Wong TY, Rosamond W, Chang PP, Couper DJ, Sharrett AR, Hubbard LD, Folsom AR, Klein R. Retinopathy and Risk of Congestive Heart Failure. JAMA. 2005;293(1):63–69. doi:10.1001/jama.293.1.63
Author Affiliations: Centre for Eye Research
Australia, University of Melbourne, Victoria (Dr Wong); Singapore Eye Research
Institute, National University of Singapore (Dr Wong); Department of Epidemiology
(Drs Rosamond and Chang), Division of Cardiology (Dr Chang), and Department
of Biostatistics (Dr Couper), University of North Carolina, Chapel Hill; Department
of Epidemiology, Johns Hopkins University, Baltimore, Md (Dr Sharrett); Department
of Ophthalmology, University of Wisconsin, Madison (Mr Hubbard and Dr Klein);
and Division of Epidemiology, University of Minnesota, Minneapolis (Dr Folsom).
Context Congestive heart failure (CHF) affects a substantial proportion of adults
including those without preexisting coronary heart disease. The pathogenesis
of CHF is uncertain, but microvascular disease has been hypothesized as a
Objective To determine the relationship of retinopathy, a marker of systemic microvascular
disease, to risk of CHF.
Design, Setting, and Participants Population-based, prospective 7-year cohort study in 4 US communities
using the Atherosclerosis Risk in Communities Study database. Participants
(n = 11 612, aged 49 to 73 years) had retinal photographs taken
between 1993 and 1995. The photographs were graded according to a standardized
protocol for the presence of retinopathy (eg, microaneurysms, retinal hemorrhages,
soft exudates), arteriovenous nicking, focal arteriolar narrowing, and generalized
Main Outcome Measures Association between retinopathy and incident CHF, identified from hospitalization
and death records.
Results The 7-year cumulative incidence of CHF was 5.4% (492 events). Participants
with retinopathy had a higher incidence of CHF compared with those without
retinopathy (15.1% vs 4.8%, P<.001). After controlling
for age, sex, race, preexisting coronary heart disease, mean arterial blood
pressure, diabetes, glucose level, cholesterol level, smoking, body mass index,
and study site, the presence of retinopathy was associated with a 2-fold higher
risk of CHF (relative risk, 1.96; 95% confidence interval, 1.51-2.54). Among
participants without preexisting coronary heart disease, diabetes, or hypertension,
retinopathy was associated with a 3-fold higher risk of CHF (relative risk,
2.98; 95% confidence interval, 1.50-5.92).
Conclusions Retinopathy is an independent predictor of CHF, even in persons without
preexisting coronary heart disease, diabetes, or hypertension. This suggests
that microvascular disease may play an important role in the development of
heart failure in the general population. Some asymptomatic persons with retinopathy
on an ophthalmologic examination may benefit from further assessment of CHF
Congestive heart failure (CHF) is a leading cause of morbidity, hospitalization,
and mortality in the United States.1 Despite
recent advances in research, the pathogenesis of CHF remains uncertain.2 Microvascular disease has long been suggested as a
possible pathogenic factor in the development of different cardiac diseases.3 There is increasing evidence that coronary microvascular
dysfunction may be one underlying mechanism in patients with symptoms and
signs of myocardial ischemia without angiographically detectable coronary
artery disease, and even in asymptomatic patients with cardiovascular risk
factors.4,5 Microvascular processes
have also been implicated in left ventricular dysfunction and subsequent heart
failure,6,7 particularly in people
with diabetes8,9 and hypertension,10- 12 as well as in patients
with dilated13- 18 or
hypertrophic19 cardiomyopathy. However, the
majority of these studies have been cross-sectional and have focused on small
samples of highly selected patients. In addition, since the coronary circulation
cannot be visualized in vivo, what is known about microvascular mechanisms
in the pathogenesis of CHF has been derived indirectly from studies of functional
parameters such as myocardial blood flow and coronary flow reserve.20,21 We are unaware of prospective or
population-based studies examining whether microvascular disease predicts
risk of CHF in the general community.
