OUP user menu

Editor-in-Chief

Laurence J. Egan, Ireland

Associate Editors

Maria T. Abreu, USAShomron Ben-Horin, IsraelSilvio Danese, ItalyPeter Lakatos, HungaryMiles Parkes, UKGijs van den Brink, NLSéverine Vermeire, Belgium

3.562
4.023

Published on behalf of

Clostridium difficile infection is associated with worse long term outcome in patients with ulcerative colitis

Udayakumar Navaneethan, Saurabh Mukewar, Preethi GK Venkatesh, Rocio Lopez, Bo Shen
DOI: http://dx.doi.org/10.1016/j.crohns.2011.09.005 330-336 First published online: 1 April 2012

Abstract

Background: Clostridium difficile infection (CDI) is becoming prevalent in general population as well as in patients with inflammatory bowel disease (IBD).

Aim: The aim of the study was to investigate the long-term impact of CDI in patients with ulcerative colitis (UC).

Methods: UC inpatients or outpatients who had positive results for stool C. difficile toxins A and B between 2002 and 2007 were identified (N = 45). The 45 patients were matched for age and gender to UC patients who were negative for C. difficile and had never been diagnosed with CDI (N = 101). The primary Colectomy within 12 months of C. difficile testing was the primary outcome patients with CDI and no-CDI.

Results: Forty-five patients were CDI positive and 101 were negative. Patients who were CDI positive had significantly more UC-related emergency room visits in the year following initial infection (37.8% vs. 4%, p < 0.001) than those without CDI. One year following the index infection admission, CDI patients also had a significantly higher rate of colectomy than controls (35.6% vs. 9.9%, p < 0.001). Among patients with CDI, 55.8% of patients had an escalation in medical treatment in the year after CDI as compared to the prior year of 12.9%, p < 0.0001. CDI (odds ratio (OR) 10.0, 95% confidence interval CI: 2.7, 36.3, p < 0.001) and severe disease on endoscopy (OR 16.7, 95% (CI): 4.1, 67.9, p < 0.001) were found to be independently associated with colectomy within 1 year on logistic regression analyses.

Conclusions: CDI appears to be associated with escalation of medical therapy in the year following infection. CDI and severe disease on endoscopy appear to be associated with an increased risk for subsequent colectomy on long-term follow up.

Keywords
  • Clostridium difficile
  • Colectomy
  • Ulcerative colitis
  • Inflammatory bowel disease

1 Introduction

Clostridium difficile infection (CDI) is the leading identifiable cause of antibiotic-associated diarrhea1, causing substantial morbidity and mortality.2 Although a growing body of knowledge on the epidemiology, pathogenesis, risk factors, and management of CDI has been obtained over the last decade, the increased incidence and severity of CDI continue to pose challenge to medical community.3,4 The expected health care costs due to CDI alone are estimated up to 3.2 billion dollars per year in the US.5 Clearly the impact of CDI on the health care system continues to grow with emergence of community-acquired CDI.6,7

Inflammatory bowel diseases (IBD), comprising Crohn's disease (CD) and ulcerative colitis (UC) are chronic relapsing inflammatory conditions that frequently require long-term medical therapy, periodic hospitalizations, and even surgery. Chronic use of antibiotics, corticosteroids, and immunomodulators has been shown to increase the risk of CDI in IBD patients.8,9 However, the use of biological agents did not appear to increase the risk of acquisition of CDI.10 Recently, two single-center studies and two national inpatient database studies have reported rising rates of CDI among IBD patients and their contributions of increased rates of hospitalizations and mortality.8,9,11,12 The risk of CDI in IBD appears to persist even after colectomy. For example, small bowel involvement with enteritis can occur due to CDI.13 CDI has also been reported in UC patients with restorative proctocolectomy and ileal pouch anal anastomosis (IPAA). 14,15

Approximately 5%–19% of patients admitted for relapsing IBD tested positive for C. difficile toxins in stools.16,17 Studies have been performed to identify risk factors for CDI in patients with IBD.8,9,11,12 Patients with UC appear to be at a higher risk for CDI than CD and patients with UC and CDI had a higher mortality and a higher risk for colectomy.8,9,11,12 Colectomy in UC patients even without CDI has been shown to be associated with a higher in-hospital mortality.18

There are limited published studies on the long-term outcome of CDI in patients with IBD. In a retrospective case control study, UC-CDI patients had worse clinical outcome than UC patients without CDI, with a follow-up of up to a year after CDI.19 On the other hand, the study did not discriminate between recurrent CDI versus worse IBD disease activity because of the retrospective nature and study design. In another case–control study comparing the disease course for 1 year before and 1 year after the initial CDI in 87 patients with IBD, colectomy occurred in only 10.3% of patients (9/87) following CDI.20 Because of limited data on long-term outcomes of UC-CDI patients, we wanted to analyze the long-term impact of CDI in UC patients in this study.

