Impact of specialized inpatient IBD care on outcomes of IBD hospitalizations: A cohort study

Abstract

Background

The management of inflammatory bowel diseases (IBD; Crohn’s disease (CD), ulcerative colitis (UC)) is increasingly complex. Specialized care has been associated with improved ambulatory IBD outcomes.

Aims

To examine if the implementation of specialized inpatient IBD care modified short and long-term clinical outcomes in IBD-related hospitalizations.

Methods

This retrospective cohort study included IBD patients hospitalized between July 2013 and April 2015 at a single tertiary referral center where a specialized inpatient IBD care model was implemented in July 2014. In-hospital medical and surgical outcomes as well as post-discharge outcomes at 30 and 90 days were analyzed along with measures of quality of in-hospital care. Effect of specialist IBD care was examined on multivariate analysis.

Results

A total of 408 IBD-related admissions were included. With implementation of specialized IBD inpatient care, we observed increased frequency of use of high-dose biologic therapy for induction (26% vs. 9%, odds ratio (OR) 5.50, 95% CI 1.30 – 23.17) and higher proportion of patients in remission at 90 days after discharge (multivariate OR 1.60, 95% CI 0.99 – 2.69). While there was no difference in surgery by 90 days, among those who underwent surgery, early surgery defined as in-hospital or within 30 days of discharge, was more common in the study period (71%) compared to the control period (46%, multivariate OR 2.73, 95% CI 1.22 – 6.12). There was no difference in length of stay between the two years.

Conclusions

Implementation of specialized inpatient IBD care beneficially impacted remission and facilitated early surgical treatment.

INTRODUCTION

Inflammatory bowel diseases (IBD; Crohn’s disease (CD), ulcerative colitis (UC)) affect an estimated 1.5 million Americans, 2.2 million individuals in Europe, and several thousands more worldwide^{1, 2}. Characterized by a protracted relapsing-remitting course, both CD and UC frequently result in hospitalization or require surgical intervention for medically refractory disease or disease-related complications^3–6. The past two decades have witnessed a revolution in the management of IBD with availability of effective therapies targeting inflammation through different mechanisms⁷. Furthermore, treatment strategies have evolved to include early aggressive use of biologic therapy⁸, use of combination immunosuppression⁹, and personalized optimization through therapeutic drug monitoring and dose escalation in those with suboptimal treatment response^{10, 11}. As a result, management of IBD has become increasingly complex with rapid evolution of treatment algorithms, strategies, and targets¹⁰.

Several prior studies have examined the effect of specialized care in the management of IBD^12–16. It has been previously shown that particularly for complex IBD admissions requiring surgery, admission to a high-IBD volume hospital is associated with better outcomes in both CD and UC^{12, 13, 15}. In a population-based study from Manitoba province in Canada, early gastroenterologist care was associated with reduced need for surgery within two years after diagnosis of CD¹⁵. Indeed even among hospitalized patients, non-gastroenterologist care was associated with greater in-hospital mortality and at one year¹⁴. At many specialized high-volume centers, outpatient IBD care is concentrated in the hands of gastroenterologists with particular expertise in IBD. However, inpatient IBD care is frequently under the care of a general gastroenterologist or non-gastroenterologist physicians. The impact of specialist IBD inpatient care on patient outcomes has not been examined previously.

In our institution, beginning July 1, 2014, we initiated a specialized inpatient IBD service for management of IBD-related hospitalizations. This was attended on by one of eight gastroenterologists with expertise in IBD care and a high volume outpatient IBD practice. We hypothesized that provision of such specialized inpatient care would improve in-hospital and post-discharge outcomes. Therefore, the aims of this study were: (1) To compare in-hospital medical and surgical outcomes of IBD-related hospitalizations between July 2013 – June 2014 (control year without specialist IBD care) and July 2014 – April 2015 (specialist IBD inpatient care) ; (2) To compare post-discharge outcomes at 30 and 90 days including escalation of medical therapy, remission status, readmissions and surgery, and (3) To examine influence of specialist IBD care on measures of quality of in-hospital IBD care.

