2058 adults age 60–85 years from the 1999–2002 NHANES were used in this analysis. MAP and PP were calculated from brachial BP as 1/3 systolic + 2/3 diastolic BP, and systolic – diastolic BP, respectively. Cognition was assessed using the digit-symbol substitution test (DSST).
Unadjusted multiple regression revealed although PP independently predicted DSST score (β = − 0.21, p < 0.05), MAP was only predictive when entered simultaneously with PP (β = 0.06, p < 0.05). Further adjustments for sociodemographics and medical history significantly improved prediction of DSST score (∆ R2 = 0.37, p < 0.05), and removed the association between MAP and DSST score, although PP remained a significant predictor (β = − 0.07, p < 0.05).
PP, but not MAP, was significantly and inversely related to cognitive function among a representative sample of adults aged ≥ 60 years after controlling for sociodemographic information and medical history.