Antihypertensive Treatment and Secondary Prevention of Cardiovascular Disease Events Among Persons Without Hypertension:  A Meta-analysis FREE

Cardiovascular disease (CVD) is the leading cause of death in the United States and globally, representing 30% of all deaths worldwide.¹ Prospective cohort studies have established a strong, graded, and independent positive association between blood pressure levels and risk of CVD, stroke, and premature death.^2- 3 Increased CVD risk begins at systolic blood pressure levels as low as 115 mm Hg, with 54% of stroke and 46% of ischemic heart disease events occurring in persons with blood pressures in this range.⁴ In persons with prehypertension, 90% have at least 1 risk factor above optimal levels for heart disease or stroke, and 68% have at least 1 clinically high-risk factor for heart disease or stroke.⁵

Among adults 35 years and older, more than 17% of those with normal blood pressure and 37% of those with blood pressure in the prehypertensive range (130-139 mm Hg systolic, 86-89 mm Hg diastolic) progress to overt hypertension within 4 years without changes in lifestyle or pharmacological intervention.⁶ In adults 55 years and older, lifetime risk of developing hypertension is greater than 90%.⁷ Recent national surveys report that more than 30% of the general adult population in the United States, Korea, and China has prehypertension.^8- 10

Clinical trials have documented that lowering blood pressure reduces cardiovascular mortality among patients with hypertension.^3,11 Several randomized controlled trials of blood pressure lowering for the prevention of CVD have demonstrated benefit among persons with prehypertension or normal blood pressures,^12- 13 while others have not shown benefit.^14- 15 Given these conflicting results, a meta-analysis of randomized controlled trials that examine antihypertensive treatment among persons with blood pressures in the prehypertensive or normal range for the primary or secondary prevention of CVD may help clarify this issue. The objective of this meta-analysis is to evaluate the association between antihypertensive treatment and secondary prevention of CVD events and all-cause mortality among persons without clinically defined hypertension (≥140 mm Hg systolic or ≥90 mm Hg diastolic and/or use of antihypertensive medications or history of hypertension).

We searched online databases including MEDLINE (1950 to week 3 of January 2011), EMBASE, and the Cochrane Collaboration Central Register of Controlled Clinical Trials using the following terms as Medical Subject Headings and keywords: hypertension or blood pressure or normal blood pressure or prehypertension or pre-hypertension or pre-hypertensive or normotensive and antihypertensive agents, and cardiovascular disease. No language restrictions were applied. Searches were limited to randomized clinical trials in human participants 19 years or older. A manual examination of references in selected articles was also performed.

The titles and abstracts of 874 potentially relevant references were identified through the literature search and reviewed independently by 3 investigators (A.M.T., T.H., C.L.E.) to determine whether they met eligibility criteria for inclusion. Discrepancies regarding whether to include or exclude a study were resolved by consensus with other investigators (J.H., L.A.B.).

Studies were eligible for inclusion if they were randomized controlled trials of antihypertensive treatment among persons with blood pressure less than 140 mm Hg systolic or less than 90 mm Hg diastolic for the prevention of CVD events (fatal or nonfatal stroke, fatal or nonfatal myocardial infarction [MI], congestive heart failure [CHF], or CVD mortality). For studies that produced multiple publications, data from the most recent or most complete publication were included in the analysis.

Studies were excluded if CVD events were not reported by hypertension status in studies that included participants with and without hypertension; the study population did not include persons with blood pressure in the normal or prehypertensive ranges; the study population did not include persons with preexisting CVD or CVD equivalents, such as diabetes; antihypertensive treatment was not part of the intervention; treatment allocation was not random; a measure of variance (P value or confidence interval [CI]) was not reported or could not be calculated from the information provided; participants were younger than 18 years; or there were differences between intervention and control groups other than antihypertensive treatment.

All data were independently abstracted by 3 investigators (A.M.T., T.H., C.L.E.) using a standardized data collection form. Discrepancies were resolved through discussion with other investigators (J.H., L.A.B.) and through reference to the original articles. We attempted to contact study authors for additional information when necessary. Trial characteristics abstracted included design of the randomized controlled trial, type of control, number of treatment groups, description of treatment regimens, description of inclusion and exclusion criteria, numbers of fatal and nonfatal events, definition of participants without hypertension, and demographic characteristics of study populations at baseline. The outcomes recorded included incidence of stroke, MI, CHF events, composite CVD events (as defined by the study), CVD mortality, and all-cause mortality.

