Functional Status Modifies the Association of Blood Pressure with Death in Elders

Editor's Note: Commentary based on Wu C, Smit E, Peralta CA, Sarathy H, Odden MC. Functional status modifies the association of blood pressure with death in elders: health and retirement study. J Am Geriatr Soc 2017;65:1482-9.

Rationale/Background: The randomized controlled Systolic Blood Pressure Intervention Trial (SPRINT) demonstrated the benefits of a lower blood pressure (BP) goal for the prevention of the composite outcome of myocardial infarction, other acute coronary syndromes, stroke, heart failure, or death from cardiovascular causes.1 However, the risks of this lower BP goal in older frail adults continues to be a concern among many in the geriatrics and geriatric cardiology communities.2 Understanding how frailty (as assessed by gait speed and grip strength) modifies the effects of systolic (SBP) and diastolic blood pressure (DBP) on mortality in a real-life setting may help clinicians navigate BP management in older adults.

Study Question: Do grip strength and gait speed modify the associations between SBP and DBP for the outcome all-cause mortality in older adults?

A Brief Digression on the Meaning of Interactions:3,4

Definition of interaction: When the extent of association between exposure (SBP and DBP) and disease (all-cause mortality) substantially/statistically differs as a function of a third variable (grip strength and gait speed).

Synonyms for the term interaction: Effect-measure modification (preferred), effect modification, heterogeneity of effect, heterogeneity of measure, effect variation.

Scales of measurement: Multiplicative scale (relative risk, prevalence ratios, odds ratios, additive scale (absolute proportions, incident rate).

Types of interactions: Statistical interaction (scale used is based on underlying reason for study; often not particularly informative), public health interaction (additive scale), mechanistic interaction (additive scale), biological/physical interaction (realm of physical laboratory science).

Study Design/Methods/Study Population:

Funding: Study supported by National Institute on Aging K/R01 grant.

Study Population: Health and Retirement Study. Population based study of non-institutionalized adults ≥51 years of age.

Source Population: Pooled data from two collection dates (2006 and 2008). During these periods, face-to-face interviews were conducted and blood pressure and physical function data were collected. Inclusion criteria for present study: 1) age ≥65 years; 2) average of three sitting BP readings (45-60 seconds apart); 3) completed grip strength and gait speed measurements; 4) availability of vital status data.

Outcome: All-cause mortality.

Predictors: SBP and DPB (modeled as continuous variables and in categories).

Effect Modifiers: Grip strength (kg, measured by a spring-type hand dynamometer), gait speed (meters/second, calculated by converting the average of two measurements of the time to walk ~2.5 meters at usual pace).

Covariates: Demographics, behavioral characteristics, clinical measures.

Statistical Analysis: Cox models accounting for nested data structure. Effect modification was tested between grip strength and both grip strength and gait speed with SBP and DBP using Wald test for interaction. Interactions were presented in both the multiplicative scale (relative risk) and the additive scale (risk difference).

Results: Total cohort was 7,492 participants. 15.9% (1,077) were classified as weak grip strength (<16 kg for female and <26 kg for male). The average follow-up time was 6.0 years with 1,870 (24.9%) deaths during follow-up. At baseline, SBP did not differ between the weak and the normal groups. However, DBP was slightly lower in the weak group compared with normal grip strength (76.1±12.1 vs. 76.8±11.4, p < 0.001). Those classified as weak were older, more often female, and had higher comorbidity and anti-hypertensive medication burdens.

We will summarize the data mainly in relation to SBP.

In the multiplicative scale:
In participants with normal grip strength and SBP ≥ 150 mmHg there was a 47% (95% CI 31-65%, Pint=0.03) higher unadjusted relative risk and a 24% (95% CI, 7–43%, Pint=0.07) higher adjusted relative risk for all-cause mortality compared to SBP < 150 mmHg.

In participants with weak grip, the investigators did not find a statistically significant association between SBP and mortality.

