Epidemiological Association Between Systemic Diseases and Age-Related Macular Degeneration: The Korea National Health and Nutrition Examination Survey 2008–2011

Abstract

Purpose.: We examined the epidemiological association between systemic diseases and age-related macular degeneration (AMD) in the general Korean population.

Methods.: This cross-sectional study involved nationally representative data obtained from the 2008 to 2011 Korea National Health and Nutrition Examination Surveys. A total of 14,352 subjects aged ≥40 years participated in standardized health interviews regarding physician-diagnosis of several systemic diseases as well as physical examinations, including fundus photography for the evaluation of AMD.

Results.: The overall prevalence rates of early, late, and any AMD were 6.0%, 0.6%, and 6.6%, respectively. In univariate logistic regression analyses adjusted for age and sex as well as smoking in late AMD, any AMD and late AMD were less prevalent among diabetic patients and more prevalent in participants with a history of liver cancer. A history of liver cirrhosis was associated with a higher prevalence of any AMD. In the final multivariate model, the associated factors for any AMD included age (odds ratio [OR], 1.09), the presence of diabetes mellitus (DM; OR, 0.74), and a history of liver cancer (OR, 4.32). Factors associated with late AMD included age (OR, 1.09), ever-smoking history (OR, 2.45), the presence of DM (OR, 0.22), and a history of liver cancer (OR, 12.51). The presence of diabetic retinopathy was associated with a lower prevalence of any AMD (OR, 0.35).

Conclusions.: When adjusted for confounders, any AMD and late AMD were less prevalent in diabetic patients. In contrast, a history of liver cancer was associated with a higher prevalence of any AMD and late AMD.

Introduction

Age-related macular degeneration (AMD) is the leading cause of visual impairment among the elderly population in developed countries. ^1,2 The pathophysiology of AMD is yet to be unraveled, but several risk factors for the development of AMD have been identified, including smoking, hyperopia, and genetic variation in complement factor H. ^3–5 Revealing risk factors will provide valuable information to better understand AMD pathogenesis. 

Among systemic diseases, several medical conditions occasionally have been associated with risk of AMD. ^6,7 Epidemiological studies have reported hypertension, some cardiovascular diseases, and diabetes mellitus (DM) to be associated with increased AMD prevalence ^7–9 However, other studies have shown inconsistent results for these diseases, ^4,10 and have even reported an inverse association of DM with AMD. ¹¹ The association of other systemic diseases with AMD remains undisclosed due to insufficient data. Unveiling the relationship between systemic diseases and AMD would contribute to increased understanding of the pathophysiology of AMD, allow screening of patients at risk, and ultimately may help prevent AMD. 

In the present study, we investigated the association between various systemic diseases and AMD in the Korean general population. To acquire a nationally representative sample, we used data obtained from the Korea National Health and Nutrition Examination Survey (KNHANES) on behalf of the Korean Ophthalmological Society (KOS). The ethnic homogeneity of Korea is advantageous in identifying disease risk factors, because it would minimize bias resulting from interracial differences. ¹² Consequently, several systemic diseases including cardiovascular, endocrinologic, pulmonary, and arthritic diseases, as well as malignancy, were examined for association with AMD in a large-numbered study population. To our knowledge, this is the first comprehensive study on this association in Asian ethnics. 

Methods

Study Population

The data analyzed in this study were acquired from the fourth and fifth KNHANES performed from July 2008 to December 2011. The KNHANES is an ongoing nationwide cross-sectional survey that examines the health and nutritional status of the noninstitutionalized civilian South Korean population. ¹³ The survey has been conducted annually since 2007 by the Korea Center for Disease Control and Prevention (KCDC), and ophthalmologic examination was included beginning in the second half of 2008. Details of the KNHANES design and methodology have been described previously. ^12,13 To summarize briefly, the KNHANES includes approximately 4000 households annually based on national census data, using a stratified multistage cluster sampling method, which ensured that annual survey results represented the general Korean population and that the results from each year could be merged. ¹² All family members aged ≥1 year in the sampled households were included as eligible subjects. The study described here adhered to the tenets of the Declaration of Helsinki, and written informed consent was obtained from all participants. The survey protocol was approved by the Institutional Review Board (IRB) of the KCDC (IRB No. 2008-04EXP-01-C, 2009-01CON-03-2C, 2010-02CON-21-C, 2011-02CON-06-C). 

