Clostridium difficile Infection: What is the True Incidence?

Background: Clostridium difficile infection (CDI) is the leading cause of health-care associated diarrhea. We note a discrepancy between CDI incidence as observed in longitudinal, culture-based studies and that reported by observational studies based on routine clinical and laboratory testing.

Objective: To explore the nature and potential causes of the discordance between symptomatic, hospital-acquired CDI incidence as observed in longitudinal, culture-based surveillance studies and that recorded by hospital-based observational studies based on routine clinical and laboratory testing.

Methods: We performed a systematic literature review in the context of developing a computer simulation model of nosocomial CDI. We searched the Medline database for all relevant articles from 1985 – October 2010, excluding case reports and series. We also reviewed unpublished data from 120 United States Veterans' Affairs sites. Case definitions, diagnostic methods, and follow-up periods were recorded.

Results: Longitudinal culture-based studies, dating mainly from the 1990s, cite symptomatic CDI incidences among hospitalized patients of 1.7 – 10.1%, while hospital-based studies from the same period document significantly lower incidences of 0.03% – 1.5% (see table). Clinical diagnosis of CDI was defined similarly in both groups: diarrhea with a positive laboratory assay or with an imaging study suggestive of CDI. Longitudinal studies were not hospital-wide and enrolled patients who had undergone at least one surveillance test, thus potentially excluding lower-risk populations. They used highly sensitive methods for CDI laboratory diagnosis (stool culture with cytotoxin assay), while individual hospitals' assays varied but were less sensitive overall. In the hospital-based studies, diagnostic testing was clinically directed; the fraction of patients with diarrhea who were not tested for CDI was not measured. Follow-up periods differed; prospective studies generally followed patients beyond their discharge date, while hospital data included only the period of hospitalization.

Conclusions: The higher CDI incidence cited in longitudinal studies may reflect the use of more sensitive diagnostic methods. Incidence may be overestimated, however, given that these studies potentially excluded lower-risk populations and may have harbored a bias toward C. difficile as the major causative agent of nosocomial diarrhea, yielding false positives when merely colonized patients developed diarrhea as a manifestation of other disease processes. On the other hand, hospital-based studies may be underestimating CDI rates not only as a result of less sensitive diagnostic methods but also because clinically-directed diagnostic tests may be incomplete. More sensitive diagnostic tests are needed; in the meantime, simulation modeling may provide insight into the impact of these factors as well as the causal relationship between CDI and symptoms.