Introduction

The last decade has witnessed increased utilization of biologic therapies for inflammatory bowel disease (IBD) 1,2,3. Although surgery remains an essential treatment option for inflammatory bowel disease, successful implementation of effective and timely biologic therapies may reduce the need for surgery.

Twenty percent of ulcerative colitis and 80% of Crohn’s disease (CD) patients will undergo surgery during their lifetime 4. A total proctocolectomy for ulcerative colitis (UC) is an option for patients with disease refractory to medical treatment, and can be curative and improve the quality of life for UC patients 5. In addition for medically refractory Crohn’s disease (CD), surgery is also commonly employed for complications such as strictures, fistulas, and abscesses6.

With the advent of biologic therapy starting with the FDA approval of infliximab in 1998 and more aggressive treatment goals of achieving mucosal healing7, one would predict a decrease in the proportion of IBD patients requiring surgery. Yet, studies to date have lacked consensus. A study of the National Inpatient Sample (NIS) showed that surgery rates for CD and UC were unchanged from 2003 to 20128, while a Spanish study showed a decrease in the rate of surgery for CD following anti-TNF approval but no change for UC9, and a Danish study showed a decrease in the rate of surgery for both CD and UC10. This study aimed to further investigate this trend, using a large US admissions database.

Furthermore, if the rate of IBD surgery is decreasing due to biologic therapy, the hypothesis is that those requiring surgery are presenting more urgently with more severe disease, and thus have higher post-operative mortality. Post-operative mortality in the era of biologic therapy has been studied in smaller cohorts. A study of a New York State database showed an increase in mortality after nonelective surgery in UC but a decrease in CD11. A Canadian study with a post-colectomy mortality rate of 1.5% in UC, identified older age, emergent surgery, and more than 2 comorbidities as independent predictors of post-operative complications12. However, few studies have been published on the subject using US nationwide databases.

This study sought to utilize a US nationwide database to evaluate whether rates of IBD-related surgery, urgent surgery, and post-operative mortality in Crohn’s disease (CD) and ulcerative colitis (UC) have changed over the last decade in the era of biologic therapies.

Materials and Methods

The Nationwide Readmission Database (NRD) was developed by the Healthcare Cost and Utilization Project (HCUP) and provides nationally representative estimates of approximately 36 million all-payer acute care hospitalizations each year throughout the USA13. The NRD captures and links all admissions for a patient in a given calendar year within a given state, which allows for evaluation of each patient’s history of hospitalizations throughout the year. It does not capture prior hospitalizations beyond the calendar year. The NRD also includes data on demographics, diagnoses, procedures, payer mix, death, and hospital characteristics.

The NRD from all available years (2010 through 2017) was queried for patients with CD and UC using the International Classification of Diseases, Clinical Modification, Ninth and Tenth Revision (ICD-9-CM, ICD-10-CM) codes in all diagnosis positions (Supplemental Table 1). Patients with concurrent diagnoses of CD and UC were excluded. IBD-related surgeries were defined as having undergone a small bowel resection (including ileocolic resection) or colectomy (segmental colectomy, proctectomy, total colectomy with ileostomy, or total proctocolectomy) (Supplemental Table 2). The index hospitalization was defined as the first hospitalization or first hospitalization involving an IBD-related surgery (if performed) for each patient in the calendar year.

The following variables were obtained from the NRD: patient age, sex, medical comorbidities, median household income for patient’s zip code, primary payor, elective vs. non-elective admission, hospitalization day number for IBD-related surgery, and death. The medical comorbidities used to calculate the Charlson-Deyo Comorbidity Index included myocardial infarction, congestive heart failure, cerebrovascular accident, peripheral vascular disease, chronic obstructive pulmonary disease, diabetes, chronic kidney disease, dementia, peptic ulcer disease, hepatic disease, rheumatologic disease, human immunodeficiency virus infection, and cancer14. Protein calorie malnutrition was included as an additional variable and identified based on ICD-9/-10 codes. Hospital characteristics included bed size (small, medium, large), hospital control/ownership (public, not-for-profit private, investor-owned private), and teaching status (metropolitan non-teaching, metropolitan teaching, rural). The database did not include information on race, medications, endoscopy reports, histopathology, results of radiographic studies, or any IBD-specific lab results such as C-reactive protein or fecal calprotectin.

The primary outcome was all IBD-related surgery. Secondary outcomes included urgent surgery (defined as surgery within 2 days of a non-elective admission) and post-operative 30-day mortality. All of the variables listed above were included in the univariable and multivariable analyses for post-operative 30-day mortality.

