A network for students interested in evidence-based health care

Antihypertensive drugs for primary prevention – at what blood pressure do we start treatment?

Posted on 10th December 2018 by

Evidence Reviews
Blood pressure tool

The Question

Clinical guidelines generally recommend initiating antihypertensive drug therapy in most patients with systolic blood pressure (SBP) over 140 mm Hg or diastolic blood pressure (DBP) over 90 mm Hg, even if they do not have a history of cardiovascular (CV) disease, e.g., myocardial infarction, stroke, or heart failure. This is to prevent these and other complications of hypertension (primary prevention). The 2017 AHA/ACC guidelines even recommend initiation if blood pressure is over 130/80 mm Hg if their expected 10-year chance of these complications is 10%. Some guidelines, such as the 2018 Hypertension Canada guidelines, recommend the same threshold if patients have diabetes. However, a recent Therapeutics Initiative Letter concluded that “in mild hypertension (140-159/90-99) antihypertensive drugs have not been proven to reduce mortality and morbidity.” Why is there such a discrepancy?

The Evidence


From a meta-analysis of observational studies published in 2002 in the Lancet, we know that increasing blood pressure is associated with CV diseases starting as low as 115/75 mm Hg. However, this by itself does not guarantee that simply lowering blood pressure with medications will reduce risk of CV disease. After all, well-done observational studies also showed that increased vitamin E intake was associated with reduced CV disease, but later randomized controlled trials (RCTs) showed that taking vitamin E supplements had no benefit.

A meta-analysis of RCTs of antihypertensive drugs was published in the Lancet in 2016 and showed similar proportional reductions in CV risk no matter the starting blood pressure, down to a starting SBP of <130 mm Hg. In other words, as long as the starting blood pressure was greater than about 120 mm Hg, lowering SBP with medications by 10 mm Hg yielded relative risk reductions of:

  • 20% for all major CV events, 95% CI: 17% to 23%
  • 17% for coronary heart disease (CHD), 95% CI: 12% to 22%
  • 27% for stroke, 95% CI: 23% to 32%
  • 28% for heart failure, 95% CI: 22% to 33%
  • 13% for all-cause death, 95% CI: 9% to 16%

For these outcomes, relative risk reductions increase with greater BP lowering, up to at least a 20 mm Hg reduction in SBP (e.g., a 20 mm Hg reduction would give about a 40% relative risk reduction in major CV events). However, this meta-analysis included trials where patients had already had a history of CV disease or were already on a blood pressure-lowering drug before the trial began (‘at baseline’). These limit the generalizability of these findings when we are wondering when to start these drugs as a form of primary prevention. Fortunately, the meta-analysis found that relative risk reduction was similar whether or not one had a history of CV disease. However, for more direct evidence, we will consider meta-analyses that have looked at this specific population more closely.

2012 Cochrane Review and 2015 BPLTTC Update

The Therapeutics Initiative based their conclusion on a 2012 Cochrane review which looked at and pooled RCTs carried out on patients with mild hypertension with no history of CV disease or renal failure. For an RCT to qualify, >80% of patients needed to fit this description, or it had to have available individual patient data. They found four such RCTs, with 8,912 participants. However, this number was still too low to be able to reliably detect any benefit from antihypertensive medications even if there was one, which lead the reviewers to state that “more RCTs needed as a significant benefit may have been missed.”

In 2015, the Blood Pressure Lowering Treatment Trialists’ Collaboration (BPLTTC) redid the exact same review, but because the BPLTTC has access to the individual patient data from more studies, they were able to add 6,361 participants to the 8,912 included in the Cochrane review. They detected statistically significant changes in some outcomes, but because 60% of the patients in the new data were already taking antihypertensive medications at baseline, its applicability is questionable. It is encouraging, though, that a subgroup analysis did not detect a difference in outcomes based on whether patients were taking antihypertensives at baseline.

