Good Practices and Promising Interventions, Technical Series No. 8: A Cost-Effectiveness Analysis of Mainstreaming Chest X-Ray Screening with Artificial Intelligence-Powered Computer-Aided Detection or Human Readers in Public Facilities

Agency for International Development (USAID), undertook a cost-effectiveness analysis of mainstreaming chest X-ray (CXR) screening with artificial intelligence (AI)-powered computer-aided detection (CAD) or human readers in public facilities for intensified tuberculosis (TB) case finding. The study aimed to document the health and socioeconomic benefits and cost-effectiveness of mainstreaming CXR screening in public facilities. The study built on the initial pilot implementation of TB IHSS's innovative ICF model that involves CXR screening with AI-powered CAD in two large tertiary public hospitals. The study estimated the incremental cost and health impacts of mainstreaming CXR screening over a period of ten years, vis-à-vis ICF by symptom-based screening. It considered all the additional capital and recurrent requirements of mainstreaming CXR screening with either AI-powered CAD or human readers per facility. The study estimated the health impact of the two CXR screening alternatives, SA2 and SA3, based on the demand for CXR screening and TB diagnostic and treatment outcomes in pilot hospitals. Results suggest that mainstreaming CXR screening is very cost-effective. Its incremental cost-effective ratio (ICER) is PHP 43,376 per disability-adjusted life-year (DALY) averted with AI-powered CAD and PHP 47,667 per DALY averted with human readers, both below the gross domestic product (GDP) per capita in 2020 (PHP 163,701). The study also described the socioeconomic benefits of mainstreaming CXR screening with either AI-powered CAD or human readers in public health facilities. The results indicate that mainstreaming CXR screening can help reduce delayed or missed TB case detection due to prolonged turnaround time (TAT). The study estimated the additional costs and benefits of mainstreaming CXR screening, including the reduction in TB incidence, productivity losses averted, and time savings. The study found that mainstreaming CXR screening with either AI-powered CAD or human readers can lead to significant cost savings, including out-of-pocket (OOP) cost savings and productivity losses averted. The cost savings per case screened and diagnosed are PHP 370.22 with AI-powered CAD and PHP 370.22 with human readers. The cost savings as a percent of household income are 1.04% with AI-powered CAD and 1.04% with human readers. The study's results suggest that mainstreaming CXR screening with either AI-powered CAD or human readers is a promising solution to some of the key limitations of the current ICF strategy in the Philippines. The study's findings can inform policy and program decisions on the implementation of CXR screening in public facilities for intensified TB case finding.