Passive TB screening based on TBSS followed routine activities at the outpatient department (OPD) of the TB health centers [16]. Patients with TBSS seek health care at OPDs; a nurse registers them and notifies them if they have presumptive TB. A medical doctor performs a general examination and recommends that they undergo a CXR examination.

The preparation phase of the mobile app screening strategy included research and development of the mobile app, consultation for sensitivity to the community with stakeholder meetings, and health staff training. The screening phase included community and household visits by health staff to identify individuals with presumptive TB via the app, upload their data for notification and registration at the corresponding TB health center, and refer them for CXR examination via a QR code referral form. The TB team leader supervised the activities at the active TB screenings.

The preparation phase of the CXR screening strategy included consultation for sensitivity to the community with stakeholder meetings and health staff training. The screening phase included community and household visits by health staff to provide health education, identify eligible individuals aged ≥15 years within the area, and refer these individuals for CXR examination.

Individuals with presumptive TB who accepted the CXR examination were examined with digital CXRs. Those with CXR findings suggesting TB received a Gene Xpert Mycobacterium tuberculosis/resistance to rifampicin (MTB/RIF) examination to detect active TB. The diagnostic activities included CXR examination by a technician and interpretation by a medical doctor, a Gene Xpert MTB/RIF examination by a microbiologist, and treatment initiation at an OPD according to national TB treatment guidelines. Differences in TB screening strategies are presented in Multimedia Appendix 2.

The operational costs were evaluated using a standardized framework for cost evaluation of TB screening programs during a community-based TB screening survey [17]. Costs were evaluated for 6 major activities: preparation, screening, notification of presumptive TB at the OPD, CXRs, Gene Xpert MTB/RIF examination, and treatment initiation at the OPD. The resources used and unit cost (in US $) per participant are described in Multimedia Appendix 3. The CXR machine can perform a minimum of 12 examinations per hour, operating for 8 hours per day, thereby resulting in approximately 24,000 examinations per year. The Gene Xpert MTB/RIF machine can run 4 simultaneous tests every 2 hours, operating for 8 hours per day, which amounts to approximately 3840 tests per year. For the TBSS strategy, one CXR machine and one Gene Xpert MTB/RIF machine were used. For the mobile app strategy, 3 of each machine were used, and for the universal CXR strategy, 5 of each machine were used.

The cost of human resources was calculated by multiplying the total working hours of health staff spent on TB activities by their hourly wage during the community-based TB screening survey. The hourly wage was US $0.94 for health staff and registered nurses, US $1.11 for CXR and Gene Xpert MTB/RIF technicians, and US $1.36 for medical doctors.

The total costs for traveling, mobile app use, and data uploading were calculated for all participants during the community-based TB screening. The costs for a CXR referral form and digital CXRs were estimated from market-based costs obtained from the procurement section of the national TB control program conducted in 2020 under the Global Fund.

The costs during the preparation phase were taken as capital costs for the entire program, targeting a community with a population of 100,000 to be screened [17].

The annualized capital cost of building and equipment purchases was calculated by dividing the current value of the asset by an annualizing factor, which was calculated based on the discount rate and useful life of the asset. A 3% discount rate was applied for all capital costs based on values from the Central Bank of Myanmar [18]. The useful life of a building was considered to be 50 years and for hospital equipment, 10 years, according to the Myanmar Ministry of Finance [19].

The following formula was to calculate the annualization factor [20]:

Overhead costs included annual maintenance of quality control and electricity consumption (kW/h) charges for the CXR and Gene Xpert MTB/RIF machines. Annual data for the CXR and Gene Xpert MTB/RIF examinations were collected for the 2021 fiscal year (April 2020-March 2021). All local costs in Myanmar kyats were converted to US $ at an exchange rate of 1 US $= 1464 Myanmar kyats (the exchange rate on January 1, 2021).

The patient costs included productivity loss (average US $1 per patient) and traveling to the CXR and Gene Xpert examinations (average US $0.05 per 1 kilometer) during the community-based survey.

