From the Department of Medicine (EGS) and the Departments of Pediatrics and Epidemiology and the Children's Research Center (EDS), Yale University School of Medicine, New Haven, CT.
Checking for direct PDF access through Ovid
Because of widespread public concern about Lyme disease and the erroneous belief that it commonly may present with vague, nonspecific symptoms without accompanying objective physical signs, the use of serologic tests to rule out Lyme disease has become very common in the evaluation of patients with such nonspecific symptoms as fever, malaise, arthralgia and fatigue.1 Many studies document both interlaboratory and intralaboratory variability in the results of widely used serologic tests for Lyme disease.2-6 These reports emphasize the need for serologic tests with excellent sensitivity, specificity and reproducibility. Less well-documented in the literature are the pitfalls that arise from overuse of serologic tests in patients with a low probability of having Lyme disease, even tests performed at reference laboratories, that are more accurate and more reproducible than those done by most laboratories that use commercial kits. A number of investigators have documented the high proportion of patients seen at referral centers that are misdiagnosed as having Lyme disease;1, 7 this can be attributed at least in part to the indiscriminate use of serologic tests. The purpose of this report is to illustrate how wide-spread use of serologic tests for Lyme disease can frequently lead to misdiagnosis.Sensitivity is the proportion of persons with a positive test among persons with disease; specificity is the proportion of persons with a negative test among persons without disease. A reasonably good serologic test for Lyme disease might have a sensitivity of 95% and a specificity of 90%.8 Using these values we calculated the positive and negative predictive values for such a test in a theoretical sample of 10 000 persons, of whom 1% have Lyme disease. This theoretical prevalence of 1% is even higher than the annual incidence rate in most areas in which Lyme disease is endemic.The positive predictive values of a test is the proportion of persons with disease among all those with a positive test for the disease; the negative predictive value is the proportion of persons without disease among all those with a negative test for the disease.8 We also calculated the predictive values of the results of the same test in samples with higher prevalences of disease.The results are shown in Table 1. Under circumstances of a 1% prevalence of disease, the predictive value of a positive test is only 8.7%, and of all the positive tests for Lyme disease 91.3% were false positive results. The results of applying this test (with the same 95% sensitivity and 90% specificity) to samples with disease prevalences of 10 and 50%, respectively, are also shown.Erythema migrans, the characteristic rash of Lyme disease, is pathognomonic. However, Lyme disease can present with less specific, objective signs such as arthritis, neurologic abnormalities (cranial nerve palsies, meningitis) or heart block, all of which may also be accompanied by very nonspecific symptoms such as myalgia, headache and fatigue. In these cases the diagnosis of Lyme disease is supported by serologic evidence of infection with B. burgdorferi. However, because of widespread anxiety about Lyme disease, the use of serologic tests for Lyme disease has become very common in the workup of patients with a low probability of having Lyme disease. Indeed the impetus to order a serologic test for Lyme disease often may come from the patient and not from the physician. In one review of the use of serologic tests for Lyme disease in a health maintenance organization in California, 35% of the 117 tests for Lyme disease were obtained because of a request by the patient.