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.

Summary.

• Key points

  • ○
    Nearly all SPRINT participants aged 65 years or older had a predicted net benefit (i.e., the difference between benefits and harms) that favored a systolic blood pressure (BP) target of less than 120 mm Hg compared with a target of less than 140 mm Hg, but the degree of net benefit varied considerably according to predicted risks and simulated outcome preferences.
  • ○
    Compared to community‐dwelling participants who are aged 65–74 years, fit, or without polypharmacy at baseline, those who are age 75 years or older, frail, or with polypharmacy experienced more treatment‐related harms from a lower systolic BP goal (e.g., acute kidney injury and syncope). However, they also had larger absolute cardiovascular, cognitive, and mortality benefits, resulting in greater net benefits under both “harm tolerant” and “harm averse” preference scenarios.
  • ○
    When accounting for an individual's risks and preferences regarding various outcomes related to BP lowering, the benefits of a systolic BP target of less than 120 mm Hg outweigh the harms for most community‐dwelling older adults, especially among high‐risk subgroups who are often assumed not to benefit from intensive BP treatment targets.
• Why does this paper matter?

  • ○
    Most older adults have hypertension, yet the optimal blood pressure (BP) target remains controversial. Guidelines recommend a shared decision‐making approach to selecting BP targets, but no framework is available.
  • ○
    This study demonstrates that evidence about BP targets from the Systolic Blood Pressure Intervention Trial (SPRINT) can be tailored to community‐dwelling older adults with hypertension by using each individual's estimated risks and preferences for multiple outcomes related to BP lowering.
  • ○
    Nearly all community‐dwelling older adults in SPRINT had a positive net benefit (i.e., predicted benefits exceeded harms) favoring a systolic BP target of < 120 mm Hg compared to < 140 mm Hg.
  • ○
    Moreover, community‐dwelling older adults with advanced age, frailty, or polypharmacy had greater net benefits favoring a lower systolic BP goal despite experiencing more treatment‐related harms.

1. Introduction

Hypertension affects nearly 75% of U.S. adults aged 65 years or older and substantially contributes to cardiovascular disease (CVD) risk and mortality [1, 2]. Results from the Systolic Blood Pressure Intervention Trial (SPRINT) support lower blood pressure (BP) targets among noninstitutionalized, ambulatory, community‐dwelling older adults to reduce the risk of CVD and mortality [3, 4]. However, lower BP targets also confer an increased risk of treatment‐related adverse events, including acute kidney injury (AKI), hypotension, and syncope. Balancing the potential benefits and harms of lower BP targets can be challenging when making hypertension treatment decisions for older adults who are often perceived to be at higher risk of treatment‐related harms, particularly those with advanced age, frailty, or polypharmacy [5, 6, 7]. Hypertension guidelines recommend clinical judgment and shared decision‐making to personalize BP targets, but no framework exists to tailor evidence about BP targets to community‐dwelling older adults or to incorporate patient preferences into BP target decisions [8, 9].

Previous work shows that the benefits and harms of intensive BP lowering vary widely across individuals [10, 11, 12, 13, 14]. Among older adults, intensive BP lowering significantly reduces major cardiovascular events and all‐cause mortality, although this is accompanied by an increased risk of adverse events such as hypotension, syncope, electrolyte imbalances, and AKI [15, 16, 17]. However, the benefits and harms have primarily been assessed separately rather than in combination using a decision‐analytic measure such as a net benefit (difference between benefits and harms) [18, 19]. Incorporating patient preferences about the relative importance of different endpoints allows for a direct comparison of the benefit‐harm trade‐offs according to an individual's personal values [20]. Several ancillary studies from SPRINT have shown that patient preferences can influence the benefit‐harm trade‐offs of lower BP targets, but none have focused on older adults [21, 22]. In addition, prior studies did not factor in the cognitive benefits of intensive BP lowering, which may be a particularly important consideration for older adults [23, 24].

The objectives of this study were to (1) estimate individualized benefits and harms of intensive BP lowering among community‐dwelling older adults in SPRINT; (2) combine these treatment effects into an individualized net benefit using simulated preference weights that reflect the relative importance of each outcome to a participant; and (3) compare the distributions of individualized net benefits from intensive BP lowering among older adults by age, frailty, and polypharmacy.

