Impact of Prediabetes on Cardiac Function Among Primary Care Patients

Abstract

Background:

Prediabetes is a novel risk factor recently associated with changes in the left ventricle. Our aim is to determine if prediabetes is associated with heart failure (HF) and structural heart disease.

Methods:

We conducted a cross-sectional study and performed screening echocardiograms to consecutive primary care patients. We calculated the hemoglobin A1c (HbA1c) within 3 months of the echocardiogram and classified patients as having normal glucose, low-risk or high-risk prediabetes or diabetes. Our primary outcome was HF defined as an ejection fraction (EF) <50% and HF with preserved EF. Our secondary outcome was structural heart disease defined as having either a large atrium, left ventricular hypertrophy, or low EF.

Results:

We included 15,056 patients who underwent a screening echocardiogram and had a recorded HbA1c. Only 2794 patients had a normal blood glucose, 4201 had low-risk prediabetes, 2499 had high-risk prediabetes, and the remainder had diabetes. The adjusted odds ratio (ORs) of HF for low-risk prediabetes, high-risk prediabetes and diabetes were 1.38 [confidence interval (95% CI) 1.07–1.78] (P = 0.01), 1.47 (95% CI 1.05–2.01) (P = 0.01), and 1.60 (95% CI 1.16–2.01) (P < 0.01), respectively, when compared with normoglycemic patients. The adjusted OR of HF with preserved EF for low- and high-risk prediabetes and diabetes were 1.17 (95% CI 0.86–1.60) (P = 0.30), 1.60 (95% CI 1.15–2.21) (P < 0.01), and 1.63 (95% CI 1.24–2.13) (P < 0.01), respectively, when compared with normoglycemic patients.

Conclusions:

Prediabetes is a prevalent condition associated with structural heart disease and HF.

Introduction

Prediabetes is a highly prevalent condition that has been linked to cardiovascular risk.^1,2 There is growing evidence that prediabetes is significantly associated with a number of echocardiographic abnormalities such as a lower E/A ratio, a higher left atrial volume, increased left ventricular mass, and a subtle reduction in left ventricular systolic function.^3,4 Moreover, the duration of prediabetes seems to influence the occurrence of structural heart changes.

Although the exact pathophysiology still needs to be elucidated, there is sufficient preliminary data to understand that the effects of prediabetes on the function of the heart are detrimental. The prediabetes effects on the heart structure and function seem to be related to impaired endothelial function, which in turn may be due to an increase in glycogen deposition, inflammation, and apoptosis of cardiac cells.^5,6 However, the paucity of the data to date has led to a lack of guidelines on how providers should address cardiovascular risk in a population with an expanding prevalence of prediabetes.⁷

A large at-risk network of primary care clinics across seven states established a preventive model of care that favors screening of risk factors to offer primary prevention strategies to all their patients.⁸ As part of those efforts, they decided to perform a screening echocardiogram among all patients in their practice.

The purpose of this study is to expand the evidence to date and evaluate the impact of prediabetes on heart function in a large racially diverse senior patient population and evaluate how the distribution of cardiovascular abnormalities differed across the spectrum of glycemic control: from normoglycemic to type 2 diabetic patients. This will inform initiatives to improve prediabetes management and future studies on the long-term impact of preventative interventions on cardiac function and outcomes.

Methods

Study setting

Chen Neighborhood Medical Centers (CNMC)/JenCare Neighborhood Medical Center (JCNMC) are a fully capitated group network practice in seven states. Patients are insured through Medicare Advantage Plans that serve as fiscal intermediaries for processing claims. As part of a system-wide focus on prevention and wellness, patients with chronic illnesses are seen every 1–2 months by their primary care providers and have a screening echocardiogram. The study was approved by Western IRB.

