Background
An estimated 18.8 million Americans have a diagnosis of diabetes, and a further 7 million have undiagnosed diabetes [
1]. A recent analysis including diagnosed diabetes, undiagnosed diabetes, and pre-diabetes indicates that >40% of the U.S. adult population has some form of hyperglycemia [
2]. The economic burden associated with diabetes is substantial: the total cost attributable to recognized diabetes in the U.S. in 2007 was estimated at $174 billion [
3].
Cardiovascular (CV) complications are the major contributors to morbidity and mortality in patients with diabetes [
4]. The risk of mortality from cardiovascular disease (CVD) is increased by up to 5-fold in patients with diabetes [
5‐
7]. Although debated within the literature, the National Cholesterol Education Program Adult Treatment Panel III has classified diabetes as a coronary heart disease (CHD) risk equivalent [
8,
9]. Dyslipidemia and hypertension are independent predictors of future CV events, and clinical interventions that target these risk factors have been shown to reduce CV outcomes in patients with diabetes [
10‐
16]. As such, aggressive control of modifiable CV risk factors is particularly important in this high-risk population.
Current national treatment guidelines for patients with diabetes [
4] advocate a low-density lipoprotein cholesterol (LDL-C) goal of <2.59 mmol/L (100 mg/dL), with an optional goal of <1.81 mmol/L (70 mg/dL) in those with overt CVD, and a blood pressure (BP) goal of <130/80 mmHg. With respect to lipid-lowering therapy, treatment recommendations include the use of statins in addition to lifestyle modification to improve lipid profiles. For diabetic patients with overt CVD, or those without CVD but >40 years of age with ≥1 other CVD risk factor, statin therapy is recommended irrespective of baseline lipid levels.
Using National Health and Nutritional Examination Survey (NHANES) data covering the period 1999-2008, this analysis of the U.S. adult diabetic population assessed recent trends in the utilization of lipid-lowering and antihypertensive agents, as well as LDL-C and BP goal attainment rates.
Discussion
This analysis of 1999-2008 NHANES data has revealed that the prevalence of diabetes among U.S. adults increased significantly over the past decade. In 2007-2008, 11.9% of participants ≥20 years of age had either diagnosed or undiagnosed diabetes, representing an estimated 26.4 million Americans. Concomitant with the rise in diabetes prevalence was an increase in the utilization of lipid-lowering and antihypertensive medications in these individuals. Although there was a significant improvement in LDL-C goal attainment rates over the 10-year period, the proportion of participants at BP goal remained relatively stable, as did the prevalence of hypertension. Hence, despite the observed increase in the pharmacological treatment of CV risk factors within these high-risk individuals, nearly one-half of the U.S. diabetic population is currently not at recommended LDL-C or BP therapeutic goals. Moreover, only ~1 in 4 people with diabetes are achieving both the LDL-C and BP treatment targets simultaneously, with a similar proportion of diabetic participants considered to be at very high risk of future CV events (those with a history of CVD) reaching the more stringent LDL-C goal of <1.81 mmol/L (70 mg/dL).
These latest estimates of diabetes prevalence and CV risk factor control in the U.S. diabetic population extend upon previous reports derived from NHANES data to 2006. Data from NHANES I (1971 to 1975), NHANES II (1976 to 1980), NHANES III (1988 to 1994), and NHANES 1999 to 2004 showed a linearly decreasing trend of participants who reported not having been diagnosed with diabetes [
19]. For the period 2005-2006, one analysis estimated that 12.9% of U.S. adults had diabetes, which included diagnosed diabetes and undiagnosed diabetes based on either an oral glucose tolerance test or fasting plasma glucose [
2]. The prevalence of diabetes based on diagnosis or fasting plasma glucose only, which is more in line with the criteria for diabetes used in this current analysis, was ~10%. Another study, which included only individuals with diagnosed diabetes, found a prevalence rate for 2003-2006 of 7.8% [
20]. Recent analyses of NHANES data suggest that, based on fasting plasma glucose levels, an additional 2%-3% of the U.S. population have undiagnosed diabetes [
2,
3,
21]. Our extended analysis has shown that diabetes prevalence has increased by nearly 2% in the past 4 years, from 10.2% in 2005-2006 to 11.9% in 2007-2008, an increase of ~4.3 million people. Previous increases in diabetes prevalence rates have been driven by a rise in diagnosed diabetes, with the prevalence of undiagnosed diabetes remaining stable [
2,
21], and may be attributable to increased awareness of the condition among healthcare professionals and the general public. However, this analysis has shown that around one-quarter of diabetes in the U.S. adult population is currently undiagnosed, indicating the potential for further improvement in the recognition of diabetes and its associated risk factors.