2. Methods

2.1. Study Design

The design and conduct of SPRINT have been reported previously [25]. SPRINT was an NIH‐funded, open‐label clinical trial that randomized participants with hypertension to an “intensive” systolic BP (SBP) target (< 120 mm Hg) versus a “standard” SBP target (< 140 mm Hg). Inclusion criteria were age ≥ 50 years; SBP 130–180 mm Hg; and high CVD risk (defined as prevalent clinical or subclinical CVD other than stroke, chronic kidney disease (CKD) [estimated glomerular filtration rate (eGFR) 20–59 mL/min/1.73 m²], age ≥ 75 years, or 10‐year CVD risk > 15% based on the Framingham risk score). Key exclusion criteria included an expected survival of less than 3 years, a clinical diagnosis of or treatment for dementia, unintentional weight loss (> 10% of body weight) in the preceding 6 months, or residence in a nursing home. In addition, those with diabetes mellitus, prior stroke or transient ischemic attack, eGFR < 20 mL/min/1.73 m², symptomatic heart failure, or a left ventricular ejection fraction < 35% were excluded [25]. A total of 9361 participants with hypertension and high CVD risk were enrolled between November 2010 and March 2013 across 102 sites in the United States and Puerto Rico. The SPRINT study was approved by Institutional Review Boards at each study site, and all participants provided written informed consent.

This ancillary analysis included SPRINT participants who were 65 years of age or older at baseline and was approved by the committees on human research at the University of California, Davis, the University of California, San Francisco, and the San Francisco Veterans Affairs HealthCare System. The data that support the findings of this study are available from the National Heart, Lung, and Blood Institute Biologic Specimen and Data Repositories and the corresponding author upon request.

2.2. Prediction Models

As previously described, parsimonious prediction models to estimate the risk of each potential benefit and harm from intensive BP lowering were developed and internally validated in the overall SPRINT population [26]. Among 34 candidate baseline clinical variables that are commonly available in the electronic health record, 7 were selected for inclusion in each prediction model: age, sex, current smoking, history of CVD, total number of medications, eGFR calculated using the race‐free 2021 Chronic Kidney Disease‐Epidemiology (CKD‐EPI) Collaboration equation for creatinine and cystatin C [27], and urine albumin‐to‐creatinine ratio (UACR). The same predictors were included in all models to facilitate clinical application. Additional details on model development in the overall SPRINT study and internal validation of models in older adult participants of SPRINT are in the Supporting Information. The relative treatment effect of intensive versus standard BP lowering on each outcome was assumed to be constant across levels of baseline risk [28].

2.3. Age, Frailty, and Polypharmacy Subgroups

Advanced age was defined as ≥ 75 years at baseline. Frailty status was defined according to a 36‐item frailty index (FI), which was extended to include gait speed as measured by a 4‐m walk test for those aged ≥ 75 years [29]. The FI incorporates global cognitive function based on the Montreal Cognitive Assessment [30], self‐rated health from the Veterans RAND 12‐Item Health Survey [31], depression symptoms from the nine‐item Patient Health Questionnaire (PHQ‐9) [32], BP measurements, body mass index, laboratory results, and self‐reported comorbidities [29]. The FI was calculated as the sum of the score for each deficit, divided by the total number of nonmissing items [29]. Participants with > 30 nonmissing items had an FI calculated [29]. Consistent with previous studies in SPRINT, participants were classified as fit (FI ≤ 0.10), less fit (0.10 < FI ≤ 0.21), or frail (FI > 0.21) [29]. Polypharmacy was defined as ≥ 5 prescribed medications recorded at the baseline visit, including antihypertensive medications [33].

2.4. Outcomes

We evaluated adjudicated outcomes in SPRINT that were categorized as benefits or harms based on whether intensive BP treatment decreased or increased their risk. Benefit outcomes included (1) all‐cause death; (2) stroke; (3) cognitive impairment, defined as the composite of probable dementia or mild cognitive impairment; (4) acute decompensated heart failure (HF); and (5) myocardial infarction (MI) or acute coronary syndrome (ACS) not resulting in MI. Harm outcomes included SPRINT's prespecified serious adverse events of interest: (6) kidney‐related harms, including AKI or serious electrolyte abnormality; (7) hemodynamic harms, consisting of syncope, hypotension, or bradycardia; and (8) injurious falls. The definition, ascertainment, and formal adjudication of these events have been previously described in detail [25, 34].

