European Journal of Clinical Microbiology & Infectious Diseases

, 27:397

Repeated enzyme immunoassays have limited utility in diagnosing Clostridium difficile

Authors

    • Division of Geographic Medicine and Infectious DiseasesTufts-New England Medical Center
    • Sackler School of Graduate Biomedical SciencesTufts University
    • Christiana Care Health SystemCenter for Outcomes Research
  • D. R. Snydman
    • Division of Geographic Medicine and Infectious DiseasesTufts-New England Medical Center
    • Sackler School of Graduate Biomedical SciencesTufts University
  • C. E. O’Sullivan
    • Division of Geographic Medicine and Infectious DiseasesTufts-New England Medical Center
    • Sackler School of Graduate Biomedical SciencesTufts University
    • Barts and The Royal London TrustPathology and Pharmacy Building
Brief Report

DOI: 10.1007/s10096-007-0452-8

Cite this article as:
Drees, M., Snydman, D.R. & O’Sullivan, C.E. Eur J Clin Microbiol Infect Dis (2008) 27: 397. doi:10.1007/s10096-007-0452-8
  • 73 Views

Abstract

Many clinical laboratories use enzyme immunoassays (EIA) to diagnose Clostridium difficile-associated disease (CDAD). Clinicians frequently order three EIAs to “rule out” CDAD. We performed a retrospective cohort study to determine the clinical utility of repeating EIA testing to diagnose CDAD. We reviewed all EIAs performed by our laboratory during 2005, determined the total number of tests per patient and per testing episode, and calculated the relative negative predictive value (NPV) of one EIA compared to ≥2 EIAs. The laboratory performed 2,938 EIAs, of which 253 (8.6%) tests were positive. Most patients (85%) were diagnosed by the first EIA performed. Of >1,000 testing episodes that included ≥2 EIAs within 7 days, only 15 patients had a positive second or third test after negative initial testing. The relative NPV of the first EIA was 97.4%. These data suggest that using newer generation EIAs, repeated testing is of limited benefit in diagnosing CDAD.

Recently noted trends toward higher incidence and increased severity of Clostridium difficile-associated disease (CDAD)[1, 2] have made rapid and accurate diagnosis imperative. The cellular cytotoxin assay (CCA), currently the accepted gold standard, has reported sensitivities that typically exceed 90% [3], but sensitivities as low as 67% have been reported [4]. Many clinical laboratories now use enzyme immunoassays (EIA) exclusively. These tests are rapid, less expensive, and technically easy to perform, but have variable sensitivities (<60% to 99%) [4, 5]. Similar to the use of sputum analysis in diagnosing tuberculosis, many clinicians frequently order three EIAs to definitively “rule out” CDAD. However, there has been limited research evaluating the clinical performance of repeated diagnostic testing, particularly with the newer generations of EIAs [6, 7]. We conducted this retrospective cohort study to evaluate the clinical performance of the EIA used in our tertiary care academic medical center and to determine the diagnostic utility of repeating EIA testing for CDAD.

We reviewed all C. difficile EIAs performed by the microbiology laboratory during 2005 at Tufts-New England Medical Center (Boston, MA). The laboratory used the Premier Toxins A&B EIA (Meridian, Bioscience, Inc.), which has published sensitivities of 94.7–96.8% and negative predictive values of 98.9–99.8%, compared to the CCA [8, 9]. The test was performed according to the manufacturer’s directions, without any confirmatory testing. Laboratory policy included rejection of greater than two stool samples sent for C. difficile testing within a 5-day period, but was not strictly enforced. For all patients with positive EIAs, the electronic medical record was reviewed to confirm the clinical diagnosis and retrieve demographic information, comorbidities and prior history of CDAD. Repeat EIAs performed within 7 days of initial testing were considered part of a single testing episode, while EIAs performed after 7 days were considered a new testing episode, assuming that sufficient time had elapsed for new exposure to C. difficile. Because we were most interested in the utility of repeated EIAs rather than in comparing the EIA to a gold standard [from which we could calculate a negative predictive value (NPV)], we calculated a “relative” NPV for the first EIA performed compared to multiple EIAs. We defined true negatives as patients who had at least two EIAs performed and who were never diagnosed with CDAD by laboratory testing.

