Development of a clinical algorithm scoring system to diagnose smear negative pulmonary tuberculosis in resource- constraint settings: modified Delphi method

Introduction: Tuberculosis (TB) remains a major health threat globally with more than 10 million people infected and one million deaths in 2023; most cases occur in low-and-middle income (LMIC) countries. One contributing factor is the late diagnosis among patients with smear negative pulmonary TB as rapid molecular test kits are unavailable and costly. This study aimed to develop a clinical algorithm scoring system to diagnose smear negative pulmonary tuberculosis (TB) among symptomatic patients. Method: A modified Delphi method was conducted in Malaysia with a group of experts comprising pulmonologists, family medicine specialists, public health specialists, microbiologists, radiologists and a policymaker using a three-round exercise via emails from January to June 2024 to identify clinical parameters that are important for the diagnosis of TB. This was followed by a consensus meeting to finalize the list of clinical parameters and the decision on the weightings of each parameter was made by consensus agreement. Next, weightings for parameters were refined using data from 60 symptomatic individuals whose sputum smears were negative for TB. A cut-off score for ‘likely’ vs ‘unlikely’ to be smear negative TB was derived. Results: 27 experts were invited and 23 (85.2%) consented to participate. One expert did not participate in Round 2 and 3, but all gave their input in the final consensus agreement. After Round One, 54 parameters were identified; this was reduced to 26 after Round Two and 23 after Round Three. Following the consensus meeting, 21 parameters were included in the final clinical algorithm and a cut-off score of 19 was found to identify a diagnosis of “likely or unlikely TB”. Discussion: The developed clinical algorithm can hasten the diagnosis of smear negative pulmonary TB in countries that have limited access to rapid molecular tests but will require validation in future study. Funding: NIHR 132826