2.5. Statistical Analysis

2.5.1. Individualized Net Benefits and Patient Preference Simulations

Cox proportional hazards models were fit to estimate the probability of experiencing the first occurrence of each outcome, with the data administratively censored at 4 years of follow‐up. SPRINT participants were also censored at death or the last available follow‐up when the trial concluded in August 2015. For the model predicting cognitive impairment, outcomes were ascertained at an extended follow‐up visit between October 2017 and July 2018; therefore, this model was administratively censored at 5 years of follow‐up or the last available extended follow‐up visit. The individualized treatment effect for each outcome was calculated as the predicted difference in absolute risk between intensive versus standard SBP lowering using an individual's baseline characteristics, treatment assignment, and counterfactual treatment assignment [26]. The predicted absolute differences in risk for each outcome were then combined into an individualized net benefit score, which was calculated as the average of preference‐weighted risk differences across the benefit and harm outcomes, with weights corresponding to the relative importance of each outcome for an individual [20, 35]. A positive predicted net benefit indicated that the benefits of intensive BP lowering outweighed the harms.

Data on patient preference weights for BP treatment outcomes are limited. Previous studies [21, 22, 36] have either surveyed patients using a preference estimation method that asks them to rank the most and least worrisome BP treatment outcomes across a series of questions, or surveyed physicians to rank the severity of each outcome. Based on these data and consensus agreement (MSJ, RS, RLK, JHI, MGS, and SBA), two preference scenarios were considered: (1) “harm tolerant,” which describes preferences for an individual who prioritizes avoiding the disease‐related outcomes (cognitive impairment, CVD, and death) as much more important than avoiding the treatment‐related outcomes and (2) “harm averse,” which describes preferences for an individual who places similar, intermediate importance on avoiding the benefit and harm outcomes. The “harm tolerant” preference weights were 1.0 for death, 0.8 for stroke and cognitive impairment, 0.7 for HF and MI or ACS, and 0.1 for all treatment‐related harms. Preference weights in the “harm averse” scenario were 1.0 for death; 0.6 for stroke, cognitive impairment, HF, and MI or ACS; and 0.5 for each harm outcome. In both simulations, death was assumed to have a preference weight of 1.0; thus, the net benefit can be defined as the cumulative effect of intensive BP lowering across all outcomes, scaled to their importance relative to death.

For each simulation, we calculated the proportion of SPRINT participants with a predicted individualized net benefit of < 0, 0–2, and > 2 percentage points. In prespecified subgroup analyses, we assessed whether the net benefit from intensive BP lowering differed according to age, frailty, and polypharmacy using the Kruskal–Wallis test. We additionally evaluated whether the net benefit varied by frailty and polypharmacy among those aged ≥ 75 years.

In sensitivity analyses, we adjusted preference weights for treatment‐related harm outcomes under both preference scenarios to determine how different the preferences would need to be for the standard and intensive BP‐lowering treatment arms to be equivalent (defined as median net benefit = 0%) and for standard BP lowering to be superior to intensive BP lowering (defined as having the net benefit interquartile range [IQR] < 0%).

All analyses were conducted using Stata (Stata Statistical Software, release 17; StataCorp LP, College Station, TX, USA) and the R statistical computing language (R Development Core Team, Vienna, Austria).

3. Results

3.1. Study Population

Among the 5556 SPRINT participants aged ≥ 65 years at baseline, 413 (7%) were excluded because they were missing baseline data for 1 or more predictors. Among the 5143 included in the present analysis, the mean age was 74 ± 6 years, 37% were female, 23% had a history of CVD, and the mean systolic and diastolic BP at baseline were 140 ± 15 and 74 ± 11 mm Hg, respectively (Table 1). At baseline, 33% of participants were categorized as frail and 63% had polypharmacy. During follow‐up, there were 1398 mortality, CVD, and cognitive events, which included 275 (5%) deaths, 104 (2%) stroke events, 678 (14%) cognitive impairment events, 132 (3%) HF events, and 209 (4%) MI or ACS events. There were also 1186 serious adverse events, including 356 (7%) AKI or serious electrolyte abnormality events; 357 (7%) syncope, hypotension, or bradycardia events; and 473 (9%) injurious fall events. There were more events of each type among those aged ≥ 75 years versus 65–74 years, those who were frail versus less frail or fit, and those with versus without polypharmacy (Table S1).

TABLE 1.

Baseline characteristics of older SPRINT participants, stratified by age.

