As healthcare costs related to inflammatory bowel diseases (IBDs) rise, there is an increasing need to identify hospital characteristics that promote cost-effective care while minimizing adverse outcomes for this complex population. Although hospital teaching status is one characteristic that has been associated with higher overall inpatient costs [1], recent data suggest a mortality benefit for all-cause hospitalizations at these centers [1, 2]. Differences in inpatient IBD outcomes in academic versus nonacademic centers have also been assessed and notable for greater disease complexity and severity among patients receiving care at academic medical centers but similar adjusted length of stay, cost, readmission rates, and in-hospital mortality rates [3]. A summary of studies that have evaluated the association between hospital teaching status and hospitalization outcomes for IBD or other gastrointestinal conditions is presented in Table 1.

Table 1 Studies that have evaluated the association between hospital teaching status and hospitalization outcomes for inflammatory bowel diseases or other gastrointestinal conditions
Full size table

In this issue of Digestive Diseases and Sciences, Sharma et al. [4] assessed differences in mortality, length of stay, cost, and other outcomes for IBD hospitalizations by hospital teaching status using a large administrative database. Though they found no difference in overall mortality for IBD, they noted higher mortality rates for ulcerative colitis in teaching hospitals. Teaching hospitals also had greater adjusted costs, charges, and length of stay compared with their nonteaching counterparts.

To assess hospitalization outcomes, Sharma and colleagues [4] analyzed the National Inpatient Sample (NIS), a large nationwide database consisting of patient-level discharge data. The sample of nearly 200,000 patients with IBD facilitates the assessment of several outcomes of interest to IBD care. Use of only the 2016 and 2017 samples provides the most recent data available, avoiding variable definition errors related to ICD coding changes that occurred prior to 2016. The authors also adjusted for appropriate available confounders in the NIS for their multivariable models. Nevertheless, certain aspects of the study design limit the conclusions that can be drawn. Due to their inclusion of only patients with IBD as a primary diagnosis, the authors have underestimated the total inpatient IBD population. Providers may code an IBD complication such as small bowel obstruction as the primary diagnosis, with ulcerative colitis or Crohn’s disease as the secondary diagnosis. The inclusion criteria therefore introduce selection bias by excluding many individuals admitted (and coded) primarily for an IBD complication.

There are a number of limitations to the NIS that must also be considered when assessing IBD-specific outcomes. The NIS includes common comorbidities, but does not provide data related to IBD severity or complexity, such as prior abdominal surgeries, exposure to biologic agents, inpatient administration of biologic agents, or IBD duration. While some information regarding disease phenotype may be present in ICD-10 diagnosis codes, these are subject to potential coding errors. Since understanding IBD severity is essential to any analysis of this complex population, these factors should be accounted for in multivariable models that assess IBD-specific outcomes.

The NIS definition of teaching hospitals is also problematic and may not be the most relevant to IBD care. To be categorized as a teaching hospital in the NIS, only one Accreditation Council for Graduate Medical Education (ACGME)-approved residency program is required [5]. Teaching hospitals with a small number of residency programs may be less likely to have the specialist IBD care available needed to favorably impact outcomes. Assessment of hospitals by the availability of specialist IBD physicians, surgeons, and highly experienced radiologists and pathologists would be more helpful in order to inform IBD patients where they should obtain their inpatient care.

Though prior research has identified higher unadjusted costs for IBD hospitalizations and greater unadjusted mortality for UC hospitalizations [3], adjusting for metrics of disease severity eliminated these perceived differences, suggesting that differential outcomes are driven by the differential disease complexity of the IBD populations in academic versus nonacademic centers. In contrast, Sharma and colleagues [4] noted a higher adjusted odds of mortality for UC in teaching hospitals. While they controlled for demographics and comorbidities, the authors could not adjust for markers of UC severity that could confer a higher baseline risk of mortality. They acknowledge that higher inpatient mortality among patients with UC could be a consequence of the higher rates of inpatient colectomy at teaching hospitals and the likelihood of interhospital transfer of complex patients from nonacademic to academic centers. Although the NIS does not allow for identification of interhospital transfers, early inpatient colectomy and parenteral nutrition, which could serve as surrogates of baseline IBD severity, could have been included as covariates in the mortality analysis for patients with UC in order to help explain this discrepancy.

Similarly, the higher costs and length of stay at teaching hospitals identified in this study are likely due to differences in baseline IBD complexity. While a subgroup analysis of complicated IBD still demonstrated significantly higher adjusted costs at teaching hospitals, the ICD-10 coding on admission that defines this subgroup does not take into account patients’ IBD history. Many patients with complicated IBD were also likely excluded due to the inclusion criteria as previously stated. Furthermore, the possibility of interhospital transfer due to patient complexity likely affects these outcomes.

IBD complexity and severity must be considered when comparing outcomes between patient populations where the magnitude of such characteristics may differ. Future studies which are able to account for patient-level metrics of IBD severity are needed to better understand the reasons for potential differences in hospitalization outcomes. Cost effectiveness could be compared among teaching hospitals with specialized IBD physicians and surgeons, teaching hospitals without specialized IBD providers, and nonteaching hospitals. A more granular understanding of hospital characteristics that influence IBD-specific outcomes will help determine the optimal setting for patients with IBD who require inpatient care.