Subjects were asked to participate in health interviews and examinations performed by trained teams in mobile centers. Demographic variables, current and past medical conditions, and health behavior patterns were obtained using detailed standardized questionnaires in health interview surveys. Physical examinations involved body profile measurement, blood tests, routine urinalysis, and ophthalmologic examinations, including a visual acuity test, applanation tonometry, and slit-lamp biomicroscopy. A 45° nonmydriatic color fundus photograph was taken for each eye in all participants aged ≥19 years using a digital fundus camera (TRC-NW6S; Topcon, Tokyo, Japan). If the participant had a history of DM, random blood glucose level ≥200 mg/dL, or suspected diabetic retinopathy (DR) in the nonmydriatic fundus photograph, seven standard fundus photographs were taken according to the Early Treatment for Diabetic Retinopathy Study (ETDRS) protocol after pharmacologic pupillary dilatation. Ultimately, participants ≥40 years of age with ≥1 evaluable fundus photographs were included in this study. 

Evaluation of Systemic Diseases

The status of various systemic diseases was investigated using self-reported questionnaires regarding physician diagnosis in health interview surveys. Study participants were asked twice about systemic disease status: once for the existence of the disease and once for physician confirmation of the diagnosis. For example, the first question for stroke was, “Have you ever suffered from stroke?” The second question was, “Was the diagnosis of stroke confirmed by a doctor?” Only participants who answered in the affirmative to both questions were considered to have the systemic disease. Nonspecific diseases, such as “thyroid disease” or “dyslipidemia” were excluded from analyses in this study. 

In cases of DM and hypertension, multiple diagnostic criteria were combined to evaluate the disease status. The presence of DM was defined by any of the following criteria: a self-reported history of physician-diagnosed DM, a self-reported history of DM and current treatment for DM using insulin or oral hypoglycemic agent, and a fasting plasma glucose level measured by blood test of ≥126 mg/dL (7.0 mmol/L). Hypertension was defined by either a self-reported history of hypertension diagnosis and current usage of antihypertensive drugs, or a corrected measured systolic blood pressure of ≥140 mm Hg or corrected measured diastolic blood pressure of ≥90 mm Hg. 

Evaluation of AMD and DR

Each fundus image was graded twice for the presence and type of AMD. Preliminary grading was done onsite by ophthalmologists trained by the National Epidemiologic Survey Committee of the KOS using the International Age-related Maculopathy Epidemiological Study Group grading system. ^12,14 Nine retinal specialists with expertise in the grading of AMD, who were masked to the patients' characteristics, performed detailed grading for their own image subsets mutually exclusive of one another. An independent specialist (JPS) resolved any discrepancies between the preliminary and detailed grading. The presence of any AMD was defined as having either early AMD or late AMD. When one eye was not assessable, the subject was assigned the grade of the other eye. When the severity of AMD differed between both eyes, the subject was assigned the more advanced grade. The quality of grading was verified by the KOS. Grading agreement between the preliminary graders and the standard reading specialists ranged from 90.2% to 95.3%. In addition, the presence of DR among diabetic patients was evaluated in seven mydriatic standard fundus photographs according to the ETDRS protocol and was analyzed for association with AMD. 

Statistical Analysis

All estimates were acquired using sample weights adjusted for response rate, extraction rate, and distribution in the Korean population. Continuous variables were expressed as mean ± SE or mean with 95% confidence intervals (CIs). Categorical and continuous variables were examined using odds ratios (ORs) with 95% CIs. 

Before main analyses, the covariates for the presence of any and late AMD were identified by age- and sex-adjusted univariate logistic regression analyses among demographic risk factors and health-behavioral variables. Next, after adjusting for age, sex, and identified covariates, univariate logistic regression analysis was performed for each systemic disease to select candidate diseases with a threshold P value of <0.1 on the existence with AMD. Variance inflation factors (VIF) were calculated to adjust for potential multicollinearity between the candidate diseases. Those diseases with VIF ≥ 5 were excluded from subsequent analyses. Afterwards, multivariate logistic regression was used to examine the association of each type of AMD with several candidate diseases. A final model for the presence of any AMD or late AMD was built with this final set of risk diseases. Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 20.0 (IBM Corp., Armonk, NY, USA). 

