You are here

Article Text

Article menu
Original articles
Inflammatory bowel disease
Impact of gastroenterologist care on health outcomes of hospitalised ulcerative colitis patients
  1. Sanjay K Murthy1,
  2. A Hillary Steinhart1,2,
  3. Jill Tinmouth2,3,4,
  4. Peter C Austin2,4,
  5. Geoffrey C Nguyen1,2,4
  1. 1Mount Sinai Hospital IBD Centre, Department of Medicine, University of Toronto, Toronto, Canada
  2. 2Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada
  3. 3Division of Gastroenterology, Department of Medicine, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Canada
  4. 4Institute for Clinical Evaluative Sciences, Toronto, Canada
  1. Correspondence to Dr Sanjay Murthy, Mount Sinai Hospital, 445-600 University Avenue, Toronto, ON M5G 1X5, Canada; sanjay.murthy{at}utoronto.ca

Abstract

Objectives To evaluate the impact of in-hospital gastroenterologist care, relative to other provider care, on health outcomes of hospitalised Ulcerative colitis (UC) patients.

Design A population-based cohort study of 4278 UC patients hospitalised between 2002 and 2008 was conducted in Ontario, Canada. The primary outcome was in-hospital mortality risk.

Results UC patients admitted under non-gastroenterologists had a higher in-hospital mortality rate (1.1 vs 0.2%, p<0.0001) but a similar in-hospital colectomy rate (5.4 vs 4.9%, p=0.69) as compared to UC patients admitted under gastroenterologists. Following covariate adjustment, non-gastroenterologist care was associated with a greater in-hospital mortality risk relative to gastroenterologist care (adjusted OR (aOR) 3.28, 95% CI 1.03 to 10.5). This increased mortality risk was observed in patients admitted to other internists (OR 5.49, 95% CI 1.75 to 17.2) and general practitioners (OR 6.02, 95% CI 1.84 to 19.7), with a trend towards greater mortality risk among patients admitted to general surgeons (OR 3.49, 95% CI 0.90 to 13.6). Among patients who were discharged from hospital colectomy-free, those who were admitted under non-gastroenterologists had a greater one-year risk of death than patients who were admitted under gastroenterologists (adjusted HR 2.07, 95% CI 1.26 to 3.40). The type of hospital provider did not impact in-hospital or one-year colectomy risks or the risk of hospital re-admission in this cohort.

Conclusions Primary in-hospital gastroenterologist care was associated with decreased in-hospital and one-year mortality risks among hospitalised UC patients. Optimised care strategies by experienced specialists may confer important health advantages in this patient population.

  • Ulcerative colitis
  • inflammatory bowel disease
  • specialist care
  • gastroenterologist
  • health outcomes
  • mortality
  • crohn's disease
  • colorectal neoplasia
  • endoscopy
  • infliximab
  • IBD
  • clinical trials
  • IBD clinical
  • health economics
  • health outcomes
  • health disparities
  • health service research

https://doi.org/10.1136/gutjnl-2011-301978

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Significance of this study

    What is already known on this subject?

    • Specialist care, as compared to generalist care, is associated with improved health outcomes for a variety of complex medical conditions.

    • Gastroenterologist care within the first year following a diagnosis of Crohn's disease has been associated with decreased long-term surgery rates.

    What are the new findings?

    For hospitalised ulcerative colitis patients:

    • Primary in-hospital care provided by gastroenterologists is associated with lower risks of mortality during hospitalisation and over one-year following hospital discharge as compared to in-hospital care provided by non-gastroenterologists.

    • Consultant gastroenterologist care does not decrease mortality risk among patients admitted to non-gastroenterologists.

    • Colectomy and re-hospitalisations risks are not affected by the type of in-hospital physician provider.

    How might it impact on clinical practice in the foreseeable future?

    • Care of hospitalised ulcerative colitis patients may be best undertaken by gastroenterologists or specialists with expertise in the management of inflammatory bowel disease.

    • Practitioners who intend to manage complex inflammatory bowel disease should receive specific training in this area.