2 Methods

2.1 Patients

This was a historical cohort study of patients who were followed up in the IBD center of the Cleveland Clinic's Digestive disease Institute between the years 2002 and 2007. All patients with concurrent diagnosis of UC and CDI were identified from our electronic medical records. This study was approved by the Institutional Review Board. Some of the patients in this study (N = 28) were included from our prior study on the short-term outcome in CDI and UC patients.21

2.2 Inclusion and exclusion criteria

Both inpatients and outpatients with UC who had positive results for C. difficile toxins A and/or B were identified. Patients with Crohn's disease or microscopic colitis with co-existing CDI were excluded. CDI patients without a prior history of IBD or co-existing IBD were also excluded.

2.3 Laboratory test for C. difficile infection

CDI was defined by a positive enzyme-linked immunosorbent assay (ELISA) test for stool C. difficile toxins A and/or B and were considered infected if they presented with concomitant symptoms of colitis (i.e., diarrhea, increased stool frequency, rectal bleeding, urgency, cramping, and/or tenesmus). For each case with CDI, a control UC case with a negative stool C. difficile test and no prior history of CDI was matched by age and gender.

2.4 Clinical variables

Patient records were reviewed for demographic and clinical information, including age, gender, extent of the disease, severity, endoscopic findings, clinical symptoms, multiple laboratory parameters, IBD medications (5-aminosalicylic acids [5-ASA], immunomodulators [azathioprine, 6-mercaptopurine, and methotrexate]), anti-tumor necrosis factor therapy, antibiotics, corticosteroids, non-steroidal anti-inflammatory drugs (NSAIDs). Additional clinical parameters analyzed included anatomic extent of colitis (extensive colitis with involvement proximal to the splenic flexure vs. distal colitis on endoscopy and/or histology). The severity of disease was defined based on endoscopy. Disease activity in UC patients was assessed using the Schroeder (Mayo Score).22 Disease activity was scored on a scale of 0–3 (Schroeder activity index) with the index of 3 being severe disease. The determination of disease severity was done on the basis of the last colonoscopy on record at our institution.

Also information on the treatment/medications in hospital, number and types of colonic/GI surgeries performed during admission or in the year following hospitalization, number of hospitalizations at Cleveland Clinic for UC exacerbation or complication in the year following indexed hospitalization, number of emergency room (ER) visits for UC not resulting in admission to hospital at Cleveland Clinic in the year following index hospitalization.

All UC patients who tested positive for C. difficile in the setting of colitis symptoms were treated with antibiotics including oral vancomycin, or oral or intravenous metronidazole. As patients with UC who developed CDI were less likely to respond to oral metronidazole, oral vancomycin was increasingly used in our practice for the treatment of CDI, particularly those diagnosed after 2005.

2.5 Outcome measurements

The primary aim of this study was to compare colectomy within 1 year of initial CDI in UC patients and UC patients without CDI. We also studied the number of UC-related hospitalizations and ER visits in the year following index admission. Our secondary aim was to evaluate whether infection with CDI resulted in a more severe course of UC. This was defined by escalation of medical treatment with the addition of immunomodulator therapy or biological therapy to antiinflammatory treatment, or the addition of biological therapy to immunomodulatory treatment.

2.6 Treatment of CDI and UC

In treatment of UC-CDI flare, our approach was to use antibiotics with or without steroids. Immunomodulators or biologics if used were held during the episode and once patient improved, these medications were re-instituted in the outpatient setting. This was not termed as escalation if these patients were already on these medications. However if the medications were started after the CDI episode, then we termed it as escalation.

2.7 Statistical methods

Descriptive statistics were computed for all variables. These include means, standard deviations and percentiles for continuous variables and frequencies and percentages for categorical factors. A univariable analysis was done to assess differences between UC patients with and without CDI. Student's t-test was used to compare age; Pearson's chi-square tests were used for categorical factors and Wilcoxon rank sum tests were used for number of ER visits and hospital admissions. In addition, a multivariable logistic regression analysis was performed to evaluate factors associated with colectomy within 1 year of index admission. An automated stepwise variable selection method on 1000 bootstrap samples was used to choose the final model. Presence of CDI was force-included in the model and all other variables were considered for inclusion; the 3 additional variables with the highest inclusion rates were kept in the final model.

