Objectives: To develop and validate clinical guidelines for diagnosis of smear-negative pulmonary tuberculosis (TB) in developing countries with low-HIV prevalence.
Methods: We developed diagnostic guidelines for smear-negative TB. Clinical diagnoses based on these guidelines were compared with sputum culture, chest X-rays and reports of an expert panel.
Results: The guidelines achieved a sensitivity of 0.59 [confidence interval (CI) 0.46-0.66] and a specificity of 0.86 (CI 0.84-0.88) in diagnosing smear-negative TB. A total of 6.8% of patients who initially improved after a course of antibiotics were later confirmed to have TB. Clinicians detected an abnormal chest X-ray in 92% (CI 88-96%) and radiological signs of pulmonary TB in 98% (CI 94-100%) of cases.
Conclusions: Our experience highlights a number of dilemmas faced in developing, testing and implementing diagnostic guidelines in poorly resourced conditions. Using radiological criteria for TB and appropriate training can help in improving the diagnostic skills of primary care clinicians working in low-HIV settings with access to X-ray facilities. But a significant number of apparently smear-negative TB cases may in fact be smear positive and TB programmes should focus on improving the quality of direct acid-fast bacilli microscopy. The value of an antibiotic trial is questionable due to the relatively large number of false negatives generated by this approach.
Tropical Medicine & International Health (2006) 11(3) 323-331 [doi: 10.1111/j.1365-3156.2006.01559.x]