Let the user beware: specificity and sensitivity limits for in vitro diagnostic devices.

Abstract

Agrowing number of in vitro diagnostic screening tests are on the market today to diagnose numerous infectious diseases. Simpler to perform than the older methods of cultures or acute and convalescent serology, they promise immediate answers at a lower cost, which suits clinical as well as cost-containment interests in today's health care market. Their easy processing conforms to the preferences of Clinical Laboratory Improvement Amendments; the simplicity of many test kits minimizes human error. Most physicians assume that before these kits can be marketed, their reliability and validity have been proven. However, our recent experience with one new product, the Monolert,1 which is promoted as a rapid screening test for infectious mononucleosis, suggests that this general assumption is erroneous. Starting in the early 1990s, our pediatric group sent a number of blood samples to our local hospital laboratory to identify which patients with symptoms of lethargy, fever, sore throat, headache, and lymphadenopathy actually had early Lyme disease. Tests included a complete blood count, erythrocyte sedimentation rate, Lyme enzyme-linked immunosorbent assay, and a Monolert test. Over the next few years, very few children had serologic evidence of Lyme disease, but a surprisingly large number tested positive for infectious mononucleosis. This was perplexing because many of these children had illnesses that were not ultimately typical of primary Epstein-Barr virus (EBV) infection. As some children's symptoms progressed, they were found to have had a variety of other illnesses including pneumonia and pyelonephritis. Mononucleosis was unlikely to be the correct diagnosis. The Monolert product insert provided no obvious explanation for the many positive active infectious mononucleosis results. It stated that acute varicella, elevated rheumatoid factor, and acute cytomegalovirus infection could cause false-positive results; however, our patients' clinical courses were not compatible with these diagnoses. The hospital laboratory director could identify no obvious technical …