Few studies have examined LDL-C goal attainment in the general U.S. diabetic population [
20,
22], with most assessing the achievement of a total cholesterol target of <5.18 mmol/L (200 mg/dL) [
22‐
24]. An analysis of data from NHANES 1988-1994 (NHANES III) and NHANES 1999-2002, in addition to Behavioral Risk Factor Surveillance System data from 1995 and 2002, found a substantial increase in the proportion of the adult diabetic population achieving the less-stringent LDL-C target of <3.37 mmol/L (130 mg/dL) between the baseline surveys (42.4%) and later surveys (64.2%); an LDL-C <2.59 mmol/L (100 mg/dL) was attained in 10.8% and 33.8% of participants with diabetes, respectively [
22]. Estimates from more recent NHANES study cycles found that the proportion of adults with diagnosed diabetes achieving an LDL-C target of <2.59 mmol/L (100 mg/dL) increased significantly, from 36.1% in 1999-2002 to 46.5% in 2003-2006 [
20]. Our results indicate that this positive trend in LDL-C goal attainment among U.S. adults with diabetes is continuing, with around half (54.4%) of this population achieving the therapeutic target in 2007-2008.
The same cannot be said of BP goal attainment. Almost without exception [
24], previous analyses of national population-based data sets have reported a lack of significant improvement in BP control over time in those with diagnosed diabetes [
20,
22,
23]. Although we observed a numerical increase (to 55.1%) in the proportion of U.S. diabetic individuals with a BP ≤130/80 mmHg, the overall trend in BP goal attainment over the 10-year period was not significant.
Undertreatment of dyslipidemia in patients with diabetes
Despite an observed increase in the use of lipid-lowering agents by the U.S. diabetic population over the survey period, in the most recent study cycle (2007-2008) less than one-half (42.2%) of these high-risk individuals reported taking some form of lipid-modifying therapy, and just over one-half (54.4%) were achieving the LDL-C goal recommended by current national treatment guidelines for diabetes. Undertreatment of dyslipidemia in diabetes has also been reported in the real-world clinical setting: a recent evaluation of the pharmacological treatment of mixed dyslipidemia (suboptimal LDL-C, HDL-C, and/or triglyceride levels) in diabetic patients from a large U.S. managed healthcare plan found that >40% received no lipid-lowering therapy even after suboptimal lipid values were obtained [
25]. Underutilization of lipid-lowering drugs in patients with diabetes is not restricted to the U.S. but is a global problem that also affects Europe [e.g. ref. [
26]] and Asia [e.g. ref. [
27]]. An analysis of the German DUTY diabetes registry found that just 24% of primary prevention patients and 46% of secondary prevention patients received lipid-lowering therapy [
26]. Similarly, lipid-lowering therapy was prescribed in only ~29% of Chinese diabetic patients in a prospective cohort analysis of the Hong Kong Diabetes Registry [
27].
Numerous factors likely contribute to the gap between treatment recommendations for CV risk management in diabetes and real-world clinical settings. For example, while LDL-C-targeted statin therapy is advocated for the pharmacological management of dyslipidemia in patients with diabetes [
4], many U.S. physicians hold falsely elevated concerns about the perceived risk of hepatotoxicity that prevent statin prescription in patients with a clinical indication for their use [
28]. Such concerns are not supported by current evidence from randomized trials and meta-analyses demonstrating the safety and tolerability of statin therapy, particularly in patients with diabetes [
10‐
13,
29], and recommendations to discontinue the routine monitoring of liver function tests in statin-treated patients [
30].
