Social Science & Medicine 75 (2012) 179e185

Contents lists available at SciVerse ScienceDirect

Social Science & Medicine

Short report

Predictors of and health services utilization related to depressive symptoms

a r t i c l e i n f o a b s t r a c t

Article history: While the prevalence, comorbidity, risk profile and health care utilization for late-life depression have
Available online 5 April 2012 been described for many Western countries, much less is known about the recent epidemiology of
late-life depression in East Asian countries such as Korea. We investigated predictors for depressive
Keywords: symptoms and the association between depressive symptoms and the utilization of both medical care
Late-life depression and preventive services in elderly Koreans. Data were obtained from a nationally representative sample
Korea
of Koreans aged 60 and above (2226 men, 2911 women) who participated in the 2008 wave of the Korean
Predictors
Longitudinal Study of Ageing. Depressive symptoms were measured using the 10-item Center for
Health care utilization
Epidemiological Studies-Depression scale. Risk factors considered included sociodemographics, health
behaviors, chronic diseases, and physical function. Health care utilization factors included hospitaliza-
tion, outpatient clinic use and basic medical checkup. Being female, being unmarried, and having less
education, lower household income, physical inactivity and lower weight were associated with
depressive symptoms. Presence of chronic diseases and limited physical function also showed a signifi-
cant association with depressive symptoms. Depressive symptoms were associated with increased odds
of hospitalization and outpatient visits, but decreased the odds of utilization of basic medical checkup
after controlling for potential confounders. Findings on most risk factors, except lower weight, were
consistent with reports from Western countries. It is important to recognize the burden of depressive
symptoms in the elderly. The interaction of such symptoms with chronic diseases should be
acknowledged and considered in the clinical setting as well as in health care planning and policymaking.
Ó 2012 Elsevier Ltd. All rights reserved.

Background and 33% (Cho, Hahm, Jhoo, Bae, & Kwon, 1998; Kim, Shin, Yoon, &
Stewart, 2002; Lee & Shinkai, 2005; Suh et al., 1998). Moreover,
Depression is an important public health issue considering its some studies suggest that Korean elderly have a higher prevalence
high prevalence and established association with major comor- of depressive symptoms than those in other Asian countries such as
bidities such as cardiovascular disease, cerebrovascular disease and Japan, Indonesia, Vietnam, and Lao PDR (Sakagami et al., 2005).
cancer (Meeks, Murrell, & Mehl, 2000; Wells, Golding, & Burnam, Elderly Korean people of today have experienced dramatic
1989). The prevalence of clinically significant depressive symp- socio-cultural changes in their lifetimes: they were born in Korea’s
toms among the elderly in Western countries varies from 7.2% to Japanese colonization era and their period of adolescence or young
49% (Djernes, 2006), which is similar in Asia (Hahn, Yang, Yang, adulthood was interrupted by the Korean War in the 1950s. Rapid
Shih, & Lo, 2004; Jain & Aras, 2007; Wada et al., 2005). In Korea, westernization and industrialization during the 1970s and 1980s
the estimated prevalence of late-life depression is between 15.2% brought additional burdens. This unique historical background can
potentially impact mental health and possibly give rise to a high
prevalence of depression. Owing to the dramatic declines in fertility
* Corresponding author. Department of Community and Family Medicine, Duke
and increasing life expectancy, Korea is one of the most rapidly
University Medical Center, Durham, NC, USA. Tel.: þ1 919 660 0331. aging countries in Asia. In 2006, the average life expectancy in
E-mail address: truls.ostbye@duke.edu (T. Østbye). Korea was 75.7 for men and 82.4 for women. Over the past decade

0277-9536/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2012.03.012
180 J.H. Shin et al. / Social Science & Medicine 75 (2012) 179e185

