Fig 1.
Matching and imputation scheme for visit 2.
At each visit, 14,275 out of 14,348 participants were included after exclusion for missing age, sex, race, body mass index, antihypertensive medication use and antihypertensive medication class as well as if they had only 1 visit’s blood pressure (BP). Among the 14,275 participants without missing data, 9,555 participants were not taking antihypertensive medications. 1,078 of these participants were not taking antihypertensive medications with elevated BP. These untreated hypertensive participants were matched to one of the treated hypertensive participant (N = 4,720) to simulated the treated BP.
Table 1.
Summary of participant characteristics at visit 2.
Table 2.
Accuracy of imputed underlying BP to measured untreated BP.
Fig 2.
Bland-Altman plot of imputed underlying SBP at visit 2.
The measured untreated systolic blood pressure (SBP) was treated as gold standard for the Bland-Altman plots with confidence intervals at 2.5%-97.5%. Imputed underlying SBP was derived using methods 2–5 (A-D).
Table 3.
Bias in intercept and age associations from linear mixed effects models.