The retinal circulation offers an opportunity to noninvasively explore
the relationship of systemic microvascular disease to cardiovascular disorders.
Signs of retinopathy (eg, microaneurysms, retinal hemorrhages, and soft exudates
[cotton-wool spots]) are structural markers of microvascular damage from elevated
blood pressure, diabetes, and other processes,22 and
narrowed retinal arterioles have been found to predict incident coronary heart
disease in women.23 In this study, we examined
the relationship of retinopathy signs to risk of CHF in a population-based
cohort of healthy, middle-aged men and women.
The Atherosclerosis Risk in Communities (ARIC) study is a population-based
cohort study of cardiovascular disease and its risk factors.24 The
original cohort included 15 792 participants aged 45 to 64 years selected
by probability sampling from 4 US communities: Forsyth County, North Carolina;
Jackson, Miss; suburbs of Minneapolis, Minn; and Washington County, Maryland.
The Jackson sample included African Americans only; in the other field centers,
samples were representative of the populations in these communities (mostly
white in Minneapolis and Washington County, and about 15% African American
in Forsyth County). Race was determined by self-report with fixed categories
(white, African American, American Indian, Asian/Pacific Islander). While
Hispanic ethnic status was ascertained, it was not used in this analysis due
to insufficient numbers.
Initial participation rates were 46% in Jackson and approximately 65%
in the other communities. Differences between participants and nonparticipants
at the baseline examination have been presented elsewhere.25 White
participants generally reported a higher socioeconomic status, better general
health, and a lower prevalence of cardiovascular disease and associated risk
factors than did nonparticipants. Differences in these characteristics between
African American participants and nonparticipants were of smaller magnitude
or absent, particularly among African American women.
Of the 15 792 participants at baseline, 14 346 (93% of survivors)
returned for a second examination 3 years later in 1990-1992, and 12 887
(86% of survivors) returned for a third examination 3 years after the second
in 1993-1995. At each study visit, participants underwent an interview assessing
demographic characteristics and medical history, a brief clinical examination
including measurement of blood pressure and weight, and a set of laboratory
tests including levels of blood lipids and serum glucose.
We first obtained retinal photographs at the third ARIC study examination.26 Characteristics of participants with and without
gradable retinal photographs have been previously reported.26 Individuals
with gradable photographs were younger and more likely to be white, but did
not differ in sex or smoking status from participants with ungradable photographs.
Of the 12 887 participants who returned for the third examination, we
excluded 245 who had no retinal photographs, 738 with ungradable photographs,
26 with retinal vascular occlusions, 186 hospitalized for prevalent CHF (as
defined below), 38 whose race was neither African American nor white (due
to small numbers in other racial groups), and 42 African American residents
in Minneapolis and Washington County (so that race and field center could
be adjusted in 5 categories: Jackson, Forsyth whites, Forsyth African Americans,
Minneapolis, and Washington County), leaving 11 612 (aged 49 to 73 years)
who provided data for this study.
Institutional review boards at each study site approved the study. Written
informed consent was obtained from all participants at each examination.
The procedures for retinal photography and the assessment of photographs
are described in detail elsewhere.26 Briefly,
we took photographs of the retina from 1 randomly selected eye after 5 minutes
of dark adaptation. Trained graders masked to all participant characteristics
used a standardized protocol to evaluate the photographic slides for microvascular
signs.26 We recorded 4 categories of retinal
microvascular signs: retinopathy, arteriovenous nicking, focal arteriolar
narrowing, and generalized arteriolar narrowing. Retinopathy was defined as
present if any of the following lesions were graded definite or probable:
blot hemorrhages, flame-shaped hemorrhages, microaneurysms, soft exudates
(cotton-wool spots), hard exudates, macular edema, intraretinal microvascular
abnormalities, and other less-common lesions. Arteriovenous nicking and focal
arteriolar narrowing were defined as present if graded definite or probable.