Results

From 2008 to 2011, among 41,016 eligible subjects, 31,646 participated in the interview and physical examination (77.2% response rate) in the KNHANES (see Figure). Of these participants, 30,401 (84.6%) underwent ophthalmologic assessment, and 16,109 (53.0%) of them were aged ≥40 years. Among them, 14,352 (89.1%) had an evaluable fundus image of at least one eye, and were included in this study. 

Figure

 

Participation flowchart for the KNHANES 2008 to 2011.

Demographics

The average age of the study participants was 54.9 ± 0.2 years (range, 40–97 years), and 48.0% (95% CI, 47.2–48.8) were men. The overall prevalence of early, late, and any AMD was 6.0% (95% CI, 5.6–6.5), 0.6% (95% CI, 0.5–0.8), and 6.6% (95% CI, 6.2–7.1), respectively. Demographic data, including a history or presence of several systemic diseases in the target population, are shown in Table 1. 

Table 1

 

Demographics of Participants Aged ≥40 Years According to the Presenting Type of AMD in the KNHANES 2008 to 2011

	No AMD, n = 13223		Any AMD, n = 1129			Late AMD, n = 95
Risk Factors		%*		%*	P Value†		%*	P Value†
Age, y	54.3 ± 0.2		64.7 ± 0.4			65.0 ± 1.7
Male sex	5670	48.3 ± 0.4	486	44.3 ± 1.8	0.034	58	68.0 ± 5.8	0.001
Ever-smoker‡	5107	43.8 ± 0.5	432	41.1 ± 1.8	0.168	57	66.6 ± 5.9	<0.001
Current smoker	3702	32.1 ± 0.5	288	27.4 ± 1.7	0.012	39	44.3 ± 6.4	0.035
Current alcohol user§	6381	54.0 ± 0.5	437	42.0 ± 1.8	<0.001	45	51.3 ± 6.5	0.759
DM	1666	12.6 ± 0.4	142	14.1 ± 1.3	0.250	5	4.8 ± 2.5	0.041
Hypertension	4838	34.1 ± 0.5	561	48.5 ± 1.9	<0.001	55	52.5 ± 6.5	0.004
Angina	311	1.8 ± 0.1	39	3.2 ± 0.7	0.008	3	5.6 ± 3.2	0.052
Myocardial infarction	130	0.9 ± 0.1	20	2.0 ± 0.5	0.001	2	1.5 ± 1.1	0.534
Stroke	333	1.9 ± 0.1	44	3.7 ± 0.7	0.001	4	3.9 ± 2.0	0.209
Renal failure	53	0.3 ± 0.1	3	0.2 ± 0.2	0.624	1	0.9 ± 0.9	0.326
Asthma	460	3.0 ± 0.2	57	4.5 ± 0.7	0.017	7	4.9 ± 1.9	0.243
Osteoarthritis	2014	12.3 ± 0.3	263	23.3 ± 1.6	<0.001	17	13.9 ± 3.7	0.813
Rheumatoid arthritis	363	2.3 ± 0.1	42	3.4 ± 0.6	0.062	4	2.8 ± 1.5	0.752
Liver cirrhosis	37	0.3 ± 0.1	14	0.9 ± 0.3	0.001	1	0.8 ± 0.8	0.397
Hepatitis B	215	1.8 ± 0.1	17	1.5 ± 0.4	0.640	1	0.8 ± 0.8	0.410
Hepatitis C	25	0.2 ± 0.1	3	0.2 ± 0.1	0.949	0		0.781
Cancer, stomach	112	0.7 ± 0.1	19	1.5 ± 0.4	0.008	0		0.462
Cancer, liver	18	0.1 ± 0.0	9	0.8 ± 0.3	<0.001	2	4.3 ± 3.1	<0.001
Cancer, colorectal	54	0.4 ± 0.1	10	0.9 ± 0.3	0.011	1	0.9 ± 0.9	0.425
Cancer, lung	19	0.1 ± 0.0	1	0.0 ± 0.0	0.151	0		0.780

*  Weight-adjusted estimation of percentage in each group.

