Clostridioides difficile infection (CDI) remains a significant public health challenge in Manitoba and elsewhere in Canada.Reference Lambert, Dyck, Thompson and Hammond1,Reference Freeman, Bauer and Baines2 Rates of reported CDI have been increasing,Reference Lessa, Gould and McDonald3,Reference Leffler and Lamont4 and hospital and institutional outbreaks caused by CDI often cause significant disruption to the delivery of health services.Reference Freeman, Bauer and Baines2 Control of CDI is hampered by a lack of harmonized information on disease burden and time trends.Reference Dubberke, Carling and Carrico5 The interpretation of surveillance data is difficult without a clear understanding of the intensity of testing. Analyses based solely on reported cases might lead to erroneous conclusions and decisions if the prevalence and predictors of testing are poorly understood.Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox6,Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7
Although CDI has been reportable to public health authorities in Manitoba since April 2005, little is known about testing rates. A positive toxin test result should only be reported as CDI in Manitoba if it is accompanied by diarrhea. The lack of information about testing rates does not allow for an unbiased comparison of surveillance data with other jurisdictions because testing protocols differ within Canada.Reference Wilkinson, Gravel and Taylor8 We assessed the probability of testing and the toxin positivity rate for Clostridioides difficile in Manitoba hospitals.
Methods
Manitoba Health is the publicly funded health insurance agency providing comprehensive health insurance, including coverage for hospital and outpatient physician services, to the province’s 1.3 million residents. Coverage is universal, with no eligibility distinction based on age or income, and participation rates are very high (>99%). Insured services include hospital and diagnostic services.
To measure the intensity of screening, we identified all inpatient tests for C. difficile for all Manitoba hospitals between June 1, 2016, and November 30, 2018, from the Shared Health Laboratory Information System (LIS; Delphic, Sysmex, New Zealand), including the collection dates, test results, and patient demographics. Since 2002, all in-hospital C. difficile testing in Manitoba has been done by a single provincial entity; this entity is fully funded by Manitoba Health and fully accountable to Manitoba Health and is currently named Shared Health, Diagnostic Services. All requested inpatient C. difficile tests in Manitoba and all test results are registered in the LIS. During the study period, stool specimens were screened using the glutamate dehydrogenase (GDH) membrane immunoassay using the C. Diff Quik Chek test (Techlab, Blacksburg, VA) following the manufacturer’s recommendations. Specimens positive for GDH antigen were then subjected to a C. difficile toxin single nucleic acid amplification test (Illumigene C. diff; Meridian Bioscience, Cincinnati, OH).
We attributed all tests for a patient to the same episode of care if it was within 56 days of another test (in line with the provincial protocol and literature definitions of recurrent CDIReference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7). We included 1 test per episode of care in our analysis, the first positive test for positive episodes (16% of all positives were recurrent) and the first negative test for negative episodes. We calculated testing and incidence (of positive toxin) rates using the number of patient bed days from Manitoba hospitals9 as the denominator.
Because its proportion of elderly patients is relatively large, we retrospectively reviewed charts of 400 randomly selected patients discharged from the adult medicine ward at Grace Hospital in 2016, a 247-bed urban hospital in the main urban center of Winnipeg, Manitoba. A trained registered nurse reviewed the charts and extracted using a custom-designed Epi Info data form (Centers for Disease Control and Prevention, Atlanta, GA) to extract demographic information (age, gender, area of residence), diarrhea status at admission (as a symptom, regardless of whether it was the main reason for admission), diarrhea onset during hospitalization, collection of stool specimens, positive tests for C. difficile toxin, and other characteristics of the hospitalization for each patient. We summarized the extracted data using descriptive statistics.
We used Stata version 14 software (StataCorp, College Station, Texas) for all analyses. This study was approved by the University of Manitoba Research Ethics Board [no. HS 19565 (H2016:111)] and the Winnipeg West Integrated Health & Social Services Senior Leadership team.
