Discussion

To understand diagnosis and testing, this review distinguished between CDI vs. C. difficile colonization (which is the main reason for positive NAAT tests in the absence of disease).

For this topic, initial estimates were made from the literature and a modified Delphi panel then reviewed these, made changes and reached consensus on guidance. (Baghdadi et al. under submission)

Incidence and risk factors:

We found no studies describing specific risk by healthcare setting. Incidence was estimated as follows. 50% of CDI is healthcare associated. Only 7.3% of the approximately 320 million Americans are hospitalized. Therefore, CDI incidence in one year among people with at least one hospitalization is approximately 1.0%, nursing home patients were estimated to have incidence of 2.0% and CDI incidence over one year among people with no hospitalizations 0.05%.[3,4]

The primary risk factor for CDI is antibiotics. The impact of these was found to be an increase in odds by 5-7 in a cohort study.[5] The impact could be as high as 10, based on an estimate in outpatients that CDI prevalence is 0.23% in those receiving antibiotics vs. 0.023% in those not receiving antibiotics.[6] Based on a modified Delphi panel, antibiotic risk was defined by antibiotics as per table. Use of PPIs or prior hospitalization or having cancer were also risk factors with estimates coming from a cohort study.[5]

Symptoms/Syndromes:

Diarrhea is the primary symptom of CDI. Almost every patient being considered for CDI have loose stools (if not, testing will generally not be performed by microbiology). However, the presence of diarrhea new or unexplained severe diarrhea, >10 BMs/ goes beyond this minimum standard. Expert opinion was that presence of diarrhea confers a LR of 5 (Baghdadi et al. under submission).

Other general indicators of infection are associated with CDI including severely elevated WBC. Notably, an expert panel believed fever alone was not predictive of CDI (LR=1).

Test sensitivity & specificity:

Assays to detect C. difficile were compared to clinical diagnosis of CDI that required presence of diarrhea to best represent clinical sensitivity and clinical specificity. The best study to identify these clinical metrics was from 2011.[7] There was some variability between assay companies, but a NAAT assay was identified to have ~99% sensitivity (90%-100%) and ~89% specificity (78-91%). Similar numbers were identified by a previous study although with some bias likely inflating the specificity of PCR.[8] ESCMID guidance provided an exhaustive review of many tests providing estimates within these ranges.[9]

A toxin EIA was ~85% sensitive (65%-98%) and 94% specific (85-97%).

This is consistent with IDSA guideline recommendations. And ESCMID guidance.[9,10]