The potential of a multiplex high-throughput molecular assay for early detection of first and second line tuberculosis drug resistance mutations to improve infection control and reduce costs: a decision analytical modeling study

van’t Hoog, AH; Bergval, I.; Tukvadze, N.; Sengstake, S.; Aspindzelashvili, R.; Anthony, RM; Cobelens, F.

doi:10.1186/s12879-015-1205-4

Table 2 Model parameters for cohort proportions, diagnostic test performance and costs

From: The potential of a multiplex high-throughput molecular assay for early detection of first and second line tuberculosis drug resistance mutations to improve infection control and reduce costs: a decision analytical modeling study

		PE	Range		Source, scenario
Cohort proportions
	Proportion of PTB patients who are sputum-smear positive	0.64	0.62	0.65	[29] α
	Susceptible to all first line drugs, or resistance to either streptomycin, ethambutol or pyrazinamide, or combinations of those.	0.71	0.75	0.55	[29] α
	IHN mono resistance, which may or may not include resistance to other first line drugs streptomycin, ethambutol, and/or pyrazinamide (poly resistance), but not rifampicin	0.127	0.128	0.122	[29] α
		0.127	0.128	0.122	Footnote (a)
	Rifampicin resistance with- or without INH resistance, without additional resistance to 2nd line drugs. Resistance to ethambutol and/or pyrazinamide may or may not be present.	0.108	0.073	0.212	[29] α
	MDR with additional resistance to ≥1 fluoroquinolone(s) but not to second-line injectable drugs (pré-XDR)	0.008	0.006	0.015	[29] α
	MDR with additional resistance to ≥1 SLID but not fluoroquinolones (pré-XDR)	0.044	0.032	0.087	[29] α
	XDR: MDR with additional resistance to ≥1 fluoroquinolones and ≥1 SLID.	0.008	0.007	0.014	[29] α
Diagnostic accuracy parameters
	Sensitivity of molecular tests in detecting rifampicin resistance (assumed to be the same as LiPA)	0.99	0.96	1.00	[28] α
	Sensitivity of molecular tests in detecting INH resistance (assumed to be the same as LiPA)	0.96	0.93	1.00	[28] α
	Sensitivity of molecular tests in detecting resistance to fluoroquinolones	0.831	0.787	0.867	[9] δ
	Sensitivity of molecular tests in detecting resistance to SLID, taken as the sensitivity of LiPA sl to detect capreomycin resistance	0.795	0.583	0.914	[9] δ
	Specificity of molecular tests in detecting rifampicin resistance (assumed to be the same as LiPA)	0.99	0.98	1.00	[28] α
	Specificity of molecular tests in detecting INH resistance (assumed to be the same as LiPA)	1.00	0.99	1.00	[28] α
	Specificity of molecular tests in detecting resistance to fluoroquinolones	0.977	0.943	0.991	[9] δ
	Specificity of molecular tests in detecting resistance to SLID, taken as the sensitivity of LiPA sl to detect capreomycin resistance	0.958	0.934	0.973	[9] δ
	Sensitivity and specificity of DST for resistance to 1st and 2nd line drugs	1	-		model assumption β
Repeat testing (Proportion of tests with invalid results requiring repeat testing, for:)
	Xpert MTB/RIF	0.011	0.0004	0.020	[30] γ
	LiPA	0.027	0.007	0.068	[3] γ
	mycobacterial culture	0.052	0.048	0.057	[31] γ
	high-throughput MRD-assay	0.027			same as LiPA; δ model assumption
	phenotypic DST	0			model assumption; β footnote (b)
Median number of days to result^a		days	sd		source
	MTBDRplus assay (LiPA)	3.0	1.7		[4]; γ footnote (c)
	LJ culture	34.1	11.3		[4] β
	MGIT culture	8.9	3.9		[4] γ
	LJ DST	67.5	15.0		[4] β
	MGIT DST	21.6	9.3		[4] β
	high-throughput MRD assay in scenarios A and C	6.0	3.0		Model assumption; footnote (d)
	high-throughput MRD assay in scenario B	3.0	1.5		Model assumption; footnote (e)
	Xpert MTB/RIF	0			Model assumption; γ footnote (f)
Median days from lab result until clinical review and treatment initiation
	for a standard treatment regimen (1st line or empirical 2nd line)	1			[32] α
	for an individualized regimen (assuming additional consultation)	4			Model assumption α
Median days from treatment initiation to sputum culture conversion
	in patients with susceptible TB or INHmono resistance (days, sd)	34	26		[33–35] α
	in patients with MDR-TB on an appropriate regimen, (days, 95 % CI)	61	59	67	[13] α
	in patients with XDR-TB in high-throughput RMD scenario (days, 95 % CI)	75	60	90	[14] δ
	Increase in duration of préXDR (SLID res) in baseline	0.55			[15] β
	Increase in duration of préXDR (FQ res) in baseline	0.72			[15] β
Time to failure
	Months to failure on a first-line regimen	5			(15;30) α
	Months to failure on a second-line regimen	4			(15;30) α
	Infectious time in XDR patients who fail	24			Model assumption (duration of treatment) α
Per-test unit cost for diagnostic tests US$ 2013 (min, max)
	Sputum smear [2]	3.34	2.42	5.08	[17] α
	Xpert PEPFAR pricing	17.29	15.66	18.92	[18] γ
	high-throughput MRD-assay - ratio compared to per-test unit costs of LiPA	2	0.5	4	model assumption δ
	DST 1st line (MGIT)	44.56	40.05	49.07	[18] β
	DST 2nd line (LJ)	25.35	20.68	30.02	[18] β
	Line Probe Assay (LiPA)	21.32	18.45	24.18	[18] γ
	LJ culture	18.48	11.08	33.30	[17] γ
	MGIT culture	18.48	11.08	33.30	[17] γ
Treatment cost parameters US$ 2013 (min, max)
	First-line treatment course^b	945	629	1419	[19] α
	Second-line treatment course for MDR	4176	2341	7449	[19] α
	Ratio of pré-XDR regimen cost compared to MDR regimen cost	2			[20] α
	Ratio of XDR regimen cost compared to MDR regimen cost	3			[20] α
	Hospitalization for MDR/XDR, cost per day^b	67			[1] α

