Epidemiology of Heart Failure in Diabetes

Association of Diabetes and Incident Heart Failure

Multiple epidemiologic studies have demonstrated that diabetes increases the risk for the development of HF ( Table 23-1 ). In the first 20 years of follow-up in the Framingham Heart Study, diabetes was associated with an almost twofold increased risk of HF in men and a fourfold increased risk in women independent of other risk factors (age, systolic blood pressure, tobacco use, cholesterol, and left ventricular (LV) hypertrophy). Multivariable analyses revealed that diabetes had a high population attributable risk for HF in the Framingham Heart Study, accounting for 6% of cases in men and 12% in women. In the Multi-Ethnic Study of Atherosclerosis (MESA) study of 6814 individuals free of symptomatic cardiovascular disease (CVD) at baseline, diabetes was associated with an almost twofold increased risk for the development of HF, independent of other established risk factors, including baseline LV function (hazard ratio [HR] 1.99, 95% confidence interval [CI] 1.08-3.68). In the National Health and Nutrition Examination Survey (NHANES) Epidemiologic Follow-up Study, the multivariable adjusted relative risk associated with diabetes for the development of HF was 1.85 (95% CI 1.51-2.28; P < 0.001), and the population attributable risk for diabetes was 3.1%. Several other epidemiologic studies have also confirmed that DM is associated with a 2-fold to 3.5-fold increased risk for the development of incident HF compared with the risk in people without diabetes in the general population.

Other studies in populations with greater baseline risk of developing HF have also demonstrated that diabetes is independently associated with incident HF. For example, in the Cardiovascular Health Study of people 65 years of age or older, the multivariable adjusted relative risk of HF in people with diabetes compared with those without diabetes was 1.74 (95% CI 1.38-2.19). In the Cardiovascular Health Study, the incidence rates of HF in men and women with diabetes were 44.6 and 32.5/1000 person-years, respectively, and were markedly greater than in those without diabetes (see Table 23-1 ). In patients with established CAD, diabetes also remains a powerful risk factor for incident HF. In the Heart and Soul Study of 839 individuals with stable CAD, individuals with diabetes had a threefold increased risk of HF compared with those without diabetes (adjusted HR 3.34, 95% CI 1.65-6.76). The incidence rate of HF in the Heart and Soul Study was 36.6/1000 person-years and 17.9/1000 person-years in individuals with and without diabetes, respectively. Diabetes was associated with a more-than-doubled risk of development of HF in patients with stable CAD enrolled in the PEACE clinical trial (HR 2.16, 95% CI 1.67-2.79). Finally, among 2391 women with established CAD who were free of HF at baseline and who were enrolled in the Heart and Estrogen/Progestin Replacement Study (HERS), diabetes was the strongest risk factor for the development of HF (adjusted HR 3.1, 95% CI 2.3-4.2).

In addition to overt diabetes, epidemiologic studies have also demonstrated that milder abnormalities of glucose regulation (below the diagnostic threshold for diabetes) are associated with increased rates of HF. In a community-based, observational cohort of 1187 elderly men without congestive heart failure (CHF) and valvular disease at baseline, parameters of insulin resistance (clamp glucose disposal rate and fasting proinsulin level) predicted CHF incidence independently of established risk factors including clinical diabetes. Similarly, in participants without diabetes or HF at baseline in the Atherosclerosis Risk in Communities (ARIC) study, incident HF rates increased in a stepwise manner with increasing hemoglobin A1c (HbA1c) when compared with the reference group (HbA1c 5.0% to 5.4%) ( Fig. 23-1 ).

Risk factors for incident HF in patients with diabetes are similar to those in individuals without diabetes (see also Chapter 25 ). Cohort studies of individuals with diabetes have shown that risk factors for the development of HF include older age, the presence of ischemic heart disease and CAD, peripheral vascular disease, nephropathy and renal insufficiency, metabolic complications of diabetes, retinopathy, diabetes duration, obesity, and hypertension. ^{, ,} In addition, multiple studies have demonstrated that worsened glycemic control is associated with greater risk for the development of HF in individuals with diabetes. Several studies have demonstrated that for each 1% increase in HbA1c in individuals with diabetes, the risk of incident HF increases by 8% to 36%. ^, The relationship between HbA1c and incident HF in a community-based study of diabetic individuals with and without baseline coronary heart disease is shown in Figure 23-2 .