To estimate the generalized retinal arteriolar narrowing, photographs were
digitized and diameters of individual retinal vessels coursing through a specified
area were measured using a computer and summarized as the arteriole-to-venule
ratio.26 A smaller ratio represents narrower
arterioles (since venule diameters vary little), and generalized arteriolar
narrowing was defined as the lowest quintile of the sample distribution of
the arteriole-to-venule ratio.26 Intragrader
and intergrader κ statistics for various signs of retinopathy ranged
from 0.61 to 1.00.26 For arteriole-to-venule
ratio, intragrader and intergrader reliability coefficients were 0.84 and
We assessed CHF events by contacting participants annually by telephone
to identify self-report of all hospitalizations during the previous year,
by ongoing surveillance of community hospital discharges, and by surveying
death certificates from state vital statistics. Medical records from all hospitalizations
were obtained and discharge codes investigated by trained medical record abstractors.
A hospitalization was considered a CHF event if it contained a hospital discharge
diagnosis code of CHF (International Classification of Diseases,
9th Revision, Clinical Modification [ICD-9-CM]
code 428 or 518.4). Death certificates with an underlying cause of death coded
as heart failure (ICD-9-CM code 428 or ICD-10 code I50) and not identified through the above process were
also included as events. Because retinal photography was performed at the
third examination, incident CHF was defined as new hospitalizations for CHF
or death from CHF subsequent to this examination. Prevalent CHF cases that
occurred up to and including the third examination (n = 186) were
excluded from this study.
We evaluated all participants for cardiovascular risk factors at each
examination.27 Ascertainment and definition
of coronary heart disease in the ARIC study followed a standardized protocol.28 In this study, patients were defined as having preexisting
coronary heart disease if they had had acute myocardial infarction, silent
infarction, or coronary revascularization procedures (eg, coronary angioplasty)
in the period up to and including the third examination. A history of angina
or use of antianginal medications in the absence of the above events was not
considered a criterion for coronary heart disease.
Blood pressure was measured with a random-zero sphygmomanometer according
to a standardized protocol.27 The mean arterial
blood pressure was computed as two thirds of the diastolic value plus one
third of the systolic value. We used the average of this over the first 3
examinations (6-year mean arterial blood pressure) to assess blood pressure–independent
association of retinal signs and incident CHF. Hypertension was defined as
systolic blood pressure of 140 mm Hg or greater, diastolic blood pressure
of 90 mm Hg or greater, or use of antihypertensive medication during the previous
2 weeks. Diabetes mellitus was defined as a fasting glucose level of 126 mg/dL
(7.0 mmol/L) or greater, a nonfasting glucose level of 200 mg/dL (11.1 mmol/L)
or greater, or a self-reported history of or treatment for diabetes. Height
and weight were measured with participants wearing scrub suits, and body mass
index (BMI) was calculated as weight in kilograms divided by the square of
height in meters. Blood collection and processing for levels of total cholesterol,
low-density lipoprotein cholesterol, triglycerides, and glucose are described
elsewhere.27 Education, cigarette smoking,
and alcohol consumption status were ascertained from interview. All covariates
were based on data collected at the third examination, except for education
(first examination) and 6-year mean arterial blood pressure (all 3 examinations).
We estimated the 7-year cumulative incidence of CHF (defined as 100 × [1 – Kaplan-Meier
cumulative CHF-free survival at 7 years]) according to presence or absence
of retinal microvascular signs. Follow-up time was defined as the number of
days from the third examination visit to the date of the first CHF hospitalization,
death, last contact, or December 31, 2000. The relative risk (RR) for CHF
associated with presence of retinal signs was derived from Cox proportional
hazards regression. We initially adjusted for age, sex, race, and examination
center. In multivariable analyses, we further adjusted for education (up to
and including grade school, high school without diploma, high school graduate,
vocational school, college graduate, graduate school, or professional school),
preexisting coronary heart disease, 6-year mean arterial blood pressure, use
of antihypertensive medications, diabetes, cigarette smoking status (never,
current, past), glucose level, low-density lipoprotein cholesterol level,
and BMI. Analyses were repeated stratifying the population by sex, race, preexisting
coronary heart disease, diabetes, and hypertension. We also formally tested
for interaction by adding cross-product terms of these variables into the
models of the total population.