In addition, a paired analysis was done to compare number of ER visits, hospital admissions and increase in medications during 1 year before and after infection in subjects with CDI. McNamara's tests were used to compare % of subjects with increase in medications and Wilcoxon signed rank used for number of visits and admissions. A p < 0.05 was considered statistically significant. All analyses were performed using SAS (version 9.2, The SAS Institute, Cary, NC).

3 Results

A total of 153 were studied, including 52 in the study group with UC and CDI and 101 in the control group with UC without current or history of CDI. 7 (13.5%) of patients with CDI had recurrent CDI within the next year. These patients with recurrent CDI were excluded and a total of 45 patients were included in the study group.

3.1 Risk factors for CDI in patients with ulcerative colitis

Table 1 summarizes the demographic and clinical characteristics of UC patients with and without CDI. The clinical and laboratory parameters listed in the table were analyzed.

View this table:
Table 1

Demographic and clinical characteristics.

FactorNo CDI (N = 101)CDI (N = 45)p-Value
Male67(66.3)28(62.2)0.63
Age at the infection*44.5 ± 15.748.9 ± 18.00.14
Duration of UC (years)2.0[1.00,5.0]3.0[1.00,8.0]0.15
Rectum4(4.0)2(4.4)0.89
Left sided11(10.9)7(15.6)0.43
Extensive86(85.1)31(68.9)0.023
Severe disease32(31.7)23(51.1)0.025
Baseline ASA use76(75.2)24(53.3)0.008
Baseline immunomodulators4(4.0)4(8.9)0.23
Baseline biologics0(0.0)1(2.2)0.13
Colectomy at 3 months9(8.9)3(6.7)0.65
Colectomy at 1 year10(9.9)16(35.6)< 0.001
Number of ER visits within 3 months after0.006
099(98.0)39(86.7)
12(2.0)6(13.3)
Number of ER visits within 1 year< 0.001
097(96.0)28(62.2)
10(0.0)12(26.7)
2+4(4.0)5(11.1)
Number of hospitalizations before 1 year0.003
089(88.1)30(66.7)
18(7.9)11(24.4)
2+4(4.0)4(8.9)
Number of hospitalizations within 3 months0.45
085(84.2)40(88.9)
116(15.8)5(11.1)
Number of hospitalizations within 1 year0.25
083(82.2)34(75.6)
118(17.8)7(15.6)
2+0(0.0)4(8.9)
Increase in meds within 3 months13(12.9)0(0.0)0.012
Increase in meds within 1 year13(12.9)25(55.6)< 0.001
Prednisone: 1 year before31(30.7)16(35.6)0.56
Prednisone: 1 year after*20(44.4)

Values presented as Mean ± SD with t-test, Median [P25, P75] with Wilcoxon rank sum tests or N (%) with Wilcoxon rank sum test for number of ER and hospital admissions and Pearson's chi-square test otherwise. * In patients with no CDI, their age at infection denote the time of data entry.

    Patients with UC and CDI infection were less likely to have pancolitis, were less likely on ASA agents, more likely to have severe disease on endoscopy, had higher number of ER visits within 3 months and 1 year of index admission, higher number of hospitalizations in the year before CDI, escalation in medications within 1 year of index admission, and were more likely to land up in colectomy when than patients without CDI.

    3.2 Univariate analysis for risk factors of colectomy

    Table 2 summarizes the results of univariate analysis for risk factors associated with colectomy. The presence of severe disease on endoscopy, CDI, use of biologics, higher number of ER visits during the year prior to CDI and a higher number of hospitalizations within 3 months of CDI were found to be significantly associated with an increased risk for colectomy.

    View this table:
    Table 2

    Univariable analysis of factors associated with need for colectomy.

    FactorNo colectomy (N = 120)Colectomy (N = 26)p-Value
    Male75(62.5)20(76.9)0.16
    Age at infection*46.5 ± 17.342.8 ± 12.20.31
    Duration of UC (years)2.0[1.00,5.0]3.5[3.0,7.0]0.034
    Proctitis4(3.3)2(7.7)0.31
    Left sided colitis14(11.7)4(15.4)0.6
    Extensive colitis98(81.7)19(73.1)0.32
    Severe disease32(26.7)23(88.5)< 0.001
    Baseline ASA81(67.5)19(73.1)0.58
    Baseline immunosuppressants5(4.2)3(11.5)0.13
    CDI29(24.2)16(61.5)< 0.001
    Baseline use of biologics0(0.0)1(3.8)0.031
    Number of ER visits within 3 months after0.59
    0114(95.0)24(92.3)
    16(5.0)2(7.7)
    Number of hospitalizations in the year before0.032
    0102(85.0)17(65.4)
    111(9.2)8(30.8)
    2+7(5.8)1(3.8)
    Number of hospitalizations within 3 months< 0.001
    0109(90.8)16(61.5)
    111(9.2)10(38.5)
    Increase in meds within 3 months10(8.3)3(11.5)0.6
    Prednisone: 1 year before37(30.8)10(38.5)0.45

