Appendix

(1)

Manuscript title

Cost-effectiveness of testing for latent tuberculosis infection in people living with human immunodeficiency virus

Running head: Testing for LTBI in people living with HIV

Peter E. AUGUSTE, Hema MISTRY, Noel D. McCARTHY, Paul A. SUTCLIFFE, Aileen E.

CLARKE

(2)

Decision tree structure: screening strategies

TST: People being screened with TST, may or may not return to have their test result interpreted.

People with positive TST results (induration ≥ 5mm) are assessed for initial active TB by a chest x- ray and sputum examination. People with a positive result on the chest x-ray and sputum examination are treated for active TB. We assumed that the chest x-ray and sputum examination is 78% sensitive and 51% specific for diagnosing people who have initial active TB.^[18] People who adhered to TB treatment may develop isoniazid-induced hepatitis and can survive or die from this adverse event.

People with a negative result on the chest x-ray and sputum examination (LTBI) can either accept or refuse to be treated for LTBI. People who have accepted LTBI treatment may adhere/not adhere to treatment. If the TST is not read or negative, the model assumes that people are not followed-up.

IGRA (QFT-GIT/T-SPOT.TB): People being screened with an IGRA may have a determinate or an indeterminate result. People with a positive result are assessed for initial active TB by a chest x-ray and sputum examination. People with a positive result on the chest x-ray and sputum examination are treated for active TB. People who have a negative result on the chest x-ray and sputum examination can either accept or refuse to be treated for LTBI. People who have accepted LTBI treatment can adhere or not adhere to treatment. People who adhered to TB treatment may develop isoniazid- induced hepatitis and can survive or die from this adverse event. People with an indeterminate IGRA result receive a second IGRA test which is the same as the initial IGRA. If the IGRA is negative, the individual is not followed-up.

Sequential: When screening with a sequential strategy, people receive the QFT-GIT test first. If the result is positive, people receive a chest x-ray and sputum examination to detect initial active TB. If there is a positive result on the chest x-ray and sputum examination people are treated for active TB.

If the result is negative, people can accept or refuse treatment for LTBI. People who have accepted and adhered to LTBI treatment may develop isoniazid-induced hepatitis and can survive or die from this adverse event. People with negative QFT-GIT results undergo a TST test, and they follow similar pathways for those who received the TST alone strategy. People with an indeterminate QFT-GIT

(3)

3 result receive a second QFT-GIT test, which is the same as the initial QFT-GIT. These people follow similar pathways as those who received one QFT-GIT test. At most, people will receive two QFT-GIT tests. People with negative QFT-GIT or second indeterminate results, negative TST result or the TST result has not been read, the individual is not followed-up.

No testing (or treatment): This strategy does not consider any screening for LTBI/initial active TB or their treatment. People with LTBI that is likely to progress to active TB will eventually progress and those who progress to active TB will remain in this health state until they die.

The model begins with people receiving one of these testing strategies and further being categorised as having LTBI that progress to active TB/no LTBI, followed by the probability of test results.

However, in clinical practice, the test result is known before LTBI is diagnosed. Modelling the test result first followed by disease category or vice versa makes no mathematical difference in terms of the expected values calculated for each diagnostic strategy.^[19]

Markov model

Stage two is a disease progression model, which models the progression between no TB, LTBI that will progress to active TB, and active TB. The illustrative model structure is shown in Figure 2. This illustrative structure is identical for people who were/were not treated for latent/active TB (in stage 1), but with different transmission probabilities in each of these cases. The outputs (proportions and outcomes) of stage one of the model are used to determine the starting proportions of people in each state, specifically: 1) No LTBI/TB, 2) LTBI, 3) Active TB, 4) Resolved TB- treated for active TB.

This stage of the model shows disease progression of LTBI to TB in the absence of any intervention.

People incur costs (at stage 1) and accrue benefits, measured as quality-adjusted life years (QALYs) according to the length of time occupied in each health state.

