Supported in part by the “Talpiot” Medical Leadership grant from Sheba Medical Center (to S.B.H.).
The optimal management of Clostridium difficile infection (CDI) in flaring inflammatory bowel disease (IBD) patients has not been defined. Limited data suggest that coadministration of immunomodulators (IM) with antibiotics (AB) results in a worse outcome. We investigated the prevalent practice among North American gastroenterologists in this scenario.
A structured questionnaire presented the clinical cases of two hospitalized patients with ulcerative colitis and concomitant CDI, either with or without prior IM treatment. The questionnaire was distributed to a sample of gastroenterologists at medical centers across North America. Respondents were requested to denote their therapeutic choices for these patients.
The survey included 169 gastroenterologists, 122 from the US and 47 from Canada, with an average of 12 ± 10 years of experience in gastroenterology. Forty-two (25%) of the respondents were IBD experts. Seventy-seven (46%) respondents elected to add an IM in combination with AB, whereas 82/169 (54%) treated the flare with AB alone (P = NS). The rate of administering combined AB+IM was similar for the IBD experts and the non-IBD experts. Only 11% of respondents withdrew maintenance azathioprine upon the diagnosis of CDI. More IBD experts stopped azathioprine treatment compared to the non-IBD experts (12/42 versus 6/127, P < 0.001). Overall, 65% of surveyed gastroenterologists stated they believe these patients are afflicted by two simultaneous but separate disease processes.
There is significant disagreement among gastroenterologists on whether combination AB+IM or AB alone should be given to IBD patients with CDI-associated flares. Controlled trials are needed to investigate the optimal management approach to this clinical dilemma. (Inflamm Bowel Dis 2010;)
Clostridium difficile infection (CDI) has been increasingly recognized to be associated with disease flares in patients with inflammatory bowel disease (IBD).1–3 The exact pathogenic role of C. difficile in these clinical settings remains poorly defined. Conceptually, it can either produce symptoms that mimic an exacerbation of IBD, precipitate a genuine flare, or merely be an innocent colonizing bystander organism.4 Regardless of its exact pathogenic role, the incidence of CDI in hospitalized IBD patients is increased compared to the general population.3, 5 Furthermore, it was shown that patients with a dual diagnosis of CDI and IBD during hospitalization suffered increased rates of mortality and morbidity compared to patients hospitalized with only one of these diagnoses.3, 6
The mainstay of treatment of CDI is antibiotics (AB), most commonly metronidazole or vancomycin.7 There are limited data supporting the superiority of vancomycin in patients with severe CDI,8 but whether this agent is also superior to metronidazole for the treatment of IBD patients with CDI is unknown. Moreover, when a flaring IBD patient is found to be infected with C. difficile, the clinician is confronted by the dilemma of whether to withhold all immunosuppression and prescribe AB alone or, conversely, administer AB along with immunosuppressants to treat a possible concurrent exacerbation of underlying IBD.9, 10 In a recent retrospective European study, a trend towards a more adverse outcome was observed among patients treated by combination therapy compared to those treated by AB alone.11 However, neither that study nor others have directly addressed the common practice among gastroenterologists in prescribing one of those two opposing regimens for IBD patients who have evident CDI.
Therefore, the primary aim of the present study was to investigate the common practice among North American gastroenterologists confronting the clinical dilemma of prescribing AB alone or combination AB and immunomodulators (IM) for flaring IBD patients with CDI. As a secondary goal, we also aimed to explore treatment preferences regarding the choice of the first line antibiotic agent in this scenario.
