Table 2.

Optional Triage approaches and key characteristic assumptions

Triage label	Description of triage approach	Sensitivity	Specificity	Additional cost per testb
T1- Base case	No triage
T2- Cough 1 week	Respiratory symptom of cough > 1 week [18]	88%	19%	US$0
T3 Cough 3 weeks	Respiratory symptom of cough > 3 weeks [18]	61%	51%	US$0
T4- Clinical Score	Scorecard based on aggregating scores assigned to respiratory symptoms including chest pain, cough, sputum expectoration, hemoptysis, night sweats, fever, shortness of breath and weight loss [18].	83%	52%	US$2
T5- ANN	Artificial Neural Network (ANN) based on using a multilayer perceptron (MLP) approach [19] to infer the probability of a patient having active pulmonary-TB from personal data and clinical symptoms i.e. age, gender, cough, fever, weight loss, smoker, night sweats, hospitalisation, chest pain, dyspnea, and hemoptysis.	98%a	32%a	US$2
T6- TPP (optimal)	A theoretical optimal target product profile (TPP) as proposed by Denkinger et al. [21]	95%	80%	US$2
T7- TPP (minimal)	A theoretical target product profile (TPP) with the minimum characteristics required to be useful as proposed by Denkinger et al. [21]	90%	70%	US$2

^athe sensitivity and specificity figures are taken from unpublished research in Brazil

^bThe additional cost per triage test is assumed to be low as the characteristics are those which clinicians will already consider today. An additional allowance (US$2) has been made if some computation is required in line with the costs proposed by Denkinger et al [20] for the TPP’s