    Values presented as Mean ± SD with t-test, Median [P25, P75] with Wilcoxon rank sum tests or N (%) with Wilcoxon rank sum test for number of ER and hospital admissions and Pearson's chi-square test otherwise.* In patients with no CDI, their age at infection denote the time of data entry.

      3.3 Multivariate analysis for risk factors of colectomy

      Table 3 summarizes the results of multivariate analysis of risk factors associated with colectomy in UC patients within 1 year. CDI (odds ratio (OR) 10.0, 95% confidence interval (CI): 2.7, 36.3, p < 0.001) and severe disease on endoscopy (OR 16.7, 95%: 4.1, 67.9, p < 0.001) were found to be independently associated with colectomy within 1 year.

      View this table:
      Table 3

      Multivariate logistic regression analysis of factors associated with colectomy

      FactorOR (95% CI)p-Value
      CDI10.0 (2.7, 36.3)< 0.001
      Severe disease on endoscopy16.7 (4.1, 67.9)< 0.001
      Number of ER visits within 3 months5.7 (1.5, 21.7)0.011
      Use of ASA3.3 (0.99, 11.0)0.051

      OR: odds ratio; CI: confidence interval.

        3.4 Pre/post CDI (sub-group analysis)

        There was no significant difference in number of ER visits (p = 0.21) or number of hospital admissions (p = 0.73) during the year before and after CDI. Among patients with CDI, 55.8% of patients had an escalation in medical treatment in the year after CDI as compared to the prior year of 21.2% (p < 0.0001). Specifically, 23 subjects had no increase in medications at either time, 11 had an increase both before and after CDI and 18 had an increase in medications only during the year after CDI. Among patients with CDI, 48.1% of patients had an escalation in prednisone dosage in the year after CDI as compared to the prior year of 36.5% (p = 0.03). Specifically, 26 patients had no escalation with prednisone at either time, 18 had an inceased used both before and after CDI and 7 had an increased use only during the year after CDI (Fig. 1).

        Figure 1

        Comparison of patients before and after CDI.

        3.5 Pre/post sub-group analysis in the non-CDI group

        There was no significant difference in number of ER visits (p = 0.13) or number of hospital admissions (p = 0.82) during the year before and after the index C. difficile toxin testing in the non-CDI group. 12.9% of patients had an escalation in medical treatment in the year after CDI as compared to the prior year of 4.6% (p = 0.10).

        3.6 CDI treatment

        Nine patients were treated with a combination of metronidazole and vancomycin, while six patients were treated with metronidazole alone. The remaining 30 patients were treated with oral vancomycin alone.

        4 Discussion

        The present study attempted to assess the long-term impact of CDI in UC patients on adverse outcome defined by colectomy within 12 months of CDI. Patients who had CDI had significantly more UC-related ER visits in the year following the index infection (38.5% vs. 4% than those without CDI. One year following the index infection admission, patients with CDI also had a significantly higher rate of colectomy than compared to C. difficile-negative patients (36.5% vs. 9.9%). Among patients with CDI, 55.8% of patients had an subsequent escalation in medical treatment in the year after CDI as compared to the prior year of 12.9%. CDI (OR 10.0, 95%: 2.7, 36.3) and severe disease on endoscopy (OR 16.7, 95% CI: 4.1, 67.9) were found to be independently associated with colectomy within 1 year.

        We have earlier reported the short-term clinical outcome of CDI in UC patients with colectomy at 3 months as a primary end point.21 CDI did not seem to have a negative impact on clinical outcome (colectomy) in UC patients in our previous study. Even in the present study, there was no significant difference in the colectomy risk at 3 months between the two groups. We speculate that identification of CDI during UC flare up may be “beneficial”, as some patients with refractory UC might have been due to superimposed CDI and treatment of CDI may help inducing clinical remission of UC. However we observed that CDI-UC patients on long-term follow up have an increased risk for colectomy. Also these patients might have a worse clinical course of their underlying UC as patients with CDI recurrence were excluded from the study group. CDI-UC patients also had more ER visits than those without CDI.