    Introduction

    Ulcerative colitis (UC) is a chronic inflammatory bowel disease (IBD) characterised by episodic bloody diarrhoea, abdominal pain and weight loss. Chronic disease activity and recurrent disease flares can considerably impair quality of life among affected individuals. This disorder may also be associated with a number of serious adverse health events, including toxic megacolon, bowel perforation, systemic infections and venous thromboembolism, any of which may ultimately result in death or the need for colectomy.1–4 Between 3 and 6% of UC patients are hospitalised for a UC-related indication in any given year.5 It has been estimated that the inpatient costs incurred from managing hospitalised IBD patients accounts for about half the cost of managing this disease in the UK.6

    The contribution of physician training and experience in IBD care to adverse outcome rates and health resource utilisation among hospitalised UC patients is unknown. As the medical management of these patients, who typically have severe colitis, can be particularly challenging, the care of such patients may be best undertaken by specialists with advanced knowledge in the care of complex IBD patients. However, these patients are often looked after by a variety of healthcare providers in the hospital setting, including gastroenterologists, general internists, general surgeons and family physicians, each of whom may have varying levels of proficiency in IBD patient care. Therefore, we sought to investigate whether the specialty of the primary in-hospital provider impacts in-hospital and long-term outcomes in these patients by specifically evaluating the impact of in-hospital care provided by gastroenterologists relative to other specialists.

    Methods

    Data sources

    Data for the present study was obtained from multiple population health administrative databases from Ontario, Canada, including: (1) The Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD) specific to Ontario, which contains demographic and medical information from all hospitalisations throughout Ontario; (2) The Ontario Health Insurance Plan (OHIP) physicians' claims database; (3) The Ontario Registered Persons Database (RPDB), which contains personal information and vital status about Ontario residents; (4) The Ontario Cancer Registry (OCR); and (5) Census databases. All data was accessed through the Institute for Clinical Evaluative Sciences (ICES), in which patient information can be linked via unique patient identifiers, permitting consolidation of demographic, clinical and health services information, as well as longitudinal assessment of patient health outcomes. Information about ICES and the surveyed databases may be found at http://www.ices.on.ca. The CIHI-DAD was the main data source used in this study.

    Patients and study design

    A population-based retrospective cohort study was conducted of all adult UC patients (≥18 years of age) admitted to Ontario hospitals between 1 March 2002 and 1 March 2008 who had a ‘Most Responsible Diagnosis’ of UC (K51.x in ICD-10) recorded for a given hospitalisation in the CIHI-DAD. The Most Responsible Diagnosis in CIHI-DAD corresponds to the acute illness that most profoundly affected a patient's treatment course or else accounted for the greatest length of hospital stay (LOHS).

    Patients who had a gastroenterologist recorded as the ‘Most Responsible Provider’ (MRP) during hospitalisation were compared to patients who had a non-gastroenterologist as their hospital MRP on a number of adverse health events occurring during hospitalisation and up to 1 year following hospital discharge. The MRP in CIHI-DAD designates the provider service under which a patient was admitted for the longest duration while in hospital. In the evaluation of in-hospital colectomy, the admitting physician was preferentially designated as the MRP in situations in which these providers differed, to ensure that the studied provider was most responsible for a patient's care prior to the occurrence of colectomy.

    MRP was further stratified into gastroenterologist, other internist, general practitioner or general surgeon for the purposes of comparing patient health outcomes. ‘Other internist’ and ‘general surgeon’ categories comprised all specialists who were certified in these respective specialties or one of their designated subspecialties by the Royal College of Physicians and Surgeons of Canada (RCPSC). ‘General practitioner’ category included all family medicine, community medicine and emergency medicine physicians.

    A subgroup analysis was also conducted to evaluate the impact of provider care among patients who did not undergo colectomy in hospital, as ‘semi-elective’ admissions of patients who were less acutely ill in order to expedite surgery had the potential to variably affect in-hospital mortality risks within the different provider categories. Additionally, the impact of gastroenterologist consultant care on health outcomes of UC patients admitted to non-gastroenterologists was assessed.

    Patients were excluded from the study for the following indications: (1) Inception hospitalisation for elective colectomy (based on the ‘Admit Category’ field in CIHI-DAD); (2) Previous partial or total colectomy (using a five-year look-back window in the CIHI-DAD); (3) Diagnosis of colorectal cancer from 5 years prior to index hospitalisation to the end of the study period; and (4) Admission to a provider service that was not under evaluation in this study (∼0.3% of eligible patients).