When displayed in the additive scale (the appropriate scale for public health considerations), the incident mortality rate is as shown below:

In participants with normal grip strength and:
SBP ≥ 150 mmHg, the incident mortality rate was 47.0/1000 patient-years
SBP < 150 mmHg, the incident mortality rate was 32.3/1000 patient-years

In participants with weak grip strength and:
SBP ≥ 150 mmHg, the incident mortality rate was 92.8/1000 patient-years
SBP < 150 mmHg, the incident mortality rate was 83.5/1000 patient-years

Finally, the mortality rate was highest in those with SBP <150 mmHg who were slow walkers with weak grip strength: 131.1/1000 patient-years

Conclusions: The main take home point from the current study was that the association between SBP and mortality differed when a third variable, frailty (measured by grip strength and gait speed), was present. This can be restated as: an interaction was present between SBP and frailty (or frailty and SBP) and the presence of this interaction altered the relative risk or incident rate of mortality.

Specifically, the investigators found that in those with normal grip strength and SBP ≥ 150 mmHg there was a statistically significant increase in the relative risk of all-cause mortality compared to SBP < 150 mmHg. In those with weak grip strength and SBP ≥ 150 mmHg the association was attenuated and was not statistically significant when risk adjusted.

However, the incident mortality rate (an additive scale) showed that participants with SBP ≥ 150 mmHg with either a normal or weak group, had a higher incident mortality rate. Lastly, mortality was highest in those with SBP < 150 mmHg and slow walkers with weak grip strength.

Limitations: Residual confounding cannot be excluded in observational studies and categorization of continuous variables may obfuscate the true relationship with outcomes of interest.

Geriatric Perspective for the Cardiovascular Clinician: The SPRINT trial found that in participants with a SBP between 130 to 180 mmHg without diabetes mellitus or prior stroke who met SPRINT inclusion criteria, a lower BP target decreased the incidence of the primary outcome. However, this was at a small cost of increasing the number of serious adverse events such as syncope, hypotension, electrolyte abnormalities, and acute renal insufficiency. Based on this finding, the geriatric cardiology and geriatrics communities have raised strong concerns in regards to the SPRINT-like BP targets in older hypertensive adults with geriatric syndromes such as frailty, polypharmacy, and multimorbidity.

The study by Wu et al.5 posed the question of whether frailty (as measured by grip strength and gait speed) modifies the relationship between blood pressure and all-cause mortality. Using the Health and Retirement Study population, they found that the presence of frailty substantially altered the relationship between BP and mortality. This interesting study, despite its limitations, brings to the forefront the concept that an interaction exists between blood pressure and frailty. This means that changes in one can modify the effect of the other on the risk of all-cause mortality. Translating this evidence to daily care could mean that aggressive blood pressure control in frail, older adults may not be beneficial and could be harmful. While additional research is needed, the implications of this study are that practitioners, including cardiologists, should routinely assess frailty in their older patients and consider the modulating impact of frailty on clinical outcomes when discussing treatment for hypertension as well as other cardiovascular disorders.


  1. SPRINT Research Group, Wright JT, Williamson JD, et al. A randomized trial of intensive versus standard blood-pressure control. N Engl J Med 2015;373:2103-16.
  2. Kim DH. Intensive vs standard blood pressure control for older adults. JAMA 2016;316:1921.
  3. VanderWeele TJ, Knol MJ. A tutorial on interaction. Epidemiologic Methods 2014;3:33-72.
  4. Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol 2012;41:514-20.
  5. Wu C, Smit E, Peralta CA, Sarathy H, Odden MC. Functional status modifies the association of blood pressure with death in elders: Health and Retirement Study. J Am Geriatr Soc 2017;65:1482-9.

Clinical Topics: Acute Coronary Syndromes, Geriatric Cardiology, Heart Failure and Cardiomyopathies, Prevention, Acute Heart Failure, Hypertension

Keywords: Geriatrics, Antihypertensive Agents, Blood Pressure, Comorbidity, Proportional Hazards Models, Frail Elderly, Acute Coronary Syndrome, Hypertension, Blood Pressure Determination, Systole, Hypotension, Heart Failure, Myocardial Infarction, Stroke, Diabetes Mellitus, Acute Kidney Injury, Polypharmacy, Syncope

< Back to Listings