An assessment of nucleic acid amplification testing for active mycobacterial infection

Outcome-state probabilities

The results of the decision tree analysis are presented in Figure (AFB model arm) and Figure (AFB plus NAAT model arm), Appendix . The probability at each decision tree terminal is derived from a composite of the prevalence of TB in the tested population, clinical suspicion and test accuracy parameters. The difference between the model arms is the incorporation of NAAT accuracy parameters into the intervention arm.

The proportion of patients managed per each outcome state, as derived from the decision tree analysis, is presented in Table .

Table Outcome state probabilities, TB mixed scenario

True status	Treated status	AFB	AFB + NAAT	Difference
No TB	Untreated	70.6%	71.3%	0.69%
No TB	Standard treatment	7.4%	6.7%	–0.70%
No TB	MDR treatment	0.0%	0.0%	0.01%
TB	Untreated	3.4%	1.1%	–2.33%
TB	Standard treatment	18.1%	20.3%	2.14%
TB	MDR treatment	0.0%	0.2%	0.18%
MDR-TB	Untreated	0.1%	0.0%	–0.05%
MDR-TB	Standard treatment	0.4%	0.1%	–0.30%
MDR-TB	MDR treatment	0.0%	0.3%	0.35%
TOTAL	-	100%	100%	0%

Note: The probabilities associated with the correct treatment are highlighted.

AFB = acid-fast bacilli; MDR = multidrug-resistant; NAAT = nucleic acid amplification test; TB = tuberculosis

The highlighted rows in Table reflect the relevant optimal treatment strategy. NAAT is associated with improvements in each of these strategies, due to improved sensitivity of NAAT in conjunction with AFB (fewer false-negative results) and the ability of NAAT to identify and treat MDR-TB earlier. A net reduction in false-positive results is also observed, due to false-positive results associated with NAAT in AFB true-negative patients being outweighed by AFB false-positive patients correctly identified using NAAT.

Incremental costs

The incremental cost of NAAT, broken down by the source of the cost, is presented in Table . NAAT is associated with an incremental cost ($85.11) that is largely driven by the cost of NAAT, offset by reduced costs associated with TB transmissions (due to the identification of MDR-TB and fewer false-negative results) and reduced hospitalisations (due to fewer patients with false-positive results who require hospital isolation).

Table Breakdown of incremental costs, TB mixed scenario

Cost	AFB	AFB + NAAT	Increment
Treatment	$607.05	$631.19	$24.14
Treatment of AEs	$3.76	$3.70	–$0.06
Management	$192.72	$190.79	–$1.93
Hospitalisation	$1,490.04	$1,457.36	–$32.68
TB transmissions	$103.40	$69.04	–$34.36
NAAT cost	$0.00	$130.00	$130.00
TOTAL	$2,396.97	$2,482.08	$85.11

AEs = adverse events; AFB = acid-fast bacilli; NAAT = nucleic acid amplification test; TB = tuberculosis

Incremental QALYs

The incremental costs by outcome health state are derived from the product of the outcome-state probabilities (Table ) and the overall utility, by outcome state as presented in Table . These are presented in Table . Overall NAAT is associated with a small incremental QALY gain (0.001), driven by a shift from initially untreated TB (or standard treatment in the case of MDR-TB) to correct treatment.

In part the small difference results from the assumption that the correct diagnosis is achieved in all patients after 2 months, and that this 2-month delay in treatment is not assumed to affect patient mortality rates or the severity of disease. Whereas a utility penalty has been applied for the transmission of active TB, none has been applied for the transmission of latent TB.

Table Weighted utility by outcome state, TB mixed scenario

True status	Treated status	AFB	AFB + NAAT	Increment
No TB	Untreated	0.992	1.002	0.010
No TB	Standard treatment	0.104	0.094	–0.010
No TB	MDR treatment	0.000	0.000	0.000
TB	Untreated	0.046	0.015	–0.031
TB	Standard treatment	0.248	0.277	0.029
TB	MDR treatment	0.000	0.002	0.002
MDR-TB	Untreated	0.001	0.000	–0.001
MDR-TB	Standard treatment	0.004	0.001	–0.004
MDR-TB	MDR treatment	0.000	0.004	0.004
TOTAL	-	1.395	1.396	0.001

Note: The outcomes associated with the correct treatment are highlighted.

