We evaluated trends in the treatment and mortality of patients with and without diabetes mellitus and acute myocardial infarction over the last decade. Despite improvements in the provision of evidence-based care, patients with diabetes did not derive improvements in long-term survival.
During the last decade, major advances in the spectrum of evidence-based strategies available to improve outcomes during and after acute myocardial infarction (MI) have been made. Recent work has demonstrated the success of the medical profession in providing such treatments, with a corresponding decrease in mortality.1 However, the evidence for many such therapeutic options is derived from highly selected populations that may not reflect the high-risk populations treated in everyday practice. Patients with diabetes mellitus currently constitute approximately a quarter of patients with acute MI, and global rates of diabetes are estimated to double in the next 20 years.2,3
Most evidence for improving outcomes after acute MI in these patients is derived from subgroup analyses, although some diabetes-specific trials are currently ongoing. It is therefore unclear if patients with diabetes are benefiting from our efforts to use widely accepted management strategies. The goal of our study was to compare changes in baseline risk, treatment, and mortality of patients with and without diabetes with acute MI.
Patients and methods
We retrospectively analyzed data from the Evaluation of Methods and Management of Acute Coronary Events (EMMACE) I and II prospective observational studies, conducted in 1995 and 2003, respectively.4,5 These studies included consecutive unselected acute coronary syndrome patients admitted to multiple adjacent hospitals (including 1 tertiary referral center) in West Yorkshire, United Kingdom. In view of the evolving definition of acute coronary syndrome, we studied comparable patients by selecting only acute MI patients as defined by the World Health Organization diagnostic criteria.6 Validation of acute MI required any 2 of the following criteria: ischemic chest pain, cardiac enzyme/biomarker elevation, and serial electrocardiographic changes. This resulted in study cohorts of 1762 and 1642 in 1995 and 2003, respectively.
For each patient, demographic characteristics, relevant comorbid conditions, and indices of index event risk were recorded. Both inpatient treatment strategy and pharmacotherapy prescribed at discharge from the hospital were documented. Mortality data up to 18 months after the index event were provided by the United Kingdom Office of National Statistics.
The characteristics of cohorts with and without diabetes in 1995 and 2003 are shown in the Table. The prevalence of diabetes increased from 12.5% to 16.6% between study periods (P < .001). The age and sex of cohorts with and without diabetes did not change significantly between study periods; the age of the patients with diabetes and acute MI increased (P = .04 in 2003). Cardiovascular risk factors, such as hypertension and hyperlipidemia, became more prevalent in both study groups, although this may, in part, relate to evolving diagnostic thresholds.7,8 The comorbid conditions of cardiac failure and ischemic heart disease became less prevalent only in the group of patients without diabetes. Importantly, the proportion of acute MI patients with non—ST-segment elevation MI (NSTEMI) increased significantly in both groups, although the increase was greater in patients with diabetes. Correspondingly, Killip class, a measure of acute cardiac failure, decreased between 1995 and 2003 in both groups; cardiac arrest rates also decreased by about 25%, although this only achieved significance for cohorts without diabetes.
(n = 221)
(n = 272)
(n = 1541)
(n = 1370)
Age, years, mean (SE)
Male sex, % (n)
Smoking, % (n)
Hypertension, % (n)
Hyperlipidemia, % (n)
Heart failure, % (n)
Ischemic heart disease, % (n)
Cerebrovascular disease, % (n)
Systolic BP, mm Hg, mean (SE)
Heart rate, bpm, mean (SE)
ST-segment elevation, % (n)
Killip class, mean (SE)
Cardiac arrest, % (n)
Reperfusion, % (n)
Revascularization, % (n)
Cardiology input, % (n)
Antiplatelet agent, % (n)
Statin, % (n)
ACE inhibitor, % (n)
β blockers, % (n)
BP indicates blood pressure; bpm, beats/minute; ACE, angiotensin-converting enzyme. (Reprinted with permission from Cubbon RM, Wheatcroft SB, Grant PJ, et al. Temporal trends in mortality of patients with diabetes mellitus suffering acute myocardial infarction: a comparison of over 3000 patients between 1995 and 2003. Eur Heart J. 2007;28:540-545.)
Reassuringly, the management of acute MI has increasingly been coordinated by dedicated cardiology specialists; in 2003, approximately 70% of patients in both groups received care from cardiologists. This is mirrored by improvements in the provision of acute care and evidence-based secondary prevention strategies. Provision of reperfusion therapy (predominantly thrombolysis) decreased in both groups, although this is expected, given the increased prevalence of NSTEMI. Importantly, though, the proportion of patients referred for early revascularization increased substantially for patients with and without diabetes—this is vital, given the increasing contribution of NSTEMI. Prescription of secondary preventative therapies (antiplatelet agents, statins, angiotensin-converting enzyme [ACE] inhibitors, and beta-adrenoreceptor antagonists) assessed at hospital discharge (or death, if sooner) improved in both groups. Particularly striking was the more than 10-fold increase in the use of statins, although ACE-inhibitor and beta blocker use also doubled.