Because of the strong associations of CHF with preexisting coronary
heart disease, and of retinopathy with diabetes and hypertension, we also
performed 3 post hoc analyses. We repeated the analyses excluding participants
with preexisting coronary heart disease at the third examination (n = 396),
stratified according to diabetes and hypertension status; performed an analysis
excluding participants who first developed incident coronary heart disease
and subsequently developed CHF; and used a standard formula29 to
estimate the population-attributable fraction for CHF associated with retinopathy.
Power calculations were performed to determine the lowest RR detectable
for these associations. For the whole cohort, we had approximately 80% power
to detect an RR of 1.4 and 90% power to detect an RR of 1.46.
All analyses were performed with SPSS version 10.0 (SPSS Inc, Chicago,
Ill) and P<.05 was set as the threshold for statistical
There were 719 persons with retinopathy at baseline and 492 incident
CHF events over a mean follow-up of 6.2 years (cumulative incidence, 5.4%).
Of the 492 cases, 20 (4.1%) were based on death certificate data. In general,
persons with retinopathy were older and more likely to be African American;
to have a history of coronary heart disease, hypertension, or diabetes; to
have higher systolic and diastolic blood pressure, glucose level, triglycerides
level, and BMI; and to report less alcohol use (Table 1). Persons who had an incident CHF event were older and more
likely to be men; to be current cigarette smokers; to have a history of coronary
heart disease, hypertension, or diabetes; to have higher systolic blood pressure,
glucose level, triglycerides level, and BMI; and to report less alcohol use.
The 7-year incidence of CHF was higher in persons with retinal microvascular
lesions compared with those without the retinal lesions (15.1% vs 4.8%, P<.001) (Table 2).
After controlling for age, sex, race, and field center, the RR of CHF ranged
from 1.20 (95% confidence interval [CI], 0.95-1.52) for focal arteriolar narrowing
to 4.02 (95% CI, 2.75-5.86) for soft exudates. After further adjustment for
prevalent coronary heart disease, mean arterial blood pressure, diabetes,
glucose level, and other risk factors, retinopathy remained independently
related to incident CHF (any retinopathy: RR, 1.96 [95% CI, 1.51-2.54]; microaneurysms:
RR, 2.20 [95% CI, 1.61-3.01]; retinal hemorrhages: RR, 1.89 [95% CI, 1.37-2.60];
and soft exudates: RR, 1.87 [95% CI, 1.22-2.84]). The associations for arteriovenous
nicking and for focal and generalized arteriolar narrowing were not statistically
significant after multivariable adjustment.
The pattern of association of retinopathy and CHF was similar in men
and women (RR, 1.85 [95% CI, 1.31-2.61] vs 2.10 [1.40-3.15]), in whites and
African Americans (1.76 [1.26-2.44] vs 2.28 [1.47-3.53]), and in persons with
and without preexisting coronary heart disease (1.57 [0.81-3.06] vs 2.01 [1.51-2.68]).
Interaction terms for these variables were not statistically significant (P>.20 in all models).
There were significant associations between retinopathy and incident
CHF in persons without preexisting coronary heart disease, stratified according
to diabetes and hypertension status (Table 3).
The strongest association was seen among diabetic persons without hypertension
(RR, 4.32; 95% CI, 2.13-8.76), although there was also a substantial association
in persons without either diabetes or hypertension (RR, 2.98; 95% CI, 1.50-5.92).
However, unlike diabetes, hypertension alone did not contribute to the increased
risk of CHF in patients with retinopathy. The population-attributable fraction
was highest (30.5%) in the subgroup with diabetes but no hypertension.
