SPRINTING toward the optimal blood pressure target for hypertensive patients

Identifying the optimal blood pressure (BP) target for treatment of adults with high BP is a priority for health care providers and patients because of the high prevalence of hypertension in the population and the abundant evidence that lowering BP reduces cardiovascular disease (CVD) events and death. Treatment to a systolic BP (SBP) level below the traditional target of <140 mm Hg is desirable in most adults with hypertension who are at high risk for atherosclerotic CVD (ASCVD).[1] At least 18 randomized controlled trials (RCTs) have compared CVD outcomes in participants allocated to different BP targets, and 32 RCTs have compared CVD outcomes in participants allocated to more or less intensive antihypertensive drug therapy. Meta-analyses of these trials have identified statistically significant and clinically important reductions in CVD events in those whose BP was treated to a lower goal compared to less treatment.[2] Prior to publication of the Systolic Blood Pressure Intervention Trial (SPRINT) results, meta-analyses had reported benefit to an average achieved SBP of 133 mm Hg.[3] SPRINT, alone and in combination with the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, provides compelling evidence in support of a SBP target <120 mm Hg in adults with hypertension who are at high risk for ASCVD.[4–5] As is common with landmark trials, the SPRINT results and their relevance to clinical practice have been questioned by some, as in the Viewpoint by Kjeldsen et al in the current issue of Circulation Research [6]. In this Counterpoint Viewpoint, we address some of the more frequent criticisms of SPRINT.

The relevance of SPRINT has been questioned on the basis that the method of measuring BP in the trial was unconventional and not applicable to ordinary office practice.[6,7] Specifically, the perceived requirement for a 5-minute period of quiet rest without staff in the room prior to BP recording in SPRINT has been depicted as unprecedented in outcome trials of BP treatment and impractical for usual clinical practice. Concern has been expressed that there is a large difference between attended and unattended BP measurements in the clinic and in RCTs and thus the BP readings obtained in SPRINT were substantially different from those obtained in usual office practice or in other RCTs of BP treatment. However, at least six previous trials have examined the effect of staff attendance on BP, and in every instance the difference between average SBP and diastolic BP for the attended compared to unattended measurements was <2 mm Hg.[8]

Knowing how BP is measured is important to understanding BP control and guiding clinicians in appropriate management of hypertension. The SPRINT trial used programmable automated oscillometric devices (Omron Professional Digital Blood Pressure Monitor; Model HEM-907XL) to measure BP in an effort to standardize BP measurements among the 102 clinical sites. SPRINT BP measurements used methods recommended by professional societies and BP guidelines committees at the time the trial started. These recommendations emphasized the importance of the BP measurement methods, but did not state whether the patient should be attended or unattended, and the SPRINT protocol did not require that BP measurements be unattended by staff.[9] Other hypertension trials using a wide variety of BP measurement protocols have not raised questions about the generalizability of the BP measurement technique or its influence on the study’s outcome.[10] SPRINT is the first RCT where the issue of staff presence or absence during measurement and its effect on BP values has been questioned.[6,7]

In response to these concerns, the SPRINT clinical sites were surveyed immediately after the completion of trial closeout visits to inquire whether BP measurements were usually attended or unattended by staff.[9] There were 38 sites (4082 participants) that measured BP after leaving the participant alone during the entire measurement period; 25 sites (2247 participants) that had personnel in the room the entire time; 19 sites (1746 participants) that left the participant alone only during the 5-minute rest period, and 6 sites (570 participants) that left the participant alone only during the BP readings. Similar BPs within randomized groups were noted during follow-up at the majority of visits in all four measurement categories. There was no evidence that unattended BP measurements led to lower SBP at baseline or during follow-up compared with attended BP measurements. Further, similar CVD risk reduction was observed in the intensive treatment group in SPRINT, whether the BP measurement technique used was primarily attended or unattended [9]. Accordingly, the discussion of BP measurement technique in the Kjeldsen et al Viewpoint is misleading. In particular, the argument that BP targets in SPRINT should be translated into ordinary clinical practice by adding an arbitrary number of 10–20 mm Hg is poorly justified and likely to lead to error in clinical decision-making and under-treatment of hypertension, thus increasing the risk of CVD and total mortality in persons at high risk for ASCVD.

In adults with hypertension who are at high risk for ASCVD, use of the SPRINT SBP goal of <120 mm Hg[4] or the more conservative <130 mm Hg goal recommended in the 2017 ACC/AHA BP guideline[1] is sensible when coupled with the BP measurement techniques used in SPRINT. The latter include use of a validated automated BP device, staff training to allow for a 5-minute quiet rest period prior to the BP measurements, proper positioning of the patient’s arm and body, use of the correct cuff size, no talking during the measurements, and averaging of multiple measurements. The BP measurement technique per se is far more important than whether the BP measurement is conducted in the presence or absence of an observer.