METHODS

Study population

This study included all IBD-related hospitalizations at Massachusetts General Hospital, a tertiary referral center for IBD care, between July 1, 2013 and April 30, 2015. Potential cases were identified using the Research Practice Data Registry which has been described previously in publications from our group^17–19. In brief, it is a continually populated data warehouse of all inpatient and outpatient contacts within the Partners Healthcare network and consists of discharge summaries, office visits, surgical and other operative procedures, radiology, endoscopy, pathology, and electronic prescription information. First, we identified eligible patients with a broad filter of at least one International Classification of Diseases, 9^th edition, clinical modification (ICD-9-CM) code for CD (555.x) or UC (556.x). Each case was reviewed to confirm IBD diagnosis and IBD-related hospitalizations. Patients treated by the pediatric gastroenterology service, admissions for elective surgery or post-surgical complications, and non-IBD related hospital stays were excluded for both years of the study.

Year one of the study formed the control period with general gastroenterologist care and consisted of all hospitalizations between July 1, 2013 and June 30, 2014. Year two of the study (study period), comprised the period after implementation of a specialized IBD inpatient service, and included all hospitalizations between July 1, 2014 and April 30, 2015. The cut-off in April was selected to allow for a full 90 day follow-up for all patients in year two.

Inpatient IBD service

In the control period, all IBD-related inpatient hospitalizations were being managed either in consultation with or primarily under the care of a general gastroenterologist attending on the inpatient service (in two-week blocks). Gastroenterology fellow involvement was split among all four inpatient fellows for the year. Additional input of an outpatient IBD attending was ad-hoc and at the discretion of both the inpatient and outpatient providers. In the second year, all IBD-related hospitalizations were managed by one of eight gastroenterologists specializing in IBD care; each gastroenterologist covered a period ranging from 4 to 10 weeks and worked in conjunction with one of the gastroenterology fellows and was responsible for coordinating care with other inpatient and outpatient providers. The IBD-gastroenterologists at our institution were unchanged during the two study years, and there were no changes to the overall census of patients under the care of a gastroenterology fellow. There were no structural changes such as co-locating the IBD patients or changes in care personnel during the study period. There were no pharmacists or case managers assigned specifically to the IBD patients during either of the two study years. There were also no standardized order sets used in either year of the study and access to surgical care including surgical staff availability and operating room time was also not different between the two years.

Covariates

Data from each admission was reviewed by one of four study investigators (CCYL, SS, RR, ANA). Inter-investigator agreement on subjective patient outcomes was assessed by random blinded duplicate review of records and disagreements resolved through consensus. Information was extracted regarding age, gender, smoking status, and source of admission (Emergency department (ED), outpatient clinic, or transfer from another hospital). Location and disease behavior in CD, and disease extent in UC were classified according to the Montreal classification²⁰. Prior severity of IBD was quantified by previous medications, need for hospitalizations or surgery, while severity at the time of hospitalization was assessed based on current medications as well as admission hemoglobin, albumin, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) values. Additionally, information was obtained about measures representing quality of in-hospital care including testing for Clostridium difficile (C difficile) and cytomegalovirus (CMV) infection, appropriate use of computed tomography (CT) or magnetic resonance imaging (MRI) scans, thromboprophylaxis against venous thromboembolism (VTE), as well as timing and frequency of lower endoscopic procedures and surgical consultation. Of these, testing for C difficile and thromboprophylaxis are included within the American Gastroenterological Association IBD physician performance quality measures while the Crohn’s and Colitis Foundation of America (CCFA) recommends sigmoidoscopy to evaluate for cytomegalovirus infection in individuals not responding to corticosteroids by day 3 of hospitalization. Additional measures such as use of C-reactive protein were selected by the study investigators as representing markers of quality specialist care based on prior evidence supporting their role, but are not formally recommended as quality measures by the professional societies²¹.

Outcomes

In-hospital medical outcomes examined included overall length of stay and initiation of cyclosporine or biologic therapy for rescue in the setting of corticosteroid refractory disease. In addition to ascertaining whether anti-TNF therapy (infliximab (IFX), adalimumab (ADA), certolizumab pegol (CZP), golimumab (GLM)) was initiated while inpatient, we examined the use of high-dose biologics (defined as > 5mg/kg of IFX) or an accelerated induction schedule (two doses less than 2 weeks apart). In-hospital surgical outcomes investigated were frequency of IBD-related surgical procedures (resection, colectomy, strictureplasty, exam under anesthesia with seton placement for fistulae, surgical or endoscopic dilation, and surgical or percutaneous drainage of abscess). We calculated time to such procedures from admission, and separately analyzed resection and non-resection procedures.

