RESUMO
BACKGROUND: Most Clostridium difficile infection (CDI) surveillance programs neither specify the diagnostic method to be used nor stratify rates accordingly. We assessed the difference in healthcare-associated CDI (HA-CDI) incidence and complication rates obtained by 2 validated diagnostic methods. METHODS: This was a prospective cohort study of patients for whom a C. difficile test was ordered between 1 August 2010 and 31 July 2011. All specimens were tested in parallel by a commercial polymerase chain reaction (PCR) assay targeting toxin B gene tcdB, and a 3-step algorithm detecting glutamate dehydrogenase and toxins A and B by enzyme immunoassay and cell culture cytotoxicity assay (EIA/CCA). CDI incidence rate ratios were calculated using univariate Poisson regression. RESULTS: A total of 1321 stool samples were tested during a period totaling 95 750 patient-days. Eighty-five HA-CDI cases were detected by PCR and 56 cases by EIA/CCA (P = .01). The overall incidence rate was 8.9 per 10 000 patient-days (95% confidence interval [CI], 7.1-10.9) by PCR and 5.8 per 10 000 patient-days (95% CI, 4.4-7.4) by EIA/CCA (P = .01). The incidence rate ratio comparing PCR and EIA/CCA was 1.52 (95% CI, 1.08-2.13; P = .015). Overall complication rate was 27% (23/85) when CDI was diagnosed by PCR and 39% (22/56) by EIA/CCA (P = .16). Cases detected by PCR only were less likely to develop a complication of CDI compared with cases detected by both PCR and EIA/CCA (3% vs 39%, respectively; P < .001). CONCLUSIONS: Performing PCR instead of EIA/CCA is associated with a >50% increase in the CDI incidence rate. Standardization of diagnostic methods may be indicated to improve interhospital comparison.