During 2005, our laboratory performed 2,938 EIAs, of which 2,685 (91.4%) were negative and 253 (8.6%) were positive. The average age of those with CDAD (n = 189) was 56 years, 50% were female, and 18% had a documented prior history of CDAD. Two patients developed toxic megacolon and required colectomy, one of whom died. The majority of patients both with and without CDAD had a single test performed during 2005, but multiple testing was common (Table 1). Of the 217 positive EIAs (excluding duplicate positives during the same admission), 184 (85%) were the first EIA performed. Among the 33 patients with previous negative EIAs, 13 (6% of total) were preceded by a negative test within 7 days. The remaining 20 patients had positive EIAs more than 7 days after initially negative test results, which may indicate new infection rather than lack of detection via initial testing. Of these, during subsequent testing episodes one additional patient had two EIAs performed within 7 days where only the second test was positive, and one patient had three EIAs performed within 7 days where only the third EIA was positive. Among >2,000 testing episodes among patients whose EIAs remained negative, more than half involved two or more tests within 7 days. There were 480 patients with at least two negative EIAs. Of those with at least one positive EIA, 184 had a positive first EIA and 13 had a negative first EIA. The relative NPV of the first EIA was therefore 97.4%.
Table 1

Frequency of EIA testing in patients with and without positive test results

 

Patients with any positive EIA

Patients without any positive EIAs

N = 189

N = 1,358

n (%)

n (%)

Total number of tests performed per patient*

One

111 (58.7)

878 (64.7)

Two

54 (28.8)

259 (19.1)

Three

9 (4.8)

105 (7.7)

≥Four

15 (7.9)

116 (8.5)

 

Testing episodes in patients with positive EIAs†

Testing episodes in patients without any positive EIAs

 

N = 217

N = 2,010

Number of tests performed within each testing episode

One

202 (93.1)

960 (47.8)

Two

14 (6.9)

640 (31.8)

≥Three

1 (0.5)

410 (20.4)

*Tests may occur during more than one hospitalization

For patients diagnosed with CDAD, testing episodes during subsequent hospitalizations were included. During each hospitalization, any subsequent testing after the first positive EIA was excluded (i.e., the final EIA in each testing episode was the first positive result)

Few studies have examined the common practice of repeating diagnostic testing for C. difficile. Several have found little value in repeating testing using the CCA [10, 11], but the number of EIAs required to definitively rule out CDAD has not been studied using more recently developed assays. A 1995 study [7] found that repeating EIA testing increased yield by 12%, but a more recent study [6] found that of 396 patients with negative initial EIA results, only 1 of 78 patients who underwent repeat EIA testing had a positive test result. Our study found multiple testing within 7 days to be extremely common, but the yield of these repeated tests was low, identifying only 15 additional patients from more than 1,000 patients tested at least twice within 7 days.

This study is subject to certain limitations. The clinical suspicion for CDAD when the tests were ordered is unknown. We were unable to detect patients with CDAD but without a positive EIA result who were treated empirically. We assumed that positive EIAs indicated CDAD, although this may have overestimated cases of disease. However, we did review the medical record of all patients with positive EIAs to confirm that CDAD was also clinically diagnosed. Our microbiology laboratory did not routinely perform confirmatory testing, so a gold standard with which to calculate sensitivity, specificity and negative predictive value was not available. Instead, we calculated the relative NPV of a first negative EIA compared to multiple EIAs. While this does not answer the question of EIA sensitivity, it is useful for clinicians whose microbiology laboratories perform EIA testing exclusively. However, the interpretation of this high relative NPV, and the implications of negative initial testing, will vary based on clinical suspicion of CDAD and local disease prevalence and virulence.

Most cases of CDAD in our institution were diagnosed by the first EIA performed. We found the relative NPV of a single EIA, compared to multiple EIAs, to be quite high (97%). For difficult-to-diagnose infectious diseases such as tuberculosis, repeated testing is recommended to increase diagnostic yield [12]. We found that repeating the EIA was of little benefit in diagnosing CDAD, as few of these repeat tests were positive. This suggests either increased sensitivity of newer generation EIAs or that an alternate testing modality in addition to EIA should be used, as has been suggested [4, 13]. Avoidance of unnecessary repeat diagnostic testing for CDAD is desirable for several reasons, including the cost of additional tests as well as costs and adverse events related to empiric antibiotic use and patient isolation while awaiting these additional tests. We believe that our results support the practice of repeating diagnostic testing only when the clinical suspicion for CDAD is relatively high, and then performing no greater than two EIAs within a 7-day period. While further study of this issue is needed, repeating EIA testing for CDAD may be unnecessary for many patients.

Acknowledgements

Dr. Drees was supported by NRSA grant 5 T32 HS 000060–13. There was no external funding for this study.

The authors would like to thank Linda Perry, Director, Tufts-New England Medical Center Microbiology Laboratory, for her assistance with this study.

Copyright information

© Springer-Verlag 2007