    Outcomes and subgroup analyses

    The primary study outcome was in-hospital mortality risk. Other in-hospital outcomes of interest included colectomy risk and acute care LOHS. Additionally, one-year risks of death, colectomy, non-elective UC-specific re-hospitalisation (re-admission for a UC-related indication), and non-elective all-cause re-hospitalisation were evaluated among patients who were discharged from hospital without undergoing colectomy. General surgery admissions were excluded from the analysis of in-hospital colectomy, as there was no way to distinguish UC patients admitted to surgeons for a true disease flare from less acutely ill UC patients who were admitted ‘semi-electively’ by surgeons for the purpose of expediting surgery to treat medically-refractory disease.

    Covariates

    The association between hospital MRP and in-hospital patient mortality risk was adjusted for the effects of patient age and co-morbid illnesses (based on the weighted Charlson co-morbidity score7). All other associations between hospital MRP and outcomes were adjusted for the effects of age, sex, co-morbidities, prior UC-related hospitalisation, duration of UC, socioeconomic status (based on mean neighbourhood household income quintile), patient residential setting (rural vs urban), admitting hospital type (teaching vs non-teaching), hospital annual UC admission volume quintile and admitting hospital size (based on tertile of number of acute care hospital beds). The association of hospital MRP with acute care LOHS was additionally adjusted for the effects of in-hospital colectomy and in-hospital death. Admission year was excluded from all models as it was not associated with either provider type or health outcomes in bivariate analysis and did not improve model fit.

    Statistical methods

    Bivariate comparisons of categorical variables were conducted using the χ2 test or Fisher's exact test. The Kaplan–Meier product limit method with log-rank test was used to compare time-to-event outcomes within the first year following colectomy-free discharge. The adjusted effect of provider care on binary (in-hospital) and time-to-event (longitudinal) outcomes were estimated using logistic and Cox proportional hazards regression models, respectively. To account for clustering of patients within hospitals, generalised estimating equation (GEE) methods and marginal Cox models with robust variance estimates were used to estimate each of these models, respectively. LOHS was modelled as counts and the adjusted association between MRP and acute care LOHS was estimated using negative binomial regression analysis.

    Multivariable modelling generally incorporated all pre-specified covariates. However, as there were a limited number of deaths in the study cohort, a stepwise modelling approach that evaluated which variables had the greatest impact on the association between provider type and death outcome was adopted to decide which variables to include in the final models, ensuring a minimum of ten death events per degree of freedom. As there were only 32 in-hospital deaths, the association between provider type (gastroenterologist vs non-gastroenterologist) and in-hospital mortality risk was adjusted only for age and co-morbidity burden (notably, other covariates infuenced the estimate for this association by <10%). Additionally, no covariates could be included in the model evaluating the association between multiple provider categories (gastroenterologist vs other internist vs general practitioner vs general surgeon) and in-hospital mortality risk. A formal test for multicollinearity did not show any of the covariates to be collinear, based on a threshold value of 10 for the variance inflation factor.

    In the analysis of one-year post-discharge mortality risk, patients were censored at the time of colectomy, such that only patients at risk of experiencing UC-related death were analysed going forward from that time. In the analyses of non-elective hospital re-admission, patients were censored if they underwent colectomy on the day of re-admission, as such patients may have been admitted to hospital on an elective basis to undergo colectomy.

    Statistical significance was based on two-sided type I error rate of 5%. All statistical analyses were performed using SAS V.9.1 software (SAS Institute Inc).

    Ethics review

    This study was approved by the ICES Privacy Office. ICES is a prescribed entity in the Personal Health Information Act of Ontario that is permitted to collect and analyse health information of Ontario residents under the approval of the Information and Privacy Commissioner of Ontario.

    Results

    Of 5265 patients hospitalised for UC in Ontario between 2002 and 2008, 4278 patients met study eligibility criteria. Baseline characteristics of study patients are presented in table 1. The MRP was a gastroenterologist for 1528 patients (35.7%), other internist provider for 1080 patients (25.3%), general practitioner for 983 patients (23.0%) and general surgeon for 687 patients (16.1%). Prior to discharge from hospital, 0.75% of patients had died and 8.4% had undergone colectomy. Among patients who were discharged from hospital without undergoing colectomy, an additional 3.1% of patients died, 13.3% underwent colectomy and 21.4% were re-admitted to hospital for a UC indication over the subsequent year.