(4)

Scenario analysis results

Briefly, Auguste et al. (2019).^[12] derived alternative estimates of sensitivity and specificity by including results from single arm studies with those that compared between tests in their meta- analyses. Including this additional information resulted in the following results for TST sensitivity (0.46, 95% credible interval: 0.18, 0.80), specificity (0.87, 95%CrI: 0.68, 0.95), QFT-GIT sensitivity (0.59, 95%CrI: 0.19, 0.90), specificity (0.87, 95%CrI: 0.70, 0.95) and T-SPOT.TB sensitivity (0.57, 95%CrI: 0.23, 0.85) and specificity (0.82, 95%CrI: 0.54, 0.95). Scenario analysis results are reported in Appendix Table 4 to 6. These results show that in all strategies except screening with T-SPOT.TB, the expected total diagnostic errors reduced. Screening with T-SPOT.TB is expected to yield more false positives, and thus, an increase to the overall diagnostic errors (see Appendix Table 4). We noticed modest changes to the expected correct diagnoses (see Appendix Table 5), which had little impact to the ICERs with undominated strategies remaining cost-effective (see Appendix Table 6).

(5)

5

Tables for appendix

Cost-effectiveness of testing for latent tuberculosis infection in people living with human immunodeficiency virus

Running head: Testing for LTBI in people living with HIV

CLARKE

(6)

Table 1: Model assumptions

People are tested once unless they received an indeterminate result and are then re-tested.

People being assessed for initial active TB undergo chest radiography and, if positive, receive a sputum examination.

Sputum examination is 100% accurate when diagnosing initial active TB.

People who have been diagnosed with initial active TB accept.

Only people who adhered to TB-preventative treatment or TB treatment is at risk of isoniazid-induced hepatotoxicity.

People who do not adhere to LTBI/TB treatment discontinue after one month.

There is a 0.004 probability of isoniazid-induced hepatitis following treatment.

People who do not adhere to LTBI treatment are not at risk of developing isoniazid-induced hepatotoxicity.

Table 2: Deterministic analysis results based on cost per diagnostic error avoided (2018/19 prices) Strategy Estimated

mean diagnosis costs (£)

Incremental mean diagnosis costs (£)

False positives

False negatives

Total diagnostic errors

Incremental diagnostic error

ICER (cost per diagnostic error avoided) QFT-GIT -ve

followed by TST5mm

£276.50 - 0.2871 0.0038 0.2909 - Dominated

T-SPOT.TB £189.16 -£87.34 0.1661 0.0095 0.1756 0.1153 Dominated

TST5mm £136.95 - 0.1655 0.0095 0.1750 - N/A

QFT-GIT £163.59 £26.64 0.1466 0.0090 0.1556 0.0194 Extendedly

dominated

No screening £0.00 £136.95 0.0000 0.0000 0.0000 0.1750 £782.57

ICER, incremental cost-effectiveness ratio; QFT-GIT, quantiferon gamma-release assay; TST, tuberculin skin test

(7)

7 Table 3: Deterministic analysis results based on cost per correct diagnosis (2018/19 prices)

Strategy Estimated

mean diagnosis costs (£)

Incremental mean diagnosis costs (£)

Effects (correct diagnosis)

Incremental effects

ICER (cost per correct diagnosis)

No screening £0 - 0 - -

TST5mm £136.95 £136.95 0.7650 0.7650 £178.43

QFT-GIT £163.59 £26.63 0.8435 0.0785 £339.43

T-SPOT.TB £189.16 £25.58 0.8235 -0.0200 Dominated

QFT-GIT -ve followed by TST5mm

£276.50 £112.91 0.6578 -0.1857 Dominated

(8)

8 Table 4: Probabilistic sensitivity analysis results based on cost per QALY (2018/19 prices)

mean costs (£)

Incremental mean costs (£)

Effects (QALYs)

Incremental QALYs

ICER (cost per QALY)

No screening £0 - 21.2882 - -

TST5mm £137.12 £137.12 21.3017 0.0135 £10,157

QFT-GIT £162.46 £25.34 21.3038 0.0021 £12,067

£276.05 £113.59 21.3094 0.0056 £20,284

ICER, incremental cost-effectiveness ratio; QALY, quality adjusted life-year; QFT-GIT, quantiferon gamma-release assay; TST, tuberculin skin test

(9)

9 Table 5:Scenario analysis results based on cost per diagnostic error avoided (2018/19 prices)

Strategy Estimated mean diagnosis costs (£)

Incremental mean diagnosis costs (£)