Baseline characteristic	Age 65–74 years	Age ≥ 75 years	Total
	(n = 2664)	(n = 2479)	(n = 5143)
Demographics
Age, years	69 ± 3	80 ± 4	74 ± 6
Female	985 (37%)	934 (38%)	1919 (37%)
Black	638 (24%)	424 (17%)	1062 (21%)
Hispanic	280 (11%)	166 (7%)	446 (9%)
Clinical data
Current smoker	205 (8%)	79 (3%)	284 (6%)
BMI, kg/m2	29.8 ± 5.4	27.8 ± 4.8	28.8 ± 5.2
Prevalent CVD	570 (21%)	610 (25%)	1180 (23%)
Systolic BP, mm Hg	139 ± 15	142 ± 16	140 ± 15
Diastolic BP, mm Hg	77 ± 10	71 ± 11	74 ± 11
Statin use	1308 (49%)	1308 (53%)	2616 (51%)
No. of BP medications	2.0 ± 1.1	2.1 ± 1.1	2.0 ± 1.1
Total no. of medications	5.9 ± 3.4	6.4 ± 3.5	6.1 ± 3.5
Polypharmacy (≥ 5 medications)	1596 (60%)	1641 (66%)	3237 (63%)
Fitness index (FI)	0.17 ± 0.08	0.19 ± 0.08	0.18 ± 0.08
Fit (FI < 0.10)	477 (18%)	288 (12%)	765 (15%)
Less fit (FI 0.10–0.21)	1429 (54%)	1249 (51%)	2678 (52%)
Frail (FI > 0.21)	753 (28%)	931 (38%)	1684 (33%)
Laboratory data
eGFR, mL/min/1.73 m2	75 ± 20	63 ± 19	69 ± 20
UACR, mg/g*	9 (6, 19)	13 (7, 32)	11 (6, 25)

Note: Data displayed as mean ± SD or N (%), except where noted by *, median (IQR).

Abbreviations: BMI, body mass index; BP, blood pressure; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate using the combined creatinine and cystatin C equation; UACR, urine albumin‐to‐creatinine ratio.

3.2. Model Validation

The prediction models for each outcome derived in the overall SPRINT sample had adequate discrimination and calibration for each outcome among those aged ≥ 65 years in SPRINT (Table S2 and Figure S1).

3.3. Individualized Net Benefits in Older Adults in SPRINT

Prior to preference weighting, the combined median predicted decrease in the absolute risk of death, CVD events, and cognitive impairment with intensive BP lowering was 5 percentage points (IQR: 4–9). Meanwhile, the combined median predicted increase in the absolute risk of serious adverse events with intensive BP lowering was 6 percentage points (IQR: 4–8). When simulating “harm tolerant” outcome preferences, the predicted net benefit from intensive BP lowering was 4 percentage points (IQR: 3–6); and 100% of participants had a net benefit > 0, favoring intensive BP lowering. With “harm averse” preferences, the median predicted net benefit was 1 percentage point (IQR: 0–2); and 85% of participants had a predicted net benefit > 0. In this scenario, those who had a net benefit < 0, favoring standard BP lowering, were on average younger, more often female, had a low prevalence of CVD, had more prescribed medications, and had higher eGFR (Table S3). Under both simulated preferences, the net benefit from intensive BP lowering appeared greater among those with older age, CVD, frailty, lower eGFR, and higher albuminuria (Table S3).

Participants with advanced age, frailty, or polypharmacy experienced greater increases in each of the treatment‐related harms with intensive BP lowering, but they also derived greater absolute risk reductions for cardiovascular, cognitive, and death outcomes (Figure 1). As a result, participants aged ≥ 75 years had significantly greater predicted net benefits from intensive BP lowering compared to those aged 65–74 years in both preference scenarios (net benefit with “harm tolerant” preferences: 6 vs. 3 percentage points, p < 0.001; net benefit with “harm averse” preferences: 2 vs. 0 percentage points, p < 0.001) (Table 2 and Figure 2). Similarly, those classified as frail compared with participants who were fit or less frail had significantly greater predicted net benefits irrespective of simulated preferences (Table 2 and Figure 3). A similar pattern of results was observed across frailty categories among those with advanced age in both preference scenarios (Figure S2). Participants with polypharmacy compared to those without polypharmacy also had significantly higher predicted net benefits from intensive BP lowering in the “harm tolerant” preference scenario (Table 2). A similar pattern of greater net benefits among those with versus without polypharmacy was observed among participants with advanced age (Table S4).

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).

TABLE 2.

Individualized predicted net benefits from intensive BP lowering among older SPRINT adults, stratified by age, frailty, and polypharmacy at baseline.

