Repeat EIA Testing for Clostridium difficile Infection (CDI) Diagnosis is Not Reliable

Repeat EIA Testing for Clostridium difficile Infection (CDI) Diagnosis is Not Reliable

Background:   CDI remains an important cause of healthcare-associated (HA) disease.  Enzyme immunoassays (EIA) are the most widely used tests for detection of toxin A or A and B in the US.  To overcome the insensitivity of EIA assays, it has become common practice to test multiple stools.  However, this increases the likelihood a positive result is a false positive test (Ann Intern Med. 151:176-9, 2009).

Objective:   We evaluated the EIA test result pattern for patients with possible HA-CDI to assess the impact of multiple testing on CDI diagnosis. 

Methods:   Patients with a positive EIA test at hospital day ≥2 from October 2004 to September 2008 were selected for study.  All positive and negative test results for these patients were recorded for a 14 day period around the first positive test.  EIA testing was by the C. difficile Tox A/B II ^TM test (TechLab, Blacksburg, VA).  Statistical comparison was by the Fisher's exact test.

Results:  489 patients had a positive EIA toxin test at hospital day ≥2 in this 4 year period.  131 of the 489 patients (27%) had a first test that was EIA negative.  In these 131 patients, 2 to 7 replicate tests were done over a 14 day period.  114/131 patients had the number of negative tests equal (43) or outnumber (71) the positive tests.  Chart review was performed on a convenience sample of 101 patients to determine the likelihood of CDI and only 46 (46%) had a CDI clinical presentation.  The 101 patient records reviewed were sorted in to Groups A to D:

A.    1 negative and 1 positive test (41% true positive).

a.      16 patients with likely CDI

b.     23 patients with no clinical evidence of CDI

B.    More positive than negative tests (88% true positives)

a.      14 patients with likely CDI

b.     2 patients with no clinical evidence of CDI

C.    More negative than positive tests (39% true positives)

a.      16 patients with likely CDI

b.     25 patients with no clinical evidence of CDI

D.    5 indeterminate subjects

Group A was different from B (P = 0.0024); Group A was not different from C (P > 0.5); and Group C was different from B (P = 0.0011).  We also assessed test performance in these 96 patients with CDI or no CDI.  In the 45 cases with clinical CDI, 16 had 1 negative and 1 positive test; 13 had more positive than negative tests; and 16 had more negative than positive tests.  For the 51 cases with no clinical evidence of CDI, 13 had 1 negative and 1 positive test; 2 had more positive than negative tests; and 26 had more negative than positive tests.  Only in those cases where there were more positive than negative tests was there a difference between CDI and no CDI (P = 0.0042).

Conclusions:   Repeat stool testing to maximize the sensitivity of EIA testing is not reliable.  It results in significant misdiagnosis of diarrheal disease and falsely elevates the observed rate of HA-CDI.  The practice of repeatedly using the same type of testing on a patient to diagnose CDI should be stopped.