|Home | About | Journals | Submit | Contact Us | Français|
Objective To define the cardiovascular effects of lowering blood pressure in people with chronic kidney disease.
Design Collaborative prospective meta-analysis of randomised trials.
Data sources and eligibility Participating randomised trials of drugs to lower blood pressure compared with placebo or each other or that compare different blood pressure targets, with at least 1000 patient years of follow-up per arm.
Main outcome measures Major cardiovascular events (stroke, myocardial infarction, heart failure, or cardiovascular death) in composite and individually and all cause death.
Participants 26 trials (152290 participants), including 30295 individuals with reduced estimated glomerular filtration rate (eGFR), which was defined as eGFR <60 mL/min/1.73m2.
Data extraction Individual participant data were available for 23 trials, with summary data from another three. Meta-analysis according to baseline kidney function was performed. Pooled hazard ratios per 5 mm Hg lower blood pressure were estimated with a random effects model.
Results Compared with placebo, blood pressure lowering regimens reduced the risk of major cardiovascular events by about a sixth per 5 mm Hg reduction in systolic blood pressure in individuals with (hazard ratio 0.83, 95% confidence interval 0.76 to 0.90) and without reduced eGFR (0.83, 0.79 to 0.88), with no evidence for any difference in effect (P=1.00 for homogeneity). The results were similar irrespective of whether blood pressure was reduced by regimens based on angiotensin converting enzyme inhibitors, calcium antagonists, or diuretics/β blockers. There was no evidence that the effects of different drug classes on major cardiovascular events varied between patients with different eGFR (all P>0.60 for homogeneity).
Conclusions Blood pressure lowering is an effective strategy for preventing cardiovascular events among people with moderately reduced eGFR. There is little evidence from these overviews to support the preferential choice of particular drug classes for the prevention of cardiovascular events in chronic kidney disease.
Chronic kidney disease, most commonly defined by a reduced glomerular filtration rate (GFR) or abnormal concentrations of proteinuria, or both, is an important public health problem, affecting 10-15% of the adult general population.1 2 3 It is associated with an increased risk of kidney failure and cardiovascular disease.4 5 6 Individuals with early chronic kidney disease are more likely to experience a cardiovascular event than kidney failure,7 and precise and reliable evidence about the effects of strategies to prevent cardiovascular disease in this large population of patients is of great importance.
Blood pressure is an important determinant of the risk of cardiovascular disease in the general population.8 It is well established that interventions that lower blood pressure prevent cardiovascular events.9 10 Blood pressure is commonly raised in individuals with chronic kidney disease,5 6 and guidelines recommend lower blood pressure targets in this population than in people without chronic kidney disease.11 12 Several studies have also suggested particular benefits of drug classes acting through the renin-angiotensin system for the prevention of renal complications.13 The evidence that lowering blood pressure is beneficial for patients with chronic kidney disease as well as those without, however, remains limited, and the comparative efficacy of different regimens to lower blood pressure on the risk of cardiovascular events in patients with and without chronic kidney disease remains uncertain.14 15 16 17 18 19
The Blood Pressure Lowering Treatment Trialists’ Collaboration20 was established to perform a series of prospectively defined overviews of randomised trials to investigate the effects of blood pressure lowering drugs on cardiovascular morbidity and mortality, including assessments of the comparative effects of regimens between major patient subgroups. This analysis, prespecified in the original collaboration protocol,20 quantifies the proportional benefits of blood pressure lowering, and the comparative effects of different classes of blood pressure lowering drugs in individuals with and without chronic kidney disease.