The definition of nonhypertensive varied in each study; however, all studies included in this analysis had populations with blood pressure less than 140 mm Hg systolic, less than 90 mm Hg diastolic, or no clinical history of hypertension at baseline. The study-specific definitions of persons without hypertension and outcomes included in this analysis are provided in eTable 1 and eTable 2.

Two authors (A.M.T., T.H.) independently evaluated quality of each study using an established tool.¹⁶ Nine domains were assessed: randomization, concealment of treatment allocation, similarity of groups at baseline, eligibility criteria, blinding of outcome assessor, patient and care provider, point estimates, and intention-to-treat analysis. Disagreement was resolved through consensus and discussion.

For studies that provided an effect estimate such as a relative risk (RR) or hazard ratio, the study-provided effect estimate was directly used in the pooled meta-analysis calculations. For studies that published number of events but did not publish an effect estimate, this information was used to calculate the RR of each outcome for the intervention compared with the placebo group. We logarithmically transformed the RR and corresponding standard error to stabilize the variance and normalize the distribution. We calculated the overall pooled-effect estimates using inverse-variance weighting to calculate both fixed-effects and DerSimonian and Laird random-effects models.¹⁷ The Q test was used to assess the presence of heterogeneity and the I² index to quantify the extent of heterogeneity.^18- 19 Fixed- and random-effects models yielded similar findings, but we detected between-study heterogeneity for several outcomes; therefore, results from the random-effects models are presented. Absolute risk reductions for individual studies were calculated as the difference in event rates between treatment and control groups based on the reported or estimated number of events for each outcome. Pooled absolute risk reductions were calculated using inverse-variance weighted DerSimonian and Laird random-effects models.

To assess for publication bias, we constructed funnel plots for each outcome in which the ln(RR) was plotted against its standard error. The Begg rank correlation test was used to examine the asymmetry of the funnel plot,²⁰ and the Egger weighted linear regression test was used to examine the association between mean effect estimate and its variance.²¹ Prestated subgroup analyses were conducted to assess the influence of the presence or absence of comorbid conditions at baseline and class of antihypertensive treatment. We then conducted sensitivity analyses to examine the robustness of the results and restricted analyses by antihypertensive medication use at baseline, definition of persons without hypertension, trial size, duration of follow-up, and year of publication. Additionally, we conducted sensitivity analyses whereby each study was excluded in turn to evaluate the relative influence of each trial on the pooled estimates. P < .05 was considered statistically significant, and all tests were 2-sided. All analyses were conducted in STATA version 9.2 (StataCorp, College Station, Texas).

Of 874 potentially relevant studies identified in the initial literature search, 25 were included in the meta-analysis (Figure 1). Table 1 describes the characteristics of trials included in the meta-analysis. The class and dose of medication administered in the antihypertensive treatment group varied between studies, but for most studies it progressively increased to a defined target dose. Study duration ranged from a mean length of 1.5 to 63 months. Entry criteria also varied between studies; however, all studies required a history of CVD; clinical evidence of recent MI, CHF, coronary artery disease, or stroke; or CVD equivalent such as type 2 diabetes.

The 25 studies included in the meta-analysis incorporated data from 64 162 participants without hypertension (Table 2). The mean age of participants in the studies ranged from 55.0 to 68.0 years, and 76% of study participants were men. Clinical history of MI, CHF, diabetes, stroke, and coronary artery disease at baseline varied between studies.

Pooled overall RRs and absolute risk reductions per 1000 persons are presented in Figure 2 and Figure 3 for all study outcomes. There was a 23% reduction in risk of stroke (RR, 0.77 [95% CI, 0.61 to 0.98]), 20% reduction in risk of MI (RR, 0.80 [95% CI, 0.69 to 0.93]), 29% reduction in risk of CHF events (RR, 0.71 [95% CI, 0.65 to 0.77]), 15% reduction in risk of composite CVD events (RR, 0.85 [95% CI, 0.80 to 0.90]), 17% reduction in risk for CVD mortality (RR, 0.83 [95% CI, 0.69 to 0.99]), and a 13% reduction in risk for all-cause mortality (RR, 0.87 [95% CI, 0.80 to 0.95]). The absolute risk reduction per 1000 persons was −7.7 (95% CI, −15.2 to −0.3) for stroke, −13.3 (95% CI, −28.4 to 1.7) for MI, −43.6 (95% CI, −65.2 to −22.0) for CHF events, −27.1 (95% CI, −40.3 to −13.9) for composite CVD events, −15.4 (95% CI, −32.5 to 1.7) for CVD mortality, and −13.7 (95% CI, −24.6 to −2.8) for all-cause mortality.