Screening echocardiogram protocol

We offered a screening echocardiogram to all patients enrolled in CNMC or JCNMC in seven states who attended a clinic visit between January 1, 2014 and September 9, 2017. Patients underwent the screening echocardiogram only once during that period. Patients signed informed consent forms and had the echocardiogram performed according to American Society of Echocardiography guidelines.⁹ A single board-certified cardiologist blinded to the patient's history interpreted the echocardiograms at each site. This screening protocol did not include additional maneuvers hence we could not differentiate between stage 3 and 4 diastolic dysfunction and as a consequence all patients with grade IV were classified as a grade III for the purpose of this analysis.

Study design and study population

We conducted a cross-sectional study of all the patients who had the screening echocardiogram during the study period. We included all consecutive primary care patients who came to a visit and had a screening echocardiogram and a hemoglobin A1c (HbA1c) measured.

Definition of diabetes cohorts

We classified everyone into four mutually exclusive groups: normal glucose metabolism, low-risk prediabetes, high-risk prediabetes, and diabetes using the mean 3-month HbA1c before the echocardiogram and the presence of a diabetes diagnosis. We used ICD-9 code 250.x to define diabetes mellitus type 2.¹⁰ We considered patients normoglycemic if the HbA1c before the echocardiogram was <5.7 and there was no diagnosis of diabetes. To define prediabetes we used the American Diabetes Association (ADA) criteria that considers a HbA1c between 5.7% and 6.4% in the absence of a diabetes diagnosis (ICD-9 code 250.x) as prediabetes. The ADA further divides this group as low-risk (HbA1c 5.7–6.0) and high-risk (HbA1c 6.1–6.4) prediabetes based on the risk of conversion to diabetes for the next 10 years.^11,12 Our last group were the diabetics defined as a HbA1c ≥6.5 with a diagnosis of diabetes or use of oral antidiabetic or injectable antidiabetic medication.

We ascertained use of these antidiabetic medications by National Drug Codes.

Primary outcome

Our primary outcome was systolic heart failure (HF). We defined systolic HF as the presence of an ejection fraction (EF) of <50% in the screening echocardiogram. In addition, we defined HF with preserved EF as the presence of ICD-9 code 420.xx code with a normal EF in the screening echocardiogram.¹³

Our secondary outcome was a composite of the following findings: diastolic dysfunction, left atrial size, left ventricular size, and left ventricular mass. We used standard criteria to evaluate left ventricular diastolic function. We used the Doppler mitral flow and tissue velocities tracings (septal and lateral wall) to measure the following diastolic parameters: E-wave velocity, A-wave velocity, E/A, E-wave deceleration time. Diastolic function was classified according to recent recommendations of American Society of Echocardiography on diastolic functional evaluation.

The grading scheme was mild or grade I (impaired relaxation pattern), moderate or grade II (pseudonormalized filling), and severe (restrictive pattern) or grade III.¹⁴ Measurements of left ventricular ejection fraction were performed using the modified biplane Simpson's rule as a mean of three cardiac cycles. We measured the left atrium as the anteroposterior diameter in the parasternal long axis view. We defined structural heart disease as any abnormal left atrial size (>3.8 cm in women or >4.0 cm in men) or abnormal left ventricular size (posterior or septal thickness as >0.09 cm) or abnormal left ventricular mass indexed for body surface area.

Other variables

We collected data from the electronic medical record to ascertain the following covariates: age, gender, race, values for body mass index, and blood pressure lipid panel at the time of the echocardiogram. We also collected the diagnosis of coronary artery disease and the use of angiotensin-converting enzyme (ACE) inhibitor as probable confounders on the effect on left ventricular dynamics. We calculated the Charlson score as a measure of disease burden. This is a validated method to assess co-morbidity status.¹⁵

Statistical analysis

We compared baseline characteristics by blood glucose categories using chi-squared test and ANOVA.

We calculated the percentage of having structural heart disease by blood glucose group and compared it using chi-squared test. We also calculated the mean left atrial size, left ventricular size, and EF, and compared it between blood glucose group using ANOVA.