MATERIALS AND METHODS
A structured questionnaire was designed to present the clinical cases of two patients with ulcerative colitis, either with or without prior IM treatment, who have been hospitalized with a flare of colitis and found to be afflicted with CDI. The questionnaire was developed by two of the authors (H.Y. and S.B.H.) with assistance by a bio-epidemiologist and was pretested on a sample of six senior gastroenterologists to ensure validity and comprehensibility. The description of the cases along with the multiple-choice questionnaire appears in the Appendix. All of the surveyed physicians were board-certified gastroenterologists practicing in the US or Canada and were contacted by one of the investigators (in person or electronically) between April 1 and December 20 2009. To reduce the potential bias stemming from the low response rate that is common in random mail-based questionnaires, we deliberately opted for a direct-contact distribution of questionnaires. Accordingly, each of the investigators distributed the questionnaires to all physicians in his/her department and to community-affiliated physicians, in person or electronically. In addition, two investigators (H.Y. and L.Y.) handed out questionnaires to several randomly picked attendees of the 2009 CCFA Advances in IBD course in Hollywood, Florida. The respondents were requested to denote the therapy they would prescribe to the flaring patients described in the clinical scenarios. The rates of administering AB alone versus initiating IM in combination with AB were compared. The professional demographics of the respondents and their reasons for choosing the specific therapy were also recorded. To analyze management differences among subpopulations of gastroenterologists, we defined respondents as IBD experts if they reported working at a dedicated IBD facility and having at least 50% of their patients with IBD. The responses of the IBD experts were compared to those of the non-IBD experts.
Continuous variables were analyzed by two-tailed Student's t-test or Mann–Whitney U-test, as appropriate, and categorical variables were analyzed by Fisher's Exact test. All variables were then entered into multivariate analysis using a multiple backward logistic regression model to identify factors independently affecting dichotomous clinical outcomes. For calculation of the odds ratios, continuous variables were entered to the model as categorical variables according to their respective quartile rank in the corresponding parameter. Sample size was determined using a subject-to-item ratio of 1:10, and taking into account 10 variables, of which three had multiple subitems. For the resulting 15 variables, at least 150 subjects were needed to reduce the risk of overfitting the logistic regression model. All statistics were performed using MedCalc software (Mariakerke, Belgium). P < 0.05 was considered significant.
The study was approved by the Institutional Review Board of Sheba Medical Center, Tel-Hashomer, Israel.
Of the 186 board-certified gastroenterologists who were contacted directly by one of the participating investigators, 169 (91%) completed the questionnaire and comprised the study population. Of these, 122 (72%) were from the US and 47 (28%) were from Canada. Forty-two (25%) of the total cohort were IBD experts.
The percentage of gastroenterologists electing each of the multichoice therapeutic options for each of the two cases is shown in Figure 1.
For purpose of analysis, we grouped the physicians who advocated AB alone (whether vancomycin or metronidazole), and compared them with physicians who chose a combination of AB with IM. In virtually all cases the added IM consisted of initiating corticosteroids to treat the flare. The resulting analysis for each of the two cases separately is shown in Table 1. Averaging the results obtained for the two hypothetical patients, combination IM with AB was prescribed by 46% of gastroenterologists compared to 54% who administered AB alone (P = 0.3). The rate of administering combination therapy (AB+IM) did not differ between the IBD experts and the non-IBD experts (50% versus 44%, P = 0.6).
Table 1. Rates of Administering IM and AB (IM+AB) or AB Alone (AB Alone) for Each of the Two Presented Cases
To investigate the impact of respondents' individual characteristics on the choice to administer AB+IM in at least one of the two cases, a univariate analysis was performed for all the parameters and the unadjusted odds ratio for prescribing combination AB+IM were calculated (Table 2). All variables were also entered into a multivariate backwards logistic regression to investigate which of these parameters was independently associated with the therapeutic choice, as expressed by adjusted odds ratio (OR; Table 2). As shown, several parameters were found on univariate analysis to be associated with increased likelihood of administering combination therapy (AB+IM). However, on a multivariate analysis, believing the cause of patients' symptoms was an underlying IBD inflammatory process, either in conjunction with CDI or alone, was the parameter most strongly associated with the decision to administer combined therapy (Table 2). Not shown in the table are other reasons noted for the therapeutic choice, which included possible corticosteroids benefit in C. difficile colitis (n = 4), combinations of several reasons (n = 6), or absence of a specific explanation (n = 2). To further examine the independent role of parameters possibly influencing the decision of physicians to administer AB+IM, we also performed an additional multivariate analysis, this time incorporating into the regression model only the variables that were found to differ between the two groups with a P < 0.1 on univariate analysis. In this analysis, administering AB+IM was positively associated only with the belief that the etiology is concomitant IBD flare and infection (OR 16, 95% confidence interval [CI] 4.7–54, P < 0.001) or isolated IBD flare (OR 89, 95% CI 6.7–1185, P < 0.001) and was negatively associated with the decision to discontinue azathioprine (OR 0.07, 95% CI 0.01–0.7, P = 0.02).