alone, average life expectancy has increased by 5.7 years for men studies with the US, UK and other European countries that already
and 4.6 years for women. The proportion of elderly aged 65 and have similar longitudinal studies in place. The 2006 baseline survey
older was 7.2% in 2000, and is expected to increase to 24.4% by 2030 was conducted among Koreans aged 45 years or older and used
(Korean National Statistics Office, 2006, pp. 5e355). In terms of multistage, stratified probability sampling (based on geographical
public health planning and clinical practice, it is important to area) to select households that are representative of the national
identify and better understand groups among the elderly who are population (Korea Labor Institute, 2010). A total of 10,254 individ-
at increased risk for depression. Despite the assumed high levels of uals completed the interviews conducted by a trained interviewer
depressive symptoms among the elderly, previous studies have in 2006. In the 2008 follow-up survey, 8875 respondents from the
been largely limited to selected communities (Cho et al., 1998; Kim original sample were included (86.6% response rate), of which we
et al., 2002; Lee & Shinkai, 2005). restricted the analysis to 5137 individuals aged 60 years and older
Late-life depression also presents substantial rates of comor- who had complete data on depressive symptoms. The survey
bidity with other chronic diseases. To explain this association, some consisted of eight detailed sections: demographics, family, health
studies have suggested an immunological link between depression (including physical and psychological conditions and health
and chronic diseases (Kiecolt-Glaser & Glaser, 2002; Kobrosly & van behaviors), employment, income, wealth, subjective life expec-
Wijngaarden, 2010). Several mechanisms have also been suggested tancy, and quality of life. Sampling weights accounted for the
to explain how depression may lead to the limited use of health sampling rate and the response rate in various subgroups. Written
resources, poor adherence (behavioral pathways), biased health informed consent was obtained from all respondents prior to the
decision process (cognitive pathways), or deterioration of social interviews. The research was approved by the Duke University
networks (social pathways) (Cohen & Rodriquez, 1995). The Medical Center Institutional Review Board.
underuse of preventive services may then increase chronic disease
morbidity. However, little research has been conducted on the Measures
association between the receipt of screening for chronic diseases
and having depressive symptoms among the elderly. More specif- Outcome measures
ically, it has been reported that depression among Caucasian The short-form (10-item) Center for Epidemiological Studies-
elderly was associated with high utilization of several types of Depression (CES-D10) scale was used to assess depressive symp-
medical care including inpatient admissions, laboratory tests, toms. The CES-D 10 is a short scale developed to assess depressive
prescription medications, and outpatient visits (Katon et al., 2003; symptoms in the prior week in population studies. The 10-item
Luber et al., 2001). screening questionnaire shows good predictive accuracy when
In 1988, the Korean government established universal public compared to the full-length 20-item version of the CES-D (k ¼ 0.97,
health insurance and a medical aid policy for the poor and aged, p < 0.001) (Andresen, Malmgren, Carter, & Patrick, 1994). The
which provides basic health coverage for all Korean citizens. In an CES-D 10 includes questions about symptoms experienced during
effort to lower the prevalence of chronic diseases e including the last week, focusing on five subscales: mood (five items), energy
hypertension, diabetes and cancer e public health insurance level (two items), irritability, concentration, and sleep (one item
covers a free national health screening program every other year. each) (Irwin, Artin, & Oxman, 1999; Kohout, Berkman, Evans, &
The system consists of three parts: a screening test to detect Cornoni-Huntley, 1993). The response set is the same for each
chronic diseases, a confirmative test, and a consultation with question, with four possible answers: “less than 1 day” (0 points),
a physician if the screening test is positive or indicates that the “1e2 days” (1 point), “3e4 days” (2 points), and “5e7 days” (3
patient is at high risk for disease. As health examination agencies points). The total score of the 10 items ranges from 0 to 30, with
carrying out national heath examinations increased from about scores reversed for two positively phrased items. This score was
2000 in 2002 to 14,000 in 2010, the total screening rate for health used as the outcome measure, with higher scores indicating greater
examinations increased from 43% in 2002 to 66% in 2010 (National distress. Internal consistency of the CES-D10 was high for this
Health Insurance Corporation, 2011). In assessing the Korean sample (a ¼ 0.86). The correlation between each subscale (mood,
health care system, it is important to evaluate the availability of energy level, irritability, concentration, and sleep) and the overall
clinical preventive services as well as the predictors of their CES-D10 score were 0.37, 0.87, 0.78, 0.76, and 0.67, respectively
utilization. (p < 0.001), suggesting that energy level, irritability and concen-
The aims of the present study were twofold: to assess the key tration were contributing more strongly to depressive symptoms
predictors of late-life depressive symptoms, and to assess the than were mood and sleep.
impact of depressive symptoms on health service utilization for To estimate the prevalence of depression and investigate the
both medical care and prevention, while adjusting for medical impact of the depressive symptoms on health care utilization, we
comorbidities. In particular, this study aims to assess whether or not classified elderly with a score of 10 (cut-off 9/10) as “having
having depressive symptoms has a differential effect on medical clinically significant depressive symptoms”. Although the cut-off 9/
care and preventive services. The findings may point to an expla- 10 has not been validated for Koreans, this has been validated
nation for the excess comorbidity associated with late-life depres- among elderly in Western and other Asian countries: in one
sion and advance better public health interventions for the elderly. Western study (Andresen et al., 1994), it had a sensitivity of 96% and
a specificity of 100% with a k of 0.97 and in a Chinese study (Boey,
Methods 1999), it had a sensitivity of 82% and a specificity of 98% with a k of
0.84, when compared with the original CES-D 20.
Study population The use of health care services was arranged by self-report and
categorized as follows:
We used data from the 2008 wave of the Korean Longitudinal
Study of Ageing (KLoSA). The KLoSA is a nationally representative 1. Hospitalizations in the past 12 months;
cohort study conducted by the Korea Labor Institute. This survey 2. Outpatient clinic visits (public health clinic, doctor’s office and
collected a broad set of data including employment status, income, hospital outpatient office) for treatment in the past 12 months;
assets, family relations, health status and health utilization for 3. Basic medical checkup as covered by the National Health
middle/old-aged population. The KLoSA allows for comparative Insurance to screen for chronic diseases in the past 2 years.
 Table 1
Sample characteristics, KLoSA, 2008(%) and the CES-D 10 score by subgroups.
Predictor variables
Variables Total(N ¼ 5137) CES-D 10 score P-valueb
Demographic and socioeconomic variables included age (10-
year intervals), education (elementary school or less/middle % Mean(SD)

