RESUMEN
The objective of this study is to describe the proportion of patients with chronic cough and negative smear microscopy appropriately diagnosed as tuberculosis [TB] and to identify clinical features that could be used in developing a diagnostic scoring system for smear-negative patients. Records of patients with chronic cough and > 3 negative sputum smears for acid fast bacilli who attended a reference University hospital in south-eastern Iran and screened by culture were retrospectively reviewed. We compared confirmed smear-negative pulmonary TB [PTB; culture-positive] and unconfirmed smear-negative patients [culture-negative] to describe the appropriateness of treatment and their characteristics. Features independently predictive of smear-negative PTB [SNPTB] were entered into a logistic regression to create a diagnostic rule. This study enrolled 350 patients, of which 52 [14.8%] were culture-positive and 298 [85.2%] culture-negative. Of these, 38 out of 52 [sensitivity 73%] confirmed SNPTB were diagnosed as TB and 283 out of 298 [specificity 95%] unconfirmed sputum-negative patients were diagnosed as non-PTB. Variables associated with confirmed SNPTB were the presence of night sweats, family history of TB, typical chest radiography, erythrocyte sedimentation rate > 45 mm and white blood cell count < 11000/mL. The score constructed with these variables had a sensitivity of 94% and specificity of 74% with an area under the curve of 0.90. The clinical differences between SNPTB and control patients could be used to develop a clinical scoring system to identify patients with SNPTB