Abstract
Background
Inflammatory bowel disease (IBD) patients are at an increased risk of Clostridium difficile infection (CDI), but the impact of CDI on disease severity is unclear. The aim of this study was to determine the effect of CDI on long-term disease outcome in a matched cohort of IBD patients.
Methods
Patients who tested positive for infection formed the CDI positive group. We generated a 1:2 propensity matched case to control cohort based on risk factors of CDI in the year prior to infection. Healthcare utilization data (emergency department use, hospitalizations, telephone encounters), medications, labs, disease activity and quality of life (QOL) metrics were compared by CDI status.
Results
A total of 198 patients (66 CDI, 132 matched controls) were included (56.6% female; 60.1% Crohn’s disease, 39.9% ulcerative colitis). In the year of infection, having CDI was significantly associated with more steroid and antibiotic exposure, elevated C-reactive protein or erythrocyte sedimentation rate, low vitamin D, increased disease activity, worse QOL, and increased healthcare utilization (all p<0.01). During the next year after infection, CDI patients continued to have increased exposure to CDI targeted antibiotics (p<0.001) and other antibiotics (p=0.02). They also continued to have more clinic visits (p=0.02), telephone encounters (p=0.001), and increased healthcare financial charges (p=0.001).
Conclusions
CDI in IBD is significantly associated with markers of disease severity, increased healthcare utilization and poor QOL during the year of infection, and a five-fold increase in healthcare charges in the year following infection.
Keywords: Clostridium difficile, inflammatory bowel disease, healthcare utilization, propensity score
INTRODUCTION
Clostridium difficile is a Gram-positive, anaerobic, spore-forming bacilli and the etiological agent of antibiotic associated pseudomembranous colitis. Common risk factors for community or hospital acquired C. difficile infection (CDI) are comorbid diseases and exposure to any class of antibiotics.1,2 Comorbid inflammatory bowel disease (IBD) is considered the most significant risk factor for the acquisition of CDI in the community setting.1 Previous research has shown that the likelihood of infection is increased by many factors including, but not limited to, healthcare system contact, nutritional deficiencies, and antibiotic exposure.2–5 CDI is an increasingly prevalent infectious complication in the IBD patient population.6–8 CDI in IBD patients is associated with higher rates of hospitalization, surgery, longer hospital stays, increased healthcare charges, and, most importantly, an increased mortality.9,10
Much of what is known about CDI in IBD has been derived from large national databases and administrative healthcare data. These data have provided information about increased risks of mortality and colectomy in IBD patients with CDI and how these risks have increased over time.10,11 However, many of the findings were derived from retrospective samples and administrative data, and are unable to account for a patient’s inherent risk of CDI or their disease severity prior to infection.12
We are unaware of any studies that have used a propensity score matching approach to generate a control cohort that has similar risk factors of developing CDI in the year prior to infection. Our primary aim was to determine the impact of CDI on biomarkers of IBD severity, healthcare utilization, and patient reported outcomes compared to a matched cohort based on known risk factors for infection. Our secondary aim was to investigate if changes of healthcare utilization patterns continued into the year following infection. We hypothesized that CDI would negatively impact patient outcomes in the acute period of infection and in the long-term follow up period.
Materials and Methods and DESIGN
Study design and participants
This study was conducted as a part of the UPMC IBD research registry, which has been previously described in detail.13 Briefly, IBD patients are consented and enrolled in a prospective, longitudinal, natural history registry, which organizes real world patient care data from 2009 to the present time. All data from the registry is derived from the electronic medical record and systematically processed and transformed for research.
In this study, we included all IBD patients in the UPMC IBD registry with a definite diagnosis based on standard criteria of ulcerative colitis (UC) or Crohn’s disease (CD) for our selection of cases and controls. CDI was defined as any patient with a confirmed molecular laboratory diagnosis of Clostridium difficile from 2010 to 2014 calendar years. All confirmed molecular diagnoses were assumed to have infection. Participants with CDI also had to have clinical follow up, defined as at least one clinic visit or telephone encounter in the gastroenterology clinic over the calendar year, in the year prior to infection in order to meet inclusion criteria. Controls were selected from the remaining IBD registry participants without a history of CDI.
IBD patients in both the case and control groups were excluded if they had unclassified IBD or undefined disease type. To allow capture of data from the year before and year after CDI, IBD patients with CDI occurring in 2009 or 2015 were excluded. CDI cases were also excluded if they did not have clinical follow up in the year prior to infection. We did not exclude controls who had been tested for infection, or CDI positive participants who had multiple or relapsing infections, as this is a feature of CDI in IBD patients.14 The CDI positive cohort includes IBD patients with single and multiple positive tests for CDI documented in the medical record.
Data collection and organization
All data are prospectively collected as a part of routine healthcare visits in any UPMC affiliated hospital or clinic (comprising over 20 hospitals and 500 clinics).13 All IBD related healthcare utilization including clinic visits, telephone encounters, hospitalizations, emergency room (ER) visits, and IBD related surgeries were derived from the IBD registry and temporally organized by calendar year. Healthcare utilization was also quantified by financial charges, which includes charge data for all healthcare services including, but not limited to, gastrointestinal care. Financial charges include both inpatient and outpatient charges, but do not include pharmacy charges as prescription charges independent of the UPMC system. Labs were ordered as a part of routine care as deemed appropriate by providers, therefore, laboratory biomarkers, including C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and vitamin D and B-12 deficiencies were organized by calendar year and dichotomized as normal or abnormal based on local laboratory standards. All outpatient electronic prescriptions were organized annually for each patient. Patients were designated as having exposure to the medication if they had one or more prescriptions within the calendar year. Antibiotics only included systemic exposure. Systemic antibiotics with the exception of vancomycin and fidaxomicin were analyzed as a separate category. Within the systemic antibiotics category we looked at the subgroup of patients prescribed metronidazole, as it is indicated in the setting of CDI. However, vancomycin and fidaxomicin were analyzed alone as a separate antibiotics category, as their primary indications are for CDI.