Trials were eligible for inclusion in this prospective collaborative meta-analysis if they met one of the following criteria: patients were randomised to a blood pressure lowering drug/regimen or a control group (placebo or less intensive blood pressure lowering regimen) or patients were randomised between regimens based on different classes of drugs to lower blood pressure. Trials were also required to have a minimum of 1000 patient years of planned follow-up in each randomised arm and not to have presented or published their main results before finalisation of the overview protocol in July 1995.20 The collaboration was jointly established by the principal investigators in 1995, and the inclusion criteria for the overviews20 specified that results of trials reported only after this time could be included. In the early years of the collaboration, participants contributed aggregate trial data but, over time, agreed to provide individual patient data. New trials were identified by a range of methods including computer aided literature searches, scrutiny of abstracts and proceedings of meeting, and by inquiring among colleagues, collaborators, and the manufacturers of antihypertensive drugs. Determination of eligibility was based on a review of details of the study design and quality of the study, regardless of main results of each trial (appendix table 1). The principal investigators for eligible trials were invited to join the collaboration as they were identified. Trials for which data by kidney function had been obtained by April 2012 were included in these analyses. For this overview, data were available from 26 trials providing either individual participant data including serum creatinine concentrations (23 trials21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44) or summary data stratified by estimated glomerular filtration rate (tabular data from two trials15 45 and published hazard ratios from one trial14). Among them, 25 trials14 15 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 were included in the main analyses, while one trial45 contributed participants only to the reduced eGFR stratum and was included in sensitivity analyses. Data regarding urinary protein excretion at baseline were available for 11 trials.21 22 24 27 28 29 30 31 32 37 40 Two trials15 34 39 did not provide information regarding the combined endpoint of major cardiovascular events according to chronic kidney disease status but provided data for other endpoints. The individual participant data requested included characteristics of participants recorded at screening or randomisation, selected measurements during follow-up, and details of all outcomes during the scheduled follow-up period. We used the Cochrane Collaboration’s tool to assess the risk of bias.46
Glomerular filtration rates were estimated with the modification of diet in renal disease formula7:
eGFR=186.3×(serum creatinine/88.4)−1.154×(age)−0.203×1.210 (if black)×0.742 (if female)
where eGFR is in mL/min/1.73 m2, age is in years, and serum creatinine is in µmol/L. The participants were divided into two categories of baseline eGFR (≥60 mL/min/1.73 m2 or and <60 mL/min/1.73 m2) with established cutpoints recommended in renal guidelines.11 Published data from two trials in which kidney function was estimated with the Cockcroft-Gault formula14 or 125-iothalamate clearance45 and patients were categorised as having eGFR <65 mL/min and ≥65 mL/min were deemed sufficiently comparable and included in the main analysis. The presence of proteinuria was defined as any of the following: urinary albumin excretion rate s>200 μg/min or >300 mg/day, urinary albumin concentration >200 mg/L, urinary albumin creatinine ratio >300 μg/mg, or a urinary protein dipstick test result of 1+ or greater.7 Stages of chronic kidney disease were defined according to the guidelines from the Kidney Disease Outcomes Quality Initiative (K/DOQI).11
The six outcomes were defined according to the international classification of diseases and were prespecified in the collaboration’s protocol.20 Outcomes were stroke (non-fatal stroke or death from cerebrovascular disease), coronary heart disease (non-fatal myocardial infarction or death from coronary heart disease including sudden death), heart failure (causing death or requiring admission to hospital), cardiovascular death, and total mortality. The main outcome was major cardiovascular events comprising stroke, coronary heart disease, heart failure, and cardiovascular death. Only the first event of the relevant outcome type was included in each analysis.
The treatment comparisons tested were prespecified in the original protocol.20 In the broad group of trials comparing an active treatment and a control, we carried out separate overviews for angiotensin converting enzyme (ACE) inhibitor based regimens versus placebo; calcium antagonist based regimens versus placebo; and more intensive versus less intensive blood pressure lowering regimens. In the broad group of trials comparing different active agents, we carried out separate overviews for ACE inhibitor based regimens versus conventional treatment (diuretics or β blocker based regimens); calcium antagonist based regimens versus conventional treatment; and ACE inhibitor based regimens versus calcium antagonist based regimens.
The reduction in blood pressure in each trial arm was calculated as the mean of the difference between each participant’s mean blood pressure during follow-up and their blood pressure at baseline. The mean difference in blood pressure between randomised groups was then calculated by subtracting the values for the two arms according to chronic kidney disease status.
To investigate the effects of active treatment compared with placebo, more intensive compared with less intensive regimens, or different drug classes on the outcomes in patients with different baseline eGFR, we performed meta-analyses according to kidney function. All the meta-analyses used a two stage approach whereby the risk estimates were first summarised in each trial and then combined.