school/high school/college or more), marital status (married and Age, years <0.001
60e69 55.1 7.4(5.6)
unmarried [divorced, widowed and never married]) and household
70e79 34.4 9.1(6.0)
income. Household income was calculated as the total household 80 10.5 10.8(6.3)
income divided by the square root of the number of household Gender <0.001
members; these scores were then divided into quartiles. Health male 43.4 7.1(5.5)
behaviors included smoking (never, former, current), alcohol female 56.6 9.2(6.1)
Education <0.001
(never, former, current) and physical activity (active vs. inactive: <3 Elementary school or less 62.8 9.3(6.1)
times per week). BMI was based on self-reported weight and Middle school 13.7 7.2(5.4)
height. The subjects were classified by the International Obesity High school 16.7 6.6(5.5)
Task Force (IOTF) for Asian adults in the Asian and Pacific regions college more 6.8 5.7(4.6)
Household income categorya <0.001
(WHO, 2000) into five categories: underweight (less than 18.5 kg/
1(lowest) 35.8 10(6.5)
m2); normal weight (18.5e22.9 kg/m2); overweight (23 kg/m2 2 32.0 7.6(5.3)
and <25 kg/m2); obese class I (25.0e29.9 kg/m2); and obese class II 3 19.2 7.3(5.6)
(30 kg/m2). Self-report of eight chronic diseases reported to be 4(highest) 12.9 6.5(5.3)
diagnosed by a physician included: hypertension, diabetes mellitus, Marital status <0.001
married 71.3 7.6(5.6)
cancer, chronic lung disease, liver disease, cardiac disease, cere-
unmarried 28.7 10.0(6.4)
brovascular disease, and arthritis. Smoking <0.01
Physical functioning was assessed by seven activities of daily never 80.8 8.4(5.9)
living (ADL) items, and 10 instrumental ADL (IADL) items. For every former 3.3 8.0(6.1)
current 15.9 7.8(5.8) <0.001
activity, subjects rated their independence between 1 and 3 (1,
Alcohol consumption
completely independent; 2, needs some help; and 3, completely never 64.8 8.9(6.1)
dependent). ADL dependency was defined as either “needs some former 4.7 8.5(5.9)
help” or “completely dependent” in one or more ADLs: dressing, current 30.5 7.0(5.3)
washing, bathing, eating, getting out of bed, using toilet, and Physical activity <0.001
Inactive(less than once 66.4 9.1(6.2)
controlling urination and defecation. IADL dependency was derived
a week)
in the same way as ADL dependency: either some need or complete active 33.6 6.8(5.2)
dependency in grooming, cleaning, cooking, doing laundry, trav- BMI <0.001
eling a short distance, using public transportation, shopping, 18.5 5.5 11.0(6.9)
18.6e22.9 45.1 8.3(5.7)
managing money, making phone calls, or taking medicine.
23e24.9 26.7 7.9(5.6)
25e29.9 21.0 7.1(5.5)
Statistical analyses 30 1.7 7.7(5.7)
Chronic disease
We calculated the frequencies and weighted proportions of the Hypertension Yes 41.3 8.9(6.2) <0.001
No 58.7 7.9(5.7)
explanatory variables. Bivariate association between depressive
Diabetes mellitus Yes 17.8 9.7(6.3) <0.001
symptoms and predictors were analyzed using student t-tests and No 82.2 8.0(5.8)
one-way analyses of variance. Simple and multiple regression Cancer Yes 4 10.5(7.0) <0.001
analyses were used to further examine the contribution of individual No 96 8.2(5.9)
Chronic lung disease Yes 3.6 10.5(7.5) <0.001
predictive factors to variance in the CES-D 10 scores. Considering the
No 96.4 8.2(5.9)
dichotomous nature of the outcome (i.e., using or not using health Liver disease Yes 2.2 8.6(6.0) <0.001
care service), unadjusted and adjusted logistic regression models No 97.8 8.3(5.9)
were used to assess the impact of depressive symptoms on health Cardiac disease Yes 8.4 9.6(6.6) <0.001
care utilizations. Initially, the total CES-D10 score was used as No 91.6 8.2(5.9)
Cerebrovascular disease Yes 5.8 12.2(7.3) <0.001
a continuous variable, but, to better illustrate the association
No 94.2 8.1(5.8)
between the presence of clinically significant depressive symptoms Arthritis Yes 28 10.2(6.2) <0.001
and health care utilization, the CES-D10 was dichotomized using No 72 7.6(5.7)
a 9/10 cut-off. SPSS for Windows, version 17 (SPSS Inc., Chicago, Functional status
ADL dependency Yes 6.4 7.9(5.6) <0.001
Ill, USA) was used for data management and analysis.
No 93.6 15.0(6.9)
IADL dependency Yes 16.8 7.7(5.5) <0.001
Results No 83.2 11.8(6.7)
Health care utilization
There were 2226 (43.4%) male and 2911 (56.6%) female elderly Hospitalization Yes 15.1 10.5(6.9) <0.001
No 84.9 7.9(5.7)
enrolled in this study. The mean age was 69.6 (7.0) and the mean
Outpatient clinic Yes 75.1 8.0(6.0) <0.001
CES-D 10 score was 8.5 (6.0). The estimated prevalence of clini- No 24.9 7.6(5.2)
cally significant depressive symptoms among Korean elderly was Basic medical checkup Yes 61 7.6(5.5) <0.001
37.5%. Table 1 shows characteristics of the sample and bivariate No 39 9.5(6.4)
associations between the CES-D 10 score and each category of the KLoSA, Korean Longitudinal Study of Ageing. CES-D, Center for Epidemiological
predictor variables. Over half of the elderly were in their sixties and Studies-Depression. SD, standard deviation.
a
had lower education and household income; 41% of the elderly had Yearly household income was categorized into quartiles; 1) 850, 2) 851e2000,
3) 2001e3600, 4) >3600 (Unit: 10,000 Korean won, 1US dollar ¼ 1057 Korean won
hypertension and 28% had arthritis. Almost 15% reported having
at the time of the study).
been hospitalized and 75% had used outpatient clinics in the past 12 b
Student t-test for difference in CES-D by gender, marital status, physical activity,
months. Almost 60% had undergone basic medical checkups to chronic diseases, functional status and health care utilization; ANOVA for difference
screen for chronic diseases in the last 2 years. by age, education, household income, smoking, alcohol consumption, and BMI.
182 J.H. Shin et al. / Social Science & Medicine 75 (2012) 179e185