The modeled scenarios are: Base Case; MRD-A. Rapid MRD assay following culture; MRD-B. Improved analytical sensitivity; MRD-C. Improved clinical accuracy. The Greek symbol in the first column indicates to which scenarios the parameter apply: α to all four scenarios; β to the Base Case only; γ to the Base Case, MRD-A and -C but not to MRD–B; δ to the MRD scenarios (A, B, C) but not the Base Case
(a): low end of the range reflects the distribution among new patients and high end the distribution among previously treated patients.(b): if an isolate is obtained on culture, we assume DST will always give a valid result. (c): adjusted (was 4.2 days in the publication in a special study performing assays 2–3 times a run per week on 2–8 samples per run. We adjusted for current practice where LiPAs are run daily (50–60 samples per week). (d) assumes a batch of ± 50 once a week. (e): assumes a patient volume that requires batch testing of ± 50 every 1–2 (working) days. (f): Time counts from TB diagnosis and Rifampicin result comes at the same time as the TB diagnostic result
Abbreviations; PE point estimate, PTB pulmonary tuberculosis, MDR multi-drug resistance, defined as resistance to rifampicin and isoniazid [2], SLID second-line injectable drugs, XDR extensively drug-resistant, INH isoniazide, LiPA line probe assay, LJ Löwenstein-Jensen, MGIT Mycobacterial Growth Inhibitor Tube, DST Drug Susceptibility Testing, sd standard deviation, CI confidence interval
^aExcluding requirement for repeat testing
^bFirst-line treatment applies to catI and catII treatment; hospitalization costs are estimated from studies in the same region [21] and average number of hospital days in 2012 [36]

Back to article page

ISSN: 1471-2334

Contact us

Submission enquiries: bmcinfectiousdiseases@biomedcentral.com
General enquiries: ORSupport@springernature.com

BMC Infectious Diseases

Contact us