Risk Factors and Molecular Analysis of Clostridium difficile Infection in Hospitalized Adults

Background: Increased disease severity has been associated with hypervirulent strains of C. difficile (i.e NAP1) and community-associated disease has emerged.  The purpose of this study was to evaluate if risk factors associated with CDI have changed since the NAP1 strain has become prevalent.

Objective: The primary objective of this study was to identify risk factors associated with CDI in the era of NAP1 CDI in 2 Denver hospitals. Our secondary objective was to identify the presence of NAP1 CDI in our hospitalized patients.

Methods: A retrospective chart review of adult inpatients with a positive C. difficile toxin enzyme immunoassay or culture and signs and symptoms of CDI was performed at a city-county hospital and a veterans administration hospital. The study spanned a 1-year period from 11/1/07 -10/31/2008.  Cases were matched 2:1 to controls for date of admission and length of time at risk for CDI. Potential risk factors were recorded for both cases and controls.  Cases who acquired CDI >48H after admission to hospital were classified as hospital onset-health care facility associated (HO-HCFA); Cases who acquired CDI while in the outpatient setting but within 4 weeks of hospitalization were classified as community onset-healthcare facility associated (CO-HCFA), Cases who acquired CDI while in the outpatient setting after greater than 12 weeks from hospitalization were classified as community associated (CA); all other cases were classified as indeterminate. A subset of samples was sent for restriction endonuclease analysis (REA).

Results: Of 131 evaluable cases, 68.7% were HO-HCFA, 16.8% were CO-HCFA, 10.7% were CA, and 3.8% were indeterminate. When compared to controls, in the HO-HCFA group, risk factors such as older age, antibiotic exposure, total number of antibiotics, total days of antibiotic therapy, beta-lactam and fluoroquinolone therapy, immunosupression, proton pump inhibitor use, enteral feeding, lower albumin, intensive care stay, non-surgical invasive procedures, and transfer from other health care institutions were significantly associated with CDI, while surgical procedures, admission through the emergency department, and the number of room transfers during an admission were not. In the CA group, the only risk factor associated with CDI was older age. In the REA analysis of 20 strains from the veterans hospital, 35% were found to be NAP1 while 15% were found to belong to other epidemic groups.

Conclusions: NAP1 is prevalent in the Denver area and known risk factors for HCFA CDI should continue to be targeted in this new era of hypervirulent NAP1. Additionally, patients transferred from other health care institutions could be another population to target for prevention of CDI. Clindamycin use was not a significant risk for CDI in our patient poulations.  Patients with CA CDI do not share the same risk factors as HCFA CDI, suggesting that other unidentified risk factors may play a significant role.