Results
During the study period, 8,544 patients were tested. Among them, 11.4% (974) tested positive for C. difficile toxin (Table 1), and 37% (3,182) were tested ≤3 days from admission. Men tested positive at a slightly higher percentage than women, and the percent positive was ~25% higher for patients aged >70 years than for younger patients. Test positivity trended slightly downward during the study period, although different months were included for each calendar year.
Table 1. Number and Crude Rate of C. difficile Toxin Testing and Positivity According to Gender, Age Group, and Year
Note. CI, confidence interval.
a Per 10,000 patient bed days.
b Data only available June–December in 2016 and January–November in 2018.
The overall testing rate was 30 (95% confidence interval [CI], 30–31) per 10,000 patient bed days (Table 1). Patients aged >70 years were tested approximately two-thirds as often as younger patients, and men were tested at approximately the same rate as women. Testing rates were stable (on an annual basis) during the study period.
The incidence rate of C. difficile toxin positivity was 3.5 per 10,000 patient bed days (95% CI, 3.3–3.7) (Table 1). The incidence rate for patients aged >70 years was 20% lower than the rate of younger patients, and men had a slightly higher rate than women. The incidence was stable during the study period, although it was slightly higher for 2016, the first year of the study that covered only June through December. CDI incidence rates vary by month in Manitoba, the rates are highest from August through October and lowest from January through March.Reference Lambert, Dyck, Thompson and Hammond1
The charts of 400 hospital patients were reviewed retrospectively (representing 402 hospitalizations). Among them, 10 patients (2.5%) were admitted with diarrhea (7 males and 3 females). The vast majority of patients (80%) resided in Winnipeg, the location of the hospital where the chart review was conducted. Stool specimens were collected for 27 (6.8%) patients, 18 of whom had diarrhea: 5 were admitted with diarrhea and 13 developed diarrhea during their hospitalization. Stools were collected for 5 of the 10 patients admitted with diarrhea and 13 of the 20 patients who developed diarrhea during their hospitalization. One patient tested positive for C. difficile toxin.
Discussion
For every 10,000 patient bed days, 30 patients were tested for C. difficile toxin and 3.5 tested positive; 11% of tested patients tested positive. Hospital-associated CDI decreased in Canada between 2009 and 2017 (from 5.9 to 3.9 per 10,000 patient bed days, in line with our results for 2016–2018), although those authors observed large regional differences.Reference Katz, Golding and Choi10,11
In our chart review, 18 patients of 400 had diarrhea and had stool samples collected, and 1 patient tested positive for C. difficile toxin (1 of 18 is 6%). This finding is in line with a 6,000-patient UK study in which 90% of samples tested negative for C. difficile toxin.Reference Planche, Davies and Coen12 We found an 11% positivity rate in Manitoba. In a large European study, 58.4% of diarrhea samples were tested for CDI,Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7 which is close to the prevalence of testing for the diarrhea patients in our chart review. Almost half of samples (40%) were collected for patients without diarrhea; the laboratory rejects formed stools for C. difficile testing. Our chart review was not set up to investigate other reasons for stool sample collection or to collect diarrhea characteristics (eg, duration or timing with respect to stool collection), so we were unable to assess the appropriateness of testing these samples.
We found an overall positivity rate of 11.4%. The positivity rate was reported to be 18% in Canadian hospitals in 1997,Reference Miller, Hyland, Ofner-Agostini, Gourdeau and Ishak13 although CDI practices (both in terms of testing protocols for CDI and volume of testing) were highly variable throughout CanadaReference Wilkinson, Gravel and Taylor8 and have changed since then. In a more recent European study, positivity rates differed by country. These rates were 9.6% in France, 14.9% in Italy, and 3.0% in the United Kingdom based on the number of positive results among tested patients, as reported by Davies et al.Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7
When we combined the 60% testing rate for patients with diarrhea and the 11% positivity rate, we noted that for every 100 patients with diarrhea, ~60 patients would be tested and ~7 patients would be positive. The 2-stage testing algorithm used in Manitoba leads to lower reported CDI rates than jurisdictions where standalone assays are used.Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7,Reference Planche, Davies and Coen12 Rates of C. difficile testing and positivity, as well as positivity rates in Manitoba, resemble those in Italy and France more than those in the United Kingdom. A key difference could be earlier testing (within 48 hours of admission), which is more prevalent in the United Kingdom.Reference Davies, Davis, Barbut, Eckert, Petrosillo and Wilcox7
A major strength of this study is the availability of a high-quality, province-wide system for all hospital testing for CDI. These test protocols combine molecular and toxin immunoassay testing because relying on molecular tests alone would likely result in overdiagnosis of CDI.Reference Planche, Davies and Coen12,Reference Polage, Gyorke and Kennedy14
This study has several limitations. Two-step algorithms based on a GDH screen have a relatively low sensitivity in immunocompromised patients,Reference Ashraf, Rahmati, Bender, Nanda and She15 which may have resulted in underdiagnosis of CDI in this population. We were unable to determine rates for all community-tested CDI because we did not have data from a second laboratory that also tests these samples.