Table 2. Analysis of Background Factors Associated with the Decision to Prescribe Combination AB+IM to Treat the Flare in Either of the Two Cases
Some physicians in group-based practice noted a university hospital affiliation.
Isolated IBD flare (with C. difficile being only colonizer or false positive)
Only a minority of the respondents preferred withdrawing a regularly taken maintenance azathioprine upon diagnosis of C. difficile (Table 2). The decision to discontinue azathioprine was significantly more common in the IBD specialist group compared to the non-IBD specialist one (12/42 versus 6/127, P < 0.001). Discontinuing azathioprine was more prevalent among physicians prescribing AB alone rather than AB+IM (Table 2).
The therapeutic choice of vancomycin or metronidazole as the first-line antibiotic agent is also shown in Table 2. More IBD experts than non-IBD experts chose vancomycin as the first-line AB in both case 1 and 2 (25/42 versus 46/127, P = 0.01 and 33/42 versus 43/127, P < 0.001, respectively). However, on multivariate analysis, being an IBD expert lost its significant association with the decision to prescribe vancomycin as the first-line antibiotic. None of the other parameters tested showed association with this therapeutic decision (Table 3).
Table 3. Analysis of Background Factors Associated with the Decision to Prescribe Vancomycin as the First-line Antibiotic in Either of the Two Cases
Some physicians in group-based practice noted a university hospital affiliation.
Isolated IBD flare (with C. difficile being only colonizer or false positive)
The present study addressed the common clinical practice among North American gastroenterologists treating flaring IBD patients with CDI. The results show that there is significant disagreement between physicians on whether concomitant AB+IM or AB alone should be prescribed in this clinical scenario. Interestingly, divergent management policies were similarly found among IBD experts working in dedicated IBD centers.
IBD patients have been consistently found to be affected by C. difficile at a greater frequency than the general population.3, 5, 12, 13 This organism has been implicated in flares of disease especially in IBD patients with colonic disease, although cases of C. difficile-associated enteritis and pouchitis have also been reported.14, 15 Nevertheless, the exact pathogenic role of C. difficile in these settings has yet to be defined. In principle, C. difficile may cause an isolated infectious colitis superimposed on IBD, or conversely, may precipitate an IBD flare leading to two separate but simultaneous inflammatory processes. A third mechanistic possibility may be that IBD flare has occurred independently of C. difficile, which is merely a colonizer or perhaps even a false-positive result. The observation that most episodes of CDI in IBD patients present with bloody diarrhea and in the absence of endoscopically detectable pseudo-membranes12, 16 may augment the tendency to ascribe most of the symptoms to underlying IBD flare rather than to superimposed CDI. Indeed, more than half of the respondents postulated that two separate yet simultaneous inflammatory processes were taking place in these patients. Moreover, nearly 10% of physicians contended that the only cause of symptoms in such patients is an isolated IBD flare. This contrasted with other respondents, who incriminated infectious colitis by C. difficile as the sole cause of clinical exacerbation and prescribed AB alone. Importantly, the results of the multivariate analysis suggest that in the absence of evidence regarding the superior treatment regimen, it is the different mechanistic conceptions on the dominant causative inflammatory process that dictate the eventual therapeutic choice. Being affiliated with a university hospital, involved in research work, and having relatively fewer years of clinical experience were other parameters that were also associated with the tendency to prescribe combination AB+IM rather than AB alone (Table 2). This observation possibly supports the above notion that therapeutic decisions in this clinical scenario are mostly driven by disease mechanistic perceptions influenced by involvement in research work and by increased exposure to academic medicine. Nevertheless, on multivariate analysis only the duration of clinical experience retained a weak association with the drug regimen prescribed.