Older age, being female, having less education, having a lower (Table 2). The significant effects of certain individual predictors in
income and being unmarried were factors significantly associated unadjusted analyses, including age, smoking, alcohol consumption,
with a higher CES-D 10 score. Elderly who were not smoking or and hypertension, became non-significant in multivariate analysis.
consuming alcohol had higher mean depression scores. Those The following factors remained significant predictors of depressive
physically inactive had higher CES-D 10 scores, as did those with the symptoms even after multivariate adjustment: female gender, less
lowest BMI. All comorbid conditions including hypertension, dia- education, lower household income, unmarried status, physical
betes, cancer, chronic lung disease, liver disease, cardiac disease, inactivity, low BMI, diabetes mellitus, cancer, chronic lung disease,
cerebrovascular disease, arthritis and limited physical functions were cardiac disease, cerebrovascular disease, and arthritis, as well as
significantly associated with higher CES-D 10 scores. Elderly who had dependency in ADL and IADL.
been hospitalized or used outpatient clinics had higher CES-D 10 Through logistic regression analysis, we observed positive
scores, as did those who did not receive basic medical checkups. doseeresponse relationships between increasing the CES-D 10
The association between depressive symptoms and socio- scores and outpatient clinic visits and hospitalization, and an
demographic factors, health behaviors, chronic disease and phys- inverse doseeresponse relationship between increasing the
ical functions were investigated further in linear regression analysis CES-D 10 scores and basic medical checkups (data not shown).

Table 2
Regression models assessing the association between risk factors and number of depressive symptoms.

Variables Unadjusted Adjusted

b 95% CI P b 95% CI P
Intercept 8.14
Age
60e69 e e
70e79 1.71 1.70 to 1.72 <0.001 0.32 0.31 to 0.33 0.09
>80 3.42 3.40 to 3.44 <0.001 0.59 0.57 to 0.61 0.06
Gender
male e
female 2.16 2.14 to 2.17 <0.001 1.23 <0.001
Education
Elementary school or less e
Middle school 2.01 2.03 to 1.98 <0.001 0.66 0.67 to 0.64 <0.01
High school 2.63 2.64 to 2.62 <0.001 0.64 0.66 to 0.62 <0.01
college more 3.62 3.64 to 3.60 <0.001 1.16 1.19 to 1.14 <0.001
Household income categorya
1(lowest) e
2 2.31 2.33 to 2.29 <0.001 1.22 1.23 to 1.20 <0.001
3 2.61 2.63 to 2.58 <0.001 1.19 1.21 to 1.18 <0.001
4(highest) 3.42 3.44 to 3.40 <0.001 1.85 1.86 to 1.83 <0.001
Marital status
unmarried
married 2.35 2.37 to 2.33 <0.001 0.78 0.79 to 0.77 <0.001
Smoking
never e
former 0.43 0.46 to 0.40 0.35 0.15 0.17 to 0.12 0.74
current 0.64 0.66 to 0.65 <0.01 0.37 0.35 to 0.39 0.13
Alcohol consumption
never e
former 0.39 0.41 to 0.37 0.32 0.04 0.01 to 0.06 0.91
current 1.95 1.97 to e1.94 <0.001 0.17 0.18 to 0.15 0.41
Physical activity
Inactive(less than once a week) e
active 2.30 2.31 to 2.28 <0.001 1.17 1.18 to 1.16 <0.001
BMI
18.6e22.9 e
18.5 2.66 2.64 to 2.69 <0.001 1.31 1.28 to 1.32 <0.001
23e24.9 0.42 0.43 to 0.41 <0.05 0.05 0.03 to 0.06 0.79
25e29.9 0.65 0.66 to 0.63 <0.01 0.64 0.66 to 0.63 <0.01
30 0.67 0.64 to 0.71 0.31 0.79 0.83 to 0.76 0.19
Chronic disease
Hypertension 0.97 0.96 to 0.98 <0.001 0.18 0.17 to 0.19 0.27
Diabetes mellitus 1.64 1.63 to 1.66 <0.001 1.12 1.11 to 1.13 <0.001
Cancer 2.26 2.24 to 2.27 <0.001 2.43 2.41 to 2.45 <0.001
Chronic lung disease 2.28 2.25 to 2.31 <0.001 1.69 1.67 to 1.71 <0.01
Liver disease 0.30 0.27 to 0.34 0.61 0.41 0.37 to 0.43 0.49
Cardiac disease 1.41 1.39 to 1.43 <0.001 0.87 0.85 to 0.88 <0.01
Cerebrovascular disease 4.12 4.10 to 4.14 <0.001 2.11 2.09 to 2.13 <0.001
Arthritis 2.59 2.58 to 2.60 <0.001 1.38 1.36 to 1.39 <0.001
Functional status
ADL dependency 7.12 7.10 to 7.14 <0.001 3.55 3.52 to 3.57 <0.001
IADL dependency 4.16 4.15 to 4.17 <0.001 1.54 1.53 to 1.55 <0.001

b, regression coefficients. CI, confidence interval. Adjusted for all other variables shown in this table.
a
Yearly household income was categorized into quartiles; 1) 850, 2) 851e2000, 3) 2001e3600, 4) >3600 (Unit: 10,000 Korean won, 1US dollar ¼ 1057 Korean won at the
time of the study).
 J.H. Shin et al. / Social Science & Medicine 75 (2012) 179e185 183