Finally, we conducted an additional multivariable analysis excluding
participants with preexisting coronary heart disease and participants who
developed incident coronary heart disease and subsequently developed CHF.
The results of this analysis were essentially similar (RR for retinopathy,
2.21; 95% CI, 1.57-3.10).
In this prospective study of middle-aged persons, we found an association
between retinal microvascular signs, as quantified from photography, and 7-year
risk of CHF. Participants with signs of retinopathy (eg, microaneurysms, retinal
hemorrhages, and soft exudates) were twice as likely to develop CHF as those
without these signs, even in models that controlled for preexisting coronary
heart disease, long-term blood pressure levels, diabetes, cigarette smoking,
and other risk factors. This association was present in men and women and
in whites and African Americans. In lower-risk participants without preexisting
coronary heart disease, diabetes, or hypertension, the presence of retinopathy
was associated with a 3-fold higher risk of CHF. Weaker associations with
incident CHF were seen for other retinal arteriolar signs, and these associations
were not significant in multivariable-adjusted models.
Retinal microvascular signs examined here represent small-vessel damage
associated with increased age, hypertension, and diabetes.22 We
have previously shown that these retinal signs are also associated with systemic
markers of inflammation and endothelial dysfunction in persons without diabetes,
independent of blood pressure.30 Thus, the
association between retinopathy and incident CHF supports a role of systemic
microvascular disease in the development of heart failure, possibly related
to a combination of elevated blood pressure, hyperglycemia, inflammation,
and endothelial dysfunction. Our findings in this generally healthy, middle-aged
community population extend previous smaller studies that have implicated
microvascular mechanisms in CHF pathogenesis in subgroups with diabetic heart
heart disease,10- 12,33,34 and
nonischemic dilated cardiomyopathy.13- 18
It is important to note that there is not a perfect correlation of microvascular
disease in the retina with systemic or coronary microvascular disease. Although
some of the histopathological features of retinopathy signs (arteriolar narrowing,
intimal thickening, medial hyalination, and capillary occlusion and leakage)22 are also seen in histological studies of patients
with coronary microvascular disease and hypertension,35,36 experimental
studies indicate significant differences in the response to acute ischemia
of retinal and myocardial microcirculation, possibly related to differing
contractile properties of the capillary walls.37 In
addition, not all retinal microvascular signs were strongly associated with
CHF. In contrast to retinopathy, the associations for retinal arteriolar narrowing
and arteriovenous nicking were weaker and nonsignificant after multivariable
adjustment. The heterogeneity of these associations may reflect different
pathophysiological processes associated with specific retinal microvascular
Nevertheless, comparison of the retinopathy associations in persons
with diabetes and hypertension provides further insights into the relative
importance of microvascular disease mechanisms to the risk of CHF.33,34 We found a higher RR of CHF and a
greater population-attributable fraction associated with retinopathy in persons
with diabetes than in those with hypertension. Among persons with diabetes
but without preexisting coronary heart disease and hypertension, the population-attributable
fraction for CHF associated with retinopathy was 30.5%, which suggests that
nearly a third of heart failure cases in these patients are related to microangiopathic
processes as evident by a retinal examination. These results therefore support
the concept of a systemic microvascular effect on the myocardium in diabetes
beyond the indirect effects of accelerated atherosclerosis, concomitant coronary
heart disease, and hypertension.38 In contrast,
among persons with hypertension but without diabetes, the population-attributable
fraction was only 1.4%, suggesting that only a small proportion of CHF events
in patients with hypertension may be related to microvascular mechanisms.
The strengths of our study include a large sample size with participants
drawn from the general population rather than a specialized sample, the objective
evaluation of retinal microvascular signs, and the standardized identification
of incident CHF events. Certain limitations of this study need to be considered.