Specific glucose-lowering therapies have also been associated with incident HF. Observational studies have demonstrated that insulin use at baseline is associated with increased rates of HF in diabetic individuals, ^, but it remains unclear whether insulin use is a marker of diabetes duration and severity or contributing to cardiac dysfunction. Long-term follow-up of randomized, controlled clinical trials that have prospectively studied insulin use have not confirmed greater rates of HF associated with insulin, ^, suggesting that insulin is more likely a marker for diabetes duration and severity rather than a contributor to greater rates of HF. Thiazolidinediones (TZDs) have been associated with fluid retention and increased rates of HF in randomized controlled trials. Although the exact mechanisms of the increased HF events with TZDs are not known, the predominant proposed mechanism relates to TZD-associated volume expansion caused by increased renal sodium reabsorption rather than a direct effect on myocardial structure and function.

As expected, rates of incident HF in diabetic patients vary depending on concomitant HF risk factors present. The heterogeneity of HF event rates for several diabetic clinical trials is demonstrated in Table 23-2 . In the United Kingdom Prospective Diabetes Study (UKPDS), which enrolled individuals with newly diagnosed diabetes (of whom only 2% had macrovascular disease at baseline), the HF event rate was 3.0/1000 person-years. In diabetic participants with multiple cardiovascular (CV) risk factors or established CAD such as those enrolled in the ACCORD trial, the HF event rate was 7.7/1000 person-years. In comparison, in BARI 2D, a trial of patients with type 2 diabetes and obstructive CAD that was deemed suitable for revascularization, the HF event rate was 40.1/1000 person-years. Finally, in patients with chronic kidney disease, anemia, and type 2 diabetes enrolled in the TREAT study, the HF event rate was 44.3/1000 person-years.

Prevalence of Heart Failure and Cardiac Dysfunction in People with Diabetes

The prevalence of HF is approximately 2% in adults older than 20 years in the general population, and HF prevalence increases with age such that HF prevalence is estimated to be 4.5% in women and 7.8% in men who are 60 to 79 years old. Estimated HF prevalence in patients with type 2 diabetes is greater than in the general population, and estimates have ranged from approximately 12% in the general population ^, to 20% to 28% in diabetic individuals aged 60 years or older ( Fig. 23-3 ). ^{, ,} The prevalence of HF is also increased in patients with glycemic abnormalities below the threshold for diabetes as compared with individuals with normal glucose tolerance (see Fig. 23-3 ).

In addition to overt clinical HF, asymptomatic LV dysfunction and abnormalities of cardiac structure and function are also more commonly present in patients with diabetes than in those without diabetes. In the general adult population, the prevalence of asymptomatic LV dysfunction is approximately 3% to 6%. In patients with diabetes, the prevalence of asymptomatic LV dysfunction in the Framingham Heart Study (left ventricular ejection fraction [LVEF] < 50%) was 7%. In a study of individuals with diabetes and hypertension, the prevalence of mild (LVEF 41% to 54%) and severe asymptomatic LV dysfunction (LVEF ≤ 40%) was 12.1% and 5.1%, respectively.

Subclinical abnormalities of cardiac structure and diastolic function are also commonly present in patients with diabetes (i.e., diabetic cardiomyopathy; see also Chapter 24 ). These structural changes include diabetes-associated increases in LV mass, relative wall thickness, and left atrial size. In a large study of more than 12,000 diabetic patients without existing clinical HF, preclinical diastolic dysfunction, defined as an E/e′ (passive transmitral LV inflow velocity to tissue Doppler imaging velocity of the medial mitral annulus during passive filling ratio) greater than 15, was present in approximately 23% of patients. It is important to note that the presence of preclinical diastolic dysfunction was independently predictive of subsequent HF (HR 1.67, 95% CI 1.20-2.33, P = 0.003) and death (HR 2.14, 95% CI 1.36-3.36) ( Fig. 23-4 ).

Prevalence of Diabetes in Patients with Heart Failure

In individuals with symptomatic HF, population studies have demonstrated that the prevalence of diabetes varies from 12% to 33% depending on the population studied. ^, For example, in Olmsted County, Minnesota, approximately 20% of individuals with a new diagnosis of HF had previously recognized diabetes. In Olmsted County the prevalence of diabetes in HF patients increased markedly over a 20-year period (3.8% per year). Other studies have suggested that the prevalence of diabetes may be even greater in hospitalized patients. In a large registry of patients with acute decompensated HF, 44% of patients had a history of diabetes. Similarly, in individuals hospitalized with HF and normal ejection fraction (EF), 33% to 46% of individuals had diabetes. ^, The prevalence of DM in HF clinical trials ranges from 20% to 36%, depending on the patient population studied. It is important to recognize that some individuals with diabetes may be excluded from clinical trials because of exclusion of certain comorbid conditions (such as significant renal dysfunction).

The prevalence of diabetes in HF populations may be even greater when systematic diabetes screening occurs and individuals with previously unrecognized diabetes are identified. In a cohort of outpatients with systolic HF who underwent systematic oral glucose tolerance testing, almost 20% of individuals without a prior diagnosis of diabetes were found to have newly diagnosed diabetes.