Reply to: Systolic Blood Pressure and Mortality: Role of Reverse Causation

The appropriate treatment of hypertension in older adults is a multiple-choice question with more than one correct answer. It is unlikely that there is a universal blood pressure goal in this population, and therefore the scientific query should center on how to establish the ideal target for each patient. We appreciate the comments by Supiano et al.,¹ yet we disagree with the statement that the findings from our observational study conflict with results from randomized trials. In contrast, we believe that integrating evidence from different types of studies can help us reach our common goal of treating those patients who stand to benefit the most from blood pressure lowering.

One point on which we agree is that residual confounding is a “fundamental limitation inherent in all observational studies,” but we also assert that randomized controlled trials are fundamentally limited by a lack of generalizability, especially for those including older adults. The Systolic Blood Pressure Intervention Trial (SPRINT) and the Hypertension in the Very Elderly Trial (HYVET) were high quality randomized controlled trials and the investigators made strong efforts to ensure their study participants were as representative as possible. The inclusion of adults ≥75 years in SPRINT and ≥80 years in HYVET represent an important advance in demonstrating the effectiveness of blood pressure control in these populations. However, despite these efforts, limitations remain. Although SPRINT was inclusive of older adults, the inclusion/exclusion criteria of SPRINT apply to only 19.5% and 34.6% of U.S. adults aged 60–74 and 75+ years, respectively.² We have previously demonstrated that HYVET participants had lower levels of cardiovascular comorbidity than participants in the National Health and Nutrition Examination Survey (NHANES), despite being selected for higher levels of blood pressure.³ We agree with Supiano et al.’s statement that our finding should not deter clinicians from considering lowering blood pressure among eligible patients. Nonetheless, questions remain regarding whether ineligible patients would also benefit. Observational studies, when conducted appropriately, can add to the body of evidence in populations who have not been included in trials.

We have developed a body of work demonstrating that markers of functional status or frailty including measured gait speed,⁴ self-reported gait speed,⁵ limitation of activities of daily living,⁶ self-reported physical ability limitation,⁷ and grip strength⁸ modify the effect of blood pressure on outcomes including mortality, cardiovascular disease, and falls. We have used multivariable adjustment, stratification, and exclusion of deaths in the first years after blood pressure measurement to account for confounding. In addition, we have conducted post-hoc analyses of the Systolic Hypertension in the Elderly Program, in which we found a trend towards modification of treatment effects by physical ability limitation.⁷ Many other research groups have found similar findings. Notably, the interest and understating of heterogeneous treatment effects is growing. The top three winning submissions of the New England Journal of Medicine’s SPRINT Challenge focused on differential treatment effects in subgroups or individualized treatment decision-making tools.⁹ Taken together, these findings suggest that not all older adults may benefit from lower blood pressure; rather, different targets may be appropriate for different groups.

This line of investigation is in the relatively nascent stage, so there are several additional questions that need to be answered before this work can be translated into clinical practice. First, we do not know what markers best identify those older adults who are most likely to benefit from blood pressure control, and those who might not. Older adults are a diverse and complex population, and we have observed variability in the strongest effect modifiers across studies. Although markers of frailty and poor functional status appear to be promising stratification tools, a recent paper demonstrated the poor concordance in different measures of frailty, and found that different frailty scores are not interchangeable.¹⁰ This may also contribute to the differences between studies. Second, although we have found that lower blood pressure appears harmful in some populations, whether this will translate into harm associated with lowering of blood pressure through treatment is not certain. As this line of work develops, a clinical trial could help answer this question.

Results from HYVET and SPRINT have demonstrated that some elders will benefit from blood pressure lowering. Our work and that of others has demonstrated heterogeneous effects of lower blood pressure among elders. These findings are not in conflict, but represent complementary lines of evidence in the complex investigation of the benefits and harms of blood pressure control among older adults. We hope that our work can help inform targeted treatment recommendations to bring the best interventions to the populations who stand to benefit.