To identify if the blood glucose category was predictive of the primary outcome we used logistic regression to calculate the odds ratio (OR) and corresponding confidence interval (95% CI) of having HF and structural heart disease. The logistic regression models used blood glucose categories as a dummy variable using normal blood glucose as a reference and were adjusted for age, gender, race, Charlson score, body mass index, total cholesterol, coronary artery disease, and blood pressure.

The fitness of the data was assessed using the deviance ratio. Analyses were performed using STATA (College Station, TX), and all significance tests were two-tailed.

Results

Baseline characteristics

During our study period, 15,056 patients had a screening echocardiogram and a HbA1c measured within 3 months of the echocardiogram. Of those, 43% met a diabetes diagnosis, 17% had high-risk prediabetes, 28% had low-risk prediabetes, and 12% were normoglycemic. Table 1 shows the trend of increasing age, blood pressure, co-morbidities, weight, and use of ACE inhibitors as HbA1c increased from normal stage to prediabetes and diabetes range.

Table 1.

Baseline Characteristics by Diabetic Status

Characteristic	Normal (<5.7)	Low-risk prediabetes (5.7–6.0)	High-risk prediabetes (6.1–6.4)	Diabetes (>6.4)	P
N	2794	4201	2499	5562
Mean HbA1c	5.3 ± 0.23	5.85 ± 0.10	6.18 ± 0.08	7.98 ± 1.82	<0.01
Mean age	66.6 ± 11.5	68.4 ± 9.7	69.0 ± 8.8	69.0 ± 8.6	<0.01
Female gender, %	57	63	64	61	<0.01
Black, %	51	54	61	57	<0.01
Mean systolic blood pressure	133.9 ± 27.6	135.2 ± 26.1	136.4 ± 24.5	139.4 ± 19.2	<0.01
Mean diastolic blood pressure	75.9 ± 14	75.4 ± 6.4	75.6 ± 12	75.5 ± 9.6	<0.01
Mean body mass index	28.2 ± 6.6	29.8 ± 7.6	30.8 ± 6.8	32.9 ± 7.5	<0.01
Mean Charlson score	1.83 ± 1.45	2.08 ± 1.74	2.38 ± 1.82	3.20 ± 1.97	<0.01
Hypertension, %	55	63	70	75	<0.01
ACE inhibitor use, %	30	36	43	54	<0.01

ACE, angiotensin-converting enzyme; HbA1c, hemoglobin A1c.

Effect of prediabetes on HF

We found 1253 patients with an EF <50%. Of those 11% were normoglycemic, 19% had low-risk prediabetes, 11% had high-risk prediabetes, and 59% had diabetes. The adjusted ORs of HF for low-risk prediabetes, high-risk prediabetes, and diabetes were 1.38 (95% CI 1.07–1.78) (P = 0.01), 1.47 (95% CI 1.05–2.01) (P = 0.01), and 1.60 (95% CI 1.16–2.01) (P < 0.01), respectively, when compared with normoglycemic patients.

We found 1831 patients with HF with preserved EF. Of those 11% were normoglycemic, 18% had low-risk prediabetes, 14% had high-risk prediabetes, and 58% had diabetes. The adjusted OR for low-risk prediabetes, high-risk prediabetes, and diabetes were 1.17 (95% CI 0.86–1.60) (P = 0.30), 1.60 (95% CI 1.15–2.21) (P < 0.01), and 1.63 (95% CI 1.24–2.13) (P < 0.01), respectively, when compared with normoglycemic patients.

Effect of prediabetes on structural heart disease

We found 6008 patients with structural heart disease. Of those with structural heart disease 28% were normoglycemic, 31% had low-risk prediabetes, 32% had high-risk prediabetes, and 37% had diabetes (P < 0.01).

We found an increasing trend in the size of the left atrium and left ventricular hypertrophy and mass index and a decreasing trend in the EF as glycemic status worsened to prediabetes and to diabetes range (P < 0.01) (Table 2).