        We now subsequently studied the clinical outcome (colectomy) at 12 months following the index C. difficile testing. In our previous short term outcome study21 and another study utilizing the Health Care inpatient care database, the development of CDI was inversely related to the risk of colectomy.11 We postulate that the lower risk of colectomy with UC-CDI may be due to be the fact that patients with UC exacerbation resulting from CDI are more likely to improve with proper pathogen-directed medical therapy. Thus treating CDI in UC patients may actually prevent the need for colectomy in short-term. However in this long-term follow up study up to a year, CDI-UC patients had significantly more UC-related ER visits than non-infected UC patients. Although there was no significant difference in the hospitalizations for UC flares between these two groups, there was significant difference in colectomy risk. Similar results of worse long-term outcomes were reported in another single-center study where 44.6% of patients with CDI required colectomy within 1 year as compared to 25% of patients without CDI. Logistic regression analysis showed that CDI at index admission was a significant predictor for requiring colectomy within 1 year.19

        In another single-center study in pediatric IBD patients, 57% of CDI patients with UC were readmitted with an exacerbation of disease within 6 months of infection and 67% required escalation of therapy following CDI, compared to 30% of IBD patients without CDI.23 Although, patients with UC and CDI were less likely to have pancolitis during their index presentation, these patients had escalation in their medications within 1 year of index admission. When we compared UC patients with CDI the year before and after, patients with CDI were more likely to require escalation in medical treatment following CDI. Among patients with CDI, 55.8% of patients had an escalation in medical treatment in the year after CDI as compared to the prior year of 21.2%. Similar findings were reported in a case–control study, in which 46% of patients (40/87) had hospitalizations in the year following CDI. Also 53% (46/87) of IBD patients with CDI required an escalation in their IBD medical therapy.20 We however did not observe this phenomenon in the non-CDI group and this trend was only seen in the CDI group.

        It would be difficult to hypothesize that an infection like CDI which happened a year ago could directly increase the risk of colectomy up to 1 year. However we could hypothesize that CDI could have altered the natural history of UC by activating abnormal mucosal immune response, or CDI might be sign reflecting an aggressive form of UC course, which results in more ER visits, more hospitalizations, escalation of medical treatment and ultimately resulting in colectomy. Sicker patients might have got CDI. This might have played a major role in the colectomy risk because patients with CDI had significant active or severe disease on endoscopy when compared to patients without CDI. Also these patients with CDI had more number of hospitalizations the year before than those without CDI highlighting that these patients had more severe disease to begin with before developing CDI which could have contributed to the colectomy risk.

        There are limitations to our study. The study population was recruited from a tertiary-care IBD center. This might have resulted in referral bias. We recorded only the ER visits and hospitalizations in our institution. We could have missed hospitalizations and ER visits in other institutions. Also, the sensitivity and specificity of the immunoassay for C. difficile remains to be improved is not 100%. We have in fact replaced immunoassay with more sensitive polymerase chain reaction based assay for the diagnosis of CDI recently.24 Also it was a retrospective study of data and we did not have complete data on the type of antibiotic used which predisposed to infection nor did we have the information on the type of C. difficile strain in UC patients. We used the recent colonoscopy prior to CDI for assessment of disease severity. As patients may have had procedures either in remission or in the setting of active disease, this could have certainly introduced bias in assessment of disease activity or severity. We did not record the Charlson index in our patients.

        Nonetheless, the current study has several clinical implications. This study has highlighted that patients with UC who develop CDI may not necessarily require colectomy in the short run when adequately treated; however these patients on follow up for a year have an increased risk of colectomy. Close monitoring of disease course of UC following CDI is warranted. For example, patients with history of CDI may require more frequent stool testing with or without UC symptom flare up. A prospective cohort study of CDI in UC is required with serial measurements of disease activity to study the natural history of CDI in UC patients.

        To conclude, CDI appears to be independently associated with an increased risk of colectomy on the long-term follow up.

        Conflict of interest

        The authors declared no financial conflict of interest.

        Financial support

        This study is partially supported by a grant from Inflammatory Bowel Disease Working Group grant (U.N).

        Potential competing interests

        The authors declared no potential competing interests.

        5-ASA
        5-aminosalicylic acid
        CDI
        Clostridium difficile infection
        CD
        Crohn's disease
        ELISA
        enzyme linked immunosorbent assay
        IBD
        inflammatory bowel disease
        IPAA
        ileal pouch-anal anastomosis
        6MP
        6-mercaptopurine
        NSAIDs
        non-steroidal anti-inflammatory drugs
        UC
        ulcerative colitis

        References

        View Abstract