Decision Tree Prediction of Saline Infusion Test Outcomes in Suspected Primary Aldosteronism: A Multicenter CDM Study

Confirmatory aldosterone suppression testing is widely utilized following positive screening for primary aldosteronism (PA), yet its incremental diagnostic value remains debated. While the 2025 Endocrine Society guideline proposes a probability-based approach, validated tools to translate this concept into clinical decision-making, particularly in Asian populations, are lacking.

Using a multicenter Common Data Model (CDM) across three tertiary hospitals (2002-2020), we retrospectively identified 307 patients who underwent SIT for suspected PA. SIT positivity was defined as post-infusion plasma aldosterone concentration (PAC) ≥10 ng/dL. We developed a conditional inference classification tree using screening-stage variables and compared its performance with conventional rigid screening criteria based on PAC and the aldosterone-to-renin ratio (ARR).

Among 307 patients (mean age 51.2 years), 153 (49.8%) were SIT-positive. SIT-positive patients had higher PAC (median 32.9 vs. 12.7 ng/dL, p<0.001) and lower serum potassium levels (3.66 vs. 4.02 mmol/L, p<0.001). The decision tree identified a high-probability phenotype defined by PAC >25 ng/dL and PRA ≤0.55 ng/mL/h, with an SIT positivity rate of 96.4%. In patients with PAC 15.7-25.0 ng/dL, potassium ≤3.9 mmol/L further stratified SIT positivity to 68.8%, whereas patients with PAC ≤15.7 ng/dL had a low positivity rate. The tree-derived criteria outperformed conventional cut-offs, with an accuracy of 0.85, sensitivity of 0.83, and specificity of 0.86, and performance remained robust in internal validation and sensitivity analyses.

A screening-stage decision tree provides clinically interpretable probabilities for SIT positivity, supporting risk-stratified confirmatory testing and selective omission in high-probability patients.