Blood Pressure Medicines for Five Years to Prevent Death, Heart Attacks, and Strokes

125 for prevented death

Benefits in NNT

1 in 125 were helped (prevented death)
1 in 67 were helped (prevented stroke)
1 in 100 were helped (prevented heart attack: fatal and non-fatal myocardial infarction and sudden or rapid cardiac death)
97% saw no benefit
0.8% were helped by preventing death
1.5% were helped by preventing stroke
1.0% were helped by preventing heart attack (fatal and non-fatal myocardial infarction and sudden or rapid cardiac death)

Harms in NNT

1 in 10 were harmed (medication side effects, stopping the drug)
10% were harmed by medication side effects, stopping the drug
View As:


Efficacy Endpoints

Mortality, heart attack, stroke

Harm Endpoints

Adverse medication effects leading to drug stoppage


Hypertension (elevated blood pressure) is associated with an increased risk of cardiovascular events and mortality. However, numerous studies have shown a number of medications when given to reduce BP can reduce the risk of developing cardiovascular problems like heart attacks and strokes.

The data reviewed here are partially based on a Cochrane review, a meta-analysis of trials comparing antihypertensive medicines to placebo.1 The NNTs listed above assume that the patient is an average person enrolled in trials of thiazide diuretic medicines, a common first line drug class.

This is important because while the relative impact of different drug classes was similar, there were differences. For instance thiazide diuretics and ‘ACE inhibitors’ (ACEI) demonstrated a statistically significant reduction in overall mortality, total stroke, and most other cardiovascular outcomes, whereas calcium channel blockers (CCBs) and beta-blockers only showed a statistically significant reduction in total stroke and a limited number of cardiovascular outcomes. Neither CCBs nor beta-blockers statistically reduced deaths.(see Table 2 below) These differences may reflect, at least for the CCBs, the smaller numbers of research subjects evaluated in the meta-analysis.

What these differences suggest is that antihypertensive medicines are effective for reducing the risk of cardiovascular problems, but possibly to varying degrees. Their impact is therefore complex and multifaceted, and distilling this into a single number is not as valuable as individualizing. Therefore we are including Table 1 that offers NNTs based on demographic variations. We suggest using a calculator to customize even further, like this one. As always for NNTs, these numbers are rough estimates.

Table 1. Numbers-needed-to-treat to avoid the listed cardiovascular outcomes

5 years, systolic BP 170* Heart attacks (fatal and nonfatal) Strokes
Male 50 y/o 238 227
Female 50 y/o 568 310
Male 65 y/o 101 88
Female 65 y/o 294 120
Data use the Framingham calculator to estimate baseline risks of heart attack and stroke, and apply relative reductions from trials.(3-7, see Table 2) Calculator inputs for base-case data: *Non smoker, no diabetes, total cholesterol 200mg/dl, HDL 50 mg/dl


These are data estimates from randomized trials, which tend to represent a best case scenario for a drug’s benefits. In addition, the Framingham database over-estimates CVD risk for some populations so these benefits are, if anything, a further overestimate. However, these estimates are based on five years of treatment, and the number of heart attacks and strokes often increases linearly over time. If true for any given individual this would mean that after ten years of treatment each of the NNT numbers would be halved, and halved again after 15 years, and so on.

It is also notable that not all drugs that lower BP lead to benefits. Atenolol,2 doxazosin,3 and aliskiren4 all lower blood pressure but large RCTs have shown no heart attack, stroke, or death benefit from these agents when used to lower blood pressure. Moreover, evidence for lowering BP below 150 (systolic) with any agent has not been beneficial in trials, but does increase harms (see our other hypertension NNT review).

Importantly, the two earliest trials of blood pressure management5 6 treated patients whose average blood pressures were ~190/120 and 164/105 respectively, and demonstrated impressive and important benefits. These findings support data suggesting that the higher the blood pressure and the higher the risk, the better the NNT. This is evident in our numbers. Note, for instance, that our estimated overall NNTs (top of the page) are more favorable than the NNTs in Table 1. This is likely because patients in early thiazide trials were at higher risk, often due to existing heart disease. The Framingham calculator, which we used for calculations in Table 1, was developed for relatively healthy people and therefore assumes that patients have a lower risk at baseline than those in the thiazide trials.

Harms of BP medications are very real, but not as well documented in trials as benefits. Roughly 10% stop a drug due to intolerability (NNH* 10) and types of side effects vary between antihypertensive classes, including some that can be severe (angioedema, syncope, arrhythmias, electrolyte disturbances, etc.). Many of these side effects are dose related. Moreover, there is increasing documentation of a long held suspicion, namely that antihypertensives may increase fall risk and therefore risk of major injury, particularly for elderly patients.7 As always, in addition to side effects there is the inconvenience of, among many other issues, blood pressure measurement, dose and drug alterations, cost, and pill burden. These concerns are not addressed in trials but may impact quality of life in various ways and degrees.


James McCormack, MD; Peer-Reviewed by Rita Redburg, MD and Barbara Roberts, MD


July 21, 2014