    View this table:
    Table 1

    Baseline characteristics of study patients

    Impact of hospital MRP on patient risks of mortality and colectomy during hospitalisation

    UC patients admitted under a non-gastroenterologist had a higher rate of in-hospital death (1.1% vs 0.2%, p<0.0001) but a similar rate of in-hospital colectomy (5.2% vs 4.9%, p=0.69) as compared to UC patients admitted under a gastroenterologist. Following covariate adjustment, non-gastroenterologist care was associated with a threefold greater risk of in-hospital death, but no difference in the risk of in-hospital colectomy, relative to gastroenterologist care (table 2). A higher in-hospital mortality risk was also observed in association with non-gastroenterologist care among patients who did not undergo colectomy in hospital (aOR 3.37, 95% CI 1.06 to 10.7). Remarkably, no in-hospital deaths were observed among patients who underwent colectomy during index hospitalisation. Additional factors that were associated with in-hospital mortality risk include increasing patient age and co-morbidity burden, while factors that were associated with in-hospital colectomy risk include increasing patient age and prior UC-related hospitalisation (table 2). Notably, gastroenterologist care provided in a consulting capacity to patients admitted under non-gastroenterologists was not protective against in-hospital death in these patients (aOR 1.57, 95% CI 0.69 to 3.54). Pre-admission and post-admission co-morbidities potentially contributing to in-hospital death among patients who died are provided in supplemental table 1.

    View this table:
    Table 2

    Covariate associations with risks of death and colectomy among UC patients during index hospitalisation

    Bivariate analyses of the associations between provider type and patient mortality risk are presented in figure 1. Following covariate adjustment, a higher in-hospital mortality risk was observed for patients admitted to other internist providers (OR 5.49, 95% CI 1.75 to 17.2) and general practitioners (OR 6.02, 95% CI 1.84 to 19.7) relative to patients admitted to gastroenterologists. A trend towards higher mortality risk was also observed for patients admitted to general surgeons (OR 3.49, 95% CI 0.90 to 13.6). Furthermore, a subgroup analysis of patients who did not undergo colectomy in hospital clearly demonstrated a higher mortality risk for patients admitted under general surgeons (OR 5.21, 95% CI 1.38 to 19.8), other internists (OR 5.35, 1.73 to 16.5) and general practitioners (OR 5.87, 95% CI 1.83 to 18.8), relative to patients admitted under gastroenterologists. On the contrary, in-hospital mortality risk did not differ between patients admitted under the different non-gastroenterologist providers (p values for corrected pairwise comparisons were non-significant).

    Figure 1
    Figure 1

    In-hospital and one-year mortality rates among hospitalised UC patients based on hospital MRP. p Values are based on global differences between provider categories in bivariate analyses (*p value based on χ2 test; †p value based on log-rank test). See results section for individual ORs. GI, gastroenterologist; GP, general practitioner, GS, general surgeon; IM, internal medicine.

    Acute care LOHS did not significantly differ between patients admitted to gastroenterologists (medial LOHS 7 days, IQR 4–11 days) or non-gastroenterologists (median LOHS 7 days, IQR 4–12 days) (p=0.053 for covariate-adjusted association). Additionally, among patients who underwent colectomy during index hospitalisation, time to colectomy was not statistically different between patients admitted to gastroenterologists (median 3 days, IQR 1–7 days) or non-gastronterologists (median 2 days, IQR 1–6 days) (p=0.21).

    Impact of hospital MRP on 1 year health outcomes among patients discharged from hospital colectomy-free

    Among UC patients discharged from hospital without undergoing colectomy, those who had a non-gastroenterologist MRP during index hospitalisation had higher one-year rates of death (4.04 vs 1.41%, p<0.0001) and all-cause re-admission to hospital (42.3 vs 37.0%, p=0.0004), but similar one-year rates of colectomy (12.5 vs 14.6%, p=0.082) and UC-specific re-admission to hospital (21.1 vs 21.6%, p=0.81) as compared to UC patients whose MRP was a gastroenterologist.

    After covariate adjustment, there was a twofold greater risk of death over 1 year following discharge from hospital among patients who were cared for primarily by a non-gastroenterologist during hospitalisation (table 3). The higher post-discharge mortality risk was observed for patients admitted under other internists (aOR 1.92, 95% CI 1.06 to 3.47), general practitioners (aOR 2.18, 95% CI 1.29 to 3.69) and general surgeons (aOR, 2.34, 95% CI 1.22 to 2.54) alike. Again, this risk did not differ between patients admitted under different non-gastroenterologist providers (p values for corrected pairwise comparisons were non-significant). In-hospital MRP did not significantly influence the post-discharge risks of colectomy, UC-specific re-hospitalisation or all-cause re-hospitalisation (table 3).