Post-discharge information was collected at 30 and 90 days and included outcomes of corticosteroid use, biologic initiation, ED visits, re-hospitalizations, and IBD-related surgery. Among those who underwent surgery by 90 days of follow-up, we examined the proportion of patients who received this procedure ‘early’, defined as occurring in-hospital or within 30 days of discharge. Review of outpatient notes was performed to identify proportion of patients with active disease or remission by global physician impression. For each patient, all data (patient notes, colonoscopy, radiology, laboratory tests) were reviewed by one of three study investigators (CYL, SS, RR) who made the adjudication about active disease or remission using all data available. To ensure consistency in this adjudication, a blinded review of 30 patients each was performed by each investigator and outcomes compared; no disagreements were identified. A random review of 25 charts was also performed by the senior investigator (AA) who concurred with the outcomes assigned by the study investigators. Post discharge follow up continued under the care of their previous gastroenterologist.

Statistical analysis

Statistical analyses were performed using Stata 13.1 (StataCorp, College Station, TX). Continuous variables were summarized using means and standard deviations and compared using the t-test. Categorical variables were presented as proportions and compared using the chi-squared test with Fisher’s exact test used when necessary. Univariate logistic (or linear) regression was performed among all IBD-related hospitalizations for our outcomes of interest with the year of hospitalization and other covariates as independent variables. For outcomes where a statistically significant (p ≤ 0.05) difference between the two years was identified, multivariate regression modeling was used to identify if there was an independent effect of year of hospitalization adjusting for other covariates significant on univariate analysis or previously described in the literature to be important. A two-sided p-value ≤ 0.05 indicated independent statistical significance. For analyses within each subtype of IBD, we additionally adjusted for disease location, behavior, presence of perianal complications (CD) or extent (UC). A sensitivity analysis was performed restricted to only index hospitalizations during each study year to minimize correlation between multiple hospitalizations for a patient during each of the study years. A power analysis was not performed as this study included all IBD-related hospitalizations during the two study periods. The study was approved by the Institutional Review Board of Partners Healthcare.

RESULTS

Study Population

There were a total of 408 IBD-related admissions included in our study from 306 unique patients; of these 196 were in the control period and 212 in the study period. There was no difference in age, gender, or type of IBD between the two years (Table 1). Among those with CD, there was no difference in disease location or behavior of admitted patients while among those with UC there was a slightly greater proportion of patients with pancolitis in the study period (75% vs. 63%, p=0.05). Patients admitted from the outpatient IBD clinic constituted a larger proportion of all admissions in the study period (25%) compared to the control period (12%).

Table 1.

Demographic, disease, and admission characteristics of patients with IBD-related hospitalizations at MGH between July 2013 – April 2015

Characteristic	Year 1 (General gastroenterology care) (n= 196)	Year 2 (Specialized IBD care) (n= 212)	p-value
Demographics
Mean age (SD) (in years)	41.4 (17.9)	39.6 (17.1)	0.30
Male	53	48	0.32
Smoking
Never	58	59	0.45
Current	12	16
Former	30	25
Source of Admission
Emergency department	74	63	0.004*
Outpatient clinic	12	25
Transfer from another hospital	14	12
Previous IBD-related
hospitalization	91	87	0.17
Previous IBD-related surgery	42	47	0.25
Age of Diagnosis (in years)
≤16	21	21	0.57
17–40	55	60
>40	24	20
Disease Characteristics
Type of IBD
Crohn’s Disease	67	66	0.79
Ulcerative Colitis	33	34
Disease Location (CD)
Ileal	14	18	0.22
Colonic	20	13
Ileocolonic	66	69
Upper GI involvement (CD)	12	10	0.63
Disease Behavior (CD)
Inflammatory	22	17	0.57
Stricturing	23	25
Penetrating	55	58
Perianal involvement (CD)	36	45	0.15
Disease Extent (UC)
Left-sided colitis	38	25	0.05*
Pancolitis	63	75
Past Medications (%)
5-ASA	67	62	0.37
Immunomodulators	57	59	0.70
Biologics	54	54	0.88
Characteristics at Hospitalization
Current Medications
Immunomodulators	19	27	0.09
Biologics	30	39	0.04*
Steroids	45	45	0.94
Admission Labs (Mean(SD))
Hemoglobin (g/dL)	11.6 (2.5)	11.8 (2.3)	0.36
Albumin (g/dL)	3.6 (0.7)	3.7 (0.6)	0.21
Erythrocyte sedimentation rate (mm/hr)	38.7 (32.9)	39.3 (28.1)	0.86
C-reactive protein (mg/L)	53.8 (58.6)	62.1 (61.5)	0.22

