Table 6. One way sensitivity analysis and justification of variability for each input variable included in standard decision analysis model.

Input	Range of input values for each variable (defined using base case value reported in studies in review)	IGRA strategy (Mean expected value: total cost = $855)		TST strategy (Mean expected value: total cost = $1081)		Comment on justification for the variability of inputs used by the different studies
		SPREAD of expected values (|lowest value- highest value|)	POTENTIAL INFLUENCE (Spread/mean expected value)	SPREAD of expected values (|lowest value- highest value|)	POTENTIAL INFLUENCE (Spread/mean expected value)
Prevalence of LTBI	5% to 58%	$1,258	147%	$1,053	97%	Variability in probabilities justified: Differences can be justified by the consideration of different populations and sub-groups.
Reactivation rate (annual) in the absence of effective LTBI therapy	0.02% to 1.25%	$982	115%	$997	92%	Variability in probabilities predominantly unjustified: Differences could be justified based on study population (ie. immunosuppressed, close contacts etc.), however inputs used in 11/13 studies were for the general population. In only 2 studies of close contacts were higher reactivation rates used.
Cost of treating active TB	$5318 to $63120	$903	106%	$916	85%	Variability in costs predominantly unjustified: All studies should include similar costing components. Costs should be similar for high income settings. Some variability could be due to economic perspective (eg. Study 2 included indirect costs).
Cost of treating LTBI	$224 to $1577	$375	44%	$704	65%	Variability in costs partially justified: All studies should include similar costing components. Costs should be similar as most are high income settings. Duration of prophylactic regimen will result in some justified variability in costing (6 studies assumed 9INH, 5 assumed 6INH and 1 assumed 3HR).
Specificity of TST	15% to 99%	−	−	$635	59%	Variability in probabilities partially justified: Differences could be justified based on BCG status of study population. However, within each sub population we should see similar estimates that have been derived from meta analyses.
Probability of an adverse event from LTBI therapy	0 to 18%	$318	37%	$558	52%	Variability in probabilities predominantly unjustified: Most studies consider young populations that should experience similar types and rates of adverse events. One study considered an elderly population with justified use of higher rates of adverse events.
Completion rate for LTBI therapy	21% to 100%	$271	32%	$267	25%	Variability in probabilities partially justified: Estimates should be similar for studies set in general population. Differences based on duration of regimen justified, yet estimates with same duration should be similar.
Cost of an adverse event	$183 to $14006	$236	28%	$442	41%	Variability in costs predominantly unjustified: All studies should include similar events with similar costing components.
Cost of IGRA/TST	$21 to $219/$15 to $121	$198	23%	$106	10%	Variability in costs predominantly unjustified: All studies should include similar costing components for basic screening. Costs should be similar for high income settings.
Efficacy of LTBI therapy	65% to 90%	$114	13%	$112	10%	Variability in probabilities partially justified: Differences justified for different regimens. However, for each regimen, estimate should be similar and derived from previously published meta analyses.
Sensitivity of IGRA	76% to 99%	$32	4%	−	−	Variability in probabilities predominantly unjustified: All studies should use similar estimates derived from previously published meta analyses.
Specificity of IGRA	96% to 100%	$31	4%	−	−	Variability in probabilities predominantly unjustified: All studies should use similar estimates derived from previously published meta analyses.
Sensitivity of TST	67% to 99%	−	−	$45	4%	Variability in probabilities predominantly unjustified: All studies should include similar estimates that have been derived from meta analyses.