However, from a public health perspective, it is more important to Korean culture of having evident disparities between males and
assess how the group with or without disease affects health care females, and being ingrained with patriarchal social values, the
utilization; thus, the presence of clinically significant depressive generation of older women may sense powerlessness and
symptoms was used as a dichotomous independent variable instead discrimination, leading to depressive symptoms (Jeon, Jang, Rhee,
of a depressive symptom score (Table 3). Controlling for socio- Kawachi, & Cho, 2007). In univariate analysis, age was a signifi-
demographic factors and health-related factors including comor- cant risk factor for depressive symptoms. However, this association
bidities and functional status, having clinically significant lost significance after adjusting for health status and functional
depressive symptoms was associated with a 22% higher odds of impairment, supporting the notion that age-related effects on
outpatient clinic visits (OR, 1.22; 95% CI, 1.18e1. 27) and 19% of depression are mainly due to physical health problem and func-
hospitalization (OR, 1.19; 95% CI, 1.11e1.24). However, having clini- tional impairment rather than age per se (Roberts, Kaplan, Shema,
cally depressive symptoms was associated with an 18% lower odds of & Strawbridge, 1997).
utilization of a basic medical checkup (OR, 0.82; 95% CI, 0.77e0.86) Elderly non-smokers and non-alcohol consumers were inter-
even after controlling for outpatient clinic visits and hospitalization estingly more likely to have depressive symptoms in the univariate
as well as sociodemographic and health-related factors. analysis. Although other studies have shown that smoking
(Hämäläinen et al., 2001; Kendler et al., 1993; Klungsøyr, Nygård,
Sørensen, & Sandanger, 2006) and alcohol consumption (Blow,
Discussion
Serras, & Barry, 2007; Grant & Harford, 1995) are risk factors for
depression, it is possible that nicotine and alcohol might have some
The overall proportion of Korean elderly with clinically signifi-
positive reinforcing and rewarding effects on Korean elderly who
cant depressive symptoms was 37.5%. This proportion was some-
have experienced unique historical difficulties for a long time.
what higher than previously reported in community studies in
However, after adjusting for all covariates, smoking and alcohol
Korea: in urban residents, 18% of elderly had significant depressive
consumption were not associated with depressive symptoms. The
symptoms measured by the original CES-D (Cho et al., 1998) and
complex relationship between smoking and alcohol, and socio-
a national sample showed a prevalence of 25.3% in elderly aged
demographic factors and health status could explain this finding. A
55e69 years screened by the original CES-D (Suh et al., 1998).
limitation of the smoking and alcohol data was that the actual
However, many other Korean and Asian samples have shown rates
amounts were not reported. Physical inactivity was significantly
comparable to ours, namely 33.5% of Koreans using the Geriatric
associated with depressive symptoms. In contrast to the several
depression Scale (Kim et al., 2002), 39.1% of Japanese
studies in Western countries showing a positive relationship
(Matsubayashi et al., 2005), 27.3% neighbouring (Ishine et al., 2006)
between obesity and depression (Luppino et al., 2010; Sachs-
and 33.8% of Indonesians (Wada et al., 2005). However, these
Ericsson et al., 2007), an inverse relationship between BMI and
differences in prevalence should be interpreted with caution given
depressive symptoms was found in our study. This finding is
the different instruments and cut-off points used.
consistent with the results from two Asian countries with Korea,
Consistent with previous research in Western (McCall, Parks,
namely Japan and China (Kuriyama et al., 2006; Li et al., 2004),
Smith, Pope, & Griggs, 2002; Mojtabai & Olfson, 2004; Wilhelm,
where a higher weight has also been regarded as a sign of wealth
Mitchell, Slade, Brownhill, & Andrews, 2003) and Asian countries
and health in the past (Li et al., 2004). Finally ADL, IADL limitations
(Chou, Ho, & Chi, 2006; Hahn et al., 2004; Lee & Shinkai, 2005;
were also associated with depressive symptoms.
Malhotra, Chan, & Østbye, 2010), having less education and lower
Our study confirmed that most chronic diseases, except hyper-
household income were the demographic factors strongly associ-
tension and liver disease, were associated with depressive symp-
ated with depressive symptoms. Being unmarried may be associ-
toms. Although causality in the relationship between depression
ated with a lack of social support and a possible risk factor for
and chronic diseases is ambiguous, it is possible that elderly people
depressive symptoms as reported in other studies (Djernes, 2006;
with depressive symptoms are less likely to adhere to regular
Hahn et al., 2004; Jain & Aras, 2007). Our results were consistent
screening for chronic diseases; this missed screening opportunity
with most studies from elsewhere showing that females are at
can lead to the development of medical comorbidities of depressive
higher risk for depressive symptoms than males (Minicuci, Maggi,
symptoms. Little is known about screening among depressed
Pavan, Enzi, & Crepaldi, 2002; Zunzunegui, Béland, Llácer, & León,
elderly, and the findings from the few available studies were
1998), however, it should be noted that this difference has not
inconsistent and limited to the screening for specific cancers and
been universally observed (Barefoot, Mortensen, Helms, Avlund, &
influenza immunization (Lasser et al., 2003; Peytremann-
Schroll, 2001; Malhotra et al., 2010). Considering the distinctive
Bridevaux, Voellinger, & Santos-Eggimann, 2008; Pirraglia, Sanyal,
Singer, & Ferris, 2004). We attempted to address this issue by
Table 3 considering a wide range of chronic diseases including hyperten-
Regression models assessing association between the presence of ‘clinically signif- sion, diabetes, dyslipidemia, liver disease, chronic kidney disease
icant depressive symptoms’ and medical care and preventive services utilization. and chronic lung disease, in addition to breast and cervix cancer,
Unadjusted Adjusted which are included in Korea’s national health screening program.
model The 61% screening rate in the total sample was very similar to the
Model 1a Model 2b
60.2% rate reported in the total population (National Health
Medical care service
Outpatient clinic visit 1.32 (1.25e1.34) 1.25 (1.22e1.28) 1.22 (1.18e1.27)
Insurance Corporation, 2011). We found that depressive symp-
Hospitalization 1.84 (1.81e1.89) 1.64 (1.62e1.68) 1.19 (1.11e1.24) toms were associated with less frequent screening for chronic
Preventive service diseases. On the other hand, despite controlling for chronic
Basic medical checkup 0.60 (0.56e0.63) 0.71 (0.68e0.73) 0.82 (0.77e0.86)c diseases, depressive symptoms were associated with increased
Each health care and preventive service was entered into a regression model hospitalization and use of outpatient medical services; findings
separately. Values are odds ratios with 95% confidence intervals. which are consistent with studies performed in Western countries
a
Adjusted for demographic variables including age, gender, education, household (Katon et al., 2003; Luber et al., 2001; Peytremann-Bridevaux,
income and marital status.
b
Adjusted as for Model 1 plus health-related variables including smoking,
Voellinger, & Santos-Eggimann, 2008).
alcohol, physical activity, BMI, chronic diseases and disability. Considering that the national health screening program for
c
Adjusted as for Model 2 plus outpatient clinic visit and hospitalization. chronic diseases in Korea focuses on practical barriers to screening
184 J.H. Shin et al. / Social Science & Medicine 75 (2012) 179e185