Patient reported disease activity and quality of life (QOL) metrics were collected during clinical visits to the UPMC Digestive Disorders Clinic as a part of routine care, and entered into the electronic medical record. QOL was measured by the Short Inflammatory Bowel Disease Questionnaire (SIBDQ).15 Disease activity was measured by the Harvey Bradshaw Index (HBI) for CD and Ulcerative Colitis Activity Index (UCAI) for UC.16,17 “Active disease” was defined as annual mean UCAI score ≥4 or annual mean HBI scores ≥5 during the study period. Disease phenotypic characterization was performed in both CD and UC patients using the Montreal Classification at initial presentation.18
Propensity score matching
To build a comparable control cohort at baseline we utilized nearest neighbor propensity score matching.19 We generated the propensity score for CDI with the covariates listed in Supplementary Table 1, using logistic regression. The propensity score is considered the calculated “likelihood” of infection given a patient’s particular set of covariates.20 All covariates were chosen from hypothesis driven clinical parameters that may influence a patient’s risk for CDI. Most importantly we included all encounters with the healthcare system and antibiotic exposures in the year prior to infection, which have been implicated in risk for infection.2,3 We also included all antibiotic exposures in the year prior to infection, age, and vitamin D deficiency, all which have been linked to risk of infection.5,7 Patients were matched using the protocol outlined in FIGURE 1, which features a rolling propensity score matching process over time beginning with study participants who had their first CDI event in 2010. Matching was done without replacement to build a 1:2 (cases:controls) cohort. Any controls matched to cases in previous years were excluded from any subsequent control population selection pool. Covariate balance in the year prior to infection was examined following matching.
Figure 1.
Schematic of propensity score matching. Repeated nearest neighbor propensity score matching by year without replacement.
Statistical analysis
We used chi-square analyses for categorical variables, Student’s t-test for normally distributed continuous variables, and the Wilcoxon rank-sum test for nonparametric continuous variables to assess differences and balance between groups at baseline. To account for matching, outcomes in the year of infection and year after infection were assessed using conditional logistic regression for binary outcomes, and fixed effects regression for continuous variables. Counts of healthcare utilization (hospitalizations, ER visits, telephone calls, clinic visits, radiologic procedures, endoscopies) were initially evaluated using fixed effects Poisson regression and significance was ultimately reported using conditional negative binomial regression due to over dispersion of zeros. Financial charges were transformed to natural log for normality prior to regression. All statistical tests were evaluated with an alpha = 0.05, and were completed in StataSE (v.14, StataCorp, College Station, TX).
Ethical Considerations
All participants were enrolled in the IBD Research Registry using informed consent. The IBD Research Registry (Protocol #0309054) and the current analysis (Protocol #15010214) were both approved by the University of Pittsburgh Institutional Review Board. All authors had access to the study data and reviewed and approved the final manuscript.
Results
A total of 198 patients (66 CDI, 132 matched controls) were included (56.6% female; 60.1% CD, 39.9% UC) (TABLE 1). Infection and control groups did not significantly differ in terms of baseline disease characteristics in the year prior to infection for all available metrics (TABLE 2). Study groups did not differ in regard to contact with the healthcare system including hospitalizations, ER visits, endoscopies, radiologic studies, clinic visits, and total financial charges. Additionally, the groups did not differ in terms of the proportion of patients who were exposed to antibiotics or the number of times they were prescribed antibiotics (TABLE 2).
Table 1.
Baseline demographics of inflammatory bowel disease patients included in the study
Infection Status | ||||
---|---|---|---|---|
Total Study Population n= 198 |
Clostridium difficile positive n= 66 |
Controls n= 132 |
p-value | |
Age (mean years ± SD)* | 45.4 ± 15.2 | 44.3 ± 14.8 | 45.9 ± 15.5 | 0.486 |
Female, (n, %) | 112 (56.6) | 34 (51.5) | 78 (59.1) | 0.311 |
Race/Ethnicity, (n, %) | ||||
Caucasian | 190 (96.5) | 62 (93.9) | 128 (97.7) | 0.249 |
Black | 6 (3.1) | 3 (4.6) | 3 (2.3) | |
Other or unknown | 1 (0.5) | 1 (1.5) | 0 (0.0) | |
Hispanic/Latino | 0 (0.0) | 0 (0.0) | 0 (0.0) | 1.00 |
Marital Status, (n, %) | ||||
Married or significant other | 120 (60.6) | 41 (62.1) | 79 (59.9) | 0.447 |
Single | 58 (29.3) | 21 (31.8) | 37 (28.0) | |
Divorced, widowed, separated | 17 (8.6) | 4 (6.1) | 13 (9.9) | |
Unknown | 3 (1.5) | 0 (0.0) | 3 (2.3) | |
Employment status (n, %) | ||||
Full time or self-employed | 103 (52.0) | 36 (54.6) | 67 (50.8) | 0.937 |
Full time student | 12 (6.1) | 5 (7.6) | 7 (5.3) | |
Part-time | 7 (3.5) | 2 (3.0) | 5 (3.8) | |
Retired | 17 (8.6) | 6 (9.1) | 11 (8.3) | |
Not employed | 43 (21.7) | 13 (19.7) | 30 (22.7) | |
Unknown | 16 (8.1) | 4 (6.1) | 12 (9.1) |
SD - standard deviation;
Age of study participants as of January 1, 2015.