The unit cost for preparation for screening was calculated based on the targeted population of 100,000. The unit costs for human resources, recurrent items, and patients were computed from the total costs divided by the total number of participants in the survey. The annualized capital and overhead costs were converted into the unit cost by dividing by the total number of examinations in a year.

The compliance with CXR examination was 38.2% for routine passive TB screening based on TBSS and 88% for active screening based on CXRs according to the recent 2018 National TB Prevalence Survey in Myanmar [21]. Active TB screening via the mobile health app had 71.1% CXR compliance for presumptive cases of TB detected by the app [8].

The incremental cost effectiveness ratio (ICER), which is the additional cost for additional DALYs averted, was calculated by using the following formula:

Theoretically, the ICER calculated from our sample is subject to uncertainty when the result is generalized to a large population. To calculate the 95% CI of this value, it is necessary to assume a certain type of distribution for the parameters. Beta and gamma distributions were used to align with established methodologies in prior studies [10,13]. Specifically, a beta distribution was used to estimate the 95% CI for continuous random variables with a range from 0 to 1, including the prevalence of TB, sensitivity and specificity of diagnostic methods, and screening compliance assessment. Meanwhile, a gamma distribution was applied to estimate the 95% CI for parameters with nonnegative values, spanning from 0 to positive infinity, such as costs and DALYs averted [26]. Based on these distributions, probabilistic sensitivity analysis of a Monte Carlo simulation was performed for comparing the 2 screening strategies [26].

The parameters input into the Monte Carlo simulation are outlined in Multimedia Appendices 5 and 6. After conducting the simulations, 100,000 pairs of cost and effectiveness values were obtained for each comparison. These data points were plotted on a plane, with the incremental DALYs averted on the x-axis and the incremental costs on the y-axis, which we termed the cost-effectiveness plane. There are 2 sets on the plane, each of which had 100,000 points. The gray points simulate the ICER of adding universal CXR compared to TBSS screening. The black points represent screening with the mobile app compared to screening with TBSS. The diagonal lines denote the WTP thresholds of 1, 2, and 3 times GDP per capita for each DALY averted. Above the 3-times-GDP line, the ICER is beyond the WTP threshold, that is, it is not cost-effective. The proportion of points below the threshold lines is the probability of the additional effort being cost-effective. These probabilities vary by the change in the slope of the threshold lines. For example, if the WTP threshold increases, the line will become steeper and the probability of the additional effort being cost-effective will increase.

As the WTP threshold varies, the relationship between the probability of the additional efforts being cost-effective was then plotted against the WTP threshold and shown as the cost-effectiveness acceptability curve [10]. R (version 4.0.0; R Foundation for Statistical Computing) was used for data analysis and graphs.

Table 2 shows the ICERs for the mobile app and universal CXR strategies vs TBSS screening. For the mobile app, the estimated ICER was US $1064 (72% of GDP per capita) per 1 DALY averted. The universal CXR strategy had an estimated ICER of US $3143 (2.1 times GDP per capita) per 1 DALY averted.

Figure 1 shows the relationship between the costs and DALYs averted of the 3 choices for TB screening. The additional cost of active TB screening was US $345,942 for the mobile app strategy and US $1,810,712 for the universal CXR strategy. The additional effectiveness was 325 DALYs averted with the mobile app strategy and 576 with universal CXR. Therefore, the universal CXR strategy resulted in an additional 77.2% DALYs averted compared to the app (576 – 325 / 325 DALYs averted). The TBSS screening strategy had the lowest costs and lowest DALYs averted. Universal CXR had the highest costs and DALYs averted. The mobile app was between the 2 strategies. The three sloped lined represent the WTP thresholds for 1 DALY averted per 1, 2, and 3 times GDP per capita. The slope for the ICER of mobile app screening was the steepest, indicating that it had the highest ICER below 1 times GDP per capita. The ICER slope for CXR was less steep than the ICER for the mobile app for 1 DALY averted per 1 and 2 times GDP per capita, but steeper at 1 DALY averted per 3 times GDP per capita.