OUP user menu

Blood pressure reduction and renin–angiotensin system inhibition for prevention of congestive heart failure: a meta-analysis

Paolo Verdecchia, Fabio Angeli, Claudio Cavallini, Roberto Gattobigio, Giorgio Gentile, Jan A. Staessen, Gianpaolo Reboldi
DOI: http://dx.doi.org/10.1093/eurheartj/ehn575 679-688 First published online: 23 January 2009


Aims It is unclear whether prevention of congestive heart failure (CHF) by drugs that inhibit the renin–angiotensin system (RAS) occurs over and beyond the reduction in blood pressure (BP) achieved by these drugs.

Methods and results We conducted a meta-analysis of trials comparing angiotensin-converting enzyme inhibitors (ACEIs), angiotensin-receptor blockers (ARBs), or calcium-channel blockers (CCBs), with diuretics, β-blockers, or placebo in hypertensive or high-risk subjects without CHF at entry. Both fixed- and random-effect models were used. In trials vs. placebo, the risk of CHF was reduced by 21% with ACEIs (P = 0.007), whereas the effect of ARBs and CCBs was not significant (random-effect models). Thus, CCBs did not increase the risk of CHF. In trials vs. diuretics/β-blockers, no differences were found between ACEIs and comparators [odds ratio (OR) 1.02; 95% confidence interval (CI) 0.84–1.24], whereas CCBs were associated with an 18% higher risk of CHF (OR 1.18; 95% CI 1.00–1.39; P = 0.048). Therefore, ACEIs were not superior to diuretics/β-blockers for the prevention of CHF. Because heterogeneity between trials was significant, we investigated potential sources of heterogeneity by meta-regression. The risk of CHF decreased by 24% (P < 0.001) for each 5 mmHg reduction in systolic BP. The risk of CHF was 19% less with ACEIs/ARBs than CCBs (P < 0.001) and 16% less in studies without multiple risk factors required for entry (P = 0.009).

Conclusion BP reduction is beneficial for the prevention of CHF. Over and beyond BP reduction, the protective effect of ACEIs and ARBs is greater than that of CCBs.

  • Hypertension
  • Therapy
  • Congestive heart failure
  • Myocardial infarction
  • Sudden cardiac death
  • Unstable angina
  • Stroke
  • Prognosis
  • Prevention
  • Meta-analysis
  • Meta-regression
See page 638 for the editorial comment on this article (doi:10.1093/eurheartj/ehp064)


Although the prevention of congestive heart failure (CHF) remains a major therapeutic challenge,1,2 the established association between blood pressure (BP) reduction and the prevention of stroke, coronary heart disease, and cardiovascular death in hypertensive or high-risk subjects35 has not been confirmed for CHF.3

In comparative trials, angiotensin-converting enzyme inhibitors (ACEIs) were more effective than calcium-channel blockers (CCBs) for the prevention of CHF, but not more effective than diuretics/β-blockers.35 Angiotensin-receptor blockers (ARBs) reduced the risk of CHF by 16%.3 However, it is difficult to discriminate the BP-related and unrelated effects of the above classes of antihypertensive drugs for the prevention of CHF. In a previous meta-analysis, we found that over and beyond the BP drop, ACEIs were superior to CCBs for the prevention of coronary heart disease (P = 0.028), whereas CCBs were superior to ACEIs for the prevention of stroke (P = 0.042).6 Following a similar analytic plan, we undertook a meta-regression analysis to investigate the BP-related and unrelated effects of ACEIs, ARBs, and CCBs in the prevention of CHF in patients with hypertension or high cardiovascular risks.


We extracted data from randomized controlled outcome trials, which met all of the following pre-specified selection criteria: (i) comparison between old antihypertensive drugs (diuretics and β-blockers) or placebo with new drugs (ACEIs, ARBs, or CCBs); (ii) publication before 1 September 2008 in peer-reviewed journals indexed in Medline; (iii) inclusion of patients with hypertension or high cardiovascular risk, but without overt heart failure at entry (i.e. heart failure at entry had to be an exclusion criterion); (iv) CHF as a pre-specified and well-defined endpoint; (v) measurement of systolic BP at baseline and follow-up; (vi) median or average follow-up of at least 2 years; and (vii) sample size of 100 subjects or more. We searched (see Appendix for searching criteria) for eligible studies through MEDLINE and the Cochrane Library using research Methodology Filters.7