	Harm tolerant		Harm averse
	65–74 years	≥ 75 years	65–74 years	≥ 75 years
	(n = 2664)	(n = 2479)	(n = 2664)	(n = 2479)
Median net benefit a (%)	2.7 (2.0, 3.7)	5.9 (4.2, 8.8)	0.3 (−0.01, 0.9)	2.0 (1.0, 3.7)
Proportion with net benefit > 0 (%)	100	100	74	97

	Fit	Less fit	Frail	Fit	Less fit	Frail
	(n = 765)	(n = 2678)	(n = 1684)	(n = 765)	(n = 2678)	(n = 1684)
Median net benefit b (%)	2.7 (1.9, 3.8)	3.5 (2.5, 5.6)	5.7 (3.5, 9.2)	0.5 (0.2, 1.1)	0.8 (0.2, 1.9)	1.6 (0.4, 3.7)
Proportion with net benefit > 0 (%)	100	100	100	85	84	86

	No polypharmacy	Polypharmacy	No polypharmacy	Polypharmacy
	(n = 1906)	(n = 3237)	(n = 1906)	(n = 3237)
Median net benefit c (%)	3.2 (2.2, 5.0)	4.5 (2.9, 7.3)	0.9 (0.4, 1.8)	0.9 (0.1, 2.5)
Proportion with net benefit > 0 (%)	100	100	95	80

Note: Data displayed as median (interquartile range) or percentage.

^a The predicted net benefit of intensive versus standard BP lowering was significantly greater among those with age ≥ 75 years versus 65–74 years (p < 0.001 for both preference scenarios).

^b The predicted net benefit of intensive versus standard BP lowering was significantly greater among those who were frail (FI > 0.21) versus less fit (0.10 < FI ≤ 0.21) versus fit (FI ≤ 0.10) (p < 0.001 for both preference scenarios).

^c The predicted net benefit of intensive versus standard BP lowering was significantly greater among those with polypharmacy (≥ 5 prescribed medications recorded at the baseline visit) versus without polypharmacy (less < 5 prescribed medications recorded at the baseline visit) in the harm tolerant model (p < 0.001 for harm tolerant scenario and p = 0.10 for harm averse scenario).

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.

In sensitivity analyses that varied the “harm tolerant” simulation, the standard and intensive BP lowering treatment arms were equivalent (median net benefit = 0%) when the preference weights for the treatment‐related harms were increased to 0.8, and standard BP lowering was superior (net benefit IQR < 0%) when the preference weights were set as high as 0.95. For the “harm averse” simulation, preference weights of 0.7 for the treatment‐related harm outcomes resulted in equivalence between treatment arms, and weights of 0.82 resulted in superiority of the standard BP lowering arm (Table S5).

4. Discussion

In this ancillary analysis of community‐dwelling older adults with hypertension in SPRINT, we evaluated the net benefit of intensive versus standard BP lowering by combining treatment effects across multiple benefit and harm outcomes using simulated preferences that reflect the relative importance of each outcome. Most older SPRINT participants were predicted to have a positive net benefit with intensive BP lowering (SBP < 120 mm Hg) compared to standard BP lowering (SBP < 140 mm Hg), irrespective of whether they were simulated to have “harm tolerant” or “harm averse” outcome preferences. In sensitivity analyses, participants would have needed to prioritize the avoidance of potential BP treatment‐related harms more than the avoidance of CVD events and cognitive impairment for our analyses to favor standard BP lowering over intensive BP lowering. Although persons with advanced age, frailty, or polypharmacy at baseline experienced greater harms from intensive BP lowering, they also derived greater absolute CVD, cognitive, and mortality benefits compared to their counterparts. This resulted in significantly greater predicted net benefits from intensive BP lowering. These findings indicate that when accounting for an individual's estimated risks and outcome preferences, the benefits of intensive BP lowering outweigh the harms for most SPRINT‐eligible, community‐dwelling older adults with hypertension, especially among high‐risk subgroups who are often assumed to neither tolerate nor benefit from intensive BP lowering.