Thomopoulos et al. 2014 and 2017

A similar meta-analysis published in 2014 in the Journal of Hypertension excluded trials where antihypertensive drug use at baseline was prevalent, although it did not exclude a history of CV disease. In a secondary analysis limited to trials or subgroups where the control arm had a 10-year CV death risk of 4.5% (roughly 15-20% risk of any CV event, correlating to primary prevention), they still found statistically significant reductions in risk of stroke, CHD, and all-cause death. CV death was borderline significant (RR = 0.57, 95% CI 0.32-1.02). This meta-analysis did not include the results of the HOPE-3 trial, which was published in 2016. This trial too showed that lowering SBP by a mean of 6 mm Hg yielded a reduction (RR = 0.73, 95% CI 0.56-0.94) in the composite outcome of CV death + nonfatal myocardial infarction + nonfatal stroke in the pre-specified subgroup with SBP >143.5 mm Hg (mean 154.1 mm Hg) of an entirely primary prevention population where roughly 22% of the trial participants were on an antihypertensive drug at baseline (although this percentage may have been higher with SBP >143.5 mm Hg). The rate of CV events in the placebo arm was 6.5% over 5.6 years.

A 2017 meta-analysis by the same group looked at the benefit of lowering blood pressure when starting SBP was below 140 mm Hg or starting DBP was below 90 mm Hg, again with minimal antihypertensive medications being taken at baseline. Patients could not have had recent myocardial infarction or heart failure, but a past history of CV disease was not excluded. Overall, a benefit was detected only in the composite outcome of stroke + CHD, mostly due to stroke prevention. However, no benefit was detected in the subgroup with 10-year CV death risk of 2.4%, which correlates best to a primary prevention population. The confidence intervals are relatively wide (e.g., for all-cause death: 0.93 to 1.31, for a SBP reduction of 5 mm Hg), meaning they are consistent with both the benefit seen at other baseline blood pressures as well as harm. In other words, this analysis may have been underpowered to detect an effect.

The Bottom Line

Overview meta-analyses of RCTs have shown that proportionate benefits of blood pressure reductions with drugs in outcomes such as CHD, stroke, all-cause mortality, or CV death, do not depend on baseline blood pressure, going down to <130 mm Hg. Sensitivity analyses did not show difference whether baseline blood pressure had been lowered with drugs or whether there was a history of CV disease. Combining this with the observational data that CV risk increases linearly with increasing blood pressure from 115/75 mm Hg gives us extrapolated evidence for risk reduction in all major outcomes even at low starting blood pressures.

More direct meta-analyses confirm that this is true for initial untreated blood pressure in the range 140-160/90-99 mm Hg in the outcomes of CHD, stroke, all-cause mortality, and likely also CV death, at least in patients with intermediate CV event risk (15-20% 10-year risk). When starting SBP is below 140 mm Hg, direct evidence of benefit is only available for stroke in patients with a 10-year CV death risk of >10% and a history of CV disease. In other words, direct evidence is absent for initiating antihypertensive drugs when blood pressure before any drug therapy is below 140 mm Hg in a primary prevention setting. However, clinicians may feel the extrapolated evidence is strong enough to consider it in high-risk primary prevention patients.

In deciding when to begin antihypertensive drug therapy, decision making should be aided by considering the absolute risk reductions in outcomes, an overall estimate of which can be calculated by multiplying the relative risk for major CV events found in meta-analyses (e.g., from the 2016 Lancet meta-analysis: 0.8 for a 10 mm Hg reduction in SBP) by a predicted absolute CV risk like the Framingham Risk Score, which has been validated for that region. A patient with mild hypertension could very well also have relatively low CV risk (e.g., 5% over 10 years) and may decide that a 1% absolute risk reduction over 10 years is too little for starting drugs to be worth it. It should also be noted that a 10 mm Hg reduction in SBP is harder to achieve when starting blood pressure is already relatively low, so even this risk reduction may be an overestimate unless multiple medications are used. Different patients will make different decisions.



Giorgio Karam

My name is Giorgio Karam and I am a pharmacy student at Dalhousie University set to graduate in 2020. View more posts from Giorgio

Leave a Reply

Your email address will not be published. Required fields are marked *

Subscribe to our newsletter

You will receive our monthly newsletter and free access to Trip Premium.