ALT-triggered reflex AST testing in health check-ups: A decision curve analysis

Aspartate aminotransferase (AST) is frequently co-ordered with alanine aminotransferase (ALT) in routine health check-ups, although universal AST testing may have a low clinical yield in asymptomatic individuals. We evaluated ALT-triggered reflex AST testing via decision curve analysis (DCA) to identify implementable screening policies.

We retrospectively analysed adult health check-up records with paired ALT and AST results. Reflex strategies were simulated in which AST was measured only when ALT exceeded predefined thresholds (15-60 U/L). Pathological AST was defined via sex-specific upper reference limits (URL). Clinical utility was assessed via DCA, where threshold probability (pt) represents the trade-off between unnecessary testing and missed pathological results. The primary analysis prespecified pt. = 1%, with pt. = 1%-5% examined in sensitivity analyses.

Among 46,059 participants (20,489 females; 25,570 males), AST ≥ URL was observed in 5.52% of females and 7.35% of males. At pt. = 1%, DCA supported sex-specific ALT triggers of ≥19 U/L for females and ≥ 26 U/L for males. These policies reduced AST testing to 34.8% and 46.7%, respectively, while maintaining sensitivities of 97.9% and 98.0%, respectively. The number of individuals with pathological AST who would be missed under the reflex policy was 11.7 and 14.9 per 10,000 screened. Across low pt. ranges, reflex strategies consistently achieved greater net benefits than did universal testing.

ALT-triggered reflex AST testing can substantially reduce low-yield AST measurements during health check-ups with minimal loss of screening sensitivity. The DCA provides a transparent framework for selecting sex-specific, screening-oriented policy thresholds.