Table 2.

Structural Heart Disease By Diabetic Status

Diabetic status	Mean atrial size	Mean left ventricular size	Mean ejection fraction	Mean left ventricular mass
Normal	3.58 ± 0.76	1.07 ± 0.20	58.6 ± 6.2	85.5 ± 26.1
Low-risk prediabetes	3.64 ± 0.75	1.08 ± 0.19^*	58.3 ± 6.4^*	86.2 ± 26.0
High-risk prediabetes	3.67 ± 0.74	1.10 ± 0.19	58.3 ± 6.3	87.7 ± 26.6
Diabetes	3.92 ± 0.68^*	1.12 ± 0.20^*	58.3 ± 6.9	89.4 ± 27.4^*

P < 0.01.

The adjusted OR of structural heart disease for low-risk prediabetes, high-risk prediabetes, and diabetes were 1.14 (95% CI 0.99–1.25) (P = 0.05), 1.24 (95% CI 1.05–1.43) (P = 0.01), and 1.46 (95% CI 1.33–1.61) (P < 0.01), respectively, when compared with normoglycemic patients.

Discussion

In a large multisite network of clinics serving older adults, screening echocardiograms and HbA1c levels revealed that most patients with structural heart disease and/or HF had either prediabetes or diabetes. Furthermore, a third of the practice met HbA1c criteria for prediabetes and these patients were more likely to have HF with and without preserved EF, left atrial enlargement, and left ventricular hypertrophy when compared with normoglycemic patients. The association between abnormal glucose metabolism and cardiac end organ damage is supported by the dose–effect relationship seen between increasing HbA1c levels and worsening of echocardiographic parameters. These data suggest that patients with prediabetes are more likely to develop heart abnormalities that could be improved with adequate management of risk factors.

In recent years we have seen growing evidence of a correlation between prediabetes and cardiovascular abnormalities such as diastolic dysfunction, increased left atrial size, or left ventricular mass.⁵ Studies evaluating the possible pathophysiology of this association have found that endothelial dysfunction, inflammation, and oxidative stress are possible mediators between glycogen deposition and heart disease.

Our study contributes to this body of evidence supporting the causal relationship between glucose metabolism and heart disease by demonstrating the dose–effect between the different stages of glycemic control and systolic HF and structural heart disease even when adjusting for traditional risk factors.

Our study has significant strengths. The most relevant is its high internal and external validity. The study included the performance of a standardized screening echocardiogram to all consecutive patients seen at multiple sites across seven states for >2 years. In addition, >70% had a HbA1c measured within 3 months and the cardiologist performing the procedure was blinded to the aims and outcomes of the study. This diminishes the impact of selection and information bias. An additional advantage is having access to all data in the electronic health record facilitating the collection of confounder variables.

Our study has significant public health implications considering that a growing percentage of the U.S. population meets prediabetes criteria based on either HbA1c, fasting glucose, or glucose tolerance test.⁷ Furthermore, there is growing evidence that the increasing prevalence of prediabetes will lead to excess cardiovascular morbidity and mortality.² Our study suggests that a screening echocardiogram can detect early changes in heart structure and function that can be used to tailor and monitor the therapeutic management of prediabetes and other cardiovascular risk factors commonly associated with the prediabetes condition. Although, in our study, the yield of performing screening HbA1c and echocardiograms was high (a third of all patients had prediabetes and a third of those had structural heart disease), the long-term cost-effectiveness of such an approach is not known and should be the focus of future studies.

Our study has several limitations that need to be mentioned. First, we determined diabetic state using a combination of ICD-9 codes, medications and HbA1c. By using ICD-9 codes we could have introduced information bias. Also, we did not use fasting glucose and oral glucose tolerance tests, which could have given more information regarding glucose tolerance and the effect on structural heart disease. Second, even though we included hypertension, diabetes, and coronary artery disease as predictors of left ventricular hypertrophy we did not have complete information on other predictors of structural heart disease such as valvular disease.