All figures represent percentages unless indicated otherwise

SD – standard deviation; IBD – inflammatory bowel disease; CD – Crohn’s disease; UC – ulcerative colitis; 5-ASA – 5-aminosalicylic acid; g/dL – gram/deciliter; mm/hr – millimeter/hour; mg/L – milligram/liter

^*p ≤ 0.05

There were no systematic differences in severity between the two years as measured by the proportion of patients with previous hospitalizations, surgery, or past IBD-related medications (all p > 0.10). However, there was a greater proportion of patients who were current users of biologics at the time of hospitalization in the study period (39%) compared to the control period (30%, p=0.04) and a trend towards more current users of immunomodulators (IMM). However, there were no statistically significant differences in admission laboratory values.

In-hospital Quality Measures

Table 2 compares various in-hospital quality and process measures between the two years. There was a uniformly high rate of testing for C difficile and thromboprophylaxis against VTE in both years. However, more objective biomarker monitoring was performed after implementation of a specialized inpatient IBD service as evidenced by the higher rates of CRP testing on admission (82% vs. 71%) and at discharge (25% vs. 13%) (p < 0.05). A total of 42% and 44% of patients were evaluated using lower endoscopy during hospitalization in the control period and study period respectively, and surgery was consulted in 45% and 47% of admissions. There was no significant difference in time to endoscopy or obtaining surgical consultation.

Table 2.

Comparison of measures of in-hospital quality of care before and after implementation of specialized inpatient IBD care

Characteristic	Year 1 (General gastroenterology care) (n= 196)	Year 2 (Specialized IBD care) (n= 212)	p-value
VTE prophylaxis	81	79	0.61
Testing for C. difficile	68	71	0.53
CMV testing	30	32	0.66
CRP obtained on admission	71	82	0.008*
CRP obtained at discharge	13	25	0.003*
Sigmoidoscopy or colonoscopy during hospitalization	42	44	0.68
Mean days to endoscopy (SD)	2.6 (2.2)	2.5 (2.8)	0.94
CT scan obtained during hospitalization	59	54	0.27
MRI obtained during the hospitalization	9	11	0.58
Surgical consult obtained	45	47	0.65
Mean days to surgical consult (SD)	1.7 (3.2)	1.7 (3.1)	0.92

All figures represent percentages unless indicated otherwise

VTE – venous thromboembolism; C. difficile – clostridium difficile; CMV – cytomegalovirus; CRP – C-reactive protein; SD – standard deviation; CT – computed tomography; MRI – magnetic resonance imaging

^*p ≤ 0.05

In-hospital Medical and Surgical Outcomes

Overall, we observed no differences in length of stay or rates of administration of in-hospital biologic therapy (either new or continued therapy) between the two years (Table 3). However, among those who received biologic therapy in hospital, high-dose biologic therapy (IFX dose >5mg/kg) was used more frequently after specialized inpatient IBD service implementation (26% of all new IFX infusions vs. 9%, p=0.04), a difference that remained significant on multivariable analysis (Odds ratio (OR) 5.50, 95% confidence interval (CI) 1.30 – 23.17). The mean time from admission to biologic therapy was 5.6 days and 5.2 days respectively (p=0.64). Administration of accelerated induction therapy was numerically but not statistically greater in the study period (13% vs. 4%).

Table 3.