Results

There were 1,072,110 patients hospitalized with CD and 723,156 with UC between 2010 and 2017. Patients who underwent surgery were younger with fewer comorbidities, more often malnourished, from higher income households, and more likely to have private insurance than those who did not have surgery (Table 1). Compared to the non-surgical group, the proportion of men was greater in the surgical group for both CD and UC. Among the patients who had surgery a higher proportion were at larger hospitals, hospitals with private nonprofit ownership, teaching hospitals, and in more populated areas than those who did not undergo surgery (Table 1). For both UC and CD, 59% of the patients who underwent surgery were admitted electively (Table 1). The annual number of patients hospitalized with CD increased from 126,337 to 143,700 and from 82,713 to 97,597 in UC (Fig. 1). The absolute number of surgeries performed for CD remained stable from 12,708 in 2010 to 12,697 in 2017 and for UC increased from 6409 in 2010 to 7403 in 2017 (Fig. 2). Over the study period, the proportion of hospitalized CD patients undergoing surgery decreased from 10 to 8.8% (p < 0.001) and from 7.7 to 7.5% for UC patients (p < 0.001) (Fig. 3). The proportion of urgent surgeries (surgery within 2 days of non-elective admission) stayed stable from 25% in 2014 to 24% in 2017 for CD (p = 0.16) and dropped from 21 to 14% for UC (p < 0.001) (Fig. 4). Patients who underwent urgent surgery for both UC and CD were older with more comorbidities, more likely to have Medicare or Medicaid, from lower income households, and more often malnourished than those who had elective surgery. Those who had urgent surgery for Crohn's were more likely to be male, and for UC were more likely to be female. Among those who had urgent surgery a higher proportion were at government or private investor owned hospitals, non-teaching hospitals, and in less populated areas than those who underwent elective surgery (Table 2). Post-operative 30-day mortality in urgent and elective cases fluctuated between 1.2 and 1.6% in CD and decreased from 5.8 to 2.3% in UC (p < 0.001) (Fig. 5).

Table 1 Demographics of patients included in analysis, stratified by surgery or no surgery
Full size table
Fig. 1
figure 1

Total number of patients with Crohn’s disease and ulcerative colitis who are hospitalized each year. Multiple hospitalizations for the same patient within a given year are counted as one hospitalization

Full size image
Fig. 2
figure 2

Absolute number of surgeries per year for patients with Crohn’s disease and ulcerative colitis

Full size image
Fig. 3
figure 3

Overall surgical rates for patients with Crohn’s disease (p < 0.001) and ulcerative colitis (p < 0.001)

Full size image
Fig. 4
figure 4

Trends in rates of urgent surgery for Crohn’s disease (p = 0.16) and ulcerative colitis (p < 0.001)

Full size image
Table 2 Demographics of patients undergoing elective vs. urgent surgery
Full size table
Fig. 5
figure 5

Post-IBD surgical mortality within 30 days for Crohn’s disease (p < 0.01) and ulcerative colitis (p < 0.001)

Full size image

Multivariable analysis showed older age, higher Charlson Comorbidity Index, lower income, and non-elective admission to be predictors of post-op mortality for UC and CD, while rural hospital was an additional predictor for post-op mortality in CD (Table 3).

Table 3 Multivariable Cox proportional hazard analysis of post-op mortality
Full size table

When post-operative mortality was broken down by urgent or elective surgery, multivariable analysis showed that age and Charlson Comorbidity Index were risk factors for post-operative mortality for all surgeries, UC or CD and urgent or elective (Table 4). For elective UC surgery, income, malnutrition, and admission at a private non-profit hospital were also risk factors for post-operative mortality. For elective CD surgery, income and malnutrition were the only additional risk factors, and for urgent CD surgery, admission at a private non-profit or private investor–owned hospital, admission at a rural hospital, and malnutrition were also risk factors for 30-day post-operative mortality.

Table 4 Multivariable Cox proportional hazard analysis of post-op mortality by urgent vs. elective surgery
Full size table

Discussion

Infliximab was approved to treat moderate to severe Crohn’s disease in 1998 and ulcerative colitis in 2005. Between 1999 and 2012, 59,875 infusions of infliximab were administered in the Crohn’s Therapy, Resource, Evaluation, and Assessment Tool (TREAT) Registry15. Examining Johnson and Johnson sales data, the sale of infliximab increased incrementally every year of the study from $3099 million in 2010 to $4525M in 201716. A 2018 study of US insurance data showed an increase in biologic use in CD from 20 to 40% between 2009 and 2015 and 5 to 16% in UC patients17. This nationwide trend suggests that the use of these medications became more entrenched into the framework of IBD treatment for the general population, and not just specialized centers. The greater use of biologic therapy indicates that they provided an important alternative to surgery.

A study of the National Inpatient Sample (NIS) from 1998 to 2005 showed an increase in hospitalizations for both CD and UC with no change in the rates of surgical care18. Bewtra et al. showed that in the US, over the 14-year period from 1990 to 2003, there was no significant change in the annual rates of bowel resection surgeries for IBD19. A US nationwide study to evaluate surgical trends in the subsequent decade has not been performed.

Despite an increasing incidence of IBD20 and increasing rates of hospitalizations21, our study shows that during the past decade, surgery rates for IBD have declined. The results are similar to those of Ahmad et al. in the United Kingdom’s Hospital Episode Statistics (HES) database, where a significant decrease in the total and urgent surgery rates for CD occurred between 2003 and 2004 and 2012 and 2013 and the Canadian study Ma et al. which showed a decrease in both total and urgent surgery rates for CD between 2002 and 2011 22, 23.