Figure. Kaplan-Meier survival curves from 1995 and 2003 are shown for cohorts with and
without diabetes mellitus. The highlighted area refers to survival improvement within each
group between 1995 and 2003.
Mortality data for both cohorts are shown in the Figure. In both study periods, mortality was greater for patients with diabetes compared with those without diabetes. More striking was the convergence of mortality curves for patients with diabetes in 1995 and 2003, which contrasts with the parallel curves of cohorts without diabetes. Between study periods, mortality rates 30 days after the index event decreased by a relative 40.4% and 29.3% for patients with and without diabetes, respectively. However, 18-month mortality rates only improved for patients without diabetes (P = .003 by log-rank test); patients with diabetes continued to experience an 18-month absolute mortality rate of 36.4%, representing a nonsignificant decrease of 1.6%. For patients with and without diabetes, the differential improvement in 18-month mortality rates was confirmed by a significant interaction between diabetes status and the study period.
Of course, complex changes have occurred in the baseline risk and treatment of study cohorts between 1995 and 2003. To account for these, further adjusted analyses were performed using Cox proportional hazards analysis. After adjusting for the changes in baseline risk and index event severity shown in the Table, the interaction between 18-month mortality and diabetes status remained significant (P = .017). Further adjustment with the addition of the treatment variables shown in the Table resulted in persistent significance of the interaction term (P = .028). This suggests that changes in baseline risk, the index event risk, or treatment factors did not account for the lack of improvement in 18-month mortality of patients with diabetes. Similar improvements in 30-day mortality for patients with and without diabetes were confirmed by a nonsignificant interaction term, which remained after the adjustments described previously.
The most important finding of this study was that despite recent improvements in the use of evidence-based management strategies, patients with diabetes have not benefited from reductions in long-term mortality. This contrasts with the definite improvement in outcome of patients without diabetes, who benefited from similar advances in the provision of high-quality care between 1995 and 2003. However, both groups have at least benefited from reductions in early mortality assessed at 30 days after the index event. The reason patients with diabetes lost such improvements in early mortality, despite demonstrable improvements in the use of revascularization and secondary preventative pharmacotherapy, is unclear. Perhaps 1 reason long-term outcomes have not improved for patients with diabetes may be that their baseline risk (related to comorbidity) or that of the index event worsened between 1995 and 2003. Although factors such as hypertension and hyperlipidemia have increased in both groups, our adjusted analyses suggest that changes in comorbidity do not account for this. Equally, differential changes in management for groups with and without diabetes do not appear to account for our observations.
Interestingly, the improvements in short-term mortality were significant, regardless of diabetes status. This is mirrored by decreases in Killip class and cardiac arrest rates, both important factors in predicting mortality early after acute MI. Recurrent ischemic events are important in modulating longer-term outcome, and although we noted improvements in referral for revascularization in both groups, this may be less effective in the setting of diabetes.9 The same remains true of secondary preventative pharmacotherapy; for example, the Antithrombotic Trialists, Collaboration suggested that antiplatelet agents offer less benefit to patients with diabetes.10 The increasing prevalence of NSTEMI, noted particularly in patients with diabetes, may also be important, given the poorer long-term outcome associated with these syndromes compared with ST-segment elevation acute MI.11
Attention to glycemic control is also likely to be relevant to the outcome of patients with diabetes, both in the short and long term. Although hyperglycemia is a strong predictor of outcome, it remains controversial as to whether acute insulin infusion is able to improve outcome.12 It is unclear how the controversy surrounding such matters over the study period has affected the changes in outcome we have observed. We are also unable to comment on whether changes in glycemic control, or the means of achieving it, have affected mortality trends. However, our observations suggest that attention to diabetes-specific factors (such as hyperglycemia, insulin resistance, inflammation, and oxidant stress) may be important in addressing the poor improvement in outcome in this group.13 Also important is the lower-than-expected rate of diabetes noted in our study. The work of Norhammar and colleagues has shown that even in acute MI patients with random glucose levels < 11.1 mmol/L, one third exhibited diabetes and another third had impaired glucose tolerance.3 If we are to improve the outcomes of acute MI patients with diabetes, it is first important to recognize these patients, and greater attention to screening appears warranted.
We have shown that despite improvements in the quality of care of all patients with acute MI, with corresponding short-term survival gains, only patients without diabetes have benefited from long-term mortality reduction. Patients with diabetes continue to experience 18-month mortality rates similar to those of a decade ago and similar to many solid-organ malignancies. If we are to address this issue in the ever-growing population of patients with diabetes, we must immediately prioritize research into the interactions between diabetes mellitus and cardiovascular disease.
AcknowledgementsDrs Cubbon, Rajwani, and Abbas are supported by British Heart Foundation clinical PhD studentships. The EMMACE projects were funded by contributions from Astra Zeneca, Beckman Coulter Ltd, the British Heart Foundation, Merck Sharpe & Dohme, and the United Kingdom National Health Service Research and Development program.