Efficacy and impact of polymerase chain reaction in the diagnosis of extrapulmonary tuberculosis

OBJECTIVE: To evaluate the efficacy of PCR in the diagnosis of extrapulmonary tuberculosis (TB) and its impact in the management of patients in a tertiary-care center in Mexico City. METHOD: We conducted a retrospective study based on 40 clinical charts of patients to whom nested PCR was performed for the diagnosis of TB from June 1999 to December 2000. We reviewed the medical notes of 10 days before and 10 days after the PCR study to analyze its impact in the management of the patient. Also, we reviewed the rest of the chart to decide if the patient suffered from TB or not (gold standard). The categories of diagnosis were definitive case of TB, probable TB and no TB. We calculated the sensitivity, specificity, and predictive values. RESULTS: The PCR was positive in 45% of the cases. The sensitivity of PCR to diagnosis TB was 50%, specificity 59%, and the positive and negative predictive value were 35% and 72%, respectively. If just spinal fluid was included, the sensitivity and the negative predictive value increased to 75% and 63%, respectively. The PCR had an impact in the management of 13% of the patients. CONCLUSIONS: The PCR for the diagnosis of extrapulmonary TB has limited efficacy, which improves when the test is done in spinal fluid samples. The impact of the result of PCR in the clinical management of the patients was poor.