Predictive value of Ankle-Brachial Index variability for symptom progression in peripheral arterial disease: a prospective cohort study.
The ankle-brachial index (ABI) is an important indicator for assessing the severity of peripheral arterial disease (PAD), but the limitations of single measurements are increasingly recognized. This study aimed to explore the predictive value of diurnal ABI variability for PAD symptom progression and to develop a prediction model incorporating ABI variability.
This prospective cohort study enrolled patients diagnosed with PAD, stratified by Fontaine staging. All patients underwent ABI measurements at three time points (8:00, 12:00, 18:00) at baseline, with calculation of ABI mean, range, and coefficient of variation. The primary endpoint was symptom progression during follow-up. Generalized additive models (GAM) were used to evaluate the relationship between ABI variability and symptom progression, with comparison to restricted cubic splines (RCS) and linear models.
The final analysis included 680 PAD patients followed for 40 months, with 225 (33.1%) experiencing symptom progression. Diurnal ABI variability (range, standard deviation, and coefficient of variation) showed differences between progression and non-progression groups, but these were not statistically significant (P=0.079). GAM modeling revealed that age (P<0.001) and mean ABI (P=0.032) were significant predictors of symptom progression, while ABI range alone was not significant as a predictor (P=0.214). The model demonstrated good calibration (Hosmer-Lemeshow Test P=0.1327) and moderate discrimination (AUC=0.688, 95%CI: 0.643-0.733). Patients were stratified into three risk groups based on predicted risk: low-risk (≤18%), medium-risk (19-29%), and high-risk (≥30%), with actual event rates of 13.2%, 23.7%, and 36.8%, respectively, showing good agreement with predicted risks. Kaplan-Meier analysis revealed significant differences in progression-free survival when stratified by ABI severity (P=0.013), but not when stratified by ABI variability (P=0.79).
Mean ABI and age are the primary predictors of PAD symptom progression, while diurnal ABI variability may provide additional predictive value in specific patient subgroups. Prediction models incorporating multiple time-point ABI measurements facilitate risk stratification and individualized management of PAD patients.
This prospective cohort study enrolled patients diagnosed with PAD, stratified by Fontaine staging. All patients underwent ABI measurements at three time points (8:00, 12:00, 18:00) at baseline, with calculation of ABI mean, range, and coefficient of variation. The primary endpoint was symptom progression during follow-up. Generalized additive models (GAM) were used to evaluate the relationship between ABI variability and symptom progression, with comparison to restricted cubic splines (RCS) and linear models.
The final analysis included 680 PAD patients followed for 40 months, with 225 (33.1%) experiencing symptom progression. Diurnal ABI variability (range, standard deviation, and coefficient of variation) showed differences between progression and non-progression groups, but these were not statistically significant (P=0.079). GAM modeling revealed that age (P<0.001) and mean ABI (P=0.032) were significant predictors of symptom progression, while ABI range alone was not significant as a predictor (P=0.214). The model demonstrated good calibration (Hosmer-Lemeshow Test P=0.1327) and moderate discrimination (AUC=0.688, 95%CI: 0.643-0.733). Patients were stratified into three risk groups based on predicted risk: low-risk (≤18%), medium-risk (19-29%), and high-risk (≥30%), with actual event rates of 13.2%, 23.7%, and 36.8%, respectively, showing good agreement with predicted risks. Kaplan-Meier analysis revealed significant differences in progression-free survival when stratified by ABI severity (P=0.013), but not when stratified by ABI variability (P=0.79).
Mean ABI and age are the primary predictors of PAD symptom progression, while diurnal ABI variability may provide additional predictive value in specific patient subgroups. Prediction models incorporating multiple time-point ABI measurements facilitate risk stratification and individualized management of PAD patients.