In 2005–2006, half of CDI infections were healthcare-acquired in Manitoba, an additional 15%–20% were hospital diagnosed and were community acquired or from an indeterminate origin.Reference Lambert, Dyck, Thompson and Hammond1 A more recent Manitoba study found similar proportions by onset for patients with inflammatory bowel disease (IBD) and their matches.Reference Singh, Nugent, Yu, Lix, Targownik and Bernstein16 We may have underestimated the postdischarge burden of community-onset, healthcare-associated CDI because we lacked information on testing after hospitalization. In 2011, two-thirds of CDIs in the United States were healthcare associated, although only a quarter had their onset during hospitalization.Reference Lessa, Mu and Bamberg17
We lacked some important clinical information. We could not ascertain the province-wide prevalence of C. difficile testing for hospital patients with diarrhea, and we had no information about comorbidity (which may be particularly relevant for older patients). Patients with IBD have a higher risk of CDIReference Singh, Nugent, Yu, Lix, Targownik and Bernstein16 and might undergo more intense screening. We also lacked information on prior healthcare encounters to ascertain the onset of CDI. Because of these limitations, we were unable to assess the predictors and appropriateness of testing, which did not allow us to estimate the full burden of CDI in Manitoba. Based on our chart review, testing rates for C. difficile appeared to be low; only 50% of hospitalized patients with diarrhea were tested. This low rate of testing could explain some of the variability in CDI incidence rates. A better understanding of testing is important because C. difficile is the most common cause of hospital-acquired diarrhea and is treatable, and missing an early diagnosis can contribute to nosocomial spread.
In conclusion, increasing age was related to a lower rate (per patient bed day) of toxin testing and positivity. C. difficile incidence rates were in line with the literature, although the context of testing is essential to comparing data between jurisdictions. Although laboratory protocols have been standardized, initiation of testing is not standardized. The resulting variability in testing rates limits our understanding of CDI and ways to control it.
Acknowledgments
We thank Alun Carter for his expertise and assistance with extraction of the laboratory data. We thank Barbara Fletcher for her expertise and assistance in conducting the chart reviews. We acknowledge Shared Health, as well as Winnipeg West Integrated Health and Social Services, which includes Grace Hospital and ACCESS Centre, for providing data for this study. The results and conclusions are those of the authors and no official endorsement by the data providers or Manitoba Health is intended or should be inferred. The sponsor had no role in the design or conduct of the study, including but not limited to, data identification, collection, management, analysis, and interpretation, or preparation, review, or approval of the final results. The opinions presented in the report do not necessarily reflect those of the funders.
Financial support
This work was supported by an unrestricted grant from Sanofi Pasteur Global Epidemiology to the International Centre for Infectious Disease (Winnipeg, Canada). S.M.M.’s work is supported, in part, by funding from the Canada Research Chair Program.
Conflicts of interest
S.M.M. has received unrestricted research grants from GlaxoSmithKline, Merck, Sanofi Pasteur, Pfizer, and Roche-Assurex for unrelated studies. S.M.M. and G.W.H. have received fees as advisory board members for Sanofi Pasteur. None of the other authors has any conflicts of interest to disclose.
