methodology allows for more broadly detecting infectious complications such as anastomotic leak, a limitation they acknowledge in their work.9
When looking at all UC patients regardless of surgery type, steroid exposure was associated with higher odds of cardiopulmonary complications. The odds were highest among patients undergoing TAC with end ileostomy. Although the odds of cardiopulmonary complications were lowest in the IPAA group, even among IPAA patients steroid exposure was associated with a nearly five-fold higher odds of cardiopulmonary complications compared to non-exposed IPAA patients. Steroid-exposed IPAA patients also had three-fold higher odds of thromboembolic events compared to non-steroid exposed IPAA patients. This suggests that physicians should have an increased concern for cardiopulmonary complications when operating on steroid-exposed UC patients, regardless of surgery performed, and for thromboembolic events after IPAA surgery. The prophylactic anticoagulation status for these patients is unknown, but certainly could play a role in these findings.
The effects of biologic medications on perioperative outcomes differ between Crohn’s disease and ulcerative colitis. Studies in CD more consistently show an increased risk of postoperative infectious and non-infectious complications after biologic-exposure.7,27,28 While we do not know why biologics have more of a detrimental effect on postoperative complications in CD, it may be because colectomy in UC is curative of IBD, but patients with CD still have IBD after surgery. In UC, results of prior studies regarding the effects of biologic medications on postoperative complications are mixed, but the evidence increasingly shows no effects on infectious complications.6–10 In contrast to other large series that show an increase in complications during the “biologic era,” our work directly evaluated complications and biologic-exposure on a per patient basis. We suspect that patients compared by era may differ in other meaningful ways such as more advanced disease severity when undergoing operation in the “biologic era” compared to the “pre-biologic era.” Our study design allows for direct attribution of complications to a patient exposed or not-exposed to biologics. It was unexpected to have lower odds of infectious and wound complications among the steroid-exposed group. Using this claims database, we were unable to evaluate the dose of steroid, but most (81%) of steroid exposed patients were on prednisone. Using a different insurance claims database, no effect of steroids on infectious complications has previously been shown.9 We suspect this is a limitation of using claims database in that steroid exposure chronicity and dose could not be determined. Patients may have been on low dose prednisone or short-term steroids and thus had minimal impact on infectious complications.
This study has several limitations. The outcomes reported here are from a privately insured population, and thus may not be generalizable to uninsured or government-insured patients. In using an insurance claims database, administrative inaccuracies in billing codes can affect the integrity of the data and selection bias from clinical decision making cannot be eliminated. It is possible that pharmaceutical claims do not reflect actual drug use, as not all patients who fill a prescription use the medications as prescribed. However, work by Tkacz et al show that, for rheumatoid arthritis, anti-TNFa adherence is high.17 Although we matched for anemia and malnutrition as markers of disease severity, these parameters are unlikely to fully represent disease severity. We are also unable to account for whether surgeries were elective or emergent. In the matching, we chose to minimize imbalance while maintaining sample size. After matching, the biologic-exposed group was younger and had a lower CCI than the non-exposed group. To account for the remaining imbalance between cohorts, we used multivariate regression to adjust for age and CCI. Despite adjustments, we acknowledge that these cohort differences may be important markers of residual confounding and could be influential. As this study was one of the first exploratory observational studies of UC complications in a large private-payer claims database, p-values were not adjusted for multiple comparisons in consideration of the balance between type I and type II error risk. Although testing multiple outcomes without p-value adjustment increases the risk of type I errors, or false positive results, the alternative, adjusting p-values for multiple comparisons, would increase the risk of type II errors (false negative results), thereby missing potentially important findings.29–31 Our work supports that of others that biologics may not increase risk of infectious postoperative complications in ulcerative colitis. We hope this work will spur further research to investigate even more granular clinical data which may be difficult to study through insurance claims.
CONCLUSION
Our study provides further evidence that preoperative exposure to biologic medications is not associated with more healthcare utilization or with increased postoperative complications in ulcerative colitis patients undergoing curative surgery. These findings may aid patients with ulcerative colitis and their physicians through the complex decision-making process to determine timing and type of operation.
Supplementary Material
Supplementary Table 1: Codes for diagnosis of ulcerative colitis.
Supplementary Table 2: Codes for determination of surgery type. Surgery types were determined by combined ICD-9 or 10 codes as shown or by CPT codes for the specific surgery type. ICD coding: All patients with ICD-9 or ICD-10 codes for total abdominal colectomy (TAC) plus code for ileal pouch anal anastomosis (IPAA) were designated total proctocolectomy (TPC) with IPAA. Patients with code for TAC plus proctectomy plus ileostomy were designated TPC with end ileostomy. The remaining patients who did not meet any of the above designations were designated TAC with end ileostomy. CPT coding: CPT codes for the specific surgery types are shown in the last column.