such as geographic factors and access to health care, our findings Chou, K. L., Ho, A. H., & Chi, I. (2006). Living alone and depression in Chinese older
adults. Aging & Mental Health, 10(6), 583e591.
have important public health implications. First, many interven-
Cohen, S., & Rodriquez, M. S. (1995). Pathways linking affective disturbances and
tions have been tested to increase the rates of clinical preventive physical disorders. Health Psychology, 14(5), 374e380.
service delivery and lower the medical cost and prevalence of Djernes, J. K. (2006). Prevalence and predictors of depression in populations of
chronic diseases among the elderly. However, before interventions elderly: a review. Acta Psychiatrica Scandinavica, 113(5), 372e387.
Grant, B. F., & Harford, T. C. (1995). Comorbidity between DSM-IV alcohol use
that target the elderly can be effective, it is necessary to further disorders and major depression: results of a national survey. Drug and Alcohol
study the mechanism through which risk factors such as depressive Dependence, 39(3), 197e206.
symptoms may result in missed opportunities for clinical preven- Hahn, C. Y., Yang, M. S., Yang, M. J., Shih, C. H., & Lo, H. Y. (2004). Religious atten-
dance and depressive symptoms among community dwelling elderly in Taiwan.
tive services. In addition, physicians responsible for primary care e International Journal of Geriatric Psychiatry, 19(12), 1148e1154.
having more opportunities for early detection through their Hämäläinen, J., Kaprio, J., Isometsä, E., Heikkinen, M., Poikolainen, K., Lindeman, S.,
outpatient clinic visits e need a well-established set of interven- et al. (2001). Cigarette smoking, alcohol intoxication and major depressive
episode in a representative population sample. Journal of Epidemiology &
tions that is integrated within routine medical checkups. This Community Health, 55(8), 573e576.
would enable better screening for chronic disease among elderly Ishine, M., Sakagami, T., Sakamoto, R., Wada, T., Khampitak, K., Fushida, M., et al.
patients with depressive symptoms. (2006). Comprehensive geriatric assessment for community-dwelling elderly in
Asia compared with those in Japan: VII. Khon Khen in Thailand. Geriatrics
The strengths of the study include its large, nationally repre- Gerontology International, 6(1), 40e48.
sentative sample of Korean elderly, and its inclusion of a wide range Irwin, M., Artin, K. H., & Oxman, M. N. (1999). Screening for depression in the older
of demographic factors, health behaviors, chronic diseases and adult: criterion validity of the 10-item Center for Epidemiological Studies
Depression Scale (CES-D). Archives of Internal Medicine, 159(15), 1701e1704.
physical functions that could be evaluated as risk factors for
Jain, R. K., & Aras, R. Y. (2007). Depression in geriatric population in urban slums of
depressive symptoms. To our knowledge, this is the first study Mumbai. Indian Journal of Public Health, 51(2), 112e113.
demonstrating the association between depressive symptoms and Jeon, G. S., Jang, S. N., Rhee, S. J., Kawachi, I., & Cho, S. I. (2007). Gender differences in
less frequent receipt of screening for chronic diseases in Korean correlates of mental health among elderly Koreans. The Journals of Gerontology:
Series B, Psychological Sciences and Social Sciences, 62(5), S323eS329.
elderly. Katon, W. J., Lin, E., Russo, J., & Unutzer, J. (2003). Increased medical costs of
Several limitations of this study should also be noted. Although a population-based sample of depressed elderly patients. Archives of General
the short-form CES-D scale is a widely used screening tool for Psychiatry, 60(9), 897e903.
Kim, J.-M., Shin, I.-S., Yoon, J.-S., & Stewart, R. (2002). Prevalence and correlates of
depressive symptoms, it cannot be used to confirm diagnosis of late-life depression compared between urban and rural populations in Korea.
clinical depression. The reference to symptoms in the past week International Journal of Geriatric Psychiatry, 17(5), 409e415.
also does not allow us to draw any inference about the duration or Kendler, K. S., Neale, M. C., MacLean, C. J., Heath, A. C., Eaves, L. J., & Kessler, R. C.
(1993). Smoking and major depression. A causal analysis. Archives of General
chronicity of symptoms, although some have argued that the CES-D Psychiatry, 50(1), 36e43.
captures both state and trait aspects of depression (Spielberger, Kiecolt-Glaser, J. K., & Glaser, R. (2002). Depression and immune function: central
Ritterband, Reheiser, & Brunner, 2003). Moreover, epidemiological pathways to morbidity and mortality. Journal of Psychosomatic Research, 53(4),
873e876.
studies using screening instruments tend to report higher preva- Klungsøyr, O., Nygård, J. F., Sørensen, T., & Sandanger, I. (2006). Cigarette smoking
lence of depressive symptoms than studies using clinical diagnostic and incidence of first depressive episode: an 11-year, population-based follow-