    View this table:
    Table 3

    Association of most responsible hospital provider with risks of adverse outcomes over 1 year following colectomy-free discharge from hospital among UC patients

    Discussion

    In this population-based study of hospitalised UC patients, admission under a non-gastroenterologist was associated with a three to fourfold higher risk of in-hospital death and a twofold higher risk of death over 1 year following discharge from hospital relative to admission under a gastroenterologist. This increased mortality risk was observed even after controlling for patient age and co-morbidity burden, which are factors that are typically associated with mortality. Furthermore, in-hospital and post-discharge mortality risks were similarly higher among patients admitted to other internists, general practitioners, or general surgeons, relative to patients admitted to gastroenterologists, although the association between in-hospital mortality risk and general surgery admission just failed to reach statistical significance. Semi-elective admissions of less acutely ill UC patients by surgeons in order to expedite surgery may have lowered the mortality risk in this analysis. Thus, when patients who underwent in-hospital colectomy were excluded, a higher in-hospital mortality risk was clearly observed with admission under any of the non-gastroenterologist providers relative to admission under a gastroenterologist. Notably, gastroenterologist care provided in a consulting capacity did not reduce the risk of in-hospital death among UC patients admitted to non-gastroenterologists in this study. It is important to keep in mind that the incidence of in-hospital death in this cohort was still quite low (<1%) and that majority of patients who died had one or more co-morbid illnesses that may have contributed to their death (supplemental table 1).

    On the other hand, there were no differences in in-hospital or post-discharge colectomy risks among patients admitted to different in-hospital providers, which suggests that colectomy risk may be more influenced by patient and disease-related factors. In the present study, for example, in-hospital colectomy risk was associated with increasing patient age and prior hospitalisation for UC, while one-year post-discharge colectomy risk was associated with these variables as well as male gender, UC disease duration and rural habitation (data not presented). Other studies have demonstrated that UC severity and immunosuppressive medication use may also influence long-term colectomy risk in this population.8 The risks of re-hospitalisation for either a UC-related indication or any indication were also not influenced by hospital provider type in this study.

    Overall, the results of this study suggest that continuous close monitoring of hospitalised UC patients by gastroenterologists, as compared to other specialists from a variety of healthcare disciplines, may confer sustained health benefits to these patients. However, gastroenterologist care in a consulting capacity may not confer the same benefit to these patients. The advantages of gastroenterologist care in this setting could relate to greater training and experience in the management of complex IBD patients, prompt recognition and management of UC-related complications, appropriate use of endoscopy to help guide management and closer post-discharge patient follow-up.

    Prolonged or inappropriate use of corticosteroids and/or opioids, both of which have been shown to increase mortality risk among IBD patients, may have been partly responsible for the higher death rate among patients admitted to non-gastroenterologists in this study.9 It is noteworthy that no in-hospital deaths were observed among the 359 patients who underwent colectomy during index hospitalisation. While this finding may partly relate to admissions of less acutely ill UC patients in order to expedite surgery and/or to selection of younger patients with fewer co-morbidities for surgery, it may also suggest a protective effect of timely colectomy against serious complications stemming from prolonged disease activity and/or immunosuppressive therapy in medically-refractory patients. Accurate assessment of early response to medical therapy has been shown to be an important predictor of health outcomes among severe UC patients, as up to 40% of these patients do not respond to corticosteroids and may incur serious adverse events if treatment with these agents is unnecessarily protracted.10–15 The American College of Gastroenterology currently advocates for a change of therapy, including consideration of colectomy, in severe UC patients who have failed 3 to 5 days of corticosteroid treatment.15

    To the best of our knowledge, this is the first study to evaluate the influence of hospital-based provider care on the health outcomes of acutely ill UC patients. Nguyen and colleagues reported that gastroenterologist care within the first year following a diagnosis of Crohn's disease is associated with decreased long-term surgery rates in this population.16 Other groups have reported varying effects of specialist care on patient health outcomes for complex medical conditions across a variety of healthcare disciplines.17–26 Future studies are greatly needed to further address the impact of specialist care on health outcomes in IBD patients.