Supplementary Table 3: Complication codes grouped by clinical domain.
Supplementary Table 4: Demographic characteristics of biologic-exposed and non-biologic-exposed groups before coarsened exact matching.
ACKNOWLEDGMENTS
Data were accessed via the Stanford Center for Population Health Sciences Data Core. This core is supported by a National Institutes of Health National Center for Advancing Translational Science Clinical and Translational Science Award (UL1 TR001085) and from internal Stanford funding. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Additional support is from the Donna and Frederick Fluegel Colorectal Surgery Fund.
Funding/Support: National Institutes of Health National Center for Advancing Translational Science Clinical and Translational Science Award UL1 TR001085, Donna and Frederick Fluegel Colorectal Surgery Fund.
Footnotes
Financial Disclosures: None reported.
REFERENCES
- 1.Vester-Andersen MK, Prosberg MV, Jess T, et al. Disease course and surgery rates in inflammatory bowel disease: a population-based, 7-year follow-up study in the era of immunomodulating therapy. Am J Gastroenterol. 2014;109:705–714. [DOI] [PubMed] [Google Scholar]
- 2.Peyrin-Biroulet L, Germain A, Patel AS, Lindsay JO. Systematic review: outcomes and post-operative complications following colectomy for ulcerative colitis. Aliment Pharmacol Ther. 2016;44:807–816. [DOI] [PubMed] [Google Scholar]
- 3.Mor IJ, Vogel JD, da Luz Moreira A, Shen B, Hammel J, Remzi FH. Infliximab in ulcerative colitis is associated with an increased risk of postoperative complications after restorative proctocolectomy. Dis Colon Rectum. 2008;51:1202–1207. [DOI] [PubMed] [Google Scholar]
- 4.Abelson JS, Michelassi F, Mao J, Sedrakyan A, Yeo H. Higher Surgical morbidity for ulcerative colitis patients in the era of biologics. Ann Surg. 2018;268:311–317. [DOI] [PubMed] [Google Scholar]
- 5.Justiniano CF, Aquina CT, Becerra AZ, et al. Postoperative mortality after nonelective surgery for inflammatory bowel disease patients in the era of biologics. Ann Surg. 2019;269:686–691. [DOI] [PubMed] [Google Scholar]
- 6.Yang Z, Wu Q, Wu K, Fan D. Meta-analysis: pre-operative infliximab treatment and short-term post-operative complications in patients with ulcerative colitis. Aliment Pharmacol Ther. 2010;31:486–492. [DOI] [PubMed] [Google Scholar]
- 7.Billioud V, Ford AC, Tedesco ED, Colombel J-F, Roblin X, Peyrin-Biroulet L. Preoperative use of anti-TNF therapy and postoperative complications in inflammatory bowel diseases: a meta-analysis. J Crohn’s Colitis. 2013;7:853–867. [DOI] [PubMed] [Google Scholar]
- 8.Selvaggi F, Pellino G, Canonico S, Sciaudone G. Effect of preoperative biologic drugs on complications and function after restorative proctocolectomy with primary ileal pouch formation: systematic review and meta-analysis. Inflamm Bowel Dis. 2015;21:79–92. [DOI] [PubMed] [Google Scholar]
- 9.Kulaylat AS, Kulaylat AN, Schaefer EW, et al. Association of preoperative anti-tumor necrosis factor therapy with adverse postoperative outcomes in patients undergoing abdominal surgery for ulcerative colitis. JAMA Surg. 2017;152:e171538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Zittan E, Milgrom R, Ma GW, et al. Preoperative anti-tumor necrosis factor therapy in patients with ulcerative colitis is not associated with an increased risk of infectious and noninfectious complications after ileal pouch-anal anastomosis. Inflamm Bowel Dis. 2016;22:2442–2447. [DOI] [PubMed] [Google Scholar]
- 11.Lightner AL, McKenna NP, Moncrief S, Pemberton JH, Raffals LE, Mathis KL. Surgical outcomes in vedolizumab-treated patients with ulcerative colitis. Inflamm Bowel Dis. 2017;23:2197–2201. [DOI] [PubMed] [Google Scholar]
- 12.Park KT, Sceats L, Dehghan M, et al. Risk of post-operative surgical site infections after vedolizumab vs anti-tumour necrosis factor therapy: a propensity score matching analysis in inflammatory bowel disease. Aliment Pharmacol Ther. 2018;48:340–346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Geltzeiler CB, Lu KC, Diggs BS, et al. Initial surgical management of ulcerative colitis in the biologic era. Dis Colon Rectum. 2014;57:1358–1363. [DOI] [PubMed] [Google Scholar]
- 14.Patkar A, Fegelman E, R Kashyap S, et al. Assessing the real-world effect of laparoscopic bariatric surgery on the management of obesity-related comorbidities: A retrospective matched cohort study using a US Claims Database. Diabetes Obes Metab. 2017;19:181–188. [DOI] [PubMed] [Google Scholar]
