An assessment of nucleic acid amplification testing for active mycobacterial infection

Inputs to the economic evaluation

Prevalence

Prevalence of TB

The prevalence of TB, defined as culture-positive, in studies included for the diagnostic accuracy of NAAT conducted in a low TB incidence country (k=11) was 24% (range 660%) (Table , Appendix ).

A reliable estimate for the prevalence of TB in the population with clinical signs and symptoms of active TB in Australia was not identified during the assessment, nor were estimates of the respective prevalences where that patient group is divided into those considered to have either a high or low clinical suspicion of TB. This introduces considerable uncertainty in the economic modelling, as the cost-effectiveness is likely to be sensitive to these variables.

Given the uncertainties in prevalence estimates identified, further information was provided by the applicant (an Australian pathology provider)²⁰. It was estimated that 1020% of patients would be considered to have a high clinical suspicion of TB, of which 5070% would have TB. In those considered to have a low clinical suspicion of TB (the remaining 8090% of patients) the prevalence is estimated to be in the range 510%. Using the upper limits of these estimates provides an overall prevalence estimate of 22% (Table ). This value is reasonably similar to the prevalence of TB reported in the diagnostic accuracy studies conducted in low incidence countries (24%, Table , Appendix ), and so appears to have face validity.

It would be expected that the higher the proportion of patients with true TB that are treated based on clinical judgement, the less cost-effective NAAT will be, as there are fewer benefits of NAAT for patients managed this way; therefore, using the upper limit of these estimates is the conservative choice and will be used in the base-case analysis of the economic evaluation. However, it should be noted that if the overall prevalence of TB is an overestimate, the cost-effectiveness of NAAT may too be overestimated. Given that these are best-guess estimates, sensitivity analyses around these estimates will be presented.

Additional scenarios are presented to examine the extent to which treatment initiation decisions based on clinical suspicion affect the ICER. The base-case prevalence of 22% is maintained in these scenarios; however, all are managed as though they have either low or high clinical suspicion of TB, depending on the scenarios (Table ).

Table Prevalence estimates used in previously published economic evaluations of NAAT

Scenario	Proportion high clinical suspicion ^A	Prevalence (high clinical suspicion)^B	TB high clinical suspicion ^C(A × B)	Prevalence (low clinical suspicion) ^D	TB low clinical suspicion ^E ((1  A) × D)