    The higher observed mortality risk among patients admitted under non-gastroenterologists, in the absence of a similar increase in colectomy risk in these patients, raises the possibility that unadjusted co-morbidity bias influenced the study results. Co-morbidity burden was adjusted using the Charlson index, which is a validated and widely used tool for predicting the impact of co-morbidity burden on health outcomes among hospitalised patients with a variety of illnesses.7 ,27 Adaptations of this index have also performed well when used with Canadian hospital discharge data.28 ,29 Still, this index may not capture all relevant co-morbidities among UC patients. Notably, we did not observe a higher risk of all-cause re-hospitalisation among patients admitted to non-gastroenterologists, which might have been expected if residual co-morbidity bias was present.

    Additionally, variation in access to hospital resources between different providers may have impacted the observed outcomes in this study. Gastroenterologists are more likely to provide admitting services in high-volume referral centres, where access to consultative services, new and expensive medications and timely surgery would likely be better than in smaller community hospitals. We attempted to control for discrepancies in access to care through variables reflecting hospital teaching status, hospital experience in UC patient care and hospital size. However, it is uncertain whether inclusion of these variables would have fully adjusted for such discrepancies. Of note, the time to colectomy among patients who underwent surgery in hospital did not differ between patients admitted to gastroenterologists or non-gastroenterologists, possibly suggesting that access to timely surgery was similar between the comparator groups.

    Unadjusted differences in baseline UC disease severity and immunosuppressive therapy between patients admitted to different providers could have further biased the observed associations in this study. Notably, by virtue of being hospitalised for UC, most patients in this study likely had severe to fulminant UC. Still, even within this stratum of disease severity, there could have been variation in disease activity that influenced patient outcomes. It is difficult to predict how such discrepancies may have impacted the study results. Gastroenterologists practicing at high-volume referral centres may have been more likely to admit patients with complex UC who were failing multiple medical therapies. On the other hand, general internists and general practitioners may have more often admitted patients with recent-onset UC who did not yet have a regular gastroenterologist; disease behaviour in such patients has been reported to be more aggressive as compared to patients later on in their disease course.8 ,30 ,31 The absence of clinical and endoscopic variables used in validated indices of UC severity precluded optimal adjustment for this potential confounder.32 Still, we adjusted for other potential markers of UC severity, including prior UC-related hospitalisation and UC duration (based on the consideration that UC disease behaviour may be most aggressive in the first 2 years following diagnosis8 ,30 ,31). Notably, it would have been inadvisable to control for differences in disease severity or medication use that arose following hospital admission, as such differences may have been directly related to varying provider care strategies.

    Although the most responsible diagnosis code for UC has not been formally validated in the Ontario version of the CIHI-DAD, it has been demonstrated to have a positive predictive value (PPV) of 79.3% for identifying hospital admission for a UC flare in an analogous version of the CIHI-DAD from the Calgary Health Zone in the province of Alberta, Canada.33 A high level of coding accuracy has also been demonstrated for a number of other most responsible diagnoses, as well as for demographic data and codes for procedures, in the Ontario version of the CIHI-DAD.34 Additionally, while there may have been some misclassification of provider type in the CIHI-DAD, this misclassification would likely have been non-differential among the comparator groups and should not invalidate significant associations that were observed in this study.35

    The wide CIs surrounding the estimates of the association between MRP and in-hospital patient mortality risk in this study preclude precise determination of the true magnitude of these associations. Even then, the SEs around these estimates may be conservative as we were unable to account for the effects of patient clustering within physicians in our analyses. As such, the results of this study require validation from population-based studies in other UC cohorts as well as from randomised controlled trials.

    Despite its limitations, this study is the first to evaluate the impact of provider care on health outcomes of hospitalised UC patients. As the study includes patients admitted to academic and community hospitals in both urban and rural settings, the results of the study should be generalisable to most hospitalised UC patients across Ontario. However, these results require confirmation in other healthcare settings before they can be generalised to all hospitalised UC patients. Furthermore, specific differences in provider care strategies that impact outcomes in this population require elucidation to provide potential targets for future educational and healthcare policy initiatives. Nevertheless, until that time, it may be prudent for hospitalised UC patients to be managed by physicians with expertise in IBD.

