OUP user menu

Risks of cardiovascular events and effects of routine blood pressure lowering among patients with type 2 diabetes and atrial fibrillation: results of the ADVANCE study

Xin Du, Toshiharu Ninomiya, Bastiaan de Galan, Edward Abadir, John Chalmers, Avinesh Pillai, Mark Woodward, Mark Cooper, Stephen Harrap, Pavel Hamet, Neil Poulter, Gregory Y.H. Lip, Anushka Patel
DOI: http://dx.doi.org/10.1093/eurheartj/ehp055 1128-1135 First published online: 11 March 2009


Aims The aim of this study was to investigate serious clinical outcomes associated with atrial fibrillation (AF) and the effects of routine blood pressure lowering on such outcomes in the presence or absence of AF, among individuals with type 2 diabetes.

Methods and results About 11 140 patients with type 2 diabetes (7.6% of whom had AF at baseline) were randomized to a fixed combination of perindopril and indapamide or placebo in the Action in Diabetes and Vascular Disease: preterAx and diamicroN-MR Controlled Evaluation (ADVANCE) study. We compared total mortality and cardiovascular disease outcomes and effects of randomized treatment for 4.3 years on such outcomes between patients with and without AF at baseline. After multiple adjustments, AF was associated with a 61% (95% confidence interval 31–96, P < 0.0001) greater risk of all-cause mortality and comparable higher risks of cardiovascular death, stroke, and heart failure (all P < 0.001). Routine treatment with a fixed combination of perindopril and indapamide produced similar relative, but greater absolute, risk reductions for all-cause and cardiovascular mortalities in patients with AF, compared with those without AF. The number of patients needed to be treated with perindopril–indapamide for 5 years to prevent one cardiovascular death was 42 for patients with AF and 120 for patients without AF at baseline.

Conclusion Atrial fibrillation is relatively common in type 2 diabetes and is associated with substantially increased risks of death and cardiovascular events in patients with type 2 diabetes. This arrhythmia identifies individuals who are likely to obtain greater absolute benefits from blood pressure-lowering treatment. Atrial fibrillation in diabetic patients should be regarded as a marker of particularly adverse outcome and prompt aggressive management of all risk factors.

  • Atrial fibrillation
  • Randomized controlled trial
  • Diabetes mellitus
  • Cardiovascular events
  • Mortality
  • Perindopril–indapamide


Diabetes mellitus is a major global health problem with an estimated worldwide prevalence of 2.8% in 2000, projected to increase to 4.4% in 2030.1 People with diabetes have at least twice the risk of vascular complications and cardiovascular death, compared with those without diabetes.2,3 Prevention of vascular complications is a major goal in the management of patients with diabetes. Blood pressure lowering has been shown to be effective in reducing the incidence of macrovascular events and improving survival in such patients.4

Atrial fibrillation (AF) is commonly observed in diabetic patients, with prevalence rates estimated to be at least double than those among people without diabetes5 and up to three times higher in patients with coexistent hypertension.6 In general population studies, the presence of AF is associated with increased risks of stroke, thrombo-embolism, heart failure, and cardiovascular death.7,8 Treatment with anticoagulants and blood pressure-lowering agents are known to ameliorate these risks.9,10 Among anticoagulated AF patients, hypertension contributes to increased stroke and thrombo-embolism, with event rates markedly increasing at systolic blood pressure levels of ≥140 mmHg.11 The extent to which AF further increases the already elevated risk of cardiovascular disease in patients with diabetes has not been evaluated. It is also unknown whether the presence of AF affects the response to blood pressure-lowering treatment in patients with diabetes.

The Action in Diabetes and Vascular disease: preterAx and diamicroN-MR Controlled Evaluation (ADVANCE) study investigated the routine administration of a fixed combination of perindopril and indapamide on major vascular outcomes in patients with type 2 diabetes at elevated cardiovascular risks.12 In these analyses, we aim to quantify mortality and cardiovascular risks associated with AF in patients with type 2 diabetes and to examine whether the effects of blood pressure lowering on cardiovascular events differ between diabetic patients with and without AF.


