OBJECTIVES: Cefiderocol is a siderophore cephalosporin antibiotic approved by the FDA in 2019 for treating complicated urinary tract infections (cUTI) in patients 18 or older with limited or no alternative treatment options. Artificial intelligence (AI) is a new tool for conducting health economic evaluations. We conducted a cost-effectiveness analysis of Cefiderocol versus imipenem/cilastatin in treating cUTI in hospitalized patients using ChatGPT and TreeAge.

METHODS: We conducted the CEA from the payer's perspective. We extracted clinical cure outcomes from the FDA Review, drug costs from average sales price published by the Centers for Medicare and Medicaid Services, life expectancy data from the Centers for Disease Control and Prevention, hospital costs from the Healthcare Cost and Utilization Project, and quality of life data from published literature. The analysis included direct health care costs and clinical outcomes, including clinical cure rate and quality-adjusted life years (QALYs). Costs were converted to 2023 constant dollars. The costs and outcomes were not discounted. A multivariate sensitivity analysis using Monte Carlo simulation was also performed.

RESULTS: The incremental cost-effectiveness ratio (ICER) per QALY was $134,350 for ChatGPT, and $134,680 for TreeAge. ChatGPT and TreeAge produced similar charts and cost-effectiveness acceptability curves. The results of the analysis were not sensitive to changes in the parameters used in the model. The small differences in results were likely due to rounding in calculation.

CONCLUSIONS: The ICER of Cefiderocol exceeded the threshold of $100,000 often cited as the maximum willingness to pay for commonly used drugs. ChatGPT provided similar results to TreeAge. In addition, ChatGPT can interpret analysis results and explain its calculations.