Individualized Net Benefit of Intensive Blood Pressure Lowering Among Community-Dwelling Older Adults in SPRINT

Abstract

Background: The optimal blood pressure (BP) target for older adults with hypertension remains controversial, particularly among those with advanced age, frailty, or polypharmacy. This study estimated the individualized net benefit of intensive BP lowering among community-dwelling older adults in the Systolic Blood Pressure Intervention Trial (SPRINT).

Methods: Among 5143 SPRINT participants age ≥ 65 years, Cox models were internally validated to predict an absolute difference in risk between treating to a systolic BP target of < 120 versus < 140 mm Hg for all-cause death, cardiovascular outcomes, cognitive outcomes, and serious adverse events. Treatment effects were combined using simulated preference weights into individualized net benefits, representing the weighted sum of risk differences across outcomes. Net benefits were compared across categories of age (65-74 vs. ≥ 75 years), SPRINT-derived frailty status (fit, less fit, and frail), and polypharmacy (≥ 5 medications).

Results: When simulating preferences for participants who view the benefits of BP lowering (reduction in death, cardiovascular events, and cognitive impairment) as much more important than treatment-related harms (e.g., acute kidney injury and syncope), the median net benefit from intensive BP lowering was 4 percentage points (IQR: 3-6), and 100% had a positive net benefit favoring intensive BP lowering. When simulating benefits and harms to have similar, intermediate importance, the median net benefit was 1 percentage point (IQR: 0-2), and 85% had a positive net benefit. Participants with advanced age and frailty had greater net benefits from intensive BP lowering despite experiencing more harm in both simulations, and those with polypharmacy had greater net benefits when benefits were viewed as much more important than harms (p < 0.001 for all comparisons).

Conclusions: Among community-dwelling older adults with hypertension in SPRINT, almost all participants had a net benefit that favored a systolic BP target of < 120 mm Hg, but the magnitude of net benefit varied according to estimated risks and simulated preferences.

Keywords: frailty; hypertension; older adults; polypharmacy; shared decision‐making.

FIGURE 1

Individualized predicted difference in absolute risk of each outcome from intensive versus standard BP lowering in older persons. Individualized predicted difference in absolute risk was calculated as the difference in 4‐year predicted survival probability of the outcome using an individual trial participant's baseline characteristics, factual randomized treatment assignment, and counterfactual randomized treatment assignment. Negative values indicate an absolute risk reduction and positive values indicate an absolute risk increase with intensive versus standard BP lowering. The model for the cognitive impairment outcome calculates 5‐year predicted probabilities. Models assumed no treatment interaction with baseline characteristics. Panel A: Stratified by age (65–74 vs. ≥ 75 years); Panel B: Stratified by frailty status (fit [FI ≤ 0.10] vs. less fit [0.10 < FI ≤ 0.21] vs. frail [FI > 0.21]); Panel C: Stratified by polypharmacy status (less < 5 vs. ≥ 5 prescribed medications recorded at the baseline visit).

FIGURE 2

Distributions of individualized predicted net benefits from intensive BP lowering among older adults in SPRINT under different outcome preference simulations. The figure shows probability density function curves of the individualized predicted net benefit from intensive BP lowering under harm tolerant and harm averse simulated preference scenarios. The area under the curve to the right of 0% indicates the probability of a positive net benefit. Individualized predicted net benefit for each individual was calculated as the sum of the preference‐weighted individualized treatment effects for each outcome. Preference weights correspond to the relative importance of each outcome. Panel A shows that under “harm tolerant”¹ simulated preferences there was a 100% probability of a positive net benefit favoring intensive BP lowering among SPRINT participants aged 65–74 and ≥ 75 years. Panel B shows that under “harm averse”² simulated preferences there was a 74% probability of a positive net benefit from intensive BP lowering among SPRINT participants aged 65–74 and 97% probability of a positive net benefit from intensive BP lowering among those aged ≥ 75 years. ¹The “harm tolerant” preference weights were 1 for death; 0.8 for stroke and cognitive impairment; 0.7 for HF and MI or ACS; and 0.1 for kidney effects, hemodynamic effects, and injurious falls. ²The “harm averse” preference weights were 1 for death; 0.6 for stroke, cognitive impairment, HF, and MI or ACS; and 0.5 for kidney‐related harms, hemodynamic harms, and injurious falls.

FIGURE 3

Distributions of individualized predicted net benefits from intensive BP lowering among older adults in SPRINT, stratified by frailty. The figure shows probability density function curves of the individualized predicted net benefit from intensive BP lowering under harm tolerant and harm averse simulated preference scenarios. The area under the curve to the right of 0% indicates the probability of a positive net benefit. Individualized predicted net benefit for each individual was calculated as the sum of the preference‐weighted individualized treatment effects for each outcome. Preference weights correspond to the relative importance of each outcome. Results were stratified by frailty status (fit [FI ≤ 0.10] vs. less fit [0.10 < FI ≤ 0.21] vs. frail [FI > 0.21]). Panel A shows that under “harm tolerant”¹ simulated preferences there was a 100% probability of a positive net benefit favoring intensive BP lowering among SPRINT participants who were fit, less fit, or frail. Panel B shows that under “harm averse”² simulated preferences the probability of a positive net benefit from intensive BP lowering was 85%, 84%, and 86% among SPRINT participants who were fit, less fit, and frail, respectively. ¹The “harm tolerant” preference weights were 1 for death; 0.8 for stroke and cognitive impairment; 0.7 for HF and MI or ACS; and 0.1 for kidney effects, hemodynamic effects, and injurious falls. ²The “harm averse” preference weights were 1 for death; 0.6 for stroke, cognitive impairment, HF, and MI or ACS; and 0.5 for kidney‐related harms, hemodynamic harms, and injurious falls.