Study design and participants

ADVANCE is a factorial randomized controlled trial evaluating the effects of blood pressure lowering and intensive blood glucose control on vascular outcomes. The design has been previously published.12 Briefly, 11 140 patients with type 2 diabetes aged 55 years or older, with at least one additional risk factor for cardiovascular events, were included in the study. Approval for the study was obtained from each centre's institutional Ethics Committee, and all participants provided written informed consent.

After a 6-week active run-in period, eligible participants were randomly assigned to either a fixed combination of perindopril and indapamide (2 mg/0.625 mg) or matching placebo. The doses were doubled after 3 months, so that participants were receiving either perindopril–indapamide 4 mg/1.25 mg or matching placebo. Other treatments were continued at the discretion of the responsible physician, except that angiotensin-converting enzyme (ACE)-inhibitors other than perindopril had to be substituted with open-label perindopril at a dose of 2 or 4 mg daily, thus ensuring that the maximum recommended dose of 8 mg was not exceeded. Similarly, open-label perindopril to a maximum of 4 mg/day was the only ACE-inhibitor that could be commenced during follow-up. Use of thiazide or thiazide-like diuretics was not permitted.

Definitions of atrial fibrillation and electrocardiographic abnormalities

Twelve-lead electrocardiograms (ECGs) were obtained at baseline in all participants and reviewed by the local investigator responsible for the patient's care for the presence of AF, left ventricular hypertrophy (LVH), and pathological Q-waves. Atrial fibrillation was considered present when it was identified by the investigator on the baseline ECG or when AF confirmed by ECG had been previously diagnosed. The validity of using investigator diagnosis of AF based on the ECG was evaluated in a randomly selected sample of 1026 (9.2%) ECGs. These ECGs were independently interpreted for the presence or absence of AF by two central readers (X.D. and E.A.). Inter-observer agreement was 100%. Compared with the ECG analysis by the central study investigators, a diagnosis of AF made by the local investigator had a specificity of 98.6% and a sensitivity of 81.6%.


Participants were seen 3, 4, and 6 months after randomization, and subsequently, every 6 months. Information on clinical outcomes was collected at and between scheduled visits, with the date of each event recorded. The outcomes evaluated in this analysis were all-cause mortality, cardiovascular death, major coronary events (fatal or non-fatal myocardial infarction), major cerebrovascular events (fatal or non-fatal stroke), and heart failure (heart failure events leading to death, requiring hospital admission or resulting in an increase in New York Heart Association class). With the exception of non-fatal heart failure, all clinical outcomes were reviewed and validated by an independent endpoint adjudication committee blinded to treatment allocation, using all available documentation (including, but not limited to, inpatient or outpatient records, discharge summaries, death certificates, and autopsy reports).

Statistical analysis

Differences in variables at baseline between subgroups of the study population were tested using the Student's t-test, the Mann–Whitney test, or the χ2 test, as appropriate. The Kaplan–Meier procedure was used to produce the cumulative event curves. Participants were censored at their date of death or, for those still alive at the end of follow-up, the date of their last visit. The risks of events associated with the presence of AF at baseline were estimated using Cox proportional hazards models, with adjustment for potential confounding baseline covariates including age, sex, systolic blood pressure, history of currently treated hypertension, haemoglobin A1c, urinary albumin: creatinine ratio (UACR), estimated glomerular filtration rate (eGFR), low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglycerides, body mass index, LVH or pathological Q-waves on ECG, history of macrovascular disease, history of hospitalization for heart failure, current smoking, current alcohol intake, use of lipid-modifying treatment, use of antiplatelet drugs, and randomized blood pressure-lowering treatment. The selection of variables was based on identifying all measured clinical variables of known or suspected prognostic importance for the outcomes of interest. The assumption of the proportional hazards was checked graphically using the log cumulative hazard plot for all variables included in the Cox model. Only patients without missing data for all these variables were included in these analyses; however, this led to the exclusion of only 6.5% of the participants. Only the first event for each outcome was included in the analyses. Tests of homogeneity of the effects of AF on the risk of relevant outcomes within subgroups defined by sex and oral anticoagulant use were estimated by adding interaction terms to the Cox model.

