Derivation and validation of REASON: A risk score identifying candidates to screen for peripheral arterial disease using ankle brachial index

Background: The recommendation of screening with ankle brachial index (ABI) in asymptomatic individuals is controversial. The aims of the present study were to develop and validate a pre-screening test to select candidates for ABI measurement in the Spanish population 50-79 years old, and to compare its predictive capacity to current Inter-Society Consensus (ISC) screening criteria. Methods and results: Two population-based cross-sectional studies were used to develop (n = 4046) and validate (n = 3285) a regression model to predict ABI < 0.9. The validation dataset was also used to compare the model's predictive capacity to that of ISC screening criteria. The best model to predict ABI < 0.9 included age, sex, smoking, pulse pressure and diabetes. Assessment of discrimination and calibration in the validation dataset demonstrated a good fit (AUC: 0.76 [95% CI 0.73-0.79] and Hosmer-Lemeshow test: chi(2): 10.73 (df = 6), p-value = 0.097). Predictions (probability cut-off value of 4.1) presented better specificity and positive likelihood ratio than the ABI screening criteria of the ISC guidelines, and similar sensitivity. This resulted in fewer patients screened per diagnosis of ABI < 0.9 (10.6 vs. 8.75) and a lower proportion of the population aged 50-79 years candidate to ABI screening (63.3% vs. 55.0%). Conclusion: This model provides accurate ABI < 0.9 risk estimates for ages 50-79, with a better predictive capacity than that of ISC criteria. Its use could reduce possible harms and unnecessary work-ups of ABI screening as a risk stratification strategy in primary prevention of peripheral vascular disease. (C) 2010 Elsevier Ireland Ltd. All rights reserved.