†  Pearson χ² test for comparison with no AMD group, not adjusted for any covariate.

‡  A person who has smoked ≥100 cigarettes during the course of his life.

§  A person who drinks one or more units of alcohol in a month.

Univariate binary logistic regression analysis revealed that the presence of any and late AMD was highly associated with age (OR, 1.08; 95% CI, 1.07–1.08, and OR, 1.07; 95% CI, 1.05–1.10, respectively). Among variables regarding health behavior patterns, ever-smoking history of ≥100 cigarettes during the lifetime was significantly associated with the risk of late AMD (OR, 2.06; 95% CI, 1.16–3.64; P = 0.013), after adjusting for age and sex, and, thus, was used as a confounding variable in the following analyses for late AMD. When adjusted for age and sex, ever-smoking history, current smoking, and current alcohol consumption were not associated with the presence of any AMD (P = 0.645, P = 0.985, and P = 0.924, respectively). After controlling for age and sex, neither current smoking nor current alcohol consumption was associated with a risk of late AMD (P = 0.225 and P = 0.821, respectively). When age and sex were adjusted, duration of smoking was not associated with risk of any AMD (P = 0.678) or late AMD (P = 0.897). 

Association of Systemic Diseases With AMD

The association of various systemic diseases with AMD in univariate logistic regression analyses after controlling for age and sex as well as smoking in late AMD is summarized in Table 2. Among systemic diseases, DM was associated with a 25% reduced possibility of any AMD (OR, 0.75), and DM also significantly reduced the OR of late AMD (OR, 0.23). A history of liver cirrhosis was associated positively with any AMD (OR, 2.73), but did not significantly increase the OR of late AMD (P = 0.582). A history of liver cancer significantly increased the OR of any and late AMD (P = 0.002 and P = 0.001, respectively). All the VIFs of these variables were less than 5. The other systemic diseases yielded statistically insignificant correlations after adjustment. 

Table 2

 

Age- and Sex-Adjusted Univariate Logistic Regression Analyses for the Association of Systemic Diseases With AMD in the KNHANES 2008 to 2011

Systemic Diseases	Any AMD			Late AMD*
	OR	95% CI	P Value	OR	95% CI	P Value
DM	0.75	0.60–0.94	0.012	0.23	0.08–0.65	0.006
Hypertension	0.98	0.82–1.16	0.794	1.15	0.68–1.95	0.611
Angina	0.99	0.64–1.52	0.946	1.70	0.52–5.60	0.380
Myocardial infarction	1.49	0.79–2.79	0.215	0.78	0.18–3.51	0.748
Stroke	0.99	0.65–1.52	0.978	0.87	0.30–2.52	0.797
Renal failure	0.49	0.12–2.03	0.328	2.21	0.28–17.70	0.455
Asthma	0.98	0.69–1.41	0.924	1.11	0.49–2.50	0.798
Osteoarthritis	1.12	0.91–1.39	0.292	0.75	0.41–1.37	0.347
Rheumatoid arthritis	1.00	0.66–1.51	0.988	1.06	0.36–3.08	0.920
Liver cirrhosis	2.73	1.36–5.50	0.005	1.73	0.24–12.29	0.582
Hepatitis B	1.01	0.56–1.81	0.974	0.46	0.06–3.36	0.445
Hepatitis C	1.05	0.32–3.43	0.931	N/A
Cancer, stomach	1.29	0.71–2.34	0.409	N/A
Cancer, liver	6.55	2.04–21.03	0.002	21.08	3.35–132.74	0.001
Cancer, colorectal	1.29	0.58–2.84	0.532	1.04	0.14–7.47	0.971
Cancer, lung	0.15	0.02–1.15	0.068	N/A

*  Ever-smoking history of ≥100 cigarettes was additionally adjusted as a covariate.

Age, sex, ever-smoking history, DM, liver cirrhosis, and liver cancer were included in the multivariate logistic regression analyses (Table 3). In the final regression model, age (OR, 1.09; 95% CI, 1.08–1.09), DM (OR, 0.74; 95% CI, 0.59–0.93), and a history of liver cancer (OR, 4.32; 95% CI, 1.74–10.71) were found to be significantly associated factors for any AMD. In the final model for late AMD, age, ever-smoking history (OR, 2.45; 95% CI, 1.42–4.20), the presence of DM (OR, 0.22; 95% CI, 0.08–0.58), and a history of liver cancer (OR, 12.51; 95% CI, 1.39–112.87) were identified as associated factors. Sex and a history of liver cirrhosis did not show significant correlation with AMD. 