Several studies have addressed the impact of ongoing immunosuppression on the outcome of CDI in the non-IBD population but have yielded conflicting results. Increased rate of ICU admission, colectomy, and/or mortality in IM-treated patients was found in some studies, but not all.17–22 Thus, whether cessation of IM therapy upon diagnosis of CDI is mandatory in these patients is as yet undefined. The present study documents that many physicians go a step further and deliberately intensify the immunosuppression (along with AB) for flaring IBD patients with CDI. The divergent management policies revealed by the present findings may bear significant clinical impact given the increased morbidity and mortality of CDI in IBD.2, 6, 23 The importance of the therapeutic choice is further underscored by a recent study which suggested that combination therapy may be associated with a worse outcome, albeit at a borderline statistical significance.11 In contrast, a recent pediatric study suggested that hospitalized IBD patients with CDI stand a higher risk of IBD flares and escalation of immunosuppression within 6 months of the index hospitalization, thereby suggesting that CDI may trigger or mark a worsening underlying IBD.24 Interestingly, only a fraction of gastroenterologists in our study elected to withdraw ongoing maintenance azathioprine upon diagnosis of CDI. This was true even for some physicians who advocated an AB-alone policy for these patients and maintained that infection is the sole cause for the patients' symptoms. This unexpected finding is intriguing and seems to be peculiar to CDI in IBD, because the clinical guidelines on opportunistic infections in IBD generally recommend that IM therapy should be temporarily withheld until the resolution of active infection.25 The reasons for not withdrawing thiopurines in this context were not investigated directly, but may stem from a belief that the clearance of the drug would be slower than the resolution of the current infection. The findings of our study also reflect the current knowledge gap regarding the best first-line antibiotic agent for patients with CDI. Although the cohort as a whole was evenly divided between proponents of vancomycin and metronidazole (Table 2), IBD experts were significantly more inclined to prescribe vancomycin than non-IBD experts. This is probably an extrapolation from the limited data on the superiority of vancomycin over metronidazole in severe CDI in the non-IBD population.8 Nonetheless, further studies are pertinent to elucidate if vancomycin is also superior in the IBD-CDI population.
Several limitations of our study should be acknowledged. Not all clinical data can be provided in a vignette case description, so we cannot exclude that additional details could sway some respondents away from particular answers. In addition, we did not investigate the practice of administering biologics for remission induction in these patients. Nevertheless, an open answer option was available, and only one respondent mentioned anti-tumor necrosis factor (TNF) as a possible therapeutic option in these circumstances. Similarly, we did not present a case where CDI is diagnosed in a patient that was already treated by anti-TNF, although the prolonged half-life of these agents makes the need for proactive decision at presentation less pressing in most cases. These limitations make it hard to generalize the results to possible practice habits employed for patients treated with biologics who are concomitantly diagnosed with CDI.
In conclusion, the findings of the present study reveal significant disagreement among gastroenterologists, including self-identified IBD experts, regarding the optimal therapeutic approach for flaring IBD patients with concomitant CDI. In particular, there is little consensus on whether AB+IM or AB alone should be administered to these patients and on whether vancomycin rather than metronidazole should be the first-line antibiotic of choice. Given the increased morbidity and mortality reported in IBD patients with CDI, controlled trials are urgently needed to elucidate the optimal management approach in this clinical scenario.
The authors thank Pr. Benjamin Avidan from Sheba Medical Center for valuable statistical assistance.