Table 2.
Baseline disease characteristics from the year prior to infection
Infection Status | ||||
---|---|---|---|---|
| ||||
Total n= 198 |
CDI positive n= 66 |
Controls n= 132 |
p-value | |
| ||||
Disease category (n, %) | ||||
| ||||
Crohn’s disease | 119 (60.1) | 41 (62.1) | 78 (59.1) | 0.681 |
| ||||
Ulcerative colitis | 79 (39.9) | 25 (37.9) | 54 (40.9) | |
| ||||
Disease characteristics18 | ||||
| ||||
Crohn’s disease location, n=98 | ||||
| ||||
Ileal (L1) | 26 (26.5) | 7 (20.6) | 19 (29.7) | 0.331 |
Colonic (L2) | 31 (31.6) | 9 (26.5) | 22 (34.4) | 0.423 |
Ileocolonic (L3) | 44 (44.9) | 18 (52.9) | 26 (40.6) | 0.243 |
Upper GI (L4) | 4 (4.1) | 3 (8.8) | 1 (1.6) | 0.084 |
| ||||
Crohn’s disease behavior, n=98 | ||||
| ||||
Inflammatory (B1) | 50 (51.0) | 14 (41.2) | 36 (56.3) | 0.155 |
Stricturing (B2) | 38 (38.8) | 17 (50.0) | 21 (32.8) | 0.096 |
Penetrating (B3) | 20 (20.4) | 7 (20.6) | 13 (20.3) | 0.974 |
| ||||
Perianal disease, n=98 | 21 (21.4) | 6 (17.7) | 15 (23.4) | 0.506 |
| ||||
Ulcerative colitis extent, n=68 | ||||
| ||||
Proctitis (E1) | 4 (5.9) | 2 (9.5) | 2 (4.3) | 0.394 |
Left-Sided (E2) | 18 (26.5) | 4 (19.1) | 14 (29.8) | 0.354 |
Extensive (E3) | 49 (72.1) | 17 (81.0) | 32 (68.1) | 0.275 |
| ||||
History of IBD related surgery* | 62 (31.3) | 18 (27.3) | 44 (33.3) | 0.386 |
| ||||
Biomarkers of severity (n, %)† | ||||
| ||||
Elevated CRP | 89 (45.0) | 30 (45.5) | 59 (44.7) | 0.920 |
| ||||
Elevated ESR | 70 (35.4) | 24 (36.4) | 46 (34.9) | 0.833 |
| ||||
Low Vitamin D (<40 ng/mL) | 88 (44.4) | 31 (47.0) | 57 (43.2) | 0.613 |
| ||||
Low B-12 (<300 pg/mL) | 37 (18.69) | 12 (18.2) | 25 (18.9) | 0.897 |
| ||||
Medication use (n, %)† | ||||
| ||||
Immunomodulators | 45 (22.7) | 15 (22.7) | 30 (22.7) | 1.00 |
| ||||
Biologics | 66 (33.3) | 23 (34.9) | 43 (32.6) | 0.749 |
| ||||
Systemic steroids | 81 (40.9) | 132 (43.9) | 52 (39.4) | 0.540 |
| ||||
5-aminosalicylic acids | 55 (27.8) | 16 (24.2) | 39 (29.6) | 0.432 |
| ||||
Systemic Antibiotics | 107 (54.0) | 37 (56.1) | 70 (53.0) | 0.687 |
| ||||
Metronidazole | 44 (22.2) | 14 (21.2) | 30 (22.7) | 0.809 |
| ||||
Vancomycin | 35 (17.7) | 11 (16.7) | 24 (18.2) | 0.792 |
| ||||
Average Total SIBDQ, n=145 (median, [IQR]) | 46 [21.5] | 45.7 [18.8] | 47 [21.7] | 0.273 |
| ||||
Disease activity metrics (median, [IQR]) | ||||
| ||||
HBI, n=96 | 5.0 [6.1] | 5.0 [9.5] | 5.0 [4.7] | 0.860 |
| ||||
UCAI, n=61 | 5.0 [8.0] | 6.5 [10.0] | 4.2 [8.0] | 0.147 |
| ||||
Healthcare utilization, (median, [IQR]) | ||||
| ||||
Emergency room visits | 0.0 [2.0] | 1.0 [3.0] | 0.0 [2] | 0.075 |
| ||||
Hospitalizations | 0.0 [2.0] | 0.0 [2.0] | 0.0 [2.0] | 0.427 |
| ||||
Surgeries, (n, %) | 29 (14.7) | 10 (15.2) | 19 (14.4) | 0.887 |
| ||||
Radiologic studies | 2.0 [6.0] | 3.0 [5.0] | 2.0 [6.0] | 0.332 |
| ||||
Clinic visits | 2.0 [3.0] | 2.0 [3.0] | 2.0 [2.0] | 0.172 |
| ||||
Telephone calls | 5.0 [7.0] | 5.0 [8.0] | 5.0 [7.0] | 0.792 |
| ||||
Financial charges ($) | 6984.75 [118713.10] | 14462.00 [125282.00] | 5990.00 [112014.30] | 0.290 |
P-values are bolded if significant, <0.05. SIBDQ, short inflammatory bowel disease questionnaire; IBD, inflammatory bowel disease; GI, gastrointestinal; IQR, interquartile range; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; HBI, Harvey-Bradshaw Index; UCAI, ulcerative colitis activity index.