Comparison of in-hospital medical and surgical outcomes before and after implementation of specialized inpatient IBD care

Outcome	Year 1 (General gastroenterology care) (n= 196)	Year 2 (Specialized IBD care) (n= 212)	p-value
Length of stay (in days) (Mean(SD))	6.4 (5.6)	7.2 (6.1)	0.17
Use of intravenous corticosteroids	60	51	0.09
New initiation of biologic therapy	23	16	0.11
Time to new biologic initiation (in days) (mean (SD))	5.7 (4.7)	5.6 (5.1)	0.95
Accelerated biologic dose†	4	13	0.24
High-dose biologic therapy†	9	26	0.04*
New initiation of IMM	7	6	0.84
Cyclosporine therapy	0.5	1	1.00
Antiviral therapy	5	2	0.08
Any IBD-related surgical procedure	13	19	0.07
Time to surgery (in days) (mean(SD))	7.6 (5.6)	4.5 (5.7)	0.04*
Bowel resection	6	7	0.84
Time to resection (in days) (mean(SD))	9.8 (5.0)	9.1 (7.4)	0.77
Time to non-resective operative intervention (in days) (mean(SD))	5.3 (1.6)	2.1 (2.1)	0.01

All figures represent percentages unless indicated otherwise

SD – standard deviation; IMM – immunomodulator

^†- Among those receiving biologic therapy

^*p ≤ 0.05

There was a trend towards more IBD-related surgical procedures in the study period (19% vs. 13%, multivariate OR 1.58, 95% CI 0.88 – 2.83) with a shorter time to such procedures (4.5 days vs. 7.6 days; p=0.04). Multivariable analysis controlling for IBD type, index admission/readmission, age, gender, and previous IBD-related surgery yielded a 4 day shorter interval to such procedures after implementation of the specialized IBD service (regression co-efficient –4.07, 95% CI −6.86 to −1.29). This difference was striking for non-resection procedures such as abscess drainage and was not significantly different when examining bowel resection surgeries alone (p=0.77).

Post-discharge Medical and Surgical Outcomes

Table 4 presents the 30 and 90 day outcomes. A total of 45% of patients hospitalized in the study period were deemed to be in remission at 90 days, compared to 34% of patients in the control period (multivariate OR 1.63, 95% CI 0.99 – 2.69, p=0.05). There was also a non-statistically significant trend towards remission at 30 days (p=0.13) and lower use of corticosteroids at 30 days (p=0.04). There was no difference in initiation of biologics by 30 or 90 days, ED visits, readmissions or surgical procedures at 30 or 90 days. However, among all those who underwent surgery, early surgery defined as in-hospital or within 30 days of discharge was more common in the study period (71%) compared to the control period (46%, multivariate OR 2.73, 95% CI 1.22 – 6.12).

Table 4.

Comparison of post-discharge medical and surgical outcomes before and after implementation of specialized inpatient IBD care among index and repeat hospitalizations

Outcome	Year 1 (General gastroenterology care) (n= 196) (%)	Year 2 (Specialized IBD care) (n= 212) (%)	p-value	Multivariate adjusted Odds ratio (95% CI)‖
At 30 days
Biologic use	44	50	0.25	1.20 (0.77 – 1.86)
Corticosteroid use	71	60	0.04	0.60 (0.37 – 0.98)
Remission	29	36	0.13	1.35 (0.82 – 2.25) ‡
Emergency department visit	6	5	0.78	0.81 (0.34 – 1.96)
Re-admission	14	22	0.05	1.50 (0.87 – 2.59)
IBD-related surgery	20	23	0.58	1.23 (0.76 – 2.00)
Early surgery (In hospital or within 30 days of discharge)†	46	71	0.007	2.73 (1.22 – 6.12)
At 90 days
Biologic use	53	58	0.34	1.20 (0.74– 1.94)
Corticosteroid use	44	41	0.66	0.87 (0.54– 1.39)
Remission	34	45	0.05	1.63 (0.99– 2.69) ‡
Emergency department visit	6	6	0.98	0.91 (0.38– 2.20)
Re-admission	26	29	0.47	1.02 (0.58– 1.78)
Surgery between 30 and  90 days	28	27	0.87	0.54 (0.26– 1.12)

^‖- adjusted for age, gender, index hospitalization or readmission, type of IBD, previous IBD-related surgery

^‡- Additionally adjusted for use of steroids, biologics, or surgery by 30 or 90 days

^†- Early surgery is defined as those surgical procedures occurring during the index hospitalization or within 30 days of discharge among those who eventually underwent surgery.