Managing hypertension in older adults is controversial due to their high risk of treatment‐related adverse events. Thus, the clinical application of SPRINT's findings to older adults requires several important considerations. First, SPRINT excluded individuals with limited life expectancy, dementia, active cancer, those residing in institutional settings such as skilled nursing facilities, and those with a standing SBP < 110 mm Hg after 1 min. Older adults who do not meet SPRINT eligibility criteria may have competing health priorities or an impaired quality of life that could lead them to deprioritize optimizing BP control. These individuals may also be at a greater risk of CVD events and BP treatment‐related harms. For example, a recent study by Dave et al. reported a 2.4‐fold increased risk of falls, fractures, and syncope among nursing home residents aged ≥ 65 years who initiated antihypertensive treatment compared with similar residents who were not started on antihypertensive medications [37]. This underscores how our findings are most generalizable to noninstitutionalized, ambulatory, community‐dwelling older adults. Second, BP was measured in SPRINT using a standardized protocol that adheres to the American College of Cardiology/American Heart Association (ACC/AHA) guideline recommendations, and BP measured at SPRINT trial visits was generally lower than BP measured in routine clinical practice [8, 38]. This highlights the importance of proper BP measurement techniques when treating older adults with hypertension. Third, careful monitoring of older adult patients during antihypertensive initiation or dose escalation is warranted since the risk of treatment‐related adverse events may be highest during these periods [39, 40].

In the overall SPRINT trial and among the subgroup aged ≥ 75 years, targeting an SBP of < 120 mm Hg versus < 140 mm Hg led to a significant decrease in CVD, cognitive, and mortality events [16, 24]. Similarly, the Strategy of Blood Pressure Intervention in the Elderly Hypertensive Patients (STEP) trial showed that targeting a SBP of 100–129 vs. 130–149 mm Hg led to CVD and mortality benefits among hypertensive Chinese adults aged 60–80 years [41]. Recently, the Effects of Intensive Systolic Blood Pressure Lowering Treatment in Reducing Risk of Vascular Events (ESPRIT) trial demonstrated that, among a broader population of Chinese adults aged ≥ 50 years with hypertension and at high baseline cardiovascular risk, including a large proportion with diabetes mellitus or a history of stroke, targeting an SBP of < 120 mm Hg, compared with < 140 mm Hg, improved CVD outcomes [42]. An individual participant‐level meta‐analysis also found that lower BP targets confer larger absolute CVD benefits among older adults aged 65–84 years with hypertension compared with adults younger than 65 years [43].

The present study provides a novel extension to previous work on BP targets among older adults by individualizing the effect of intensive BP lowering across multiple potential outcomes using risk estimates and simulated outcome preferences. This approach facilitates how to apply the evidence about BP targets to SPRINT‐eligible older individuals with hypertension who are faced with higher risks for CVD, cognitive impairment, and death, as well as higher risks for the potential adverse effects associated with intensive BP lowering. Prior studies have used a similar approach to tailor the results from SPRINT to individuals, but to our knowledge, this is the first study to apply this framework to older adults and to evaluate those with advanced age, frailty, or polypharmacy [21, 22, 26]. This analysis also incorporates the cognitive benefits from intensive BP lowering that were demonstrated in SPRINT MIND [24].

Our findings may help address the discordant guideline recommendations for BP targets for older adults with hypertension. Among adults aged ≥ 70 years, current recommended BP targets range from < 130 mm Hg per ACC/AHA, to < 140 mm Hg per European Society of Hypertension and American Academy of Family Physicians, and to < 150 mm Hg per U.S. Department of Veterans Affairs [8, 9, 44]. Our finding that nearly all SPRINT‐eligible older adults are predicted to have a positive net benefit from intensive BP lowering, regardless of their preferences or baseline risk, supports guideline recommendations for lower BP targets. In fact, the present analysis suggests that those with the highest baseline risk, as assessed by advanced age, frailty status, or polypharmacy, are likely to derive the greatest net benefits from intensive BP lowering. Although hypertension guidelines propose different BP targets for older adults, they uniformly recommend individualizing BP targets for older adults through shared decision‐making. However, there are no available frameworks to personalize the evidence about BP targets to older adults. Our study illustrates how to incorporate patient preferences about BP treatment outcomes and individualize the magnitude of net benefit from a lower versus higher BP target, which could help facilitate shared decision‐making conversations.