History of any gastrointestinal surgery prior to 2009.
Immunomodulators include 6-mercaptopurine, azathioprine, methotrexate. Biologics include anti-tumor necrosis factor agents (infliximab, adalimumab, and certolizumab), anti-integrin therapy (vedolizumab and natalizumab). Biologics include all anti-tumor necrosis factor agents and anti-integrin therapies. Systemic antibiotics include all systemic antibiotic exposure, excluding vancomycin and fidaxomicin.
Year of infection
In the year of CDI, follow up occurred in 93.4% (n=185) of the cohort, and rates of follow up did not differ between infection and control groups (95.5% CDI vs. 92.4% controls). Having CDI was significantly associated with increased medication exposure including steroids (49.2% CDI vs. 28.7% controls, p=0.005) systemic antibiotics (excluding vancomycin) (90.5% CDI vs. 50.8% controls, p<0.001), and vancomycin exposure (73.0% CDI vs. 12.3% controls, p<0.001) (TABLE 3). Neither group had any exposure to fidaxomicin or fecal microbiota transplantation. The CDI group also had a greater total number of antibiotic prescriptions (median: 3 CDI vs. 1 controls, p<0.001). The two groups did not differ in their exposure to 5-aminosalicylic acids, immunomodulators, or biologic medications (TABLE 3).
Table 3.
Disease severity, medication exposure and healthcare utilization from the year of Clostridium difficile infection
Infection Status | ||||
---|---|---|---|---|
Total n=185 |
CDI positive n=63 |
Controls n=122 |
p-value | |
Medication use (n, %)† | ||||
Immunomodulators | 45 (24.3) | 13 (20.6) | 32 (26.2) | 0.399 |
Biologics | 54 (29.2) | 19 (30.2) | 35 (28.7) | 0.812 |
Systemic steroids | 66 (35.7) | 31 (49.2) | 35 (28.7) | 0.005 |
5-aminosalicylic acids | 41 (22.2) | 17 (27.0) | 24 (19.7) | 0.355 |
Systemic antibiotics | 119 (64.3) | 57 (90.5) | 62 (50.8) | <0.001 |
Metronidazole | 48 (26.0) | 27 (42.9) | 21 (17.2) | <0.001 |
Vancomycin | 61 (33.0) | 46 (73.0) | 15 (12.3) | <0.001 |
Biomarkers of severity (n, %) | ||||
Elevated CRP | 76 (40.1) | 37 (58.7) | 39 (32.0) | 0.002 |
Elevated ESR | 49 (26.0) | 26 (41.3) | 22 (18.0) | 0.002 |
Low Vitamin D (<25 ng/mL) | 42 (22.7) | 24 (38.1) | 18 (14.8) | 0.001 |
Low Vitamin D (<40 ng/mL) | 96 (50.3) | 42 (66.7) | 51 (41.8) | 0.001 |
Low B-12 (<300 pg/mL) | 21 (11.4) | 11 (17.5) | 10 (8.2) | 0.027 |
Average Total SIBDQ, n=130 (median, [IQR]) | 48.3 [20.3] | 43.2 [19.9] | 53.5 [22.0] | 0.003 |
Active disease (n, %)* | 65 (48.9) | 33 (63.5) | 32 (39.5) | 0.016 |
Disease activity metrics, (median, [IQR]) | ||||
Harvey-Bradshaw Index, n=92 | 3.5 [5.6] | 5.5 [6.5] | 2.5 [5.0] | 0.006 |
UCAI, n=51 | 5.0 [7.0] | 6.2 [7.0] | 4.0 [6.0] | 0.960 |
Healthcare utilization, (median, [IQR]) | ||||
Emergency room visits | 0.0 [1.0] | 0.0 [4.0] | 0.0 [1.0] | <0.001 |
Hospitalizations | 0.0 [1.0] | 1.0 [4.0] | 0.0 [1.0] | <0.001 |
Surgeries, (n, %) | 18 (9.73) | 6 (9.52) | 12 (9.84) | 1.00 |
Radiologic Procedures | 1.0 [4.0] | 2.0 [5.0] | 1.0 [3.0] | <0.001 |
Endoscopies | 1.0 [1.0] | 1.0 [1.0] | 1.0 [1.0] | <0.001 |
Clinic visits | 2.0 [3.0] | 3.0 [3.0] | 2.0 [2.0] | <0.001 |
Telephone calls | 5.0 [8.0] | 9.0 [12.0] | 3.0 [7.0] | <0.001 |
Financial charges ($) | 6805.00 [94059.00] | 28433.88 [149294.70] | 4989.00 [65498.25] | <0.001 |
P-values are bolded if significant, <0.05. SIBDQ – short inflammatory bowel disease questionnaire; IBD – inflammatory bowel disease; GI – gastrointestinal; IQR – interquartile range; CRP – C-reactive protein; ESR – erythrocyte sedimentation rate; UCAI – ulcerative colitis activity index.