Reducing CV risk in diabetes: evidence from randomized controlled trials
Although suboptimal management of CV risk factors persists in U.S. adults with diabetes, the observed increases in the use of lipid-lowering agents and LDL-C goal attainment, particularly in the latter half of the past decade, may correspond to the publication of several large CV outcomes trials supporting the use of statins in this high-risk population. In 2003, a subgroup analysis of patients with diabetes enrolled in the Heart Protection Study (HPS) [
10] reported that treatment with simvastatin reduced the relative risk of first major CV events by 22%. Reductions in relative risk of up to one-third were observed in diabetic patients without a diagnosis of CHD at study entry and/or whose pretreatment LDL-C level was <3.00 mmol/L (116 mg/dL) [
10], suggesting that lipid-lowering therapy is beneficial for patients with diabetes even in the absence of overt CHD or elevated lipid levels. The 2004 publication of the Collaborative Atorvastatin Diabetes Study (CARDS) [
11], which investigated the effectiveness of atorvastatin for primary prevention of CV events in patients with diabetes and baseline LDL-C levels ~3.03 mmol/L (117 mg/dL), reinforced this notion. Randomization to atorvastatin was associated with a 37% reduction in the relative risk of major CV events; the relative risk of stroke was reduced by 48% [
11]. More recent subgroup analyses from other large statin trials, such as the Anglo-Scandinavian Cardiac Outcomes Trial-Lipid-Lowering Arm (ASCOT-LLA) [
12] and the Treating to New Targets (TNT) study [
13], have continued to provide evidence on the benefits of statin therapy in reducing CV complications in diabetic individuals.
Similarly, CV outcomes trials of antihypertensive agents demonstrating the clinical benefit of aggressive treatment to BP targets in patients with diabetes may also have contributed to the significant increase in the use of BP-lowering medications by this group in the past decade. In 1998, the U.K. Prospective Diabetes Study (UKPDS) compared tight control of BP to <150/85 mmHg versus less-tight control of BP to <180/105 mmHg in hypertensive diabetic patients [
14]. The observed reductions in relative risk in patients allocated to tight BP control were 24% for diabetes-related end points (which included sudden death, myocardial infarction, angina, heart failure, and stroke), 32% for deaths related to diabetes (which included death due to myocardial infarction, sudden death, stroke, and peripheral vascular disease), and 44% for stroke [
14]. A subgroup analysis of the diabetic cohort within the Hypertension Optimal Treatment (HOT) trial, where hypertensive patients were randomly assigned to a target diastolic BP, found that the risk of major CV events was halved in those randomized to ≤80 mmHg versus ≤90 mmHg [
15]. However, while trials such as these have generally indicated that aggressive BP control is effective in reducing the risk of CV complications of diabetes, recent results have questioned the clinical benefit of lowering BP levels to within the normotensive range. The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, where diabetic patients were targeted to a systolic BP <120 mmHg versus <140 mmHg, did not demonstrate a significant reduction in the relative risk of major CV events but did show a significant reduction in stroke risk [
31]. Similarly, a subgroup analysis from the International Verapamil SR-Trandolapril Study (INVEST) found that, among diabetic patients with CHD, strict control of systolic BP (<130 mmHg) was not associated with improved CV outcomes in comparison with usual control (130 mmHg to <140 mmHg) [
32]. Nevertheless, despite recent debate around intensive BP control, the finding from our analysis that nearly half of all U.S. adults with diabetes are not meeting the current recommendations around LDL-C or BP treatment goals is of concern, and may have serious implications for the future CVD burden in this population.