On the basis of the above criteria, the final search identified 31 trials838 that fulfilled all inclusion criteria. Two of us (P.V. and F.A.) independently extracted the data in duplicate on the basis of an intention-to-treat approach, and inconsistencies were discussed and resolved in conference. We accepted the definition of CHF as reported in the individual reports. Reference treatment refers to old antihypertensive drugs or placebo and experimental treatment to new antihypertensive drugs (ACEIs, ARBs, or CCBs). Our meta-analysis was undertaken according to the Quality of Reporting of Meta-analyses (QUORUM) statement.39 To assess the methodological quality of the studies included in our systematic review, we used the Jadad's score. Studies are scored (range 0–5) based on three key methodological features such as randomization, blinding, and accountability of all patients, including withdrawals.39 Figure 1 shows the flow diagram with information about the selected, included, and excluded trials.

Figure 1

Flow diagram with information about selected, included, and excluded trials.

We calculated odds ratios (ORs) and 95% confidence intervals (CIs) for CHF for each trial separately and for combination of studies according to fixed- and random-effect models. A continuity correction was used for trials without events in one arm. As suggested by Sweeting et al.,40 instead of using the conventional fixed value of 0.5, we conducted a specific sensitivity analysis across a number of correction factors using constants of various sizes (e.g. 10−8, 10−4, 0.01, etc.). We tested the null hypothesis of homogeneity across individual studies by the Q-test. Pooled estimates were assessed for heterogeneity by computing the I2 and τ2 statistics. We explored heterogeneity by conducting the analysis in pre-planned subgroups defined by the drug class used in the study (ACEIs, ARBs, or CCBs) and the type of comparator (old drug or placebo). We assessed the influence of individual studies on pooled estimates by computing a pooled estimate while excluding one study at the time.41 We tested for publication bias by visual inspection of the funnel plot (Figure 2) and, more formally, with a modified regression test based on sample size.42 For all tests, a probability level less than 0.05 was considered significant. All tests were two-sided. Analyses were performed using Stata, version 10 (StataCorp LP, College Station, TX, USA) and R version 2.6.2 (R Foundation for Statistical Computing, Vienna, Austria).

Meta-regression analysis

We used meta-regression analysis to test the relation between outcome and explanatory variables.43 We tested the following potential effect modifiers: (i) the baseline-corrected differences in achieved systolic BP (follow-up minus baseline) between experimental and reference treatment; (ii) drug regimen (ACEIs vs. ARBs vs. CCBs); (iii) the interaction term between the change in systolic BP and the drug regimen in relation to outcome; (iv) duration of follow-up; (v) year of publication; (vi) study design; (vii) inclusion based on the presence of one or more additional cardiovascular risk factors besides the principal one as defined in the study protocol; and (viii) trial quality score.

The fit of each model was assessed using the percentage of among-study variance explained [(1−(τ2 in final model/τ2 in model without covariates))×100], together with a significance test for each covariate. The final model included only the covariates that significantly reduced the between-study variance, as estimated by the change in τ2 statistics and by formally testing the hypothesis of τ2 =0, using a likelihood-ratio test.


Table 1 shows the main characteristics and the quality score of the 31 eligible trials, which included 225 764 patients and 6469 incident cases of CHF.

View this table:
Table 1

Trials comparing angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers or calcium-channel blockers with diuretics/β-blockers or placebo