The effects of randomized blood pressure-lowering treatment on events were calculated using univariate Cox proportional hazards models, according to the principle of intention-to-treat. Differences in blood pressure during follow-up between randomized treatment groups were estimated using linear-mixed models in patients with and without AF. These models included randomized treatment group, blood pressure level at randomization, visits for blood pressure measurements, AF status, and relevant interaction terms, with overall estimates weighed by the number of months between each visit. Tests of the homogeneity of treatment effects between subgroups of patients with and without AF were performed by adding an interaction term to the statistical model. All P-values were calculated from two-tailed tests of statistical significance with a type I error rate of 5%. All analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, NC, USA).


Study participants

Of 11 140 randomized participants, 847 (7.6%) had AF at baseline. The characteristics of patients with and without AF are summarized by randomized treatment in Table 1. Overall, participants with AF at entry were older, were heavier, had higher levels of blood pressure and UACR, but lower eGFR, and were more likely to have established macrovascular disease, history of hospitalization for heart failure, treated hypertension and LVH on ECG, consume alcohol, or have pathological Q-waves on electrocardiography. Patients with AF were also more likely to be taking antiplatelet therapy, but were less likely to be current smokers. Twenty-six per cent of the patients with AF were taking oral anticoagulant therapy, compared with 2% of patients without AF. Except for a slightly higher body mass index in AF patients assigned perindopril–indapamide than in those assigned placebo, there were no other differences in baseline characteristics between randomized groups either for patients with or without AF.

View this table:
Table 1

Clinical characteristics of patients with or without atrial fibrillation at study entry

Impact of atrial fibrillation on the risks of serious clinical outcomes

Over a mean follow-up of 4.3 years (range from <1 month to 5.6 years), a total of 879 patients died. Fifty-three per cent of these deaths were due to cardiovascular causes, and 15% of the deaths occurred in patients with AF. Patients with AF at baseline were at significantly higher risks of cardiovascular events and all-cause mortality compared with those without AF at baseline (Figure 1). After adjusting for a range of covariates, AF was found to be associated with greater risks of death due to any cause [hazard ratio (HR) 1.61; 95% confidence interval (CI): 1.31–1.96; P < 0.0001], cardiovascular death (HR 1.77; 95% CI: 1.36–2.30; P < 0.0001), and heart failure (HR 1.68; 95% CI: 1.27–2.21; P = 0.0002). Atrial fibrillation at baseline was also associated with subsequent major coronary events on univariate analysis, but with adjustment, this association was of borderline statistical significance (HR 1.27; 95% CI: 0.97–1.66; P = 0.09). Patients with AF also had a higher risk of major cerebrovascular events than those without AF (HR 1.68; 95% CI: 1.24–2.26; P = 0.0008), with the HRs associated with AF being similar for ischaemic (HR 1.78; 95% CI: 1.28–2.48; P = 0.0006) and haemorrhagic (HR 1.69; 95% CI: 0.58–4.97; P = 0.34) subtypes. All associations were similar when adjusted only for age and sex. There was no evidence of heterogeneity in the association between AF and any of these outcomes within subgroups defined by the baseline use of oral anticoagulant treatment (all P-value for homogeneity >0.11). Among patients on oral anticoagulant therapy at baseline, the adjusted HRs associated with AF were 2.16 (95% CI: 1.15–4.07; P = 0.02) and 2.32 (95% CI: 1.06–5.12, P = 0.04) for all-cause mortality and cardiovascular death, respectively.

Figure 1

Cumulative incidence of (A) all deaths, (B) cardiovascular deaths, (C) major coronary events, (D) major cerebrovascular events, and (E) heart failure in patients with (red line) and without atrial fibrillation (blue line). P-value was calculated using a univariate Cox model.

The association between AF and cardiovascular death was significantly stronger in women compared with men (Table 2). As a result, the crude annual rate of cardiovascular death was similar among men and women with AF. The estimates of relative risks associated with AF for total mortality, coronary events, and cerebrovascular events were also higher in women than in men, but none of these differences reached statistical significance (Table 2).