For each trial and each outcome, we estimated the risk estimates separately for each subgroup, according to the intention to treat principle. We computed hazard ratios or odds ratios using a Cox proportional hazard regression model (20 trials providing time to event data21 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 44) or a logistic regression model (three trials in which time to event data were not available22 23 43) in the 23 trials for which individual patient data were available. Among the three remaining trials for which data were obtained from the investigators or published reports, we estimated the hazard ratios and variances with a Cox model for one trial14 and calculated the relative risks and its variances from first principles using tabular data for two trials.15 45 Although hazard ratios, odds ratios, and relative risks are not the same, we have shown the risk estimates as hazard ratios because these risk estimates have similar values under the condition of low incidence rates of outcomes and most risk estimates in this study were hazard ratios. Additionally, we carried out sensitivity analyses to examine the impact of removing the five trials15 22 23 43 45 in which the hazard ratios were unavailable for the risk estimates from all analyses. Pooled estimates of hazard ratios and 95% confidence intervals were calculated separately for each subgroup by a random effects model with the method of DerSimonian and Laird. The “metan” routine in Stata was used for pooled estimates. The heterogeneity in treatment effects between trials included for each comparison was estimated with Cochrane’s Q and I2. The consistency of treatment effects across the subgroups of chronic kidney disease status was tested with χ2 tests of homogeneity. The potential for publication bias across included trials was evaluated with Begg’s test and Egger’s test for funnel plot asymmetry with the “metabias” routine (with P<0.10 taken as significant) and the impact of any asymmetry evaluated with the trim and fill analysis47 with the “metatrim” routine in Stata.
For the meta-analysis weighted blood pressure reduction in the placebo controlled trials, we calculated pooled estimates per 5 mm Hg reduction in systolic blood pressure over time as
where SBPi is the mean difference in systolic blood pressure over time between randomised treatment groups and logHRi and Wi are natural log transformed hazard ratios of randomised treatment on each outcome and its inverse variance in trial i.48 49 For the head to head comparisons, where much smaller differences in blood pressure were observed, we calculated similar blood pressure weighted estimates per 1 mm Hg difference between the randomised groups.
We tested for interactions between eGFR fitted as a continuous variable and the effects of treatment on the risk of cardiovascular outcomes to maximise statistical power to detect possible differences in the effects of the various regimens according to baseline eGFR. Interactions between blood pressure lowering treatment and eGFR taken as a continuous variable were investigated with Cox proportional hazard regression models or logistic regression models including study treatment, continuous eGFR values, and their interaction term. The regression coefficient for the latter term estimates the log ratio of hazard ratios for comparing treatments, which shows the effect of a unit increase in eGFR on the treatment effect. Twenty three trials contributing individual participant data were included in these analyses. The log ratios of hazard ratio were pooled with a random effects model. The pooled summary was exponentiated to arrive at the overall estimated ratio of hazard ratio for a 10 mL/min/1.73 m2 decrement in eGFR for each outcome.
Meta-regression analyses were conducted to explore the association between the difference in systolic blood pressure at follow-up between randomised groups and the risk estimate for cardiovascular events in different subgroups, where the response variable was the natural log transformed hazard ratio and the explanatory variable was the effect of treatment on blood pressure. This was investigated across trials by using random effects meta-regression models with inverse variance weighting. Analyses were carried out with the “metareg” routine in Stata. We fitted separate regression lines for each subgroup and compared the slopes of these lines to test for a differential effect of systolic blood pressure reduction on risk reduction between subgroups for each outcome by adding an interaction term between subgroup and systolic blood pressure reduction during follow-up to the model.
As trial participants could contribute only once to a given meta-analysis with regard to the comparison of active treatment versus placebo and a given meta-regression analysis, trials with randomisation to three treatment arms21 35 38 contributed results from only two of the possible three treatment comparisons, with the control arm participants divided randomly between the two comparisons. For similar reasons, for factorial trials that included randomisation to different intensities of blood pressure lowering and randomisation to different drug treatments,31 32 34 45 we included only the results of the randomisation to different intensities of blood pressure lowering.
All statistical analyses were conducted with SAS software for Windows, version 9.1 (SAS Institute, Cary, NC) or Stata, release 9.2 (StataCorp, College Station, TX). In every analysis a P<0.05 was taken to indicate that a result was possibly not due to chance, although each case required careful interpretation, given the large numbers of comparisons made.
Twenty five trials,14 15 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 including 152290 individuals, provided data for these analyses (tables 1, 2, and 33 ).). The trials involved had a low risk of bias (appendix table 1).