StudyReference drugNumber of patients (exp/ref)Study designFollow-up (years)Publication yearPatient populationAdditional CV risk factors requiredTrial quality (Jadad's score)
ACEIs as experimental drugs
 ALLHAT8Diuretics9054/15 255RCT4.92002HTN1+ ARF5
 ANBP29Diuretics3044/3039PROBE4.12003HTNNo ARF2
 CAMELOT31Placebo673/655RCT22005CADNo ARF4
 CAPPP10Diuretics/β-blockers5492/5493PROBE6.11999HTNNo ARF2
 DIABHYCAR36Placebo2443/2469RCT42004DM1+ ARF3
 DREAM38Placebo2623/2646RCT32006AGTNo ARF5
 STOP2/ACE-I11Diuretics/β-blockers2205/2213PROBE51999HTNNo ARF2
 UKPDS3912β-blockers400/358RCT8.41998DM1+ ARF3
 EUROPA13Placebo6110/6108RCT4.22003CADNo ARF5
 HOPE14Placebo4645/4652RCT4.52000HR1+ ARF5
 PART-215Placebo308/309RCT4.72000CAD1+ ARF5
 PEACE16Placebo4158/4132RCT4.82004CADNo ARF5
CCBs as experimental drugs
 ACTION30Placebo3825/3840RCT4.92004CADNo ARF5
 ALLHAT8Diuretics9048/15 255RCT4.92002HTN1+ ARF5
 ASCOT33Diuretics9639/9618PROBE5.52005HTN1+ ARF3
 CAMELOT31Placebo663/655RCT22005CADNo ARF4
 CONVINCE17Diuretics/β-blockers8179/8297RCT32003HTN1+ ARF4
 FEVER32Diuretics4841/4870RCT3.32005HTN1+ ARF4
 IDNT28Placebo567/569RCT2.62003DM1+ ARF5
 INSIGHT18Diuretics3157/3164RCT3.52000HTN1+ ARF5
 INVEST19Non-CCBs11 267/11 309PROBE2.72003HTN1+ ARF3
 MIDAS20Diuretics442/441RCT31996HTNNo ARF5
 NICS21Diuretics204/210RCT4.21999HTNNo ARF4
 NORDIL22Diuretics/β-blockers5410/5471PROBE4.52000HTNNo ARF3
 PREVENT29Placebo417/408RCT32000CADNo ARF4
 STOP2/CCB11Diuretics/β-blockers2196/2213PROBE51999HTNNo ARF2
 SYST-EUR23Placebo2398/2297RCT21999HTNNo ARF5
 SHELL24Diuretics942/940RCT2.72003HTNNo ARF2
 STONE25Placebo817/815RCT2.51996HTNNo ARF1
 SYST-CHINA26Placebo1253/1141RCT31998HTNNo ARF0
 VHAS27Diuretics707/707RCT21997HTNNo ARF2
ARBs as experimental drugs
 IDNT30Placebo579/569RCT2.62003DM1+ ARF5
 LIFE34β-blockers4605/4588RCT4.82002HTN1+ ARF5
 RENAAL35Placebo751/762RCT3.42001DM1+ ARF4
 TRANSCEND37Placebo2954/2972RCT4.92008HR1+ ARF5
  • ACEIs, angiotensin-converting enzyme inhibitors; CCBs, calcium-channel blockers; ARBs, angiotensin-receptor blockers; CHF, congestive heart failure; BP, blood pressure; CV, cardiovascular; HTN, hypertensive; CAD, coronary artery disease; DM, diabetes mellitus; AGT, abnormal glucose tolerance; HR, high CV risk; RCT, randomised controlledtTrial; PROBE, prospective randomized open, blinded endpoint; ARF, additional CV risk factors.

  • See appendix for acronyms of trial.

We calculated pooled estimates for pre-defined subgroups: ACEIs vs. placebo, ACEIs vs. old drugs, ARBs vs. placebo, ARBs vs. old drugs, CCBs vs. placebo, and CCBs vs. old drugs (Figure 3). ACEIs were associated with a significantly lesser risk of CHF compared with placebo in both fixed-effect and random-effect models (all P < 0.001). The odds for CHF were higher with ACEIs than with diuretics/β-blockers in a fixed-effect model (P < 0.05), but not in a random-effect model (P = n.s.). ARBs were associated with a non-statistically significant lesser risk of CHF when compared with placebo both in random-effect and fixed-effect models (both P = n.s.), whereas in the sole study of ARBs vs. old drugs (LIFE34), losartan did not reduce the risk of CHF.

Figure 2

Funnel plot for publication bias. Each study is plotted by its effect size [natural log of the odds ratios (ORs)] on the horizontal axis and its precision (inverse of the standard error of the log OR) on the vertical axis. In the absence of publication bias, the plot resembles an inverted funnel with less-precise studies having lesser precision scattered at the bottom to either side of the more precise studies.

Regimens based on CCBs were associated with slightly lesser risk of CHF when compared with placebo in a fixed-effect model (OR 0.83; P = 0.042), but not in a random-effect model (OR 0.83; P = n.s.) and with a significantly higher risk of CHF when compared with regimens based on diuretics/β-blockers in both fixed-effect and random-effect models (both P < 0.05).

The pre-planned comparisons of ACEIs, ARBs, and CCBs vs. control treatments (active treatment plus placebo) are reported in Figure 4. There was a marked heterogeneity across the groups (P < 0.001). None of the comparisons was significant in random-effect models. In a fixed-effect model, CCBs were associated with a 17% higher risk of CHF than comparators (P < 0.001).

Figure 3

Effect of treatment on congestive heart failure in trials comparing angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, or calcium-channel blockers with old drugs (diuretics/β-blockers) or placebo. Solid squares represent the odds ratios in individual trials, with a size proportional to the inverse of variance. Bars and diamond denote the 95% confidence intervals for individual trials and pooled estimates, respectively.