I² values were calculated to quantify heterogeneity between studies. The I² values were 26.5% (P = .24) and 0.0% (P = .85) for MI and CHF events, indicating low heterogeneity between studies. Moderate heterogeneity was detected for stroke events (I² = 61.9% [P = .02 from Q test]), composite CVD events (I² = 35.4% [P = .10]), CVD mortality (I² = 43.6% [P = .12]), and all-cause mortality (I² = 46.1% [P = .03]).

We found no evidence of publication bias as indicated by Begg rank correlation test for any outcome examined. However, possible publication bias was detected for stroke (P = .04) using Egger linear regression tests. Applying the trim and fill adjustment method produced no change in the overall effect estimate for stroke. Exclusion of any single study did not change the significance of the pooled estimates for CHF events, composite CVD outcomes, and all-cause mortality. After individual exclusion of the SOLVD, ABCD, PEACE, PROGRESS, or PATS studies, treatment with antihypertensive medications no longer showed a statistically significant benefit for the outcome of stroke. After exclusion of the SOLVD study, antihypertensive treatment for the prevention of MI no longer showed statistically significant benefit. For the prevention of CVD mortality, the benefit of antihypertensive treatment among persons without hypertension was no longer statistically significant after omission of TRACE, AIRE, or SAVE.

We conducted sensitivity analyses to examine the robustness of the results for the composite CVD outcome and all-cause mortality (eTable 3). Sensitivity analyses were not conducted for the outcomes of stroke, MI, CHF, and CVD mortality because of the small number of studies and events. Results did not differ according to any of these criteria. On a 9-point scale, our quality assessment scores ranged from 7.0 to 9.0 for all studies included. The median score was 9.0 points, and these studies were considered to be excellent quality. There was no difference in the association of antihypertensive treatment and composite CVD outcome or all-cause mortality after exclusion of studies that scored fewer than 9 points (MIS and BHAT received 8 points each; MPI, ASPS, and ABCD received 7 points each).

Additionally, we conducted subgroup analyses to examine whether the association of antihypertensive treatment differed among persons with clinical history of MI or coronary artery disease, those with preexisting CHF, and those with history of diabetes or class of antihypertensive medication (eTable 4). There was little change in the overall effect estimates by clinical history for any of the outcomes, with the exception of diabetes. For prevention of composite CVD outcomes and all-cause mortality, no statistically significant benefit of antihypertensive treatment was reported in trials conducted exclusively in patients with diabetes; however, these results should be interpreted cautiously because of the limited number of trials.

Blood pressure change from baseline to follow-up was available for nonhypertensive participants in 3 studies.^14,37,54 The blood pressure difference between the treatment and placebo groups at the end of the intervention period was significantly different only for those in the ABCD normotensive study.¹⁴

This meta-analysis is unique in that, to our knowledge, it is the first to focus on the association of antihypertensive medication use and secondary prevention of CVD events and all-cause mortality among persons without clinically defined hypertension. Our results show that persons with a history of CVD but with blood pressures in the normal and prehypertensive ranges can obtain significant benefit from antihypertensive treatments. The overall pooled results for antihypertensive treatment compared with control showed a significant reduction in risk for fatal or nonfatal stroke, CHF events, composite CVD events, an all-cause mortality. For fatal and nonfatal MI and for CVD mortality, the pooled relative risk reduction was significant but the pooled absolute risk reduction did not achieve statistical significance. This discrepancy reflects the increased variance of the absolute measures compared with the variance of the relative measures. Results for the outcomes studied were consistent across subgroups and did not differ significantly by trial characteristics.

Risk for CVD increases monotonically at all blood pressure levels in the normotensive and prehypertensive range.^2- 3 Although prehypertension affects nearly 70 million adults in the United States and is associated with an increased risk of CVD similar to that seen for those with hypertension, the use of antihypertensive treatment among persons with blood pressures less than 140/90 mm Hg has been debated.^66- 72 According to the current algorithm for treatment of hypertension in persons with compelling indications (CHF, post-MI, high coronary disease risk, and recurrent stroke prevention), pharmacological treatment is indicated for those whose blood pressure is not controlled to less than 140/90 mm Hg with lifestyle intervention alone.³ Hypertension precedes the development of CHF in the majority of patients and increases risk for MI and CHF.³

The results of this meta-analysis suggest that persons with these compelling indications but without hypertension may also benefit from reduced morbidity and mortality attributable to CVD events when treated with antihypertensive medications. In persons 40 years and older with prehypertension, more than 90% have at least 1 above-optimal risk factor, and more than 68% have at least 1 clinically high risk factor for heart disease or stroke.⁵ Although pharmacological treatment for all individuals in this population would not be economically feasible, a more reasonable strategy might be to identify groups within the prehypertensive population who would obtain the greatest benefit from early pharmacological intervention.