There were 121995 individuals (80%) with eGFR ≥60 mL/min/1.73 m2 (mean eGFR 81 (SD 17) mL/min/1.73 m2) and 30295 individuals (20%) with eGFR <60 mL/min/1.73 m2 (mean 52 (SD 7) mL/min/1.73 m2) at baseline (table 44).). Only 439 individuals (0.3%) had eGFR <30 mL/min/1.73 m2 at baseline. Individuals with reduced eGFR tended to be older (68 v 63) and were more likely to be women (60% v 40%) (table 4). Data on urinary protein excretion were available for 37161 individuals, mostly from trials that compared ACE inhibitor based regimens with placebo. Proteinuria was present in 2500 (7%) of these individuals, and patients with proteinuria were more likely to be men (66% v 55%) and tended to have a higher serum creatinine concentration at baseline (91 μmol/L v 86 μmol/L). The baseline characteristics of each trial included were well balanced between the randomised treatment allocations (appendix table 2). There was some evidence of publication bias (Egger’s test P=0.05) across the studies (appendix fig 4). A trim and fill analysis,47 however, did not change the pooled estimate.
For major cardiovascular events, allocation to an ACE inhibitor based regimen reduced the risk of a major cardiovascular event by about a sixth (hazard ratio 0.81, 95% confidence interval 0.73 to 0.90). This reduction seemed similar irrespective of baseline eGFR (0.81 (0.72 to 0.91) for eGFR of ≥60 mL/min/1.73m2 v 0.81 (0.73 to 0.89) for eGFR of <60 mL/min/1.73m2; P=0.99 for homogeneity; fig 11).). There was significant heterogeneity in treatment effect between trials for the subgroup with eGFR >60 mL/min/1.73 m2 (Q=19.4, df=8, P=0.01, I2=59%) but not in the subgroup with reduced GFR (see appendix fig 2). Sensitivity analyses that excluded the three trials15 22 23 in which the hazard ratios were unavailable for the risk estimates did not make any material difference to the findings (hazard ratio 0.80 (0.68 to 0.92) for eGFR ≥60 and 0.76 (0.68 to 0.85) for eGFR<60).
Calcium antagonist based regimens likewise reduced the risk of major cardiovascular events by about a third, once again with similar effects among individuals with and without reduced eGFR (fig 11).). Similar findings were observed for secondary outcomes (appendix table 3).
To account for different magnitudes of reduction in blood pressure with different drug classes, we also performed a meta-analysis weighted for blood pressure reduction (fig 22).). A 5 mm Hg reduction in systolic blood pressure had almost identical effects on major cardiovascular events at different eGFR, with no difference between ACE inhibitor and calcium channel blocker based regimens compared with placebo (fig 22).). As the results for each class of drugs were similar, we undertook a pooled analysis, which showed that these regimens reduced the risk of major cardiovascular events by 17% per 5 mm Hg reduction in systolic blood pressure in both those below and above eGFR of 60 mL/min/1.73m2 (hazard ratio 0.83 (95% confidence interval 0.76 to 0.90) and 0.83 (0.79 to 0.88), respectively). The proportional reductions in major cardiovascular events were also similar when participants were divided into three eGFR categories (≥60, 45-59, and <45 mL/min/1.73 m2).
We investigated the absolute effects of active treatment in patients with eGFR above and below 60 mL/min/1.73m2 on the basis of the incidence rates for major cardiovascular events in the placebo groups and the blood pressure weighted hazard ratios of active treatment. Overall, 21 major cardiovascular events were prevented for every 1000 participants treated (ACE inhibitor or calcium channel blocker based regimens) compared with placebo (number needed to treat (NNT) for preventing one event over an average of four years was 47). Although the magnitude of the relative effects of active treatment were similar across eGFR, the absolute benefit for major cardiovascular events was 50% higher at lower eGFR, with 28 events prevented for every 1000 patients with eGFR below 60 mL/min/1.73m2 (NNT=35) compared with 19 for patients with eGFR above 60 mL/min/1.73m2 (NNT=53).
As shown in figure 33,, for the trials targeting different blood pressure goals there were wide confidence intervals but no evidence of heterogeneity of treatment effects on major cardiovascular events according to baseline eGFR (hazard ratio 0.87 (95% confidence interval 0.73 to 1.03) for eGFR of ≥60 mL/min/1.73m2 v 1.24 (0.62 to 2.48) for eGFR of <60 mL/min/1.73m2; P=0.33 for homogeneity). The effects of more intensive compared with less intensive regimens for secondary outcomes are shown in appendix table 4. A sensitivity analysis that included one further trial45 that comprised only patients with reduced eGFR made little difference to the findings (1.10 (0.76 to 1.60) in patients with eGFR <60 mL/min/1.73m2; P=0.24 for heterogeneity).
