Biostatistics Case Studies



CASE 5: Test Sensivity and Specificity


Test Sensitivity and Specificity:

Clinical History:

A 24-year-old woman is pregnant for the 2nd time. She visits you for a prenatal checkup. You order laboratory tests, including TORCH titers, HBsAg, and HIV. Your state requires a specific consent form for HIV testing. She questions you about the need for HIV testing.

Laboratory Findings:

A CBC shows Hgb 13.3 g/dL, Hct 40%, MCV 85 fL, platelet count 244,400/microliter, and WBC count 9070/microliter with differential count of 65 segs, 3 bands, 22 lymphs, 9 monos, and 1 baso. Serum chemistries show sodium 143 mmol/L, potassium 4.3 mmol/L, chloride 107 mmol/L, CO2 27 mmol/L, creatinine 1.0 mg/dL, and glucose 110 mg/dL. The HBsAg test is negative. She is rubella immune. CMV IgG titer is increased, but CMV IgM is not increased. Toxoplasma and HSV I and II titers are not increased. Her initial screening HIV EIA test is positive, with a positive confirmatory EIA test.

The diagnostic sensitivity and specificity of the standard enzyme immunoassay (EIA) initial screening test on patient blood for HIV is illustrated by the following study: In an adult population, 200,000 persons are tested for HIV infection with an EIA test on patient blood for HIV. Of these, 600 are found to test positive by EIA. Of these, 200 are found on subsequent confirmatory EIA testing to really be infected. Furthermore, follow-up of the original group of patients reveals that there was 1 person who really was infected, but was missed by the initial screening EIA test.

Questions:

  1. Why should the HIV test be ordered in this case?

    It is recommended that all pregnant women be tested for HIV because the congenital transmission of HIV can be reduced with intervention. The HIV test is perhaps the best test in the laboratory, based upon sensitivity and specificity, over 99% in most studies.

  2. What is the test sensitivity and specificity calculate from the information given above?

  3. In this example, the calculations are as follows:

    Sensitivity = true positives / (true positives + false negatives)
                = 200 / (200 + 1)
                = 99.5%
    
    Specificity = true negatives / (true negatives + false positives)
                = 199,399 / (199,399 + 400)
                = 99.8%