Table 3

 

Multivariate Logistic Regression Model for the Association of Systemic Diseases With AMD in the KNHANES 2008 to 2011

AMD Type	Risk Factors	Model 1			Model 2*
		OR	95% CI	P Value	OR	95% C	P Value
Any AMD	Age	1.09	1.08–1.09	<0.001	1.09	1.08–1.09	<0.001
	Male sex	1.00	0.85–1.18	0.958
	DM	0.74	0.59–0.93	0.011	0.74	0.59–0.93	0.011
	Liver cirrhosis	1.95	0.83–4.609	0.125
	Liver cancer	3.77	1.41–10.08	0.008	4.32	1.74–10.71	0.002
Late AMD	Age	1.09	1.06–1.12	<0.001	1.09	1.06–1.11	<0.001
	Male sex	1.33	0.73–2.43	0.360
	Ever-smoker†	1.99	1.07–3.69	0.029	2.45	1.42–4.20	0.001
	DM	0.22	0.08–0.59	0.003	0.22	0.08–0.58	0.002
	Liver cancer	11.81	1.28–108.96	0.029	12.51	1.39–112.87	0.024

*  Final multivariate model consisted of risk factors of P < 0.05.

†  A person who has smoked ≥100 cigarettes during the course of his life.

Association of DR With AMD

The presence of DR was evaluated in 1699 diabetic patients. When adjusted for age and sex, DR was associated with any AMD (P = 0.028; OR, 0.35) among diabetic patients, but not with late AMD (P = 0.342). The association of DR with late AMD remained insignificant when age, sex, and an ever-smoking status were adjusted (P = 0.336). 

Discussion

The present study examined the association of several systemic diseases with AMD using nationally representative data from approximately 15,000 subjects from the KNHANES 2008 to 2011. Age and sex were considered confounders, and history of ever-smoking was regarded as a confounder only for late AMD. After controlling for these covariates, DM was associated with a lower prevalence of any AMD and late AMD. In contrast, a history of liver cancer was correlated with a higher prevalence of any AMD and late AMD. 

A remarkable finding in this study was the inverse association between DM and AMD. Although previous studies have reported a positive correlation ^8,15 or an insignificant association, ^3,16,17 several researchers have described the uncommon coexistence of advanced AMD and DM or DR. ^10,11,18,19 The 10-year incidence of neovascular AMD was lower in patients with newly diagnosed type 2 DM than in control subjects, ²⁰ and the prevalence of neovascular AMD also was lower in patients with DR than in the general population. ¹⁹ The prevalence of neovascular AMD was much lower in those who received retinal laser photocoagulation to treat DR. ¹⁹ The prevalence of early AMD also was significantly lower in diabetic patients (2.51%) than in the general population (11%). ²¹ The Los Angeles Latino Eye Study showed a decreased OR for early and late AMD in diabetic patients, although the observation was not statistically significant due to an insufficient number of subjects. ²² The current study included a large number of subjects and, showed a statistically significant association between the presence of DM and decreased prevalence of any AMD (OR, 0.74) and late AMD (OR, 0.22). 

The underlying mechanisms of this inverse association between DM and AMD are unclear. Ocular changes caused by DM include narrowing of capillaries in the retina and choroid, decreased choroidal blood flow, thickening of the basement membrane of the choriocapillaris, and proliferation of endothelial cells. ^23,24 Among these, Zylbermann et al. ¹⁹ suggested that vascular compromise and alteration of metabolic activity in the retina, RPE, and choroid might reduce production of debris that forms drusen in AMD. They also proposed that the advanced alteration in choroidal vasculature and modified Bruch's membrane structure typical of DR may interfere with the deposition of drusen that marks the progress to late AMD. ¹⁹ Additionally, the damaged inner blood–retinal barrier observed in patients with DR induces a signal that upregulates the transport function of the RPE ²⁵ and possibly delays deposition of drusen. ¹⁰  

