University of Michigan
Department of Pediatrics
Evidence-Based Pediatrics Web Site

Question

• What are the characteristics of the RSV EIA test and will they help decide whether a child has RSV?

Clinical Bottom Lines

1. For the population tested at the University of Michigan, the sensitivity of the RSV TestPack is not significant enough to effectively reduce pretest probability to a posttest probability consistent with not having RSV (i.e. < 10%).
2. Therefore, the RSV TestPack is not useful for cohorting patients seen in the ER at the time of admission, based on the test's ability to "rule out" a presumed diagnosis of RSV bronchiolitis in a single patient.
3. Whether the RSV TestPack affects the heath care system as a whole to reduce overall risk of unwanted RSV exposure is unknown.

Summary of Key Evidence

1. The Abbott TestPack RSV assay was compared to culture and direct immunofluoresence (DFA) of nasopharyngeal swabs for the presence of RSV in swabs obtained from 234 infants.¹
2. A total of 122 specimens were considered positive by culture, DFA, or blocking assay, for a prevalence of 52%.
3. The sensitivity and specificity in this population were 89% and 96%, respectively.  These characteristics result in a LR+ of  22.25 and a LR- of 0.11, under ideal circumstances.
4. The following table shows the paper's data for the EIA compared to DFA alone:



---

Test Results             DFA+                  DFA-                 LR


---

EIA                         #         %               #        % 


---

Positive                   89    86%              21    18%             4.9 
Negative                 14    14%              97    82%             0.2 


---

totals                    103  100%            118  100% 
 
5. The next table shows UM data comparing the EIA to indirect immunofluoresence (IFA).  In this data, the IFA was not done (presumed positive) if the EIA was positive.² 



---

Test Results             IFA+                    IFA-                   LR 


---

EIA                        #         %                #        % 


---

Positive                  45    60%               3      2%             30.6 
Negative                30    40%            150    98%              0.4 


---

totals                    75  100%            153  100% 
 
6. During the first three weeks of February 1999, an average of 8.5 patients per day were tested by IFA for RSV, based on clinical suspicion.  The prevalence of RSV in this population was 32%.³  Thus, with a LR- of 0.4, the post-test probability of a child having RSV despite a negative EIA is still 15.8%.

Additional Comments

• In comparison with previous years at the University of Michigan, the RSV TestPack has lost sensitivity (1994-5 sensitivity 79%).  This change is independent of the prevalence of RSV and reflects a true change in the test's effectiveness.  This could be due to changes in specimen collection or other unknown factors.
• Though the data do not support the usefulness of the TestPack EIA in a single individual, the effect seen on the risk of nosocomial transmission of RSV cannot be determined from this simple analysis.

Citation

1. Swierkosz, Flanders R, Melvin L, Miller JD, and Kline MW. Evaluation of the Abbott TESTPACK RSV enzyme immunoassay for detection of respiratory syncytial virus in nasopharyngeal swab specimens. Journal of Clinical Microbiology 1989; 27:1151-1154.
2. Pierson C. Personal communication regarding EIA IFA testing results from 11/1/98 to 2/1/99.  Original data gathering did not include IFA testing for samples that were EIA positive.
3. Pierson C. Personal communication regarding IFA results for 2/1/99 to 2/19/99.

CAT Author: David Myers, MD

CAT Appraisers: Robert Schumacher, MD

Date appraised: February 24, 1999