Active disease defined as annual mean ulcerative colitis activity index score ≥4 or annual mean Harvey-Bradshaw Index scores ≥5 during the study period.
Immunomodulators include 6-mercaptopurine, azathioprine, methotrexate. Biologics include anti-tumor necrosis factor agents (infliximab, adalimumab, and certolizumab), anti-integrin therapy (vedolizumab and natalizumab). Biologics include all anti-tumor necrosis factor agents and anti-integrin therapies. Systemic antibiotics include all systemic antibiotic exposure, excluding vancomycin and fidaxomicin.
Infection was also significantly associated with elevated inflammatory biomarkers including CRP (58.7% CDI vs. 32.0% controls, p=0.002) and ESR (41.3% CDI vs. 18.0% controls, p=0.002), low vitamin D (p=0.001), and low vitamin B-12 (p=0.02) (TABLE 3). Using patient reported metrics, infection was associated with lower QOL scores (p=0.003) and self reported active disease (p=0.02) (TABLE 3).
Patients with CDI experienced significantly increased healthcare utilization across all measured metrics except the proportion of patients requiring surgery during the year of infection (TABLE 3). Those with CDI had an increased number of radiographic studies, endoscopies, clinic visits, and telephone encounters (all p<0.001). They also had more unplanned care including ER visits (mean: 3.7 CDI vs. 1.1 controls, p<0.001) and hospitalization (mean: 2.2 CDI vs. 0.8 controls, p<0.001). Patients with CDI had increased financial healthcare charges in the year of infection (p<0.001) (TABLE 3).
Year after infection
In the year after infection, follow up occurred in 77.8% (n=154) of the original study group. The CDI group includes patients with single or multiple positive tests for CDI. CDI patients were significantly more likely to follow up in the year after infection (CDI 90.9%; controls 71.2%) compared to controls (p=0.003), (TABLE 4).
Table 4.
Risk of future disease severity and healthcare utilization in year after infection.
Infection Status | ||||
---|---|---|---|---|
n=154 | Total | CDI Positive n=60 |
Controls n=94 |
p-value |
Follow up year after infection, n (%) | 154 (77.8) | 60 (90.9) | 94 (71.2) | 0.003 |
Biomarkers of Severity | ||||
Elevated CRP | 45 (29.2) | 18 (30.0) | 27 28.7) | 0.797 |
Elevated ESR | 36 (23.4) | 16 (26.7) | 20 (21.3) | 0.328 |
Medications† n (%) | ||||
Biologics, n (%) | 49 (31.8) | 23 (38.3) | 26 (27.7) | 0.260 |
Immunomodulators | 49 (31.8) | 16 (26.7) | 33 (35.1) | 0.487 |
Prednisone | 48 (31.2) | 23 (38.3) | 25 (26.6) | 0.067 |
5-aminosalicylic acids | 34 (22.1) | 15 (25.0) | 19 (20.2) | 0.490 |
Systemic antibiotics | 81 (52.6) | 39 (65.0) | 42 (44.7) | 0.023 |
Metronidazole | 22 (14.3) | 8 (13.3) | 14 (14.9) | 0.931 |
Vancomycin | 34 (22.1) | 23 (38.3) | 11 (11.7) | 0.001 |
Average Total SIBDQ, n=102 (median, [IQR]) | 49.8 [19.0] | 47.0 [22.0] | 51.5 [19.5] | 0.078 |
Disease activity, (median, [IQR]) | ||||
HBI (Crohn’s disease), n=73 | 3.7 [6.0] | 4.0 [6.0] | 3.0 [5.0] | 0.698 |
UCAI (Ulcerative colitis), n=39 | 2.0 [5.0] | 3.3 [5.0] | 2.0 [5.8] | 0.931 |
Healthcare utilization, (median [IQR]) | ||||
Telephone calls | 4.0 [7.0] | 5.5 [8.5] | 3.0 [5.0] | 0.001 |
Office visits | 2.0 [2.0] | 2.0 [3.0] | 1.0 [1.0] | 0.023 |
Radiologic procedures | 1.0 [4.0] | 1.0 [7.0] | 1.0 [4.0] | 0.552 |
Endoscopies | 1.0 [1.0] | 1.0 [1.0] | 1.0 [1.0] | 0.874 |
Surgery, n (%) | 8 (5.2) | 4 (6.7) | 4 (4.3) | 0.778 |
Hospitalizations | 0.0 [1.0] | 0.0 [2.0] | 0.0 [0.0] | 0.400 |
Emergency room visits | 0.0 [2.0] | 0.0 [3.0] | 0.0 [1.0] | 0.087 |
Financial charges ($) | 11309.00 [98577.50] | 51146.00 [182517.00] | 8120.50 [63850.25] | 0.001 |
SIBDQ, short inflammatory bowel disease questionnaire; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; IBD, inflammatory bowel disease; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; HBI, Harvey Bradshaw Index; UCAI, Ulcerative colitis activity index; IQR, interquartile range
Bolded p-values are statistically significant (p<0.05)
Immunomodulators include 6-mercaptopurine, azathioprine, methotrexate. Biologics include anti-tumor necrosis factor agents (infliximab, adalimumab, and certolizumab), anti-integrin therapy (vedolizumab and natalizumab). Biologics include all anti-tumor necrosis factor agents and anti-integrin therapies. Systemic antibiotics include all systemic antibiotic exposure, excluding vancomycin and fidaxomicin.