Sensitivity Analysis

We performed sensitivity analysis including only the index admissions (Supplementary Table 1). In this analysis, we observed a greater proportion of patients in clinical remission at 30 days (OR 2.44, 95% CI 1.34 – 4.44) and 90 days (OR 1.94, 95% CI 1.11 – 3.40), and similar to the overall analysis, a nearly four-fold increase in likelihood of receiving surgical intervention early (74% vs. 44%; OR 3.70; 95% CI 1.46 – 9.39) without an overall increase in rate of surgery by 90 days.

DISCUSSION

The management of IBD has become a rapidly changing and complex field, with many established and emerging therapies, and evolving treatment algorithms and therapeutic targets^{10, 22}. Consequently, increasingly specialized knowledge and expertise may be required for effective management²³. In a systematic evaluation of the impact of specialized inpatient IBD care, we demonstrate that several outcomes including rates of remission at 30 and 90 days and provision of surgical intervention early were positively impacted by this change in delivery of care.

Dedicated inpatient IBD care may facilitate the delivery of an optimal and personalized therapeutic approach. One of the key findings from our study was a higher likelihood of remission at 90 days among all IBD-related hospitalizations after implementation of a specialized inpatient IBD care delivery model, and a statistically significantly greater likelihood of remission at 30 and 90 days among index hospitalizations. This difference was not due to greater corticosteroid use or surgical treatments as these proportions were similar across both years. There are a few possible explanations for this effect. First, it is possible that implementation of specialized IBD care allowed for better optimizing of medical therapy on discharge and facilitated closer follow-up with the outpatient IBD physician. The lag in effect of this dose optimization may explain the greater magnitude of change at 90 days compared to 30 days. Secondly, though there was no difference in overall rates of surgery, receiving surgical intervention early may have contributed to earlier restoration of quality of life and improved rates of remission at 90 days.

Though literature in this field is sparse, a few previous studies have looked at the effect of specialist care in IBD. In a small study by Mawdsley et al., patients attending specialist IBD clinics were more likely to be adherent to quality measures including appropriate blood tests during immunosuppression therapy, bone protection with oral corticosteroids, and colorectal cancer screening²⁴. Reddy et al. also demonstrated that patients referred for a second opinion to a referral IBD center were often under dosed, not receiving topical therapy or had prolonged use of steroids without an attempt to start steroid sparing medications²⁵. Some of the best evidence of the impact of specialized care comes from Canada where Nguyen et al. examined the outcomes of 3,403 patients diagnosed with CD between 1988 and 2008¹⁵. Visit to a gastroenterologist within the first year of diagnosis was associated with a reduced need for surgery. Corresponding data among hospitalized patients similarly suggests that care at a high-volume hospital was associated with reduced mortality among patients with CD or UC^{12, 13, 16}. Gastroenterologist care for hospitalized UC patients did not reduce colectomy rates but decreased in-hospital and post-discharge mortality at one year¹⁴. Thus, our findings are broadly consistent with the body of literature in this area but also suggest that in a cohort of patients with severe disease, only a few outcomes may have a window of opportunity to be impacted by specialist care.

Use of high dose biologics (> 5mg/kg of IFX) was more common after implementation of specialized IBD care. Growing evidence suggests a benefit to high dose induction therapy with biologics in severe disease due to potentially greater levels of fecal drug loss²⁶ and accelerated drug clearance in severe inflammation. In our study, although use of high dose biologics (and a non-significant trend towards increased use of accelerated induction) did not translate into shorter hospitalizations or reduced need for surgery, the number of patients was small and we lacked sufficient power to demonstrate if these practices translated into objective benefit. Thus, while representing a practice difference that we observed during the two years, the use of high-dose biologics or accelerated induction does not have enough data supporting efficacy for inclusion as a marker of ‘high-quality’ inpatient care and cannot yet be recommended for routine management of acute severe IBD exacerbations for all patients. Benefits with this practice, may however, be seen in a subset of patients with severe inflammation and fecal loss of infliximab²⁷. Use of objective biomarkers such as CRP was more common in the intervention year, perhaps suggesting greater familiarity with such tools among specialist IBD gastroenterologists. Monitoring of biomarkers has demonstrated utility in assessing response to treatment and can help guide personalized titration of therapy^{28, 29}.