In the trials comparing blood pressure lowering regimens based on different drug classes, there were no clear differential effects on major cardiovascular events (fig 44)) between groups with different eGFR for any comparison (all P>0.6 for homogeneity). The results of the comparisons between calcium antagonist based regimens and conventional treatment after exclusion of one trial43 in which the hazard ratios were unavailable for the risk estimates (hazard ratio 1.04 (95% confidence interval 0.98 to 1.11) for eGFR ≥60 and 1.05 (0.95 to 1.16) eGFR <60) and the results of the comparisons among different drug classes in participants with eGFR <60 after inclusion of one trial45 that contributed patients only to the reduced eGFR stratum (1.00 (0.91 to 1.10) for ACE inhibitor v conventional; 1.03 (0.93 to 1.14) for calcium antagonist v conventional; 0.96 (0.86 to 1.08) for ACE inhibitor v calcium antagonist) remained substantially unchanged. The findings were also similar in the blood pressure weighted comparisons (appendix fig 3). For the secondary outcomes, there was heterogeneity in the effect of ACE inhibitor based regimens against heart failure compared with calcium antagonist based regimen by eGFR, with evidence of greater benefit for ACE inhibitors among participants with eGFR >60 (appendix table 5).
We found no evidence of a significant interaction between baseline eGFR and treatment effect for major cardiovascular events (all P>0.18, fig 55).). For the secondary outcomes, we observed just two significant interactions between eGFR and treatment among 30 comparisons, suggesting they might have arisen by chance (appendix fig 4).
The association between blood pressure lowering achieved and the proportional reduction in risk of cardiovascular events achieved was similar in participants with reduced compared with normal eGFR when assessed with meta-regression. A 5 mm Hg reduction in systolic blood pressure during follow-up was associated with a relative risk reduction of 14.2% (95% confidence interval 8.6% to 19.5%) in participants with normal eGFR and 12.1% (2.9% to 20.4%) for participants with reduced eGFR for the primary outcome of major cardiovascular events (P=0.69 for homogeneity) (fig 66).
There was no evidence of heterogeneity (P=0.84 for homogeneity) in the effect of ACE inhibitor based regimens on major cardiovascular events between individuals with (hazard ratio 0.94, 95% confidence interval 0.84 to 1.05) and without proteinuria (0.90, 0.63 to 1.29) per 5 mm Hg reduction in systolic blood pressure, although only limited data were available (fig 77).). There were not enough trials with data on proteinuria to allow reliable assessment of the comparative effects of calcium antagonists, different blood pressure targets, and different drug classes in patients with and without proteinuria.
These analyses, based on a large volume of data with several trials that included people who did not have hypertension, provide compelling evidence for the cardiovascular benefits of reduction in blood pressure in people with stage 1-3 chronic kidney disease. While the proportional reductions in the risk of major cardiovascular events were similar in people with and without evidence of chronic kidney disease, individuals with chronic kidney disease stood to gain much larger absolute benefits because their baseline risk was much higher. The high prevalence of chronic kidney disease in the community (10-15%) means that strategies to increase the use of blood pressure lowering treatments among this group are likely to be highly effective at reducing the burden of cardiovascular mortality and morbidity.5 6 14 16
The data provided no clear evidence that any particular drug class provided greater or lesser cardiovascular protection for patients with compared with those without chronic kidney disease. This was true both for subgroup analyses done by dichotomising patients according to usual eGFR cutpoints and for the more powerful analyses done with eGFR fitted as a continuous variable. Intriguingly, the cardiovascular benefits of ACE-I and calcium channel blockers compared with placebo were virtually identical once we accounted for the small differences in blood pressure lowering achieved in analyses weighted for blood pressure reduction, although different types of calcium channel blockers were used in these analysis, including dihydropyridines (such as amlodipine and nisoldipine) and phenylalkylamines (such as verapamil). Similar findings were observed in the meta-regression analyses and in head to head comparisons that also included diuretics or β blockers. These results suggest that the cardiovascular benefits of lowering blood pressure in people with chronic kidney disease are more dependent on the blood pressure lowering effect achieved than on the agent selected. This is an observation broadly consistent with the previously reported findings across the trial populations,9 although small benefits independent of blood pressure reduction have been reported for ACE inhibitors and coronary heart disease.50 Additionally, a recent randomised control trial in patients with hypertension showed that treatment with benazepril plus amlodipine was superior to that with benazepril plus hydrochlorothiazide in preventing cardiovascular events, despite no clear difference in the achieved blood pressure during follow-up.51 Several studies have also suggested that drugs acting via the renin-angiotensin system provide greater protection against renal outcomes, but we did not assess these outcomes.13 It is therefore possible there might be additional benefits (or harms) independent of blood pressure associated with individual classes of drugs, but this analysis highlights the key role of blood pressure lowering as a driver of reduction in cardiovascular risk. Future analyses including data from completed and ongoing trials should clarify whether greater effects on renal outcomes achieved by these drug classes in patients with proteinuria will translate into greater protection against major vascular events
There was no clear benefit for more intensive compared with less intensive blood pressure lowering regimens in people with chronic kidney disease, although there was only limited power for these analyses and little capacity to detect whether the effects of treatment varied according to kidney function; lower blood pressure targets have, nonetheless, been suggested to improve renal outcomes in people with proteinuric chronic kidney disease,52 53 providing some support for the earlier recommendation that blood pressure be lowered more intensively in these high risk patients. The effects of more intensive blood pressure lowering on cardiovascular outcomes in people with kidney disease remains to be demonstrated by future trials, such as the ongoing Systolic Blood Pressure Intervention Trial (SPRINT, NCT01206062).