Those with prior CDI continued to have increased exposure to vancomycin (p<0.001) and other systemic antibiotics (p=0.02). Neither group had any exposure to fidaxomicin or fecal microbiota transplantation in the year after infection. All other medication exposures including biologics, systemic steroids, immunomodulators, and 5-aminosalicyclic acid agents did not differ between groups, although exposure to systemic steroids nearly met significance (p=0.07) (TABLE 4). In the year following infection, CDI patients continued to have more clinic visits (p=0.02), and telephone encounters (p=0.001). CDI patients also had significantly more financial healthcare charges in the year after infection (median $51,146.00 CDI vs. $8,120.50 controls, p=0.003). However, patient reported disease activity, QOL (p=0.08), and biomarkers of severity including ESR and CRP were not significantly different between the two groups. Other metrics of healthcare utilization including radiologic studies, endoscopy, surgery, hospitalizations, and ER visits were not significantly different (TABLE 4).
Year after infection, excluding patients with multiple infections
Of the 60 CDI patients who had follow up in the year after infection, there were 18 patients (30%) who had more than one CDI infection documented with molecular testing during their participation in the UPMC IBD research registry. We performed a subgroup analysis of the patients who followed up the year after infection and did not have multiple documented infections (n=136: 42 CDI, 94 controls). When patients who had more than one documented CDI were excluded, we observed that higher exposure to vancomycin in the CDI group remained significant (33.3% CDI vs. 11.7% controls, p=0.03) in the year after infection. Meanwhile, the exposure to all other classes of antibiotics was no longer significantly different (p=0.64) between the two groups.
After excluding those with multiple CDI, we observed that the number of clinic visits (median [IQR]: 2 [3] CDI vs. 1 [1] controls) between groups is no longer significantly different (p=0.18), but CDI patients continued have significantly more telephone encounters (median [IQR]: 5 [8] CDI vs. 3 [5.5] controls, p=0.04) in the year after infection. CDI patients also continued to have significantly more healthcare related charges (median [IQR]: $40,865.25 [$149,684.90] CDI vs. $7,775.08 [$74,816.48] controls, p=0.018), after excluding patients with multiple positive CDI molecular tests.
Discussion
In this propensity score matched analysis of CDI patients compared to controls, participants were matched on risk factors for Clostridium difficile in the year prior to infection. While groups did not differ at in any measured metrics at baseline, those who developed CDI in the following year demonstrated significantly increased biomarkers of inflammation (CRP and ESR), increased patient reported metrics of disease severity, increased medication exposure, decreased QOL, and significantly increased inpatient and outpatient healthcare utilization compared to controls. Interestingly, the increase in healthcare utilization and antibiotic exposure extended into the year after infection for patients with CDI. Differences in QOL just failed to reach statistical significance in the year after infection. This could be due to the statistically lower follow up in the control group in the year after infection. The poor follow up in the control group results in fewer QOL scores completed from patients who are likely feeling well. Overall, the findings suggest CDI has a lasting and measureable impact on IBD patients beyond the acute care period.
Previous research has shown that CDI in IBD is associated with systemic inflammation and disease activity, which we validated in this propensity matched cohort study.21,22 In addition to association with measures of disease severity, we observed that patients with CDI were more frequently prescribed and exposed to systemic steroids during the year of infection. This could be due to worsening of symptoms initially thought to be a flare of IBD and may have been attributable to infection, or CDI that precipitates an IBD flare. These data highlight the difficulty of diagnosis and importance of proper management of CDI in the setting of IBD, as both infection and disease flare present with similar symptoms of elevated inflammatory biomarkers and diarrhea.
A recent meta-analysis demonstrated that CDI is a significant risk factor for colectomy in patients with IBD.11 Our study failed to find differences in the proportion of patients requiring surgery in the year of infection and the year after. This could be due to the baseline matching, which selected for a severe disease cohort from the outset, placing both groups at higher risk for surgery than the general IBD population. This is supported by the fact that a third of patients had a history IBD related surgery prior to study enrollment, and around 15% of patients had surgery in the year prior to infection. In the year of infection approximately 9% of patients had an IBD related surgery, and this was similar between groups. Other reasons for the lack of surgical endpoints in this study includes the relatively small cohort of patients, and that we did not exclude patients who had prior surgery or colectomy.
In this study, we included participants who had multiple positive molecular tests for CDI. Recurrent infection is a significant and important feature of CDI in the setting of IBD that we hoped to characterize with longitudinal observational data. Additionally, many IBD patients are treated empirically without repeat molecular testing due to the high likelihood that persistent symptoms represent CDI recurrence in the setting of IBD. Similar to other studies, we observed approximately one-third of patients experience repeat infection confirmed with molecular testing.14 To ensure that patients with documented recurrent infection were not influencing the results of the statistically significant parameters, we repeated the analysis excluding this fraction of patients and found similar significant results in relation to increased vancomycin antibiotic exposure, increased telephone encounters possibly due to continuing empiric therapy, and increased financial charges which serves as an all-encompassing healthcare utilization metric. However, the differences in the number of clinic visits were no longer statistically significant.