In this study, the presence of DR was associated with a decreased risk of any AMD in addition to the effect of DM itself. One possible explanation for this observation is that the presence of DR may indicate advanced ocular diabetic changes and, thus, could have enhanced the mechanisms described above. In addition, the presence of DR might imply prolonged illness and, therefore, more accumulated systemic changes associated with diabetes, which could have influenced the pathogenesis of AMD. Moreover, it can be difficult to discriminate small drusen from hard exudates in DR, although the fundus photographs were evaluated multiple times with relatively high score reliability. Therefore, the prevalence of early AMD presenting with drusen could have been underestimated in participants with DR. The presence of DM alone decreased the risk for late AMD (OR, 0.22); the additional influence of DR was not statistically significant. The reasons for this observation require further exploration. To validate the findings in this study and better understand the association between AMD and DM or DR, further prospective clinical studies will be required, with fasting plasma glucose and glycated hemoglobin levels, DM duration, and DM treatment methods as potential covariates. 

Recently, Park et al. ²⁶ also reported an inverse association between DM and early AMD using the same dataset as this study; however, they excluded DM from their ultimate model, presumably because they had preferentially included variables with lower P values such as education and occupation. The presence of DM was not associated with late AMD in their study (P = 0.153), a difference that might be attributed to the different criteria used to determine the presence of DM. In this study, we specifically focused on the association of systemic diseases with AMD and included only well-known risk factors like age, sex, and smoking status as covariates. We adopted strict criteria to determine the presence of DM, using physician confirmation, use of antidiabetic medication, or a high fasting plasma glucose level, as suggested by the KNHANES. As a result, we obtained multivariate models that included the presence of DM as a negatively associated factor with any AMD as well as late AMD, consistent with results of previous reports. ^{10,11,18–21}  

Another notable finding in this study was a correlation between liver cancer with any AMD (OR, 4.32) and late AMD (OR, 12.51). A recent study reported a higher prevalence of AMD in patients who underwent liver transplantation (64.6%) than in a general population of similar ethnicity and age (37.1%). ²⁷ However, to our knowledge, an association between AMD and liver cancer has not been reported previously. The mechanism behind this finding is not yet clearly understood. Data were not available in the study for histological classification. The effects of cancerous conditions or hepatic dysfunction on AMD are unclear. One possible explanation might be mediation by complement factor H (CFH), which is associated strongly with development of AMD. ^5,27 The CFH is produced primarily by the liver, ²⁷ as evidenced by high levels of CFH secreted by primary human hepatocyte cultures, and expression of CFH in human liver cDNA libraries. ^28,29 Liver cancer might alter the production or function of CFH, and consequently influence the development of AMD. Alternatively, liver cancer patients in this study might have had a higher CFH Y402H allele frequency, which is a risk factor for AMD, and reported to be more prevalent in recipients of liver transplantation than in the general population. ²⁷ Unfortunately, neither plasma CFH protein measurements nor CFH genotyping was performed in this study. 

Other hepatic diseases, such as liver cirrhosis and hepatitis B or C, could facilitate development of AMD. In a previous study, hepatitis B surface antigen (HBsAg) serum positivity was identified as a significant risk factor for AMD after adjusting for age, with a >2-fold increase in risk. ³⁰ However, in the current study, only liver cirrhosis was associated with any AMD in univariate regression analysis, and the significance was diminished in multiple regression analysis. Further investigation is required to evaluate the association of liver diseases with AMD. 

Cancer and AMD share several risk factors, such as cigarette smoking, oxidative stress, chronic systemic inflammation, and genetic predisposition. ³¹ However, several studies have analyzed the relationship between AMD and several types of cancer. One population-based study reported that early AMD was associated with lung cancer mortality in African Americans, ³¹ but most studies showed an insignificant association between AMD and cancer mortality. ^32,33 In the present study, except liver cancer, all investigated types of cancer were not associated with AMD. Considering the low prevalence of cancer, as well as late AMD, a very large number of subjects would be needed for future meaningful analysis. 