- 15.Kimball CC, Nichols CI, Vose JG, Peled AW. Trends in lumpectomy and oncoplastic breast-conserving surgery in the US, 2011-2016. Ann Surg Oncol. 2018;25:3867–3873. [DOI] [PubMed] [Google Scholar]
- 16.Lee MP, Sobanko JF, Shin TM, et al. Evolution of excisional surgery practices for melanoma in the United States. JAMA Dermatol. 2019. August 28. doi: 10.1001/jamadermatol.2019.2346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tkacz J, Ellis L, Bolge SC, Meyer R, Brady BL, Ruetsch C. Utilization and adherence patterns of subcutaneously administered anti-tumor necrosis factor treatment among rheumatoid arthritis patients. Clin Ther. 2014;36:737–747. [DOI] [PubMed] [Google Scholar]
- 18.Atzinger CB, Guo JJ. Biologic disease-modifying antirheumatic drugs in a national, privately insured population: utilization, expenditures, and price trends. Am Health Drug Benefits. 2017;10:27–36. [PMC free article] [PubMed] [Google Scholar]
- 19.Desai RJ, Gagne JJ, Lii J, Liu J, Friedman S, Kim SC. Comparative risk of incident venous thromboembolism in patients with inflammatory bowel disease initiating tumour necrosis factor-α inhibitors or nonbiologic agents: a cohort study. CMAJ. 2017;189:E1438–E1447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rezaie A, Quan H, Fedorak RN, Panaccione R, Hilsden RJ. Development and validation of an administrative case definition for inflammatory bowel diseases. Can J Gastroenterol. 2012;26:711–717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Park JE, Barbul A. Understanding the role of immune regulation in wound healing. Am J Surg. 2004;187(suppl 1):11S–16S. [DOI] [PubMed] [Google Scholar]
- 22.Jensen JS, Petersen NB, Biagini M, Bollen P, Qvist N. Infliximab treatment reduces tensile strength in intestinal anastomosis. J Surg Res. 2015;193:145–152. [DOI] [PubMed] [Google Scholar]
- 23.Shah ED, Farida JP, Siegel CA, Chong K, Melmed GY. Risk for overall infection with anti-tnf and anti-integrin agents used in ibd: a systematic review and meta-analysis. Inflamm Bowel Dis. 2017;23:570–577. [DOI] [PubMed] [Google Scholar]
- 24.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383. [DOI] [PubMed] [Google Scholar]
- 25.Blackwell M, Iacus S, King G, Porro G. CEM: Coarsened exact matching in STATA. Stata J. 2009;9:524–546. [Google Scholar]
- 26.King G, Nielsen R. Why propensity scores should not be used for matching. Polit Anal. 2019;27:435–454. [Google Scholar]
- 27.Kopylov U, Ben-Horin S, Zmora O, Eliakim R, Katz LH. Anti-tumor necrosis factor and postoperative complications in Crohn’s disease: systematic review and meta-analysis. Inflamm Bowel Dis. 2012;18:2404–2413. [DOI] [PubMed] [Google Scholar]
- 28.Yang Z-P, Hong L, Wu Q, Wu K-C, Fan D-M. Preoperative infliximab use and postoperative complications in Crohn’s disease: a systematic review and meta-analysis. Int J Surg. 2014;12:224–230. [DOI] [PubMed] [Google Scholar]
- 29.Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology. 1990;1:43–46. [PubMed] [Google Scholar]
- 30.Thomas DC, Siemiatycki J, Dewar R, Robins J, Goldberg M, Armstrong BG. The problem of multiple inference in studies designed to generate hypotheses. Am J Epidemiol. 1985;122:1080–1095. [DOI] [PubMed] [Google Scholar]
- 31.Althouse AD. Adjust for multiple comparisons? It’s not that simple. Ann Thorac Surg. 2016;101:1644–1645. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Table 1: Codes for diagnosis of ulcerative colitis.
Supplementary Table 2: Codes for determination of surgery type. Surgery types were determined by combined ICD-9 or 10 codes as shown or by CPT codes for the specific surgery type. ICD coding: All patients with ICD-9 or ICD-10 codes for total abdominal colectomy (TAC) plus code for ileal pouch anal anastomosis (IPAA) were designated total proctocolectomy (TPC) with IPAA. Patients with code for TAC plus proctectomy plus ileostomy were designated TPC with end ileostomy. The remaining patients who did not meet any of the above designations were designated TAC with end ileostomy. CPT coding: CPT codes for the specific surgery types are shown in the last column.
Supplementary Table 3: Complication codes grouped by clinical domain.
Supplementary Table 4: Demographic characteristics of biologic-exposed and non-biologic-exposed groups before coarsened exact matching.