Less than half of U.S. adults aged ≥ 65 years with a hypertension diagnosis have well‐controlled BP, defined as BP < 130/80 mm Hg [7, 45, 46]. A major factor contributing to suboptimal BP control is clinician inertia in initiating or intensifying BP treatment [47]. Clinicians may be more inclined to recommend a higher BP target for hypertensive older adults with advanced age, frailty, or polypharmacy due to concerns about their increased risk of treatment‐related harms [48, 49]. However, higher BP targets are suboptimal for preventing CVD, cognitive impairment, and death. The present study shows that although older SPRINT participants with advanced age, frailty, or polypharmacy experienced more harms with intensive BP treatment compared to lower‐risk individuals, they also had greater absolute benefits in CVD, cognitive, and mortality outcomes. This led to higher net benefits for higher‐risk subgroups in both simulated patient preference scenarios. Our findings suggest that, contrary to conventional wisdom, factors such as age, frailty, or polypharmacy should not be viewed as barriers to intensive BP lowering for community‐dwelling, SPRINT‐eligible older adults with hypertension. By demonstrating the positive net benefits of intensive BP lowering after explicitly accounting for BP treatment harms, these results could be useful for patients and providers to overcome the therapeutic inertia that commonly occurs with intensifying BP treatment in older adults [49].

Our study has several limitations. First, although the prediction models demonstrated adequate performance in SPRINT, widespread application of our findings to SPRINT‐eligible older adults in clinical settings requires external validation. Second, our findings may not generalize to older adults who did not meet SPRINT's inclusion criteria, including those with limited life expectancy, dementia, active cancer, or those who reside in institutional settings such as skilled nursing facilities, as well as those with a standing SBP < 110 mm Hg, diabetes mellitus, advanced CKD, or a history of stroke [25]. Despite these exclusion criteria, an analysis of data from the 2007 to 2012 National Health and Nutrition Examination Surveys (NHANES) indicates that nearly one‐third of all U.S. adults ≥ 75 years with hypertension are SPRINT‐eligible [50]. Third, the FI used in SPRINT was constructed retrospectively using cardiovascular risk factors, disease history, and quality‐of‐life markers, which may not align with other measures used to determine frailty in clinical practice. Fourth, the study did not account for patient preferences regarding financial costs, medication burden, and side effects associated with antihypertensive medications, which may also influence BP target decisions. Finally, patient preferences are likely to vary across different populations, settings, and over time. Additional studies are needed to develop tools that reliably measure patient preferences regarding BP‐lowering outcomes.

In summary, almost all SPRINT participants aged ≥ 65 years had a predicted net benefit that favored an SBP target of < 120 mm Hg over an SBP target of < 140 mm Hg, but the degree of net benefit varied considerably according to estimated risks and simulated preferences for multiple benefit and harm outcomes. Participants with advanced age, frailty, or polypharmacy derived greater absolute harms from intensive BP lowering, but also greater absolute CVD, cognitive, and mortality benefits, which resulted in overall greater predicted net benefits from intensive BP lowering. These findings demonstrate that for noninstitutionalized, ambulatory, community‐dwelling older adults who meet SPRINT eligibility criteria, the evidence from SPRINT can be translated to each person using individualized risks and preferences, and that consideration of advanced age, frailty, and polypharmacy alone should not deter clinicians from recommending intensive BP lowering.

Author Contributions

Study concept and design: S.B.A., M.S.J., M.G.S., R.S., and M.M.E. Data acquisition: S.B.A. and R.S. Data interpretation: S.B.A., M.S.J., M.G.S., R.S., M.M.E., J.A.d.L., J.H.I., C.G., R.L.K., D.J.T., and J.D.B. Statistical analysis: S.B.A., M.S.J., and R.S. Preparation of manuscript: M.S.J. and S.B.A.

Disclosure

The authors thank the participants and staff members of the Systolic Blood Pressure Intervention Trial, which was sponsored by the National Institutes of Health (NIH), including the National Heart, Lung, and Blood Institute (NHLBI), the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), the National Institute on Aging (NIA), and the National Institute of Neurological Disorders and Stroke (NINDS), under Contract Numbers HHSN268200900040C, HHSN268200900046C, HHSN268200900047C, HHSN268200900048C, HHSN268200900049C, and Inter‐Agency Agreement Number A‐HL‐13–002‐001. It was also supported in part with resources and use of facilities through the Department of Veterans Affairs. The SPRINT investigators acknowledge the contribution of study medications (azilsartan and azilsartan combined with chlorthalidone) from Takeda Pharmaceuticals International Inc. All components of the SPRINT study protocol were designed and implemented by the investigators. The investigative team collected, analyzed, and interpreted the data. All aspects of manuscript writing and revision were carried out by the coauthors. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, the U.S. Department of Veterans Affairs, or the United States Government. For a full list of contributors to SPRINT, please see the supplementary acknowledgment list: https://www.sprinttrial.org/public/dspScience.cfm.