In sensitivity analyses, none of the trials had a significant effect on the overall estimates for CHF (Figure 5). Visual inspection of the funnel plot (Figure 2) and a modified regression test based on sample size42 showed no formal evidence of publication bias (P = 0.281).

Figure 4

Overall comparison of angiotensin-converting enzyme inhibitors (ACEIs), angiotensin-receptor blockers (ARBs), calcium-channel blockers (CCBs), and their composite vs. old drugs (diuretics/β-blockers) or placebo. The weighted systolic blood pressure (SBP) difference was calculated by weighting the difference observed in each contributing trial by the number of randomized subjects. Negative values indicate lower mean follow-up SBP in the first listed group (ACEIs, ARBs, and CCBs).

Figure 5

Results of sensitivity analysis. The figure shows the estimated odds ratios after omitting the named study, with corresponding P-value and I2 statistics.

Because of the considerable weight of the ALLHAT study on the overall estimates, we did a second sensitivity analysis in order to evaluate the influence of ALLHAT on the overall estimate of trials comparing ACEIs and ARBs vs. other active drugs. The pooled estimate obtained after omitting this study (OR 0.93; 95% CI 0.81–1.05) remained within the 95% CIs of the overall estimate obtained without excluding the ALLHAT study (OR 1.00; 95% CI 0.85–1.20). This implies that the ALLHAT study did not exert a significant influence on the overall estimate.

Finally, we did a third sensitivity analysis to evaluate the influence of ALLHAT on the overall estimate of the trials comparing CCBs vs. other active drugs. Although the exclusion of ALLHAT did reduce heterogeneity (I2 = 39% omitting ALLHAT vs. 64% including ALLHAT), the pooled estimate obtained after omitting this study (OR 1.12; 95% CI 0.96–1.32) was still within the 95% CIs of the overall estimate obtained including ALLHAT (OR 1.18; 95% CI 1.00–1.39). Again, this implies that the ALLHAT study did not exert a significant influence on the overall estimate.

Meta-regression analysis

Because heterogeneity among trials and subgroups was substantial (I2 = 71%), we used a random-effect meta-regression analysis to investigate the role of potential effect modifiers on residual between-trial variance (see Methods). Larger differences in systolic BP predicted greater reductions in the risk of CHF (Table 2 and Figure 6). For each 5 mmHg reduction in systolic BP, there was a 24% reduction in the risk of CHF (P < 0.001). Over and beyond the BP difference, the risk of CHF was 19% lower with ACEIs/ARBs than with CCBs (P < 0.001) and 16% lower in studies in which additional cardiovascular risk factors were not required for inclusion when compared with the studies with at least one concomitant risk factor required for entry. None of the other pre-defined covariates achieved statistical significance. In particular, the interaction term between systolic BP reduction and drug treatment did not achieve statistical significance (P = 0.665), suggesting that the slope of the regression lines ran parallel. The relation between the OR for CHF and the systolic BP gradient between the groups for ACEIs/ARBs and CCBs is reported in Figure 6.

View this table:
Table 2

Risk of congestive heart failure in relation to explanatory variables

Type of eventUnit of measureOdds ratio95% confidence intervalsP-value
Congestive heart failure
 Reduction in systolic blood pressure5 mmHg0.760.69–0.85<0.001
Concomitant risk factors required for entry
 One or more1
 Calcium-channel blockers1
 ACE inhibitors or ARBs0.810.72–0.92<0.001
  • The patients' age, sex distribution, and the year of publication of the trials did not contribute to the variance explained by meta-regression.

The percentage of residual between-trial variance (final model τ2 = 0.0034 and model without covariates τ2 = 0.0474) explained by including the above covariates in the model was 92% and no longer significantly different from 0 (likelihood ratio test of τ2 = 0, P = 0.298). The inconsistency statistic (I2) was reduced to 22% (95% CI 0–44%), denoting a low-to-null, and statistically non-significant, level of heterogeneity.

Figure 6

Relation between odds ratios (ORs) for congestive heart failure and differences in achieved systolic blood pressure between randomized groups. Circles represent individual trials and have a diameter proportional to the inverse of the variance of the single ORs.


This overview shows two main findings. First, it provides clear evidence that BP reduction is important for the prevention of CHF. Secondly, it suggests that the protective effect of ACEIs and ARBs is greater than that of CCBs over and beyond the BP reduction achieved by these drugs.