    References

    1. ↵
      1. Turner D,
      2. Walsh CM,
      3. Steinhart AH,
      4. et al
      . Response to corticosteroids in severe ulcerative colitis: a systematic review of the literature and a meta-regression. Clin Gastroenterol Hepatol 2007;5:103–10.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Ekbom A,
      2. Helmick CG,
      3. Zack M,
      4. et al
      . Survival and causes of death in patients with inflammatory bowel disease: a population-based study. Gastroenterology 1992;103:954–60.
      OpenUrlPubMedWeb of Science
    3. ↵
      1. Winther KV,
      2. Jess T,
      3. Langholz E,
      4. et al
      . Survival and cause-specific mortality in ulcerative colitis: follow-up of a population-based cohort in Copenhagen County. Gastroenterology 2003;125:1576–82.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Murthy SK,
      2. Nguyen GC
      . Venous thromboembolism in inflammatory bowel disease: an epidemiological review. Am J Gastroenterol 2011;106:713–18.
      OpenUrlCrossRefPubMed
    5. ↵
      1. Longobardi T,
      2. Walker JR,
      3. Graff LA,
      4. et al
      . Health service utilization in IBD: comparison of self-report and administrative data. BMC Health Serv Res 2011;11:137.
      OpenUrlCrossRefPubMed
    6. ↵
      1. Luces C,
      2. Bodger K
      . Economic burden of inflammatory bowel disease: a UK perspective. Expert Rev Pharmacoecon Outcomes Res 2006;6:471–82.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Charlson ME,
      2. Pompei P,
      3. Ales KL,
      4. et al
      . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Sandborn WJ,
      2. Rutgeerts P,
      3. Feagan BG,
      4. et al
      . Colectomy rate comparison after treatment of ulcerative colitis with placebo or infliximab. Gastroenterology 2009;137:1250–60.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Lichtenstein GR,
      2. Feagan BG,
      3. Cohen RD,
      4. et al
      . Serious infections and mortality in association with therapies for Crohn's disease: TREAT registry. Clin Gastroenterol Hepatol 2006;4:621–30.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Carbonnel F,
      2. Gargouri D,
      3. Lemann M,
      4. et al
      . Predictive factors of outcome of intensive intravenous treatment for attacks of ulcerative colitis. Aliment Pharmacol Ther 2000;14:273–9.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Jarnerot G,
      2. Rolny P,
      3. Sandberg-Gertzen H
      . Intensive intravenous treatment of ulcerative colitis. Gastroenterology 1985;89:1005–13.
      OpenUrlPubMed
    12. ↵
      1. Lindgren SC,
      2. Flood LM,
      3. Kilander AF,
      4. et al
      . Early predictors of glucocorticosteroid treatment failure in severe and moderately severe attacks of ulcerative colitis. Eur J Gastroenterol Hepatol 1998;10:831–5.
      OpenUrlPubMedWeb of Science
    13. ↵
      1. Randall J,
      2. Singh B,
      3. Warren BF,
      4. et al
      . Delayed surgery for acute severe colitis is associated with increased risk of postoperative complications. Br J Surg 2010;97:404–9.
      OpenUrlCrossRefPubMed
    14. ↵
      1. Travis SP,
      2. Farrant JM,
      3. Ricketts C,
      4. et al
      . Predicting outcome in severe ulcerative colitis. Gut 1996;38:905–10.
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Kornbluth A,
      2. Sachar DB
      . Ulcerative colitis practice guidelines in adults: American College of Gastroenterology, Practice Parameters Committee. Am J Gastroenterol 2010;105:501–23.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. Nguyen GC,
      2. Nugent Z,
      3. Shaw S,
      4. et al
      . Outcomes of patients with Crohn's disease improved from 1988 to 2008 and were associated with increased specialist care. Gastroenterology 2011;141:90–7.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Ayanian JZ,
      2. Guadagnoli E,
      3. McNeil BJ,
      4. et al
      . Treatment and outcomes of acute myocardial infarction among patients of cardiologists and generalist physicians. Arch Intern Med 1997;157:2570–6.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Greenfield S,
      2. Rogers W,
      3. Mangotich M,
      4. et al
      . Outcomes of patients with hypertension and non-insulin dependent diabetes mellitus treated by different systems and specialties. Results from the medical outcomes study. JAMA 1995;274:1436–44.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Jollis JG,
      2. DeLong ER,
      3. Peterson ED,
      4. et al
      . Outcome of acute myocardial infarction according to the specialty of the admitting physician. N Engl J Med 1996;335:1880–7.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Liu CC,
      2. Chen KR,
