Parameter Estimates for Cost-Effectiveness Model.

TB, tuberculosis; MGIT, Mycobacteria Growth Indicator Tube; NTM, non-tuberculous.mycobacteria; HIV, human immunodeficiency virus.aThe estimated number of true-positives includes all confirmed cases, plus a proportion of culture-negative cases, assuming a positive predictive value of 52% for clinician diagnosis based on the prevalence of TB and estimated sensitivity and specificity of clinical diagnosis.bSolid media = five Lowenstein-Jensen slants per specimen; MGIT = one BBL® culture tube per specimen, re-inoculated if initially contaminated (see Methods). Sensitivity and specificity are calculated among diagnostic attempts in which at least one culture was not contaminated. Diagnostic attempts may include more than one culture specimen.cThe lower contamination rate and higher sensitivity for MGIT over solid media reflect the practice of re-processing contaminated MGIT specimens, whereas solid cultures were inoculated in parallel and not re-processed.dDisability weight of 0.136 for HIV infection without AIDS [31], +0.1 for the burden of antiretroviral therapy.eAssumes that TB treatment (lasting six months) results in 50% as much disability as active TB.

TB（Tuberculosis，结核病）；MGIT（Mycobacteria Growth Indicator Tube，分枝杆菌生长指示管）；NTM（non-tuberculous mycobacteria，非结核分枝杆菌）；HIV（human immunodeficiency virus，人类免疫缺陷病毒）。a. 真阳性病例的估计数包含所有确诊病例，以及一定比例的培养阴性病例，该推断基于临床诊断的阳性预测值为52%，此数值由结核病流行率以及临床诊断的估计灵敏度与特异度推导得出。b. 固体培养基指每份标本接种5支罗琴（Lowenstein-Jensen）斜面；MGIT指每份标本接种1支BBL®培养管，若初始培养出现污染则重新接种（详见方法章节）。灵敏度与特异度的计算基于至少完成1份未受污染培养的诊断尝试，单次诊断尝试可包含多份培养标本。c. MGIT较固体培养基污染率更低、灵敏度更高，这一现象源于对受污染MGIT标本进行复处理的操作规范，而固体培养仅平行接种且未进行复处理。d. 无艾滋病的人类免疫缺陷病毒感染的残疾权重为0.136[31]，抗逆转录病毒治疗的疾病负担额外增加0.1的残疾权重。e. 本研究假设：为期6个月的结核病治疗所带来的残疾负担仅为活动性结核病的50%。