Role of blood pressure reduction

Although the protective effects of BP reduction on myocardial infarction, stroke, and total cardiovascular events are well established,36 it is unclear whether protection from CHF is also related to BP lowering. Moser and Hebert44 found a 52% lesser incidence of CHF with active treatment compared with placebo. However, in the 2003 report of the BPLTC, the difference between randomized groups in the incidence of CHF did not show an association with the observed difference in achieved BP.3 Such a result was driven by a composite pool of three trials that compared CCBs with placebo.3 These three trials were the Systolic Hypertension in Europe trial (Syst-EUR),23 the Irbesartan Diabetic Nephropathy Trial in Patients with Type-2 Diabetes Mellitus (IDNT-2),28 and the Prospective Randomised Evaluation of the Vascular Effects Norvasc Trial (PREVENT).29 In our overview, the comparison of CCBs with placebo has been carried out on the above three trials and five additional trials published after 20033032 or meeting the pre-specified inclusion criteria for our overview.25,26 Another potentially relevant trial, the Nisoldipine in Coronary Artery Disease in Leuven (NICOLE),45,46 was excluded for lack of CHF reporting.

In our estimate, the risk of CHF did not differ between CCBs and placebo, with a non-significant trend in favour of CCBs (OR 0.76; P = 0.089). Our estimates have been obtained using a random-effects model, because of the significant heterogeneity among trials (P = 0.042) that would make the results of the fixed-effect model questionable. In view of the pathophysiological and clinical importance of the relation between BP reduction and protection from CHF, we tried to clarify the discrepancy between our findings and the BPLTC estimates. Thus, we replicated the comparison between CCBs and placebo along a similar analytic plan [i.e. analysis restricted to PREVENT, Syst-Eur, and IDNT2 trials, data extraction from the original Syst-Eur report47 instead of the update23 used in the present study, calculation of relative risk (RR), and application of a fixed-effect model3]. The resulting RR in the CCB group vs. placebo was 1.03 (95% CI 0.81–1.3), inferior to 1.21 (95% CI 0.93–1.58) reported by the BPLTC.3 As the individual estimates of IDNT2 (RR 1.30; 95% CI 0.97–1.72) and PREVENT (RR 0.20; 95% CI 0.02–1.67) were identical in our hands and in the BPLTC report available on the web,48 the discrepancy was likely accounted for by a different estimate of the RR in the Syst-Eur trial [RR 0.72 (95% CI 0.47–1.10) in our hands and 0.87 (95% CI 0.37–2.05) by the BPLTC48].

Overall, our findings remove the possibility of an excess risk of CHF associated with CCBs with respect to placebo and show that an association exists between the BP reduction and prevention of CHF. The latter finding is in line with a recent overview by BPLTC, showing that the magnitude of the risk reduction for CHF is associated with the magnitude of BP reduction.49 Furthermore, when interpreting the occurrence of CHF in studies with CCBs, it should be kept in mind that peripheral oedema could have been interpreted and coded as a manifestation of heart failure in some studies, instead as a usual side effect of some of these drugs, thus overestimating the incidence of CHF with CCBs.

Impact of renin–angiotensin system inhibition

The second original finding of the present study was that the risk of CHF was lower with drugs that inhibit the renin–angiotensin system (RAS) than with CCBs at any given level of BP reduction. These findings are consistent with the BPLTC report, in which the risk of CHF was 18% lower with ACEIs than with CCBs.3 Of note, the present study removes the potential objection that the better outcome with drugs inhibiting the RAS may have been conditioned by the BP reduction.

The benefits of ACEIs and ARBs in patients with CHF are documented.50 In the subjects included in the present overview, who had no evidence of overt CHF at entry, ACEIs could have been particularly beneficial in the subset with asymptomatic left ventricular (LV) dysfunction,51 whose risk of overt CHF is extremely high.52

Some basic mechanisms could explain a higher risk of CHF in subjects treated with CCBs. The neuro-endocrine response to CCB-induced arteriolar vasodilatation, as well as their direct negative inotropic effects on myocardium, may contribute to the development of CHF in predisposed individuals.53 In contrast, long-acting dihydropyridines and non-dihydropyridine CCBs may reduce BP and angina pectoris in patients with LV dysfunction or established CHF.53

An association between BP reduction and protection from CHF was apparent in individual studies with ACEIs or CCBs, in which the differences from comparators in the incidence of CHF were significant. For example, the significant 29% reduction in the risk of CHF with nifedipine compared with placebo in the ACTION trial30 was associated with a 6 mmHg greater BP drop in the CCB arm. In the EUROPA study,13 the significant 39% lesser risk of CHF in the perindopril group when compared with placebo was associated with a 5 mmHg greater BP drop in the ACEI arm. Conversely, the higher risk of CHF with amlodipine than with chlorthalidone in ALLHAT8 was associated with a lesser antihypertensive response in the amlodipine arm.

