Abstract
High blood pressure (BP) is a risk factor for cognitive decline. Increasingly, studies have found the relationship to be nonlinear, with low BP also indicating higher risk. This UK Biobank study examines the nonlinear relationships between BP and cognitive function, including whether the relationships differ by cognitive domain. Systolic (SBP) and diastolic BP (DBP) were measured at baseline. Cognitive domains included fluid intelligence, attention, and reaction time, measured at baseline and over time. Nonlinear mixed-effects regression models, including natural spline terms for SBP and DBP, were used to assess the relationships. Additional models evaluated interactions with age, sex, and hypertension history/antihypertensive use. There were 439 301 (mean age = 56.3, SD = 8.1, 45.1% male) included participants. Baseline SBP had significant inverted U-shaped relationships with fluid intelligence (p < 0.0001), attention (p < 0.0001), and reaction time (p < 0.0001), with substantially different ideal SBPs for each domain (118, 127.5, and 150.5 mmHg, respectively). Baseline DBP had significant relationships with fluid intelligence (p < 0.0001) and attention (p < 0.0001), again with varying ideal DBPs (57.5 and 74.5 mmHg, respectively). Higher baseline SBP had a small, inverse relation with trajectories of attention during the study (p < 0.0001), but no relationship with trajectories of either fluid intelligence or reaction time. Older, male, and untreated hypertension subgroups had significantly poorer reaction time at lower baseline SBP and DBP (p < 0.0001). The relationship between BP and cognitive function is nonlinear with the three domains optimal at differing BP levels. Older persons, males, or hypertensive patients may be particularly susceptible to negative cognitive effects of low BP.