How Strong Is the Evidence to Support Blood Pressure Treatment Goal of 130/80 mm Hg?

Sanjay Kaul, MD


Circulation. 2019;138(23):2594-2596. 

The 2017 American College of Cardiology/American Heart Association guideline for the prevention, detection, evaluation, and management of high blood pressure (BP) in adults is the first comprehensive hypertension clinical practice guideline since 2003.[1] The guideline document comprises 111 pages, 11 figures, 448 data tables, and 1000 references covering 106 recommendations (113 by my count). Only 8% (10 of 113) are supported by high-level evidence (Class I, Level of Evidence A recommendations); 36% (31 of 113) are based on Class II recommendations (22% [25 of 113] Class IIa, 14% [16 of 113] Class IIb) consistent with equivocal evidence; and 44% (50 of 113) are based on Level of Evidence C recommendations driven primarily by expert opinion. Notably, there were 8 Class I, Level of Evidence C recommendations supporting a diastolic BP goal of <80 mm Hg, for which there is little to no supporting evidence.

One of the most controversial recommendations in the new guideline is the BP treatment goal of <130/80 mm Hg across a wide spectrum of risk or comorbidities. The guideline development group conducted a structured review of the literature and commissioned a systematic review and meta-analysis from an independent Evidence Review Committee to address the optimal target for BP lowering during antihypertensive therapy.[2] For the specific evaluation of a systolic BP target <130 mm Hg compared with any higher BP target, 9 studies were identified. Data were pooled via the random-effects meta-analysis with the DerSimonian-Laird method. The key results are reproduced in the Table. The review committee concluded the following:

"SBP [systolic BP] lowering to <130 mm Hg significantly reduced the risk of major cardiovascular events (RR [risk reduction]: 0.84; 95% CI: 0.73–0.99) and stroke (RR: 0.82; 95% CI: 0.70–0.96) and risk reductions remained "marginally" significant for MI [myocardial infarction] (RR: 0.85; 95% CI: 0.73–1.00) and all-cause mortality (RR: 0.92; 95% CI: 0.79–1.06). Heterogeneity between studies was lowest for MI, stroke, and renal events and highest for major cardiovascular events and heart failure although it was not statistically significant. [2]

There are several limitations of this meta-analysis in addition to those acknowledged by the Evidence Review Committee that raise questions about its utility in reliably informing guideline recommendations. First, it was not prospectively planned to minimize potential bias. Second, it was based on study-level data, thereby precluding a deeper exploration into possible heterogeneity of treatment effect across key subgroups such as diabetes mellitus, chronic kidney disease, or age >60 years. Third, it was not adjusted for multiplicity to avoid spurious (false-positive) conclusions. Fourth, the meta-analytic P values of 0.012 for stroke reduction and 0.033 for major adverse cardiovascular event reduction are not consistent with strong evidence against the null. For example, the P value of 0.033 for major adverse cardiovascular events translates to a minimum Bayes factor of 0.104 (derived from exp[−0.5z[2]], where z for the pooled estimate is 2.13), which means that the evidence supports the null hypothesis approximately 1/10th as strongly as it does the alternative. This reduces the null probability from 50% before to 9% after the studies, which does not represent strong evidence against the null. Finally, there was sufficient clinical and statistical heterogeneity (conventional heterogeneity tests lack the necessary statistical power to capture true heterogeneity) to question pooling of the aggregate data. On the basis of these limitations, it is debatable whether this meta-analysis conforms to the criteria that would confer high credibility from a regulatory perspective.[3] Thus, it is difficult to assert that a pooled estimate of 1.1% absolute difference in major adverse cardiovascular events and a 0.5% absolute difference in stroke (both differences associated with nonrobust P values) with no evidence of benefit for all-cause or cardiovascular mortality, myocardial infarction, heart failure, or renal events constitutes strong evidence with which to formulate public policy and guide clinical practice. Furthermore, harms, including adverse events (hypotension, falls, fractures, electrolyte abnormalities, etc), cost, and inconvenience (polypharmacy, pill burden, etc), were not formally considered in the systematic review. Therefore, a transparent benefit-risk tradeoff assessment (the key driver of rating of the strength of a recommendation in the guideline grid) in the development of recommendations was not possible. Finally, factors that modify the quality of evidence were not considered in the systematic review. Important limitations such as the open-label nature of trial design (prone to bias), variation in BP measurements (prone to erroneous protocol-mandated changes in antihypertensive therapy intensity), missing data and loss to follow-up (susceptible to violation of intention-to-treat principle), and early stopping for benefit (prone to potentially overestimating benefit or underestimating harm) serve to weaken the evidence. Consequently, the American College of Cardiology/American Heart Association guideline for BP treatment goal of <130/80 mm Hg is not universally accepted by other professional society guidelines such as those from the American College of Physicians and American Academy of Family Physicians,[4] the American Diabetes Association, and the European Society of Cardiology and European Society of Hypertension. Lack of consensus across guidelines reflects uncertainty in evidence and exemplifies the challenge of public trust in guidelines.

In a sensitivity analysis that excludes the results of the SPRINT trial (Systolic Blood Pressure Intervention Trial) from the meta-analysis,[5] the pooled estimates are attenuated for most cardiovascular outcomes except for stroke and myocardial infarction, with intensive BP lowering yielding significant risk reduction only in stroke (Table). Thus, the guideline recommendation supporting a target BP of <130/80 mm Hg is driven primarily by the results of SPRINT, a trial in which the prespecified systolic BP goal in the intensive-treatment arm was <120 mm Hg. The SPRINT trial has been the focus of intense controversy, drawing criticism about the trial design and conduct. Although the effect size of the primary end point was large (25% relative risk reduction) and statistically persuasive (P=0.004), the overall quality was moderate because of, among other issues, the open-label design, premature truncation, and questions about the accuracy of BP measurement and the impact of missing data, thereby downgrading the overall strength of evidence. These limitations arguably detract from the ability of the trial to reliably guide clinical practice.

In conclusion, the overall quality and quantity of evidence are not sufficient to support the current Class I, Level of Evidence B-R recommendation for a universal BP treatment goal of <130/80 mm Hg. Given the modest and statistically nonpersuasive reduction in major adverse cardiovascular events or stroke and the limitations of study design and execution, a Class IIa (moderate), Level of Evidence C-LD (limited data) recommendation might arguably be more appropriate. Adjudication of the remaining uncertainties should be a top priority for future research.