Skip to main content
. 2018 Sep 13;362:k3503. doi: 10.1136/bmj.k3503

Fig 3.

Fig 3

Predicted probability of antihypertensive intensification by lower and higher likelihood to benefit from strict blood pressure (BP) control. Top: groups with lower likelihood of benefit. P values for comparisons: life expectancy P=0.07, history of dementia P=0.95, and history of metastatic malignancy P=0.13. Bottom: groups with higher likelihood of benefit. P values for comparison: history of myocardial infarction P=0.53, congestive heart failure P=0.01, cerebrovascular disease P=0.37, and renal disease P=0.73. Predicted probabilities for comorbidities estimated following mixed effect logistic regression accounting for age category, sex, ethnicity, income, length of stay, Charlson comorbidity indicators, primary discharge diagnosis, year, hospital training status, inpatient BP, outpatient BP, an interaction term for inpatient and outpatient BP, and random effects to account for clustering by Veterans Affairs hospital. Predicted probabilities by life expectancy generated with similar models but replacing age and Charlson comorbidity indicators with a categorical life expectancy variable calculated from age and number of comorbidities. Error bars indicate 95% CI