The present study of the American population showed a significant decrease in both the total and urgent surgical rate for UC. A likely cause for this decline is improvement in medical management. Earlier biologic administration for moderate to severe disease and more aggressive management goals towards mucosal healing may have altered the natural history of the disease, leading to fewer surgeries. For ulcerative colitis, surgery is not usually considered until a patient has failed medical management or more rarely has developed toxic colitis or megacolon, a surgical emergency. The introduction of more timely and effective biologic therapies may have provided treatment options that decreased the need for surgery.

The elective or non-elective basis of surgery is also important, as recent studies have demonstrated lower mortality for colectomies performed electively rather than urgently24. Explanations for the decreasing rates of urgent colectomy include improved ambulatory care with better options to escalate medical management, as mentioned prior, along with treatment guidelines whose goal is deep remission and mucosal healing25.

In this era of biologic therapy, a concern amongst surgeons has been that patients are spending too much time testing medical treatments to salvage severe colitis, only to arrive at surgery with higher risk of morbidity and mortality than in the prebiologic era. Encouragingly, the NRD data showed a significant decrease in post-operative mortality for UC. The decrease in mortality observed in our study could be due to improved preoperative medical management, better surgical techniques, increased use of laparoscopy, and closer post-operative monitoring. These possible contributors could be further explored with future studies. The overall post-operative mortality rates observed in the NRD database (5.8% in 2010 to 2.3% in 2017) are more promising than the 8.1% in-hospital post-operative mortality seen in a study of the New York state database from 2006 to 201326. The New York study only included patients with UC and not CD and their patient population was slightly younger with more comorbidities. Our data included a larger group of patients and included mortality not just in hospital but for the 30-day post-operative period. We followed-up the concerning findings from the New York study using a larger database with a diversity of patients that more accurately reflects the national experience.

Crohn’s disease, on the other hand, can affect any segment of the gastrointestinal tract and patients often require multiple surgeries in their lifetime, for complications such as fistula, abscess, bleeding, strictures, or perforation. More than half of CD patients will require surgery within 10 years of their diagnosis27. However, surgery in CD is associated with a high rate of recurrence at anastomoses28, and risk of short gut with repeated surgeries. This study found a decrease in the total rate of surgery without a significant change in the rate of urgent surgery or post-operative mortality.

The decrease in overall rates of surgery could be due to the increased availability and usage of biologic therapy as mentioned for ulcerative colitis, but in the case of CD could also be secondary to increased use of advanced endoscopic interventions such as balloon dilation, and more advanced percutaneous techniques that can defer or prevent surgery in fistulizing CD patients. Although our study was not able to specifically look at these procedures, a pooled analysis showed that endoscopic balloon dilation was able to postpone surgical intervention in two-thirds of patients at 2 years29, and a meta-analysis showed that 30% of patients who underwent percutaneous drainage of an intra-abdominal abscess were able to avoid surgery, with a mean follow-up of 26 months30.

When evaluating patient and hospital-based variables affecting post-operative mortality, older age, higher Charlson Comorbidity Index, lower income, and non-elective admission were all individual predictors of post-operative mortality for both UC and CD. Both older age and higher Charlson Comorbidity Index are known to be associated with mortality. Meta-analysis of population-based IBD studies have shown higher post-operative mortality after urgent (as opposed to elective) intestinal resections for both UC and CD31. And even in countries with nationalized healthcare, lower income has been shown to be associated with increased mortality in surgical patients32.

Limitations of this study included the inability to identify patients in the NRD receiving biologic therapy, as well as the retrospective and observational nature of the study. The duration of preoperative hospitalization and the events during this time could not be well described due to the limitations of the dataset. The improved outcomes may also be related to better multidisciplinary care in inpatient and outpatient settings, better preoperative optimization, and post-operative care of the patient rather than due to biologics alone. Finally, the dataset captures hospital encounters and tracks a single patient over a year in order to determine readmission, but over the multiyear time course of the study, decreased surgery may be a result of decreased reoperations in a small subgroup of patients as opposed to a generalized decrease in the risk of resection for IBD patients overall. This limitation affects the CD patients more so as a subset of patients could have required multiple resections over the study period33. The study is not able to comment on recurrent surgery or long-term need for surgery in CD or UC due to the inability of patients to be linked beyond the readmission event. These questions could be addressed in future studies.

Despite these limitations, this is the largest study to date to evaluate recent trends in IBD surgery and includes not only the overall trends in surgical rates, but more particularly urgent surgery and post-operative mortality over the last decade in the USA.

Using the NRD data, nationwide rates of bowel resections for UC and CD have significantly declined in both UC and CD patients between 2010 and 2017. There has also been a significant decrease in the rates of urgent surgery for UC, and post-operative 30-day mortality for CD and UC. This trend may be related to advances in medical therapy, more specifically the emergence and increased use of biologics.