instruments. It should also be noted that data were obtained from up study. American Journal of Epidemiology, 163(5), 421e432.
Kobrosly, R., & van Wijngaarden, E. (2010). Associations between immunologic,
a cross-sectional study, thus any causal inference drawn about the
inflammatory, and oxidative stress markers with severity of depressive symp-
relationship between the predictor variables assessed and the toms: an analysis of the 2005e2006 National Health and Nutrition Examination
depressive symptoms must be made with caution. Finally, the Survey. Neurotoxicology, 31(1), 126e133.
variables considered were based on self-reported data. Kohout, F. J., Berkman, L. F., Evans, D. A., & Cornoni-Huntley, J. (1993). Two shorter
forms of the CES-D (Center for Epidemiological Studies Depression) depression
In conclusion, this study indicates that depressive symptoms in symptoms index. Journal of Aging and Health, 5(2), 179e193.
Korean elderly are associated with being female, being unmarried, Korea Labor Institute. (2010). About KLoSA: The Korean longitudinal study of aging.
having less education, lower income, lower BMI, medical comor- Seoul: Korea Labor Institute.
Korean National Statistics Office. (2006). Statistics for the elderly. Department of
bidities and limited physical function. Our finding that depressive Social Welfare Statistics.
symptoms may act as a barrier to screening for chronic disease Kuriyama, S., Koizumi, Y., Matsuda-Ohmori, K., Seki, T., Shimazu, T., Hozawa, A., et al.
which may delay appropriate medical attention, suggests that (2006). Obesity and depressive symptoms in elderly Japanese: the Tsurugaya
project. Journal of Psychosomatic Research, 60(3), 229e235.
effective health care for the elderly with depressive symptoms Lasser, K. E., Zeytinoglu, H., Miller, E., Becker, A. E., Hermann, R. C., & Bor, D. H.
must be further investigated and developed. (2003). Do women who screen positive for mental disorders in primary care
have lower mammography rates? General Hospital Psychiatry, 25(3),
214e216.
Lee, Y., & Shinkai, S. (2005). Correlates of cognitive impairment and depressive
Acknowledgment
symptoms among older adults in Korea and Japan. International Journal of
Geriatric Psychiatry, 20(6), 576e586.
We thank the Korea Labor Institute for providing data and Li, Z. B., Ho, S. Y., Chan, W. M., Ho, K. S., Li, M. P., Leung, G. M., et al. (2004). Obesity
consultations and Marissa Stroo, for editorial assistance. and depressive symptoms in Chinese elderly. International Journal of Geriatric
Psychiatry, 19(1), 68e74.
Luber, M. P., Meyers, B. S., Williams-Russo, P. G., Hollenberg, J. P.,
DiDomenico, T. N., Charlson, M. E., et al. (2001). Depression and service
References utilization in elderly primary care patients. American Journal of Geriatric
Psychiatry, 9(2), 169e176.
Andresen, E. M., Malmgren, J. A., Carter, W. B., & Patrick, D. L. (1994). Screening for Luppino, F. S., de Wit, L. M., Bouvy, P. F., Stijnen, T., Cuijpers, P., Penninx, B. W.,
depression in well older adults: evaluation of a short form of the CES-D (Center et al. (2010). Overweight, obesity, and depression: a systematic review and
for Epidemiologic Studies Depression Scale). American Journal of Preventive meta-analysis of longitudinal studies. Archives of General Psychiatry, 67(3),
Medicine, 10(2), 77e84. 220e229.
Barefoot, J. C., Mortensen, E. L., Helms, M. J., Avlund, K., & Schroll, M. (2001). McCall, N. T., Parks, P., Smith, K., Pope, G., & Griggs, M. (2002). The prevalence of
A longitudinal study of gender differences in depressive symptoms from age 50 major depression or dysthymia among aged Medicare Fee-for-service benefi-
to 80. Psychology and Aging, 16(2), 342e345. ciaries. International Journal of Geriatric Psychiatry, 17(6), 557e565.
Blow, F. C., Serras, A. M., & Barry, K. L. (2007). Late-life depression and alcoholism. Malhotra, R., Chan, A., & Østbye, T. (2010). Prevalence and correlates of clinically
Current Psychiatry Reports, 9(1), 14e19. significant depressive symptoms among elderly people in Sri Lanka: findings
Boey, K. W. (1999). Cross-validation of a short form of the CES-D in Chinese elderly. from a national survey. International Psychogeriatrics, 22(2), 227e236.
International Journal of Geriatric Psychiatry, 14(8), 608e617. Matsubayashi, K., Ho, H. K., Okumiya, K., Wada, T., Ishine, M., Wada, C., et al. (2005).
Cho, M. J., Hahm, B. J., Jhoo, J. H., Bae, J. N., & Kwon, J. S. (1998). Prevalence of Comprehensive geriatric assessment for community-dwelling elderly in Asia
cognitive impairment and depressive symptoms among the elderly in an urban compared with those in Japan: I. Singapore. Geriatrics Gerontology International,
community. Journal of Korean Neuropsychiatric Association, 37(2), 352e362. 5(2), 99e106.
 J.H. Shin et al. / Social Science & Medicine 75 (2012) 179e185 185