Furthermore, as discussed earlier, the potential misinterpretation of peripheral oedema as a marker of incident CHF instead as a usual side effect might contribute to systematically overestimate the reported risk of CHF with CCBs.

Limitations of the study

The significant heterogeneity across the trials could be related to differences in the clinical features of subjects included in individual trials and different diagnostic criteria for CHF. A consensus definition of CHF is strongly required because of the inconsistency in CHF reporting across different trials.54 In the present study, in order to maintain a common comparator (diuretics, β-blockers, or placebo) between drugs inhibiting (ACEIs and ARBs) and not inhibiting (CCBs) the RAS, we excluded the single pair-wise comparisons between ACEIs and CCBs, or ACEIs and CCBs from the analysis. Another limitation, inherent to meta-analyses of aggregated data, is the possibility of aggregation bias or ecological fallacy, which consists in applying to single individuals the conclusions drawn from aggregate groups of studies, not from individual patient's data.55 In other terms, the relationships observed for groups may not necessarily apply to single individuals. In our study, however, the plausibility of the explored association between BP, drug treatment, and CHF incidence is strongly supported by existing evidence.35,44 In order to obviate for the lack of individual patient data, we introduced, as a further potential effect modifier, the information about whether or not concomitant cardiovascular risk factors were required for inclusion beyond the principal one. Interestingly, the multivariate analysis showed that the overall risk of CHF was lesser in the studies that did not require concomitant risk factors for inclusion than in those who did, thus suggesting a relatively less risk profile for CHF in the former. A final point to consider is the possibility of the ‘competing risk’ phenomenon as applied to our meta-analysis (i.e. the possibility that a patient may experience an event other than the one of interest, which alters the probability of experiencing the event of interest).56 In this way, censoring would not be fully independent of mechanisms that cause the patient to be censored (‘non-informative’), but actually dependent on the other case-specific event.56 Because we dealt only with the actual absolute number of events without reference to the time to event, the competing risk phenomenon would be unlikely to directly affect our estimates.


Results of our meta-analysis suggest that the risk of CHF decreases with the BP reduction and the use of drugs that inhibit the RAS (ACEIs and ARBs). RAS inhibition appears to prevent CHF, over and beyond BP reduction, in a broad population of patients with hypertension or clinical features of high cardiovascular risk (diabetes, coronary artery disease, and nephropathy), but without established CHF.


This study was supported in part by the no-profit foundation ‘Fondazione Umbra Cuore e Ipertensione—ONLUS’, Perugia, Italy.

Conflict of interest: P.V., C.C., J.A.S., and G.P.R. received consulting and lecture fees and research grants from companies manufacturing angiotensin-receptor blockers. P.V. also received a grant from the Fondazione Umbra Cuore e Ipertensione—ONLUS, Perugia, Italy, for research in cardiovascular disease prevention. G.G. and R.G. have no conflicts of interest to declare.


Medline search filter

(‘Antihypertensive Agents’[Mesh] OR ‘Angiotensin-Converting Enzyme Inhibitors’[Mesh]) OR ‘Receptors, Angiotensin’[Mesh]) OR ‘Calcium Channel Blockers’[Mesh]) OR ‘Adrenergic beta-Antagonists’[Mesh]) OR ‘Diuretics’[Mesh]) AND ‘Randomized Controlled Trial ‘[Publication Type]) NOT ‘Heart Failure’[Mesh]) NOT ‘Child’[Mesh]) NOT ‘Retrospective Studies’[Mesh]) NOT ‘Meta-Analysis ‘[Publication Type]) NOT ‘Review ‘[Publication Type]) NOT ‘Editorial ‘[Publication Type] AND ((‘1’[EDat]:’2008/08/31’[EDat])).

Cochrane central register of controlled trials search filter

(antihypertensive agents or angiotensin-converting enzyme inhibitors or receptors, angiotensin or calcium channel blockers or adrenergic beta-antagonists or diuretics).sh. and randomized controlled trial.pt.


  • This paper was guest edited by Prof. Lars Kober, The Heart Centre, University Hospital of Copenhagen, Copenhagen, Denmark


View Abstract