These analyses, based on more than 150000 trial participants and 15000 major cardiovascular events, are uniquely powerful and provide precise estimates of the effects on major cardiovascular events of the most widely used regimens to lower blood pressure in patients with and without chronic kidney disease. There are, however, several important limitations. Firstly, most participants with chronic kidney disease in this study were in stage 3a (mostly in the range of 45-60 mL/min/m2), and few participants (0.4%) had eGFR ≤30 mL/min/1.73 m2. As a result, the applicability of these results to populations with stage 4-5 disease is uncertain, although they are broadly consistent with a meta-analysis of blood pressure lowering in patients with severe kidney disease who were receiving dialysis.54 Likewise, few participants had documented proteinuria and only limited numbers of events were available in this population group. Secondly, the analyses depend on the trial data that were available at the start of this study and we could not investigate some questions of interest. These include the effects on cardiovascular outcomes of regimens based on angiotensin receptor blockers and the separate evaluation of regimens based on β blockers or diuretics alone. When possible, we contacted investigators directly to request missing data or extracted data from published reports to complement available data. Therefore, while we acknowledge this is not a complete dataset, we believe selection bias in this study is minimised by the strict pre-specification of trial eligibility criteria including those related to trial size, quality, randomisation process, outcome definition, treatment comparison, and analysis as specified the published study protocol.20 Furthermore, and in line with recommendations for reporting individual patient data meta-analyses where there are missing data,55 we formally assessed publication bias in this study. We found some evidence of publication bias, which is likely to overestimate the pooled treatment effects, although a trim and fill analysis did not change the pooled estimate. Thirdly, we collected data based on the MDRD-GFR equation from the collaborators as the CKD-EPI equation had not yet been reported when the collaboration was first established. Indeed, it has been suggested that the CKD-EPI equation might be superior to the MDRD-GFR equation for estimating GFR in people with eGFR of ≥60 mL/min/1.73 m2, but both equations have similar accuracy in people with eGFR of <60 mL/min/1.73 m2.56 As we divided into only two categories, below and above 60 mL/min/1.73 m2, rather than the detailed categories of eGFR, we believe this issue would be unlikely to have an important impact on the findings. Finally, uncertainty in estimation of the reduction in blood pressure in each trial will reduce the accuracy of estimation of the treatment effect of blood pressure lowering on cardiovascular risk.
These overviews provide clear evidence that a broad range of different blood pressure lowering regimens provide protection against cardiovascular complications in patients with and without chronic kidney disease. Specifically, blood pressure lowering per se, not the effects of a particular drug class (such as renin-angiotensin system blockade), seems to be significantly associated with lower cardiovascular risk in early stage chronic kidney disease, possibly suggesting the importance of blood pressure lowering as a driver of cardiovascular risk reduction. These findings should help to guide decision making for many physicians and their patients, given the absence of clear evidence to date supporting recommendations (for particular drug classes) to reduce the cardiovascular risk among people with early stage chronic kidney disease. As people with chronic kidney disease are at greatly increased cardiovascular risk, the benefits of blood pressure lowering in this population are large. The broader use of blood pressure lowering drugs, along with other proved treatments such as lipid lowering,57 will both improve outcomes for affected individuals and help to reduce the escalating global burden of cardiovascular disease.