First, CHF events were defined from hospitalization or death records. Self-report
may result in underreporting of hospitalization, and milder CHF cases that
did not require hospitalization were not included. The latter may limit the
applicability of our results to a wider spectrum of CHF cases not requiring
inpatient treatment. Additionally, data derived from death certificates may
have limited accuracy and it is possible that some CHF events may have been
misclassified, although less than 5% of CHF events in this analysis were based
on death certificate data. Second, retinal photography data were available
from only 1 eye for each patient, and some of the photographs were ungradeable
because of media opacity or poor pupil dilation, so that retinopathy status
may have been misclassified. However, we have no reason to expect a biased
presence of misclassification from these 2 factors. Third, although we controlled
for hypertension and diabetes, it is possible that residual confounding from
these conditions and other unmeasured CHF risk factors (such as valvular heart
disease) could have affected these associations. Furthermore, in analyses
among participants without preexisting coronary heart disease, a history of
angina and the use of antianginal medications were not used as criteria for
exclusion; thus, milder forms of coronary heart disease could be a confounder.
Some signs of retinopathy can be caused by factors other than microvascular
disease such as anemia and carotid artery disease, and it is possible that
these factors could also play a role in the development of CHF independent
of their effect on the microvasculature. Finally, diabetes can lead to both
small- and large-vessel disease and could thus promote the development of
retinopathy and CHF by separate pathways.
Our study has 2 potential clinical applications. First, our findings
support the value of specifically targeting the microcirculation in trying
to reduce cardiovascular morbidity and mortality, particularly in persons
with diabetes.39 The UK Prospective Diabetes
Study has already demonstrated the efficacy of aggressive glycemic control
in reducing microvascular events (retinopathy, nephropathy) as well as risk
of CHF in type 2 diabetes.40 There is increasing
evidence that an antihypertensive agent such as an angiotensin-converting
enzyme inhibitor may have direct beneficial effects on microvessel structure
and function beyond its primary effect on lowering blood pressure.41,42 Such agents may therefore have added
therapeutic value in preventing and treating CHF. Second, these findings raise
the possibility that a retinal examination may improve risk prediction for
CHF. In particular, patients with diabetes and signs of retinopathy may benefit
from further assessment of CHF risk, such as echocardiography to detect asymptomatic
left ventricular dysfunction, if clinically indicated.
In conclusion, we demonstrate a prospective association of retinopathy
and risk of CHF hospitalization or deaths in healthy, middle-aged people in
the community, even in those without preexisting coronary heart disease, diabetes,
or hypertension. Our study suggests that microvascular disease may play an
important role in the pathogenesis of heart failure and that some asymptomatic
persons with retinopathy on an ophthalmologic examination may benefit from
further assessment of CHF risk.
Corresponding Author: Tien Y. Wong, MD,
PhD, Centre for Eye Research Australia, University of Melbourne, 32 Gisborne
St, Victoria 3002, Australia (email@example.com).
Author Contributions: Dr Wong had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analyses.
Study concept and design: Wong, Hubbard.
Acquisition of data: Rosamond, Couper, Hubbard,
Analysis and interpretation of data: Wong,
Rosamond, Chang, Couper, Sharrett, Hubbard.
Drafting of the manuscript: Wong.
Critical revision of the manuscript for important
intellectual content: Wong, Rosamond, Chang, Couper, Sharrett, Hubbard,
Statistical analysis: Wong, Rosamond, Couper.
Obtained funding: Hubbard.
Administrative, technical, or material support:
Rosamond, Hubbard, Klein.
Study supervision: Hubbard.
Funding/Support: This study was supported by
contracts N01-HC-35125, N01-HC-35126, N01-HC-55015, N01-HC-55016, N01-HC-55018,
N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022 from the National
Heart, Lung, and Blood Institute. Additional support was provided by the Sylvia
and Charles Viertel Clinical Investigator Award and the Biomedical Research
Council Singapore (Dr Wong).
Role of the Sponsors: The organizations funding
this study had no role in the design and conduct of this study; the analysis
and interpretation of data; or the preparation, review, or approval of the
Acknowledgment: We thank the staff and participants
in the ARIC study for their important contributions.