Development of a clinical scoring system for the diagnosis of smear-negative pulmonary tuberculosis

Abstract

This study developed a clinical score based on clinical and radiographic data for the diagnosis of smear-negative pulmonary tuberculosis (SNPT). SNPT was defined as a positive culture in Ogawa in a patient with two negative sputum smears. Data from patients admitted to the emergency ward with respiratory symptoms and negative acid-fast bacilli (AFB) smears was analyzed by means of logistic regression to develop the predictive score.Two hundred and sixty two patients were included. Twenty patients had SNPT. The variables included in the final model were hemoptysis, weight loss, age > 45 years old, productive cough, upper-lobe infiltrate, and miliary infiltrate. With those, a score was constructed. The score values ranged from -2 to 6. The area under the curve for the ROC curve was 0.83 (95% CI 0.74-0.90). A score of value 0 or less was associated with a sensitivity of 93% and a score of more than 4 points was associated with a specificity of 92% for SNPT. Fifty-two point twenty-nine percent of patients had scores of less than one or more than four, what provided strong evidence against and in favor, respectively, for the diagnosis of SNPT. The score developed is a cheap and useful clinical tool for the diagnosis of SNPT and can be used to help therapeutic decisions in patients with suspicion of having SNPT.

Keywords

• Tuberculosis
• pulmonary
• regression analysis
• scoring methods