Objectives: To assess the validity of the San Antonio heart study (SAHS) diabetes prediction model in a large representative Iranian population.
Methods: A risk function derived from data in the SAHS to predict the 7.5-year risk of diabetes, was tested for its ability to predict incident diabetes in 3,242 individuals aged >or=20 years. The performance or ability to accurately predict diabetes risk, of the SAHS function compared with the performance of risk functions developed specifically from the Tehran lipid and glucose study. Comparisons included goodness of fit, discrimination, and calibration.
Results: The participants were followed for 6.3 years. The area under the receiver operating characteristic curve (AROC) for diabetes of SAHS model was 0.83 (95% CI 0.80-0.86). The model overestimated the risk of diabetes in TLGS population with the overall bias of 111%. After the recalibration, the model-predicted probability agreed well with the actual observed 6-year risk of diabetes.
Discussion and conclusion: The American SAHS was a prediction model for diabetes with good discrimination in an Iranian target population after calibration.