Third, 26% of all the patients who had a screening echocardiogram did not have an HbA1c value within 3 months of the procedure. This may introduce information bias into the study. However, we expect this risk to be small considering the small proportion of patients who did not have HbA1c data. A fourth limitation of our study was that, despite our multistate patient population was racially diverse, a large proportion of patients belonged to a racial/ethnic minority or were economically disadvantaged, somewhat limiting the generalizability of our findings to nonminority populations.

In conclusion, among racially diverse patients >65 years of age, prediabetes is common and associated to HF with and without preserved EF as well as to other structural cardiac abnormalities. Future studies should evaluate the role that early detection of structural heart disease could have on optimizing clinical management and on long-term cardiovascular outcomes.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

References

Huang

, Zhao

, Tang

, et al. Cardiac involvement in patients recovered from COVID-2019 identified using magnetic resonance imaging. JACC Cardiovasc Imaging, 2020; 13:2330–2339.

Selvin

, Lazo

, Chen

, et al. Diabetes mellitus, prediabetes, and incidence of subclinical myocardial damage. Circulation, 2014; 130:1374–1382.

Skali

, Shah

, Gupta

, et al. Cardiac structure and function across the glycemic spectrum in elderly men and women free of prevalent heart disease: The Atherosclerosis Risk in the Community study. Circ Heart Fail, 2015; 8:448–454.

Demmer

, Allison

, Cai

, et al. Association of impaired glucose regulation and insulin resistance with cardiac structure and function: Results from ECHO-SOL (Echocardiographic Study of Latinos). Circ Cardiovasc Imaging, 2016; 9. [Epub ahead of print]; DOI: 10.1161/CIRCIMAGING.116.005032.

Di Pino

, Mangiafico

, Urbano

, et al. HbA1c identifies subjects with prediabetes and subclinical left ventricular diastolic dysfunction. J Clin Endocrinol Metab, 2017; 102:3756–3764.

Liang

, Feng

, Liu

, et al. Effect of impaired glucose tolerance on cardiac dysfunction in a rat model of prediabetes. Chin Med J (Engl), 2011; 124:734–739.

Bullard

, Saydah

, Imperatore

, et al. Secular changes in U.S. Prediabetes prevalence defined by hemoglobin A1c and fasting plasma glucose: National Health and Nutrition Examination Surveys, 1999–2010. Diabetes Care, 2013; 36:2286–2293.

Tanio

, Chen

. Innovations at Miami practice show promise for treating high-risk Medicare patients. Health Aff (Millwood), 2013; 32:1078–1082.

Gottdiener

, Bednarz

, Devereux

, et al. American Society of Echocardiography recommendations for use of echocardiography in clinical trials. J Am Soc Echocardiogr, 2004; 17:1086–1119.

10.

Birman-Deych

, Waterman

, Yan

, et al. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care, 2005; 43:480–485.

11.

Albright

, Gregg

. Preventing type 2 diabetes in communities across the U.S.: The National Diabetes Prevention Program. Am J Prev Med, 2013; 44(4 Suppl 4):S346–S351.

12.

Corona

, Rastrelli

, Silverii

, et al. The identification of prediabetes condition with ARIC algorithm predicts long-term CV events in patients with erectile dysfunction. J Sex Med, 2013; 10:1114–1123.

13.

McManus

, Hsu

, Sung

, et al. Atrial fibrillation and outcomes in heart failure with preserved versus reduced left ventricular ejection fraction. J Am Heart Assoc, 2013; 2:e005694.

14.

Redfield

, Jacobsen

, Burnett

, et al. Burden of systolic and diastolic ventricular dysfunction in the community: Appreciating the scope of the heart failure epidemic. JAMA, 2003; 289:194–202.

15.

Sharabiani

, Aylin

, Bottle

. Systematic review of comorbidity indices for administrative data. Med Care, 2012; 50:1109–1118.