Identification of a Pseudo-Outbreak of C. difficile Infection (CDI): Importance of Evaluating Test Performance

Background:   We identified an increased rate of CDI at our institution in 6/06.  Despite introduction of rigorous infection control measures, the rate of CDI remained higher than baseline. In 11/06, a new toxin A/B EIA test was introduced in our microbiology laboratory. 

Objective:   To determine whether the elevated rate of CDI was partially related to test performance and to identify factors associated with false (+) toxin results.

Methods:   Surveillance for healthcare-associated (HA) CDI was performed continuously during 2008 and 2009.  The case definition for HA-CDI during the outbreak was the presence of diarrhea and a positive C. difficile test ocurring ≥48 hours after admission or ≤48 hours after discharge.  From 11/18/08 - 12/9/08, stool samples submitted for C. difficile testing were evaluated using Remel ProSpecT® toxin A/B EIA, glutamate dehydrogenase (GDH) antigen test and cytotoxicity assay.  A false (+) toxin EIA was defined as a (+) EIA and (-) cytotoxicity assay.   Retrospective medical record review was conducted of all patients.  Infection control interventions included contact precautions for CDI cases, terminal bleach cleaning of all rooms, increased hand hygiene, multidisciplinary education and physician led antimicrobial stewardship.

Results:   341 stool samples from 328 patients were tested using 3 assays.  Using the cytotoxicity assay as a confirmatory test, the toxin A/B EIA yielded a false (+) rate of 12% (n=42).  EIA sensitivity was 79%, specificity 86%, NPV 98%, and PPV 35%.  Patients with a false (+) toxin A/B EIA were more likely to reside on a specific medical ward (p≤0.0006), were more likely to be on antifungal therapy (p≤0.02), to have a higher Charlson Comorbidity Index (p≤0.005), diabetes mellitus (p≤0.01), a history of solid organ transplant (p≤0.0011), to be on immunosuppressant medications/chemotherapy (p≤0.008) or parenteral nutrition (p≤0.02). 

During 2008 there were 575 cases of HA-CDI resulting in a rate of 2.26 cases/1000 pt-days.  When the toxin A/B EIA test was supplemented with a GDH test in 1/09 (and later PCR), the rate of HA-CDI dropped to 0.99 cases/1000 pt-days.  The reduction in rate was significant (P <0.0001).  The added confirmation step increased PPV to 89.6% and maintained a high NPV of 97.2%.

Conclusions:   The increased rate of HA-CDI in our institution was partially explained by the use of an EIA test with poor positive predictive value.  Test performance must be taken into account when investigating increased rates of infections which do not respond to infection control interventions.

Table 1.  Results of GDH Antigen, Toxin A/B EIA and Cytotoxicity Tests

GDH Antigen	Toxin A/B	Cytotoxicity	Frequency	Percent
+	+	+	23	6.7%
+	-	+	6	1.8%
+	-	-	21	6.2%
-	+	-	42	12.3%
-	-	-	249	73.0%