Development And Validation Of A Prediction Equation For Practical Implementation Of Blood Flow Restriction Exercise

During exercise with blood flow restriction (BFR), cuff pressures should be based on arterial occlusion pressure (AOP). Prediction equations using anthropometric, hemodynamic, and sociodemographic variables have been used to estimate AOP. Most of these equations have been designed for use with expensive cuff systems and implementation is limited to practitioners. PURPOSE: To develop and validate an equation to predict lower-body AOP when utilizing an 18 cm wide thigh sphygmomanometer. METHODS: Healthy adults (women = 60, men = 92, age: 24 ± 5 yrs) underwent measures of thigh circumference (TC), systolic (SBP), and diastolic blood pressure (DBP). Lower-body AOP was assessed in a seated position (via Doppler ultrasound) at the post-tibial artery using an 18 cm wide thigh sphygmomanometer. Variables (TC, SBP, DBP, age, sex) were entered into a least absolute shrinkage and selection operator linear regression and used to predict AOP. The model was evaluated in the development sample and internally validated using bootstrap resampling. Performance was assessed using Bland-Altman Limits of Agreement (BA LoA), coefficient of determination (R2), root mean squared error (RMSE), and calibration intercept and slope. RESULTS: The final model included the predictors TC, SBP, DBP, age, and sex. In the development sample, the model R2 was 0.55 and RMSE was 7.37 mmHg. The calibration intercept was -0.80 and slope was 1.01. The mean difference between observed and predicted values was 0.0 mmHg, 95% CI [-0.18, 0.18]. The BA LoA were ± 18.28 mmHg, 95% CI [±15.6, ±21.0]. During internal validation, the mean model R2 across bootstrapped samples was 0.55 ± 0.1, 95% CI [0.54, 0.55] and RMSE was 7.46 ± 6 mmHg, 95% CI [7.11, 7.82]. The mean calibration intercept was -0.68 ± 0.2, 95% CI [-0.70, -0.68] and slope was 1.0 ± 0.001, 95% CI [1.004, 1.005]. CONCLUSION: Our prediction equation offers a way to estimate lower-body AOP when utilizing an 18 cm wide thigh sphygmomanometer. This work expands on a previous prediction equation designed for this type of cuff by utilizing a large sample size, implementing regularization to reduce model optimism, and performing internal validation. Results indicate that our equation provides accurate and generalizable estimates of AOP that allow for implementation of BFR without the need for specialized equipment.