Appendix: Supplementary tables and figures [posted as supplied by author]
Members of the blood pressure lowering treatment trialists’ collaboration
L Agodoa, C Anderson, FW Asselbergs, C Baigent, H Black, B Brenner, M Brown, C Bulpitt, R Byington, J Chalmers, R Collins, J Cutler, B Dahlof, B Davis, D de Zeeuw, J Dens, R Estacio, R Fagard, K Fox, T Fukui, L Hansson (deceased), R Holman, L Hunsicker, Y Imai, M Ishii, Y Kanno, J Kostis, K Kuramoto, E Lewis, M Lièvre, L H Lindholm, L Liu, J Lubsen, S Lueders, S MacMahon, E Malacco, G Mancia, M Matsuzaki, B Neal, S Nissen, T Ohkubo, T Ogihara, C Pepine, M Pfeffer, B Pitt, P Poole-Wilson (deceased), N Poulter, M Rahman, W Remme, G Remuzzi, A Rodgers, P Ruggenenti, T Saruta, J Schrader, R Schrier, P Sever, P Sleight, J Staessen, H Suzuki, K Teo, WH van Gilst, G Viberti, J Wang, P Whelton, L Wing, Y Yui, S Yusuf, A Zanchetti.
Collaboration coordinating centre
F Barzi, J Chalmers, S Heritier, N Li, S MacMahon, B Neal, T Ninomiya, V Perkovic, F Turnbull, M Woodward. Renal Working Group: C Baigent (Chair), V Perkovic, F Turnbull, K Wilson, A Cass, P Kearney, M Gallagher, L Agodoa, B Brenner, L Hunsicker, M Rahman, G Remuzzi, P Ruggenenti, P Whelton.
T Ninomiya, V Perkovic, F Turnbull, B Neal, F Barzi, A Cass, C Baigent, J Chalmers, N Li, M Woodward, S MacMahon.
Contributors: TN contributed to the study design, data management, statistical analysis, and data interpretation, and wrote the first draft of the manuscript and is guarantor. VP and CB coordinated and led the development of this paper, contributed to the study concept and design, data interpretation, and drafting of the manuscript. FT and NL contributed to the data collection, data management, and critical revision. AC contributed to the critical revision of the manuscript. FB provided support for the statistical analysis and contributed to the critical revision of the manuscript. MW contributed to the study concept, data collection, support of the statistical analysis, and critical revision of the manuscript. BN, JC, and SMacM contributed to the study concept, data collection, and critical revision of the manuscript.
Funding: This project was funded by the National Health and Medical Research Council of Australia. The funding body had no input into the design, conduct or interpretation of the study. T Ninomiya was supported by a fellowship awarded by the Banyu Life Science Foundation and by an International Society of Hypertension Visiting Postdoctoral Fellowship awarded by the Foundation for High Blood Pressure Research Council of Australia. V Perkovic was supported by a National Health and Medical Research Council program grant and a Heart Foundation of Australia Career Development Fellowship. F Turnbull was supported by a National Heart Foundation of Australia Career Development Fellowship. A Cass was supported by National Health and Medical Research Council of Australia Senior and Principal Research Fellowships. B Neal was supported by fellowships awarded by the Australian Research Council and the National National Health and Medical Research Council of Australia, and the Collaboration Coordinating Centre by Program and Project Grants provided by the National Health and Medical Research Council of Australia. N Li was supported by a Fogarty International Clinical Research Fellowship and National Health and Medical Research Council Australia-China Exchange Fellowship.
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: VP has received honoraria from Astra Zeneca, Merck, and Servier for scientific presentations and is a trial steering committee and/or advisory board member for Abbvie, Astellas, Baxter, Boehringer Ingelheim, Janssen, and Vitae: JC, MW, and SM have received honoraria from Servier for scientific presentations relating to blood pressure; SM and JC were principal investigators on ADVANCE, a blood pressure lowering trial funded by Servier and the Australian National Health and Medical Research Council; BN has received blood pressure related research support from Servier and honoraria for scientific presentations related to blood pressure from Novartis, Tanabe, and Servier.
Ethical approval: Not required.
Data sharing: Technical appendix, statistical code, and dataset are available from the corresponding author.
Cite this as: BMJ 2013;347:f5680