Several studies have reported a significant correlation of hypertension with AMD, ^4,6,34 but others have not. ^9,16 Similarly, arterial blood pressure was associated with AMD in some studies, ⁷ but not in others. ^9,17 While some studies suggest that vascular changes of the retina and choroid associated with hypertension might affect development of AMD, ⁶ the association between AMD and hypertension still is controversial. Angina and myocardial infarction (MI) also have shown inconsistent association with AMD in previous studies. ^4,9,22,34 In the current study, hypertension was not significantly associated with AMD after controlling for age and sex. We also were unable to find any association between AMD and angina, MI, or stroke, after adjusting for covariates. Renal failure also has been associated with AMD in previous studies. ^35,36 Chronic renal disease was associated positively with early AMD (OR, 1.68) as well as peripheral retinal drusen (OR, 2.01). ³⁵ Possible explanations for this may include elevated serum VEGF levels among patients with kidney disease, ³⁷ and the common genetic etiology of CFH polymorphisms. ³⁸ However, the association was not statistically significant in this study. The association between AMD and asthma also was insignificant, as reported in a previous study. ³⁹ Arthritis reportedly was more prevalent in subjects with AMD than in controls in the Age-Related Eye Disease Study, ³⁴ but in this cross-sectional study, which included 2294 patients with osteoarthritis and 409 patients with rheumatoid arthritis, the association was not significant. Smoking only affected the prevalence of late AMD in this study, and not that of any AMD; however, the reason for this difference remains unclear. 

The present study has some limitations. First, due to its cross-sectional study design, the association of factors with AMD does not imply causality. The underlying mechanisms could not be elucidated fully by the data available in the present study. To further evaluate causality and investigate the incidence of AMD in subjects with multiple systemic diseases, longitudinal cohort studies may be required. Nevertheless, the cross-sectional results of the current study will help to identify which patients with associated systemic diseases to screen for AMD. Next, this study may involve a survivorship bias caused by mortality selection induced by a certain diseases, such as DM with cardiovascular complications. This bias could result in a diminished number of patients with the disease, especially of an advanced and lethal status, and an increased number of patients with mild forms of the disease. Although we performed an age-adjustment in our analyses to minimize any mortality selection bias, it could not be completely excluded. Moreover, institutionalized persons, who may have more severe forms of disease, were not included in this survey. This may weaken the association of the disease with AMD. Third, participation rates can vary according to the prevalent diseases, which may induce additional bias. However, the nationwide random sampling design and relatively high participation rate (77.2%) could minimize selection bias in the participation process. Fourth, except for DM and hypertension, evaluation for the presence of a disease was based on patient self-reporting. Although the survey confirmed disease diagnosis by a physician through repeated questions, a patient's self-report might be inaccurate. Fifth, this survey could have selection bias from the inclusion criteria that necessitated a gradable fundus photograph of either eye. Those without a fundus image or with nongradable photographs of both eyes were excluded, leading to a study population with clear ocular media and without mature cataracts or vitreous hemorrhages. Therefore, the prevalence of AMD could be underestimated. Lastly, the associations with AMD were not analyzed in detail according to the stage or severity of each disease. The histologic type of malignancies also was not discriminated. 

However, this study also has several strengths. The data analyzed are nationally representative and include a large number of participants. Statistically well-designed sampling enabled epidemiologic associations close to the true population values. Another strength was the strict diagnostic criteria for DM and hypertension (self-reporting as well as serum glucose levels or physician confirmation) that ensured precise discernment between true patients and controls. The homogeneous tribe composition of Korea and the high degree of agreement in the rates of fundus photograph grading are additional merits of this study. 

In conclusion, this report reveals a lower AMD prevalence among patients with diabetes and a higher prevalence among patients with liver cancer in a large, nationally representative study population. To our knowledge, this is the first population-based study to describe an association of liver cancer with AMD. The results may further our understanding of AMD etiology and development, and could be helpful to plan health policies for the screening, prevention, and treatment of AMD. 

Acknowledgments

The authors thank the Epidemiologic Survey Committee of the Korean Ophthalmological Society for their dedication to the design and implementation of the Korea National Health and Nutrition Examination Survey, data acquisition and verification, and for allowing public access to the data. The authors alone are responsible for the content and writing of the paper. 

Disclosure: B.-J. Cho, None; J.W. Heo, None; J.P. Shin, None; J. Ahn, None; T.W. Kim, None; H. Chung, None