View this table:
Table 2

Associations between atrial fibrillation at baseline and the subsequent occurrence of serious clinical outcomes in men and women

Effects of randomized blood pressure-lowering treatment in patients with atrial fibrillation

During follow-up, active treatment with a fixed combination of perindopril and indapamide reduced the blood pressure by 5.3/2.3 mmHg [standard error (SE) 0.8/0.4] and by 5.9/2.3 mmHg (SE 0.2/0.1) more than placebo in patients with and without AF, respectively (P-value for homogeneity systolic = 0.58 and P-value for homogeneity diastolic = 0.90). Active treatment produced similar relative risk reductions in all-cause death, cardiovascular death, and major coronary events in participants with and without AF, without evidence of heterogeneity in any of these treatment effects (all P-value for homogeneity >0.3) (Figure 2). However, because of their higher baseline risk, the absolute benefits associated with active treatment were greater in patients with AF than in those without AF. For example, we estimate that 5 years of active treatment would prevent one death among every 39 patients with AF and one cardiovascular death among every 42 patients with AF, compared with one among 84 and 120 patients without AF, respectively. In ADVANCE, active treatment did not significantly reduce the risks of major cerebrovascular events or heart failure.

Figure 2

Effects of randomized treatment on the risks of serious clinical outcomes among patients with and without atrial fibrillation (AF). Solid boxes represent estimates of treatment effect on the risk of clinical outcomes. Centres of the boxes are placed at the estimates of effect; areas of the boxes are proportional to the number of events. Horizontal lines represent 95% confidence interval. The ‘P-value for homogeneity’ tested the consistency of the treatment effect in those with vs. those without AF after multiple adjustment.

There was very limited power to evaluate the effects of the perindopril–indapamide combination on new-onset AF during follow-up. New-onset AF was identified in 169 (3.3%) patients randomized to active treatment and in 183 (3.6%) patients allocated placebo (HR 0.92; 95% CI: 0.74–1.13; P = 0.41).


These data indicate that, among patients with type 2 diabetes, those with AF are at substantially higher risk of death of any cause, cardiovascular death, major cerebrovascular events, and heart failure, compared with those without AF. Routine blood pressure lowering using a fixed combination of perindopril–indapamide produced similar relative risk reductions in cardiovascular and all-cause mortality in patients with and without AF. However, because of their higher risk of these events, the absolute benefits of this blood pressure-lowering treatment appear much greater in patients with AF.

Comparison with previous studies

Atrial fibrillation is a relatively common arrhythmia in people with diabetes.5,6 The prevalence of AF in ADVANCE is consistent with previous estimates of AF prevalence in populations with type 2 diabetes, which have ranged between 4% in a primary care setting and 15% in hospitalized patients.5,6 In ADVANCE, participants with AF had an annual risk of dying of 3.9%, about double than that of patients without AF. After adjustment for other risk factors, AF was still independently associated with a 61% higher risk of all-cause mortality. To the best of our knowledge, there are no previous data obtained exclusively in patients with diabetes, demonstrating such heightened broad vascular and mortality risks, in addition to the known increased risks of embolic stroke.8 However, our data are consistent with those obtained in largely non-diabetic populations, in which the adjusted excess risk of all-cause mortality attributed to AF ranged from a non-significant 7% in heart failure patients on β-blockers to 132% in hypertensive patients with LVH.13,14 This divergence may be explained by the more common use of anticoagulants and (multiple) blood pressure-lowering agents in patients with heart failure than in those with uncomplicated hypertension.13,14 The level of anticoagulant use in ADVANCE was low, albeit comparable with that described in contemporary community-based studies.15,16 We found no evidence that baseline use of anticoagulant treatment mitigated the impact of AF on any of the outcomes in the current analysis, although the statistical power to examine this was limited.