Conflicts of Interest

M.M.E. has served on advisory expert panels for Boehringer Ingelheim Inc. and has a research collaborative agreement with Bayer Inc. M.G.S. has received consulting income from Cricket Health Inc.; has served on advisory panels for Boehringer Ingelheim, Astra Zeneca, and Bayer; and has received research support from Bayer. J.H.I. holds an investigator‐initiated research grant from Baxter International Inc., serves as a member of a data safety monitoring board for Sanifit Therapeutics, is a member of the scientific advisory board for Alpha Young, and has served on advisory boards for AstraZeneca and Ardelyx. J.D.B. reports grant support from the NIH, Roche Diagnostics, and Abbott Diagnostics, consulting fees from Roche Diagnostics, Astra Zeneca, and the Cooper Institute. J.A.d.L. reports grant support from Roche Diagnostics and Abbott Diagnostics, consulting fees from Roche Diagnostics, Abbott Diagnostics, Ortho Clinical Diagnostics, Beckman Coulter, Quidel Cardiovascular Inc., and Siemen's Health Care Diagnostics. He has been named a co‐owner on a patent awarded to the University of Maryland (U.S. Patent Application Number: 15/309,754) entitled: “Methods for Assessing Differential Risk for Developing Heart Failure.” The remaining authors have nothing to disclose.

Supporting information

Table S1. Number of events of each benefit and harm outcome stratified by age, frailty, and polypharmacy.

Table S2. Predictive performance of parsimonious Cox regression models predicting each outcome.

Table S3. Baseline characteristics stratified by predicted individualized net benefit when simulating a harm averse outcome preference profile.

Table S4. Distributions of individualized predicted net benefits from intensive BP lowering among SPRINT adults 75 years of age or older, stratified by polypharmacy at baseline.

Table S5. Sensitivity analyses showing the predicted net benefits of intensive versus standard BP lowering in SPRINT after re‐weighting BP treatment‐related harms.

Figure S1. Calibration plots of parsimonious prediction models for benefit and harm outcomes among older adults in SPRINT.

Figure S2. Distributions of individualized predicted net benefits from intensive BP lowering among adults 75 years of age or older, stratified by frailty.

Appendix A. Editor’s Note

Despite multiple clinical trials documenting the benefits of antihypertensive therapy in reducing cardiovascular events in older individuals, there is ongoing uncertainty about the utility of such therapy in multimorbid and frail older patients (especially nursing home residents), and about how aggressively to lower blood pressure (BP) in complex older adults. This uncertainty arises in part due to the marked under‐representation of multimorbid older adults in clinical trials, but also because no trials to date have integrated geriatrics principles, such as the 4 M's, into therapeutic decision‐making. This is critically important, as there is an inevitable tension between the potential benefits of treatment and the possible harms, including but not limited to reduced quality of life related to taking more medications.

In the present study, Jamshidian and colleagues utilized data from the Systolic Blood Pressure Intervention Trial (SPRINT) to examine the net effects (i.e., the balance between benefits and harms) of intensive BP reduction (target systolic BP < 120 mm Hg) versus standard BP reduction (target < 140 mm Hg) in adults ≥ 65 years of age and in 3 pre‐defined subgroups—those ≥ 75 years, individuals with frailty, and patients with polypharmacy (≥ 5 medications). A novel feature of the analysis was that patient preferences were modeled into 2 categories—harm tolerant (willingness to accept some risk of harm to obtain benefit) or harm averse (less willing to accept treatment‐related adverse events). The main finding was that the net benefit, which included reductions in mortality, cardiovascular events, and cognitive impairment, strongly favored the more aggressive BP target overall and in all subgroups. Moreover, although potential harms were greater in all 3 of the predefined subgroups (as expected), so too were the potential benefits, such that the net benefit was actually greater among patients in these higher risk groups.

The findings of this study have important clinical implications. While they do not obviate the need for shared decision‐making, they do provide a framework for the discussion, at least for patients similar to those enrolled in SPRINT. Such patients, including those with advanced age, frailty, polypharmacy, and/or a harm averse perspective, can be reassured that the benefits of lowering BP to < 120 mm Hg, if tolerated, outweigh the potential harms for most people. The main limitation of the study, acknowledged by the authors, was that SPRINT excluded patients with diabetes, prior stroke, decompensated heart failure within 6 months or ejection fraction < 35%, standing BP < 110 mm Hg, a clinical diagnosis of dementia, nursing home residence, or life‐expectancy < 3 years.

‐Michael W. Rich, MD