False positives

False negatives

Total diagnostic errors

Incremental diagnostic error

ICER (cost per diagnostic error avoided) QFT-GIT -ve

followed by TST5mm

£239.35 - 0.2309 0.0045 0.2354 - Dominated

T-SPOT.TB £195.62 -£43.73 0.1759 0.0095 0.1854 -0.0500 Dominated

QFT-GIT £150.68 -£88.67 0.1270 0.0090 0.1360 -0.0094 Dominated

TST5mm £105.50 - 0.1195 0.0112 0.1307 - N/A

No screening £0.00 £105.50 0.0000 0.0000 0.0000 0.1307 £105.37

(10)

10 Table 6: Scenario analysis results based on cost per correct diagnosis (2018/19 prices)

mean diagnosis costs (£)

Incremental mean diagnosis costs (£)

Effects (correct diagnosis)

Incremental effects

ICER (cost per correct diagnosis)

No screening £0 - 0 - -

TST5mm £105.50 £105.50 0.8093 0.8093 £130.36

QFT-GIT £150.68 £45.18 0.8631 0.0537 £840.72

£239.35 £88.67 0.7122 -0.1509 Dominated

(11)

11 Table 7: Deterministic sensitivity analysis results based on cost per QALY (2018/19 prices)

mean costs (£)

No screening £0 - 21.2881 - -

TST5mm £105.50 £105.50 21.2996 0.0114 £9,254

QFT-GIT £150.68 £45.18 21.3037 0.0041 £11,020

£239.35 £88.67 21.3085 0.0048 £18,473

ICER, incremental cost-effectiveness ratio; QALY, quality adjusted life-year; QFT-GIT, quantiferon gamma-release assay;

TST, tuberculin skin test

In Table 8, we report the results of the scenario analysis based on test accuracy information obtained from Kowada 2014. We chose this study because both analyses included similar testing strategies. We excluded the QFT-GIT followed by TST strategy as this information was not available. We obtained sensitivity and specificity for TST (0.43, 95%CI: 0.37, 0.50), (0.59, 95%CI: 0.46, 0.73), respectively.

Sensitivity and specificity for QFT-GIT (0.61, 95%CI: 0.54, 0.67), respectively. Sensitivity and specificity for T-SPOT.TB (0.65, 95%CI: 0.56, 0.74) and (0.98, 95%CI: 0.94, 0.99), respectively.

The results show that QFT-GIT and T-SPOT.TB were cost-effective at identify LTBI in PLWHIV, with screening with TST being dominated by T-SPOT.TB. Kowda 2014 showed that T-SPOT.TB was more effective and less costly, while our results showed that T-SPOT.TB was the most effective but more costly than QFT-GIT, which is a result of the cost of the T-SPOT.TB used our analysis.

(12)

12 Table 8: Deterministic scenario analysis results based on cost per QALY (2018/19 prices) and using test accuracy obtained from Kowada 2014

mean costs (£)

No screening 0.00 - 21.2881 - -

QFT-GIT 72.35 72.35 21.3042 0.0161 £4,499

T-SPOT.TB 92.20 19.85 21.3053 0.0011 £18,829

TST5mm 274.49 182.30 21.2988 -0.0065 Dominated

ICER, incremental cost-effectiveness ratio; QALY, quality adjusted life-year; QFT-GIT, quantiferon gamma-release assay;

TST, tuberculin skin test

(13)

13

Figures for appendix

Cost-effectiveness of testing for latent tuberculosis infection in people living with human immunodeficiency virus

Running head: Testing for LTBI in people living with HIV

CLARKE

(14)

14

Go to Active TB

Go to B

Go to C

Go to D Go to No LTBI

Appendix Figure 1: Decision tree pathway for PLWHIV undergoing LTBI testing

(15)

15 Go to B

Appendix Figure 1a: Pathway for the IGRA alone diagnostic strategy

Active TB Active TB Active TB Active TB LTBI- treated

LTBI- treated LTBI- treated LTBI- untreated

LTBI- untreated LTBI- untreated LTBI- untreated LTBI- treated LTBI- treated LTBI- treated LTBI- treated

B

(16)

16 Appendix Figure 1b: Pathway for the TST alone diagnostic

strategy C

(17)

17 Go to B

Go to C

Appendix Figure 1c: Pathway for the diagnostic strategy IGRA negative followed by TST D

(18)

18 Appendix Figure 2: Markov model

No LTBI/TB

LTBI

Active TB

Death – all cause

Death – all cause Death – TB

Death – all cause