      3. Chen HF,
      4. et al
      . Association of doctor specialty with diabetic patient risk of hospitalization due to diabetic ketoacidosis: a national population-based study in Taiwan. J Eval Clin Pract 2011;17:150–5.
      OpenUrlCrossRefPubMed
    21. ↵
      1. McAlister FA,
      2. Majumdar SR,
      3. Eurich DT,
      4. et al
      . The effect of specialist care within the first year on subsequent outcomes in 24,232 adults with new-onset diabetes mellitus: population-based cohort study. Qual Saf Health Care 2007;16:6–11.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Mitchell JB,
      2. Ballard DJ,
      3. Whisnant JP,
      4. et al
      . What role do neurologists play in determining the costs and outcomes of stroke patients? Stroke 1996;27:1937–43.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Nash IS,
      2. Nash DB,
      3. Fuster V
      . Do cardiologists do it better? J Am Coll Cardiol 1997;29:475–8.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Strauss MJ,
      2. Conrad D,
      3. LoGerfo JP,
      4. et al
      . Cost and outcome of care for patients with chronic obstructive lung disease. Analysis by physician specialty. Med Care 1986;24:915–24.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Vollmer WM,
      2. O'Hollaren M,
      3. Ettinger KM,
      4. et al
      . Specialty differences in the management of asthma. A cross-sectional assessment of allergists' patients and generalists' patients in a large HMO. Arch Intern Med 1997;157:1201–8.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Zeiger RS,
      2. Heller S,
      3. Mellon MH,
      4. et al
      . Facilitated referral to asthma specialist reduces relapses in asthma emergency room visits. J Allergy Clin Immunol 1991;87:1160–8.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. de Groot V,
      2. Beckerman H,
      3. Lankhorst GJ,
      4. et al
      . How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003;56:221–9.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Humphries KH,
      2. Rankin JM,
      3. Carere RG,
      4. et al
      . Co-morbidity data in outcomes research: are clinical data derived from administrative databases a reliable alternative to chart review? J Clin Epidemiol 2000;53:343–9.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Quan H,
      2. Parsons GA,
      3. Ghali WA
      . Validity of information on comorbidity derived rom ICD-9-CCM administrative data. Med Care 2002;40:675–85.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Langholz E,
      2. Munkholm P,
      3. Davidsen M,
      4. et al
      . Colorectal cancer risk and mortality in patients with ulcerative colitis. Gastroenterology 1992;103:1444–51.
      OpenUrlPubMed
    31. ↵
      1. Solberg IC,
      2. Lygren I,
      3. Jahnsen J,
      4. et al
      ; IBSEN Study Group. Clinical course during the first 10 years of ulcerative colitis: results from a population-based inception cohort (IBSEN Study). Scand J Gastroenterol 2009;44:431–40.
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. D'Haens G,
      2. Sandborn WJ,
      3. Feagan BG,
      4. et al
      . A review of activity indices and efficacy end points for clinical trials of medical therapy in adults with ulcerative colitis. Gastroenterology 2007;132:763–86.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Ma C,
      2. Crespin M,
      3. Proulx MC,
      4. et al
      . Accuracy of administrative data in identifying ulcerative colitis patients presenting with acute flare and undergoing colectomy. Am J Gastroenterol 2010;105:1–69. S467, A1270.
      OpenUrlCrossRef
    34. ↵
      1. Juurlink D,
      2. Preyra C,
      3. Croxford R,
      4. et al
      . Canadian Institute for Health Information Discharge Abstract Database: A Validation Study. Toronto: Institute for Clinical Evaluative Sciences, 2006.
    35. ↵
      1. Grimes DA,
      2. Schulz KF
      . Bias and causal associations in observational research. Lancet 2002;359:248–52.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Link to description of Data Integration, Measurement and Reporting Hospital Discharge. Abstract Database from Alberta Health Region (Alberta version of CIHI-DAD): http://www.health.alberta.ca/documents/Research-Health-Databases.pdf

    • Competing interests None.

    • Patient consent All data was provided de-itentified to the authors of the study. Additionally, all data is presented only inaggregate form in this manuscript. No patient identifying information is provided.

    • Ethics approval The ethics approval was provided by Institute for Clinical Evaluative Sciences Privacy Office.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    Linked Articles