This study was performed at a tertiary care center, and therefore may not be generalizable to patients in the community setting. However, the UPMC IBD Registry collects all data from the electronic medical record, which includes over 22 different hospitals and 500 clinics in the surrounding community. This analysis is restricted to only those patients who are enrolled in the UPMC IBD Research registry, and is therefore subject to participation bias. Given the strict inclusion criteria of requiring clinical follow-up in the year prior to infection, we may have missed valuable data on patients initially presenting to our tertiary care clinic for worsening disease that could be attributed to CDI, or were diagnosed with CDI on their first visit to the Digestive Disorders Center. We recognize that there is a testing bias, as only those initially tested for CDI due to clinical suspicion were included in our CDI cohort. Choice of methodology for testing was not standardized among providers and not captured in the registry data; therefore, we do not have detailed data regarding colonization as compared to infection. The size of our cohort is also relatively small, including only 66 patients with CDI; therefore, some of our measured outcomes in the year after infection may have lacked statistical power due to low sample size. Despite the small size of our study cohort, we were able to observe highly significant differences between the matched groups in the year of infection and the year after infection.
This is the first propensity score matched analysis of a CDI cohort in the setting of IBD. This approach helps to alleviate many of the caveats associated with a random sample, as those who have a history of infection may be inherently different due to risk factors associated with infection. These data are prospectively derived from the electronic medical record and represents real world care of IBD patients, as it is not collected under the standardized setting of a clinical trial. The analysis of real world data brings us closer to understanding typical care patterns and the true IBD patient experience of CDI.
In conclusion, CDI negatively impacts the clinical course of IBD in the year of infection, and also has lasting and measurable effects. CDI results in increased IBD activity, elevated biomarkers of inflammation, poor health- related QOL, and increased healthcare utilization during the year of infection, some of which extends into the year after infection. Given the dramatic impact of CDI on IBD, future studies evaluating treatment strategies of CDI in IBD are needed.
Supplementary Material
Acknowledgments
We would like to acknowledge all of the participants of the inflammatory bowel disease research registry, and all of the faculty and staff in the Digestive Disorders Clinic who continue to make this work possible.
FINANCIAL SUPPORT AND POTENTIAL CONFLICT OF INTEREST
Source of support: This work was funded by an investigator initiated research award granted to David Binion, MD from Merck & Co., Inc., (Study #GMA-FDX-14-18). Alyce Anderson (1TL1TR001858-01, PI: Kapoor) is supported by a NIH training grant. Ioannis Koutroubakis reports support by a sabbatical salary of Medical Faculty University of Crete, Greece. David G. Binion and Michael A. Dunn reports support from Grant W81XWH-11-2-0133 from the U.S. Army Medical Research and Materiel Command.
Research reported in this publication was supported by the National Center For Advancing Translational Sciences of the National Institutes of Health under Award Number TL1TR001858. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Abbreviations
- CDI
Clostridium difficile infection
- CRP
C-reactive protein
- ER
emergency room
- ESR
erythrocyte sedimentation rate
- HBI
Harvey-Bradshaw index
- IBD
inflammatory bowel disease
- CD
Crohn’s disease
- QOL
quality of life
- SIBDQ
short inflammatory bowel disease questionnaire
- UC
ulcerative colitis
- UCAI
ulcerative colitis activity index
Footnotes
The work was performed at the University of Pittsburgh Medical Center.
AUTHORSHIP ROLE
Alyce Anderson, BS – Drafting of manuscript, figure design, table creation, data collection, statistical analysis, and manuscript revisions.
Benjamin Click, MD – Drafting of manuscript, data collection, critical review of the manuscript.
Claudia Ramos-Rivers, MD – Data collection, data preparation, critical review of the manuscript.
Debbie Cheng, MD – Data collection, data preparation, critical review of the manuscript.
Dmitriy Babichenko – Data collection, data preparation, critical review of the manuscript.
Ioannis E. Koutroubakis, MD, PhD - Data collection, critical review of the manuscript.
Jana G. Hashash, MD - Data collection, critical review of the manuscript.
Marc Schwartz, MD - Data collection, critical review of the manuscript.
Jason Swoger, MD - Data collection, critical review of the manuscript.
Arthur M Barrie, III, MD, PhD - Data collection, critical review of the manuscript.
Michael A. Dunn, MD - Data collection, critical review of the manuscript.
Miguel Regueiro, MD - Data collection, critical review of the manuscript.
David Binion, MD – Critical review of the manuscript, collection of data, obtained funding, advisor to primary author and project supervision.
All authors read and approved the final manuscript.