Meeks, S., Murrell, S. A., & Mehl, R. C. (2000). Longitudinal relationships between Sakagami, T., Okumiya, K., Ishine, M., Wada, T., Kita, T., Kawakita, T., et al. (2005).
depressive symptoms and health in normal older and middle-aged adults. Comprehensive geriatric assessment for community-dwelling elderly in Asia
Psychology and Aging, 15(1), 100e109. compared with those in Japan: II. Hongchon in Korea. Geriatrics & Gerontology
Minicuci, N., Maggi, S., Pavan, M., Enzi, G., & Crepaldi, G. (2002). Prevalence rate and International, 5(2), 107e114.
correlates of depressive symptoms in older individuals: the Veneto study. The Spielberger, C. D., Ritterband, L. M., Reheiser, E. C., & Brunner, T. M. (2003). The
Journals of Gerontology: Series A, Biological Sciences and Medical Sciences, 57(3), nature and measurement of depression. International Journal of Clinical and
M155eM161. Health Psychology, 3(2), 209e234.
Mojtabai, R., & Olfson, M. (2004). Major depression in community-dwelling middle- Suh, G. H., Choi, I. G., Cho, M. J., Cho, D. Y., Rhoo, I. K., & Yeon, B. K. (1998). Prevalence
aged and older adults: prevalence and 2- and 4-year follow-up symptoms. and risk factors of depressive symptomatology among the Korean elderly.
Psychological Medicine, 34(4), 623e634. Journal of Korean Geriatric Society, 2(1), 49e60.
National Health Insurance Corporation. (2011). Analysis of health screening results. Wada, T., Wada, C., Ishine, M., Okumiya, K., Kawakita, T., Fushida, M., et al. (2005).
Korea. Comprehensive geriatric assessment for community-dwelling elderly in Asia
Peytremann-Bridevaux, I., Voellinger, R., & Santos-Eggimann, B. (2008). Health care compared with those in Japan: V. West Java in Indonesia. Geriatrics Gerontology
and preventive services utilization of elderly Europeans with depressive International, 5(3), 168e175.
symptoms. Journal of Affective Disorders, 105(1e3), 247e252. Wells, K. B., Golding, J. M., & Burnam, M. A. (1989). Chronic medical conditions in
Pirraglia, P. A., Sanyal, P., Singer, D. E., & Ferris, T. G. (2004). Depressive symptom a sample of the general population with anxiety, affective, and substance use
burden as a barrier to screening for breast and cervical cancers. Journal of disorders. American Journal of Psychiatry, 146(11), 1440e1446.
Women’s Health, 13(6), 731e738. WHO/IASO/IOTF. (2000). The Asia-Pacific perspective. Redefining obesity and its
Roberts, R. E., Kaplan, G. A., Shema, S. J., & Strawbridge, W. J. (1997). Does growing treatment. Sydney: Health Communications.
old increase the risk for depression? American Journal of Geriatric Psychiatry, Wilhelm, K., Mitchell, P., Slade, T., Brownhill, S., & Andrews, G. (2003). Prevalence
154(10), 1384e1390. and correlates of DSM-IV major depression in an Australian national survey.
Sachs-Ericsson, N., Burns, A. B., Gordon, K. H., Eckel, L. A., Wonderlich, S. A., Journal of Affective Disorders, 75(2), 155e162.
Crosby, R. D., et al. (2007). Body mass index and depressive symptoms in older Zunzunegui, M. V., Béland, F., Llácer, A., & León, V. (1998). Gender differences in
adults: the moderating roles of race, sex, and socioeconomic status. American depressive symptoms among Spanish elderly. Social Psychiatry and Psychiatric
Journal of Geriatric Psychiatry, 15(9), 815e825. Epidemiology, 33(5), 195e205.