Another key finding from our study was the earlier surgical intervention after implementation of specialized care, with a four day shorter time from hospital admission to such procedures. Importantly, although a similar proportion of patients in both groups underwent surgery by 90 days, the proportion of such procedures occurring within 30 days was significantly greater with specialist inpatient IBD care, suggesting perhaps earlier recognition of the need for surgical intervention and increased multidisciplinary collaboration. This is important in the context of prior studies that have shown delayed intervention to be associated with poor outcomes^{13, 30}. Early occurrence of inevitable surgery may reduce the likelihood of post-operative complications^{31, 32}, reduce exposure to medications and in particular corticosteroids, shorten hospitalization duration³⁰, and reduce mortality¹³. There were no changes in the surgical staff availability or operating room time between the two years, and thus the reduction in surgical time cannot be attributed to such variables.

There are several implications to our findings. Management of IBD is increasingly complex. In addition, while newer therapies targeting distinct mechanisms of action are emerging, this lends more heterogeneity to care resulting in variation even in centers with high-volume of IBD^{33, 34}. There is an important need to identify best care models for delivery of treatment in this setting to optimize patient outcomes, quality of life, and healthcare costs. While not demonstrating uniform improvement across all outcomes examined, an analysis of our one-year experience of specialist inpatient IBD care demonstrated an increased rate of remission at 90 days, and earlier indicated surgical treatment. However, other outcomes were not significant different between the two years. A few possible reasons for this exist. One reason for this could be ad hoc involvement of IBD specialists in the patient care during the control period. A second could be the high volume of IBD hospitalizations and experience of many general gastroenterologists with complex IBD-related admissions even during the control year or prior. But thirdly, and importantly, it could be that hospitalized patients represent overall a very ill group of patients with advanced or severe disease with a limited window in which to demonstrate a significant difference in outcomes. Thus, it is possible that the impact of specialized inpatient IBD care may be more prominent in centers with lower availability previously of IBD experts. Finally, implementation of a specialized IBD inpatient service is a process that is in itself made up of a number of different components including early and frequent objective assessment of response to therapy, early escalation of therapy, and multidisciplinary consultation. Thus, future studies need to examine individual components of specialized IBD care that may result in the greatest benefit. There were no structural changes such as co-locating the IBD patients, inclusion of other care personnel in the team (such as case manager or pharmacists) during the two years but plausibly such factors may further allow for delivery of uniformly high-quality care. We also did not introduce any standardized practice algorithms as there are few such care pathways available to direct management of inpatient IBD. Whether implementation of such algorithms, particularly in hospitals where specialist care may not be available, merits further study and optimization of such pathways for the spectrum of inpatient IBD care is necessary. At our institution, there was no additional compensation for the IBD specialists as a result of this care model, and the time commitment varied from 1–4 hours per day depending on the number of patients. The financial model for adoption of this may be different at other institutions where modification of outpatient commitment during the weeks on inpatient service may be necessary depending on the volume. Additionally, in non-teaching centers without fellows, it is possible that their role may be assumed by mid-level providers.

We acknowledge several limitations to our study. While our study included over 400 hospitalizations, the number of patients in specific subgroups was low and may have limited statistical power. Second, heterogeneity between different IBD specialists may contribute to variation in outcomes. Third, the study has a limited follow-up period of 90 days. Future studies are necessary to examine the long-term benefits of specialized inpatient IBD care. Fourth, we did not measure subjective patient reported outcomes such as quality of life as well as patient satisfaction, parameters that require further study. Fifth, it is possible that some of the changes identified (for example, use of high dose or accelerated biologic induction) could represent increasing familiarity with the practice over time with the evolving literature rather than being due to specialist care alone. We acknowledge that heterogeneity exists even within specialist IBD providers. Despite this, we were able to identify significant differences in some outcomes. More uniform adoption of best care practices may result in further improvement in outcomes compared to general care. Finally, as with all retrospective studies, the impact of the infrequent missing data and unmeasured confounders cannot be excluded.

In conclusion, implementation of specialized inpatient IBD care was associated with more frequent use of high dose biologics, greater rates of remission at 90 days, and early receipt of surgical intervention. However, in contrast, overall length of stay, medical and surgical re-hospitalizations were not impacted. Future investigation is required to optimize delivery of care to hospitalized IBD patients to improve patient outcomes.