The Viewpoint by Kjeldsen et al also questions the validity of the SPRINT findings based on the assumptions that: 1) heart failure (HF), one component of the primary outcome, is an uncertain endpoint in hypertension research because the signs and symptoms used to make the diagnosis are non-specific, and 2) the HF events reported in SPRINT were not HF at all, but instead artifacts of the study design, i.e., withdrawal of diuretics from the standard treatment group. HF has been an important outcome in >35 prior hypertension treatment trials, and in some has been a component of the primary outcome.[11,12] All have observed risk reductions as large as or larger than in SPRINT, and reductions in HF were strongly and directly proportional to the reductions in BP, with 28% reduction in HF events for every 10 mm Hg reduction in SBP. There was also a strong trend for treatment-related reductions in HF events even in participants with a baseline SBP <130 mm Hg, with no threshold for beneficial effects at lower BP levels.[11,12] Thus, the SPRINT HF results are concordant with a large body of evidence from prior trials.

The HF events in SPRINT were based on objective, reliable, and clinically meaningful evidence. While diagnosis of chronic, mild-moderate HF in outpatients can be challenging in clinical practice, hospitalization for acute decompensated HF (ADHF; the HF outcome in SPRINT) is far less difficult and more objective. Hospitalized ADHF has been an outcome measure in many landmark clinical trials and observational studies. ADHF events reported by SPRINT clinical sites were rigorously adjudicated by a committee that was blinded to treatment allocation, using strict criteria which required, in addition to typical symptoms and signs, documentation of a favorable response to intravenous diuretic therapy, and supported by both biochemical (B-type natriuretic peptides) and imaging (chest X-ray) data.[13] These methods for adjudication of ADHF have been validated and used in many large landmark clinical trials and population studies. SPRINT ADHF events also have strong internal validation; following an adjudicated ADHF event, participants had a 27-fold higher risk of CV death and 10-fold higher risk of all-cause death, even after adjustment for relevant covariates.[13]

Multiple lines of evidence support the conclusion that the reduction in ADHF events in SPRINT did not result from differential use of diuretics in the two treatment groups. ADHF is not merely a reflection of fluid overload. Healthy persons generally do not develop HF even with large, rapid volume infusions.[14] ADHF events require a combination of chronic and progressive impairments in function of the heart, kidney and other organs that result from chronic elevation of SBP and are retarded by BP reductions.[12,13] The preeminent role of BP as a cause of ADHF is supported by the observation in antihypertensive drug treatment trials of a direct relationship between SBP lowering and reductions in events that was as strong for HF as for stroke, and strong for both compared to myocardial infarction (MI).[11,12] To our knowledge, no antihypertensive treatment trial has reported a relationship between diuretic-induced changes in volume status and a reduction in HF events.

The fact that treatment with all classes of antihypertensive drug therapy has significantly reduced HF in primary prevention trials provides additional compelling evidence that BP lowering rather than drug-induced changes in volume underpins the beneficial effect of antihypertensive drug therapy on HF.[11] Likewise, in secondary prevention trials in patients with established HF, improved HF outcomes has resulted primarily from non-diuretic effects. For example, in the Randomized Aldactone Evaluation Study (RALES) spironolactone produced a large reduction in ADHF and mortality, with no evidence of significant diuresis or benefit due to any measure of diuresis.[15]

The type (thiazides and thiazide-like diuretics) and doses of diuretics used in SPRINT have modest effects on volume status and are ineffective treatments for HF or volume overload. In SPRINT, loop diuretics such as furosemide were not first line drugs and were used infrequently. HF events theoretically caused primarily by volume shifts from differential diuretic prescription and/or diuretic withdrawal would have been expected to occur relatively early after randomization, but most of the ADHF events occurred later in SPRINT. Separation between treatment groups in ADHF events just began to appear at 6 months, and the curves continued to diverge throughout the course of the trial.[13] Simple calculations based on data in tables 1 and 3 of the published main SPRINT outcomes paper[4] suggest that intergroup differences in diuretic prescription could potentially account for only a small percentage of the HF events, not enough to have strongly influenced the primary outcome. Many of the landmark antihypertensive drug treatment trials that have demonstrated a large treatment-related benefit for HF events, including SHEP, the MRC trial and the MRC in older adults trial were based on a two-arm parallel study design that resulted in a 100% difference in allocation to first-step therapy with a diuretic. Should the results of these trials now be discarded?

Finally, as part of exploratory analyses to confirm the robustness of the SPRINT findings, an analysis has been done with ADHF removed from the primary composite outcome. The beneficial effects of the intensive treatment goal were unchanged. The analysis will be reported in a future manuscript.

In summary, the criticisms of SPRINT presented in the Viewpoint by Kjeldsen et al are based on arguments that are not substantiated upon closer scrutiny of the evidence. We are not alone in our belief that SPRINT provides valid results. Guideline writing committees in the US[1], Canada[16] and Australia[17] have all reported that the SPRINT results were important in their deliberations leading to recommendations for SBP targets during treatment of adults with hypertension who are ambulatory but at high risk for ASCVD.