Predictive Factors of Clostridium difficile Infection in Hospital Inpatients

Background: Diagnostic testing for Clostridium difficile infection (CDI) has been revolutionized by the advent of a highly sensitive PCR assay; however, it is unable to distinguish between colonization and infection. Consequently, a positive result in the setting of a low pre-test probability may represent a false positive. Since diarrhea is common in hospitalized patients, this could have major implications for testing and treatment. The objective of our study was to identify clinical and laboratory findings that correlated with patients who had positive tests and were treated for CDI, in order to answer: “does this medical inpatient with diarrhea have C. difficile infection?”

Methods: We conducted a retrospective cohort study on the medical Clinical Teaching Units (CTU) of the Royal Victoria Hospital (Montreal, Canada). Patients were included if they had a C. difficile PCR on the CTU between January 2014 and September 2015 and their admission diagnosis was not C. difficile. CDI was defined as a patient with a positive toxin assay who received a full course of treatment. Clinical and laboratory data were extracted from hospital records. Independent predictors of CDI were determined by logistic regression.

Results: Data were available on 319 patients, of whom 274 tested negative (86%). Of the 45 patients who tested positive, 43 (95.6%) received treatment. A number of factors were independently associated with CDI, as shown in the table. The area under the receiver-operator curve (c-statistic) for the model was 0.78. An active laxative prescription was unhelpful in ruling out CDI.

Conclusion: Various clinical factors in our cohort seem promising to differentiate between those medical inpatients with CDI and other causes of diarrhea. A preliminary clinical prediction rule derived from this data (http://goo.gl/xgDao9) will need to be further refined and validated in a larger cohort.

Table – Factors independently associated with CDI in tested patients

Factor	Odds ratio	95% Confidence		P-value
Dialysis	10.0	2.7	37.1	0.001
Peripheral Vascular Disease	3.6	1.1	11.4	0.028
Warrants Empiric Therapy	3.6	1.5	8.3	0.004
Atrial Fibrillation	2.8	1.3	5.9	0.009
Not on Antibiotics (but exposed ≤ 30 days)	2.7	1.3	5.6	0.007
Previous CDI	2.3	1.0	5.2	0.039
WBC Count ≤ 10	0.5	0.2	1.0	0.042