Sex differences

The presence of AF was associated with a greater increase in the relative risk of cardiovascular death in women than in men. This has previously been observed in non-diabetic populations, in which studies have reported greater risks associated with AF for stroke and cardiovascular mortality in women, compared with men.8,17 In the Framingham cohort, the excess risk of all-cause mortality associated with new-onset AF was 50% in men and 90% in women.18 The greater risk of cardiovascular death associated with AF in women compared with men could not be explained by differences in age, cardiovascular risk factors, or use of concomitant treatments (including the use of hormone replacement therapy). Although sex differences in the association between AF and other outcomes were not clearly demonstrated in ADVANCE and thus while any such differences remain speculative, a higher propensity for thrombo-embolism in women with AF has been previously suggested.17

Atrial fibrillation and the risk of coronary events

The present analyses indicate a strong trend towards an independent association between AF and coronary events in patients with type 2 diabetes. Few studies have addressed coronary outcomes in patients with AF. The Perindopril Protection Against Recurrent Stroke Study (PROGRESS) also reported a non-significant trend towards a higher risk of coronary events in patients with established cerebrovascular disease with AF, compared with those without AF.10 In a cohort of somewhat older patients with AF without prior coronary heart disease and largely devoid of diabetes, the unadjusted annual incidence of first coronary events was as high as 3.1%, ∼50% higher than presented here.19 In the Losartan Intervention For Endpoint reduction in hypertension (LIFE) study, baseline AF was not found to be an independent predictor of non-fatal myocardial infarction in patients with hypertension.14 However, the development of AF during follow-up was associated with a 49% increased risk of myocardial infarction, which was of borderline significance (P = 0.055).20 Evidence that AF confers a hypercoagulable state—and is closely related to generalized atherothrombotic risk and atherosclerosis per se—is consistent with these and our findings.21,22

Effect of blood pressure-lowering treatment in patients with atrial fibrillation

The ADVANCE study showed that routine administration of perindopril and indapamide in the treatment of patients with diabetes produced worthwhile benefits in preventing cardiovascular outcomes, regardless of the initial blood pressure level.23 Consistent with previous studies,10,14 patients with AF derived similar relative benefits from routine blood pressure-lowering treatment, with greater absolute benefits. Atrial fibrillation is, however, frequently undetected due to the absence of symptoms or the presence of paroxysmal episodes, which may account for as much as ∼20% of all cases of asymptomatic AF.24,25 The current analysis highlights the importance of actively evaluating diabetic patients for the presence of AF, to identify those at particularly high risk of cardiovascular events. Routine administration of blood pressure-lowering treatment, as well as greater use of antiplatelet agents, statins, and oral anticoagulants, may be expected to reduce the incidence of a broad range of adverse outcomes in these patients.

Study limitations

ADVANCE was not designed to investigate the association between AF and clinical outcomes and data about AF type (paroxysmal, persistent, or permanent), or underlying causes of AF were not available. In the absence of routine Holter monitoring, the presence of paroxysmal AF in some patients may have not been identified, thus potentially underestimating the risks associated with AF. Furthermore, the diagnosis of AF relied on ECG interpretation by the local investigator rather than on central ECG analysis. This may have led to misclassification of the presence or absence of AF at baseline, which in turn would have led to the underestimation of the true impact of this condition. However, when the sensitivity (82%) and specificity (98%) estimated from the validation study are applied to the entire data set, the percentage of misclassified subjects is likely to be ∼3%, indicating that the effects of any such misclassification would be negligible. Finally, ADVANCE had insufficient power to demonstrate an effect of perindopril–indapamide on the onset of AF.


Atrial fibrillation is a strong, independent marker of overall mortality and cardiovascular events in people with diabetes. The present analyses emphasize the greater absolute benefits that would be expected from routine blood pressure-lowering treatment for patients with type 2 diabetes and coexisting AF. These findings are of direct relevance for the routine clinical management of diabetic patients and indicate that detection of AF in a patient with diabetes should prompt more aggressive treatment of all cardiovascular risk factors.


This work was supported by grants from the National Health and Medical Research Council of Australia (211086 and 358395) and Servier, France.

Conflict of interest: J.C. is a member of an advisory board for Servier and has received grant support and lecture fees from Servier. B.d.G., M.W., M.C., and S.H. have received lecture fees from Servier. N.P. and A.P. have received grant support and lecture fees from Servier. There are no other conflicts of interest relevant to this paper.


T.N. 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. A.P. was supported by a fellowship from the National Heart Foundation of Australia.


View Abstract