Table 2. Summary of the investment scenarios modelled.
| Investment | Model implementation | Parameter, events or resource changes | Assumptions | |
|---|---|---|---|---|
| Reduction in initial LTFU (in Fig 1; decision-point C and E) | All patients with positive TB test results start treatment within one month of testing—simulating a point-of-care or a track-and-trace scenario with active follow-up of people with a positive TB test result. Synergies with investment in a community health worker programme. | ptxfpos = 1—pMort_m1 (stratified by HIV and TB status) | Probability of starting treatment after positive (in month 1), from: | Monthly conditional probabilities of starting treatment from ‘out of care’ were estimated from the trial in the base scenario (reported in Table 1). In this investment scenario, patients shift from moving to the ‘out of care’ state if not started on treatment within one month, to the treatment state immediately, thus probabilities of starting treatment from ‘out of care’ approximate zero. The relative proportions of those starting various treatment types is kept the same as observed in the trial. |
| The probability of starting treatment from a positive test result was the remainder of all patients in that state after those who would die in that month had been subtracted. The mortality rate was stratified by HIV and TB status. | HIVneg: 0.882 to 1; | |||
| HIVneg_TB: 0.882 to 1; | ||||
| HIVpos: 0.802 to 1; | ||||
| HIVpos_TB: 0.802 to 1; | ||||
| ART: 0.944 to 1; | ||||
| ART_TB: 0.944 to 1 | ||||
| Empirical treatment from negative test result (in Fig 1; decision-point A) | The ability of healthcare workers to correctly act based on continued clinical symptoms, on the same day as the results visit (by giving TB treatment to those with test negative TB expressed as the sensitivity and specificity of that decision). This was based on the behaviour estimated from the microscopy arm of the model calibration and was applied to behaviour after a negative Xpert test result. | pnegpathfeg = 0 | Probability of the negative pathway after a negative test result, from: | Given the differences in health care worker behaviour after a microscopy test compared to a Xpert test result observed in the XTEND trial, we use the transition probabilities estimated from the microscopy arm of the trial [50, 51]. |
| ptreatfneg = value estimated from reported behaviour in the control arm of the XTEND study [49], under the assumption that behaviour observed after the implementation will revert back to pre-implementation levels. | HIVpos: 0.027 to 0.000 | |||
| Assumed that all have at least one visit to a public health clinic (and associated costs) after a negative test result for treatment initiation. | HIVpos_TB: 0.212 to 0.000 | |||
| ART: 0.037 to 0.000 | ||||
| ART_TB: 0.217 to 0.000 | ||||
| Probability of starting treatment after a negative test result, from: | ||||
| HIVneg: 0.002 to 0.040 | ||||
| HIVneg_TB:0.054 to 0.270 | ||||
| HIVpos: 0.009 to 0.180 | ||||
| HIVpos_TB: 0.072 to 0.360 | ||||
| ART: 0.003 to 0.060 | ||||
| ART_TB: 0.017 to 0.090 | ||||
| Improvements in the test-negative pathway (in Fig 1 decision-points B and D) | HIV-positive people with negative test results get further investigations (radiograph/culture) for TB, and a proportion are started on TB treatment, simulating additional investment in improving access to further diagnostic tests. | ptreatfneg = 0 | Probability of starting treatment after negative test result changes from: | Similar to the previous scenario, we model a healthcare worker behaviour change scenario based on the difference in observed behaviour between the microscopy and Xpert arms of the study. This scenario simulates a situation where there is an increase in the proportion of patients who receive further investigations after a negative test result. Therefore, we reduced all empirical treatment to 0 and all eligible patients received a radiograph as part of the negative pathway. |
| pnegpathfneg = 1 (stratified by HIV and TB status) | HIVneg: 0.002 to 0.000 | |||
| treatfnegpath = 0.10 (no TB); 0.80 (with TB) | HIVneg_TB:0.054 to 0.000 | |||
| The probability of starting treatment is shifted from following a negative test result to the decision to order further diagnostic tests. The probability of starting treatment after the negative pathway was 10% in those without TB, and 80% in those with TB. | HIVpos: 0.009 to 0.000 | |||
| Assumed that every person will accumulate two visits to the public clinic during the negative pathway, and that each person getting further tests will get at least one radiograph. | HIVpos_TB: 0.072 to 0.000 | |||
| ART: 0.003 to 0.000 | ||||
| ART_TB: 0.017 to 0.000 | ||||
| Probability of the negative pathway after a negative test result change from: | ||||
| HIVpos: 0.041 to 0.900 | ||||
| HIVpos_TB: 0.041 to 0.900 | ||||
| ART: 0.073 to 0.900 | ||||
| ART_TB: 0.073 to 0.900 | ||||
| Probability of treatment from negative pathway changes from: | ||||
| HIVpos: 0.027 to 0.100 | ||||
| HIVpos_TB: 0.212 to 0.800 | ||||
| ART: 0.037 to 0.100 | ||||
| ART_TB: 0.217 to 0.800 |
In the Table, the individual characteristics of the patients are labelled as HIVneg for people who are HIV negative and don’t have tuberculosis; HIVneg_TB for people who are HIV negative and have been diagnosed with tuberculosis; HIVpos for people who are HIV positive and don’t have tuberculosis; HIVpos_TB for people who are HIV positive and have been diagnosed with tuberculosis; ART represents the individuals who are HIV positive, on anti-retroviral therapy and don’t have tuberculosis; and ART_TB represents the individuals who are HIV positive, on anti-retroviral therapy and have been diagnosed with tuberculosis.