Methods: This was a retrospective data analysis of admitted patients treated at Veteran Health Administration (VHA) hospitals within the US from 2006 to 2016 using the VHA’s national Corporate Data Warehouse (CDW). All patients with stool testing for C. difficile were identified via lab codes associated with C. difficile. CDI is defined as any stool positive laboratory-identified (LabID) event for C. difficile via PCR, Toxin, GDH + (Toxin or PCR), or culture. Hospital-onset healthcare facility-associated (HO-HCFA) CDI is defined as a positive LabID event collected after 48 hours from admission. Incidence is reported as cases per 100,000 admission-days. Recurrent CDI episodes were excluded from incidence analysis. Data was extracted using SQL management studio and analyzed in Excel and JMP.
Results: A total of 389,512 patients were tested for C. difficile. Overall incidence of CDI increased from 2006 (67.6) to 2016 (127.7). This rise in total CDI incidence correlates positively with rise of PCR (p<0.0001) and 30-days CDI mortality (p<0.0001). In July 2012, VHA implemented reporting of HO-HCFA CDI. Incidence of HO-HCFA CDI and 30-day-CDI mortality increased from 2006 (45.6 and 12.3) to 2013 (69.2 and 17.1) with the rise of PCR (p<0.001) but decreased from 2013 (69.2 and 17.1) to 2016 (59.9 and 14.1) after implementation of HO-HCFA reporting (p=0.0058 and p=0.0068). The median time to testing has also been decreasing from 2006 (78.5 hours) to 2016 (45.5 hours). Amongst all patients with stool positive C. difficile LabID event, the frequency of ICD-9/10 discharge diagnosis code for CDI was 83.3%.
Conclusion: The incidence of CDI increased significantly as the use of PCR rose within the VHA. Increased incidence of CDI had a significant impact on mortality. Reporting of HO-HCFA CDI led to a downward trend in the incidence of HO-HCFA CDI and 30-days CDI mortality. Whether this is a true decrease or an improvement in testing programs is unclear as the time to C. difficile testing also declined over the study period. ICD-9/10 discharge diagnosis codes are not representative of all cases of CDI.
J. Sellick, None
K. Mergenhagen, None