AFB = acid-fast bacilli; MDR = multidrug-resistant; NAAT = nucleic acid amplification test; TB = tuberculosis

Incremental cost-effectiveness

The effect on the ICER of adding each additional group of costs is presented in Table for the TB mixed scenario. Consistent with Table , the addition of transmissions and hospitalisation costs contribute most to the final ICER. The addition of the utility penalty for transmissions decreases the ICER further. This effect is consistently observed in the additional scenarios (Table , Appendix ).

Table Stepped economic evaluation, TB mixed scenario

Utilities considered	Costs included (NAAT cost applied in AFB + NAAT arm)	ICER
Index patient utility	Treatment only	$188,307
Index patient utility	Treatment and AEs	$188,238
Index patient utility	Treatment, AEs and management	$185,882
Index patient utility	Treatment, AEs, management and hospitalisation	$145,956
Index patient utility	Treatment, AEs, management, hospitalisation and transmission	$103,978
Index and secondary case utility	Treatment, AEs, management, hospitalisation and transmission	$90,728

AEs = adverse events; ICER = incremental cost-effectiveness ratio

The incremental cost-effectiveness of NAAT in each of the additional scenarios is presented in Table .

Table Incremental cost-effectiveness ratios for additional scenarios

-	AFB	AFB + NAAT	Increment
TB low scenario	-	-	-
Costs	$2,105.43	$2,148.83	$43.39
Outcomes	1.394	1.396	0.002
ICER	-	-	$18,533
Perfect clinical judgment	-	-	-
Costs	$2,016.20	$2,119.79	$103.59
Outcomes	1.397	1.397	0.0001
ICER	-	-	$724,423
TB high scenario	-	-	-
Costs	$6,544.41	$6,692.36	$147.95
Outcomes	1.391	1.391	0.0001
ICER	-	-	$1,713,838

AFB = acid-fast bacilli; ICER = incremental cost-effectiveness ratio; NAAT = nucleic acid amplification test

NAAT is most cost-effective in the scenario in which all patients are managed as though they have a low clinical suspicion of TB. In this scenario, treatment decisions are driven by the results of AFB ± NAAT. The PASC protocol indicated that when AFB and NAAT results were discordant, treatment decisions would be based on NAAT, and so the introduction of NAAT in this scenario has the greatest ability to change patient management. NAAT is more sensitive than AFB, and this leads to improved management in patients with TB as there are fewer false-negative results. In patients without TB, change in patient management with NAAT occurs in two ways:

• 
  in patients with an initial AFB (false)-positive result, NAAT may correctly identify no TB (i.e. NAAT may improve patient management), and decreases in the specificity of NAAT lead to the same treatment outcomes as if tested with AFB
  
• 
  in patients with an initial AFB (true)-negative result, NAAT may falsely identify TB, and decreases in the specificity of NAAT lead to more false-positive results (i.e. NAAT leads to detrimental patient management).
  

Given that the pooled estimate of AFB specificity is high (98%), false-positive results are likely to be driven by the specificity of NAAT in AFB-negative results. As the pooled estimate for this parameter and that in AFB-positive results in low-incidence countries are also both high (99% and 97%, respectively), the net effect is a reduction in the number of false-positive results.

The cost-effectiveness of NAAT is least in the scenarios in which all patients (with TB, and with or without TB, respectively) are managed as though they have a high clinical suspicion of TB. In these scenarios treatment initiation decisions are based on clinical judgement, with the benefit of NAAT restricted to identifying drug resistance to initiate an appropriate treatment earlier. As the prevalence of MDR-TB in those with TB is approximately 2%, the benefits of NAAT are accrued in a very small proportion of the population tested (0.44%). As NAAT is associated with a net reduction in the number of false-positive results, the perfect clinical suspicion scenario, in which treatment initiation in true TB-negative patients is based on the results of AFB ± NAAT, is more cost-effective than the TB high scenario, in which all patients receive treatment.

Given that the TB mixed population is a combination of patients with high and low clinical suspicion of TB, it is unsurprising that the cost-effectiveness of NAAT lies between the respective estimates observed when all suspects are treated as though they have either a low or a high clinical suspicion of TB.