References
- 1.Furuya-Kanamori L, Stone JC, Clark J, et al. Comorbidities, Exposure to Medications, and the Risk of Community-Acquired Clostridium difficile Infection: a systematic review and meta-analysis. Infect Control Hosp Epidemiol. 2015;36:132–141. doi: 10.1017/ice.2014.39. [DOI] [PubMed] [Google Scholar]
- 2.Ananthakrishnan AN. Clostridium difficile infection: epidemiology, risk factors and management. Nat Rev Gastroenterol Hepatol. 2011;8:17–26. doi: 10.1038/nrgastro.2010.190. [DOI] [PubMed] [Google Scholar]
- 3.Stoica O, Trifan A, Cojocariu C, et al. Incidence and risk factors of Clostridium difficile infection in patients with inflammatory bowel disease. Rev Med Chir Soc Med Nat Iasi. 2015;119:81–86. [PubMed] [Google Scholar]
- 4.Ramos-Martínez A, Ortiz-Balbuena J, Curto-García I, et al. Risk factors for Clostridium difficile diarrhea in patients with inflammatory bowel disease. Rev Esp Enferm Dig. 2015;107:4–8. [PubMed] [Google Scholar]
- 5.Ananthakrishnan AN, Cagan A, Gainer VS, et al. Higher plasma vitamin D is associated with reduced risk of Clostridium difficile infection in patients with inflammatory bowel diseases. Aliment Pharmacol Ther. 2014;39:1136–1142. doi: 10.1111/apt.12706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hourigan SK, Oliva-Hemker M, Hutfless S. The prevalence of Clostridium difficile infection in pediatric and adult patients with inflammatory bowel disease. Dig Dis Sci. 2014;59:2222–2227. doi: 10.1007/s10620-014-3169-4. [DOI] [PubMed] [Google Scholar]
- 7.Rao K, Higgins PD. Epidemiology, Diagnosis, and Management of Clostridium difficile Infection in Patients with Inflammatory Bowel Disease. Inflamm Bowel Dis. 2016;22:1744–1754. doi: 10.1097/MIB.0000000000000793. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Issa M, Vijayapal A, Graham MB, et al. Impact of Clostridium difficile on inflammatory bowel disease. Clin Gastroenterol Hepatol. 2007;5:345–351. doi: 10.1016/j.cgh.2006.12.028. [DOI] [PubMed] [Google Scholar]
- 9.Ananthakrishnan AN, McGinley EL, Binion DG. Excess hospitalisation burden associated with Clostridium difficile in patients with inflammatory bowel disease. Gut. 2008;57:205–210. doi: 10.1136/gut.2007.128231. [DOI] [PubMed] [Google Scholar]
- 10.Ananthakrishnan AN, McGinley EL, Saeian K, et al. Temporal trends in disease outcomes related to Clostridium difficile infection in patients with inflammatory bowel disease. Inflamm Bowel Dis. 2011;17:976–983. doi: 10.1002/ibd.21457. [DOI] [PubMed] [Google Scholar]
- 11.Chen Y, Furuya-Kanamori L, Doi SA, et al. Clostridium difficile Infection and Risk of Colectomy in Patients with Inflammatory Bowel Disease: A Bias-adjusted Meta-analysis. Inflamm Bowel Dis. 2017;23:200–207. doi: 10.1097/MIB.0000000000000998. [DOI] [PubMed] [Google Scholar]
- 12.Trifan A, Stanciu C, Stoica O, et al. Impact of Clostridium difficile infection on inflammatory bowel disease outcome: a review. World J Gastroenterol. 2014;20:11736–11742. doi: 10.3748/wjg.v20.i33.11736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Anderson AJ, Click B, Ramos-Rivers C, et al. Development of an Inflammatory Bowel Disease Research Registry Derived from Observational Electronic Health Record Data for Comprehensive Clinical Phenotyping. Dig Dis Sci. 2016;61:3236–3245. doi: 10.1007/s10620-016-4278-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Razik R, Rumman A, Bahreini Z, et al. Recurrence of Clostridium difficile Infection in Patients with Inflammatory Bowel Disease: The RECIDIVISM Study. Am J Gastroenterol. 2016;111:1141–1146. doi: 10.1038/ajg.2016.187. [DOI] [PubMed] [Google Scholar]
- 15.Irvine EJ, Zhou Q, Thompson AK. The Short Inflammatory Bowel Disease Questionnaire: a quality of life instrument for community physicians managing inflammatory bowel disease. CCRPT Investigators. Canadian Crohn’s Relapse Prevention Trial. Am J Gastroenterol. 1996;91:1571–1578. [PubMed] [Google Scholar]
- 16.Harvey RF, Bradshaw JM. A simple index of Crohn’s-disease activity. The Lancet. 1980;1:514. doi: 10.1016/s0140-6736(80)92767-1. [DOI] [PubMed] [Google Scholar]
- 17.Kozarek RA, Patterson DJ, Gelfand MD, et al. Methotrexate induces clinical and histologic remission in patients with refractory inflammatory bowel disease. Ann Intern Med. 1989;110:353–356. doi: 10.7326/0003-4819-110-5-353. [DOI] [PubMed] [Google Scholar]
- 18.Silverberg MS, Satsangi J, Ahmad T, et al. Toward an integrated clinical, molecular and serological classification of inflammatory bowel disease: report of a Working Party of the 2005 Montreal World Congress of Gastroenterology. Can J Gastroenterol. 2005;19(Suppl A):5A–36A. doi: 10.1155/2005/269076. [DOI] [PubMed] [Google Scholar]
- 19.Lee J, Little TD. A practical guide to propensity score analysis for applied clinical research. Behav Res Ther. 2017 doi: 10.1016/j.brat.2017.01.005. [DOI] [PubMed] [Google Scholar]
- 20.Rosenbaum PR, Rubin DB. The Central Role of the Propensity Score in Observational Studies for Causal Effects. Biometrika. 1983;70:41. [Google Scholar]
- 21.Rao K, Erb-Downward JR, Walk ST, et al. The systemic inflammatory response to Clostridium difficile infection. PLoS ONE. 2014;9:e92578. doi: 10.1371/journal.pone.0092578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Czepiel J, Biesiada G, Brzozowski T, et al. The role of local and systemic cytokines in patients infected with Clostridium difficile. J Physiol Pharmacol. 2014;65:695–703. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.