Improving quality of care in diabetes
The frontline in the battle to improve therapeutic goal attainment within the context of diabetes is the primary-care setting; hence, quality-of-care initiatives must engage both patients and providers. National awareness campaigns, such as those by the National Diabetes Education Program, serve to increase patient understanding of CV risk factors and their control through self-management. Provider recognition programs and public reporting of performance measures may motivate healthcare professionals to improve levels of diabetes care [
33]. Other provider-level incentive strategies designed to enhance patient outcomes include educational (e.g., continuing medical education credit), practical (e.g., technical assistance for quality improvement activities), managerial (e.g., increased autonomy), and financial (e.g., pay-for-performance) components, which have been used either alone or in combination by individual health plans, employers, and government purchasers with varying degrees of success [
33]. The Physicians Quality Reporting Initiative, implemented by the Centers for Medicare and Medicaid Services, is a pay-for-participation program where providers who satisfactorily report on a set of quality measures earn an incentive payment, and is viewed by many as a framework for a nationalized pay-for-performance scheme. An example of how a government incentive program might improve evidence-based standards of care in diabetes can be found in the United Kingdom where, in 2004, the National Health Service introduced a pay-for-performance scheme to reward primary-care practices that achieve clinical targets for a number of chronic conditions. Substantial improvements in the attainment of BP (≤145/85 mmHg) and total cholesterol (≤5.05 mmol/L [195 mg/dL]) goals in patients with diabetes have been observed since the introduction of the scheme, with as many as ~80% of patients achieving these (albeit less-stringent) targets by 2007-2008 [
34]. However, further evaluation is needed to assess the contribution of this scheme to improvements in care versus underlying positive trends in quality, as well as the potential for selective inclusion or exclusion of patients and achievement-threshold effects [
34].
Study limitations
The results of our current analysis should be interpreted with consideration of the following limitations. Information on NHANES participants' drug utilization and disease history is derived from self-reported data obtained through interviews and questionnaires, and as such may be subject to recall bias. Also, laboratory measurements of glucose and lipid levels were performed on a single specimen, and relied on participants self-reporting an appropriate period of fasting, the absence of which may confound prevalence estimates and goal attainment rates. Although current national guidelines for the assessment of elevated cholesterol levels do not stipulate repeat testing for lipid measurements [
8], the diagnostic criteria for diabetes recommend that any positive test result for diabetes in asymptomatic individuals should be confirmed by a repeat test on a subsequent day to rule out laboratory error [
4]. While confirmation of a diagnosis of diabetes is of vital importance in the clinical setting, the incremental benefit derived from sequential testing of plasma glucose levels within an epidemiological context is less clear. While encouraging, the positive trend in HDL-C goal attainment rates must be interpreted with caution due to potential bias arising from the various HDL-C assays employed in NHANES during this period, despite adjustments made to HDL-C values to account for this [
18]. The low number of participants with valid data on their diabetic status, in addition to complete lipid and BP data, in NHANES 1999-2008 limited the ability to perform and interpret subgroup analyses of the data. The exclusion of institutionalized persons in NHANES, such as those residing in long-term care, may have contributed to the low number of diabetic participants identified for inclusion in this analysis. Finally, NHANES data are derived from separate cross-sectional surveys for each study cycle. Hence, while trends in prevalence estimates and goal attainment rates can be inferred from the data, conclusions as to causality of these temporal relationships are not possible.
Conclusions
Among the U.S. diabetic population, the proportion of individuals taking cholesterol- and BP-lowering medication has increased significantly in the past decade. However, while there has been a significant improvement in LDL-C goal attainment rates over time, nearly one-half of all U.S. adults with diabetes are currently not achieving LDL-C or BP treatment goals recommended by the latest clinical practice guidelines. Moreover, only ~1 in 4 people with diabetes are achieving both the LDL-C and BP treatment targets simultaneously. These nationally representative trends suggest that continued efforts by healthcare professionals are required to achieve recommended treatment goals for CV risk factors and reduce the economic and social burden of CV complications associated with diabetes.
Competing interests
A.K. and J.M. are employees of Pfizer Inc. with ownership of stock in Pfizer Inc.
Authors' contributions
AK designed the study and wrote the manuscript. JM performed the statistical analyses and wrote the manuscript. Both authors read and approved the final manuscript.