For patients with diabetes, the current algorithm for treatment of hypertension indicates pharmacological treatment for those whose blood pressure is not controlled to less than 130/80 mm Hg with lifestyle intervention alone.³ Recent findings reported from the ACCORD BP (Action to Control Cardiovascular Risk in Diabetes Blood Pressure) trial conducted in patients with diabetes demonstrated no reduction in the rate of fatal or nonfatal CVD events when systolic blood pressure was controlled to less than 120 mm Hg compared with less than 140 mm Hg.⁷³ The ACCORD BP trial included participants with systolic blood pressures of 130 to 180 mm Hg who were taking 3 or fewer antihypertensive medications at baseline. The results of our meta-analysis show that for the prevention of composite CVD outcomes and all-cause mortality, no benefit of antihypertensive treatment was seen in trials conducted in patients with diabetes and without hypertension. Our findings should be interpreted with caution because of the small number of studies in such patients.

We identified only 2 studies of antihypertensive treatment conducted in populations with blood pressures less than 140/90 mm Hg and without a history of CVD or diabetes.^74- 75 The primary objective of both trials was to examine the prevention of hypertension in persons with blood pressure in the prehypertensive range, but CVD events were also examined. Although both studies were small and had relatively few events, there was an indication of possible benefit overall. Additional studies are needed to determine if any benefit of antihypertensive treatment would be obtained in populations without hypertension or clinical history of CVD.

We were able to identify no evidence among populations with specific risk factors such as elevated lipid levels, history of smoking, or chronic kidney disease. Additionally, few studies included racial and ethnic minorities or reported results according to race/ethnicity. Because of the increased risk for CVD events in the presence of these risk factors, additional studies should be conducted to determine if there is benefit of treating prehypertension at levels less than 140/90 mm Hg in populations with these risk factors. Although antihypertensive agents, including β-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and calcium channel blockers are generally well tolerated, deleterious adverse effects are not uncommon and can be serious.

The primary strength of this meta-analysis was its inclusion of only randomized controlled trials, which are less subject to bias and confounding than observational studies. Additionally, study characteristics were very similar at baseline, lending confidence to the findings.

The primary limitation of this meta-analysis was the dearth of studies reporting the outcomes of interest for normotensive and prehypertensive participants. Few studies included in this meta-analysis presented the results by baseline blood pressure levels and treatment regimen; therefore, it was not possible to determine the dose-response relationship between baseline blood pressure and risk of first occurrence or recurrence of CVD events among persons with blood pressure less than 140/90 mm Hg. Additional studies should be conducted to examine the baseline blood pressure level at which antihypertensive treatment should begin in persons with CVD or CVD equivalents such as diabetes.

Moreover, this meta-analysis is not a mechanistic study; thus, we cannot determine whether the benefit associated with use of antihypertensive treatment was attributable to blood pressure lowering or to other tissue or neurohormonal mechanisms. Additionally, it is possible that misclassification of participants may have occurred owing to variations in methods of blood pressure measurement across studies included in the meta-analysis; however, less stringent methods of measurement may overdiagnose hypertension among participants. Because of the small number of studies included, potential publication bias and the influence of heterogeneity between studies cannot be ruled out.

Although we calculated the effect estimate from available data when it was not provided in the published data, it is possible that confounding occurred owing to differential loss to follow-up by treatment group. In addition, the statistical methods resulted in a discrepancy for the findings of 2 outcomes (MI and CVD mortality), perhaps reflecting the increased variance of the absolute measures compared with the variance of the relative measures, which may be compounded by the effect of pooling. Lastly, the total numbers of events were unavailable in some studies; therefore, the counts of events were estimated from the effect estimate and other information provided in the text of publications.^13,59 It was not possible to estimate the total number of events in the COPERNICUS or TRANSCEND studies from the information provided in the text.^15,48 A collaborative meta-analysis pooling individual-patient data could serve to eliminate many of these limitations.

Prehypertension affects nearly 30% of the adult population and carries an elevated risk for CVD incidence and mortality. To our knowledge, this meta-analysis is the first to examine the association between antihypertensive medications and CVD morbidity and mortality as well as all-cause mortality in individuals without hypertension. Among patients with clinical history of CVD but without hypertension, antihypertensive treatment was associated with decreased risk of stroke, CHF, composite CVD events, and all-cause mortality. Additional randomized trial data are necessary to assess these outcomes in patients without CVD clinical recommendations.