Journal of Psychosomatic Research 69 (2010) 435–440

Original articles

Social determinants of self-reported sleep problems inSouth Korea and Taiwan☆,☆☆

Kyoko Nomuraa,⁎, Kazue Yamaokab, Mutsuhiro Nakaoa, Eiji Yanoa

aDepartment of Hygiene and Public Health, Teikyo University School of Medicine, Itabashi-ku, Tokyo, JapanbDepartment of Technology Assessment and Biostatistics, National Institute of Public Health, Wako, Saitama-ken, Japan

Received 12 February 2009; received in revised form 22 April 2010; accepted 27 April 2010

Abstract

Objective: To clarify the social determinants of insomnia inSouth Korea and Taiwan. Methods: Cross-sectional surveys wereconducted in South Korea (n=1007) and Taiwan (n=785) in 2003.Nationwide samples of people completed a structured question-naire, based on face-to-face interviews. Outcome measures wereself-reported sleep problems, defined by at least one of three sleepsymptoms on a nightly basis for more than 2 weeks: difficultyinitiating sleep (DIS), difficulty maintaining sleep (DMS), andearly morning waking (EMW). Explanatory variables investigatedwere demographic characteristics (gender, age), socioeconomicfactors (income, education), and social capital, including norms ofreciprocity, interpersonal trust, civic association, and social support(i.e., the availability of persons to consult regarding personalproblems and important matters). Results: The prevalence of sleep

☆ Competing interests: The researchers have no financial conflictsof interest.

☆☆ Funding: This study was supported by the Ministry of Education,Culture, Sports, Science and Technology through a Grant in ScientificResearch B(2) no. 14402043 (fiscal years 2002–2004) and no. 18402001(fiscal years 2006–2008).

⁎ Corresponding author. Department of Hygiene and Public Health,Teikyo University School of Medicine, Kaga 2-11-1, Itabashi-ku, Tokyo173-8605, Japan. Tel.: +81 3 964 1211; fax: +81 3 964 1058.

E-mail address: DZB07241@nifty.ne.jp (K. Nomura).

0022-3999/10/$ – see front matter © 2010 Elsevier Inc. All rights reserved.doi:10.1016/j.jpsychores.2010.04.014

problems was 8.2% in South Korea and 9.3% in Taiwan. Stepwiselogistic regression showed that, in South Korea, sleep problemswere significantly associated with increasing age (Pb.05), lowincome (Pb.01), and having few persons with whom to consultcompared to having a spouse/partner (Pb.05); in Taiwan, sleepproblems were significantly associated with being female (Pb.05),increasing age (Pb.001), and having family members vs. a spouse/partner to consult (Pb.05). Conclusion: Self-reported sleepproblems in two northeast Asian countries were associated withcertain demographic characteristics and socioeconomic factors,which is consistent with previous results in Western countries. Inaddition, the results of this study suggested that sleep problemsmay also be associated with social support.© 2010 Elsevier Inc. All rights reserved.

Keywords: Sleep problems; Social capital; Sociodemographic characteristics; South Korea; Taiwan

Introduction

A single-day survey performed in 10 industrializedcountries estimated that one in four individuals does notsleep well [1]. Insomnia, one of the most frequently reportedhealth problems in the general population, has drawn publicattention and, in response to public concern, many scientific

studies have been conducted on the etiology of insomnia.Previous reports generally agree that psychological disorders[2] and physiological illnesses, including obstructive sleepapnea syndrome [3], obesity [4], and cardiovascularmalfunction [5], are closely related to insomnia.

In addition, several epidemiological studies have sug-gested that insomnia is associated with social factorsincluding ethnicity [6], gender (i.e., being female), age (i.e.,older age) [7–12], and socioeconomic status (i.e., low incomeor low educational attainment) [13,14]. However, themajority of these studies were conducted in Westerncountries and appropriate empirical evidence is lacking forAsia. Furthermore, the majority of these epidemiologicalstudies also failed to demonstrate the mechanisms underlyingthe reported associations. Nevertheless, two recent studieshave suggested the mechanism by which socioeconomicstatus affects sleep problems. Moore et al. [15] reported that

436 K. Nomura et al. / Journal of Psychosomatic Research 69 (2010) 435–440

higher education was associated with higher income and thathigher income was associated with better physical andpsychological health. Van Cauter and Spiegel [16] reportedthat low socioeconomic status was associated with dimin-ished opportunities to obtain sufficient sleep or withenvironmental conditions that compromise sleep quality.

Likewise, in consideration of the relationship betweensleep problems and social factors, we focused particularly onsocial capital as one of the factors mediating sleep problems.Social capital refers to social connections within andbetween human networks, and many epidemiological studiesconducted during the past few decades have suggestedthat this factor positively affects both health and well-being[17–22]. Social capital operates through civil society, whichconsists of “private organizations that are formed andsustained by groups of people acting voluntarily and withoutseeking personal profit to provide benefits for themselves orfor others” [21,23,24]. These voluntary associations alsoconnect people with one another, build trust and reciprocity,and consolidate society on the basis of a sense of altruismrather than obligation. We assumed that people becomesocially isolated under circumstances in which social capitalis less available and that such individuals are at increasedrisk for insomnia because of their limited access to resourcessuch as medical care, information, and emotional support.

Thus, the purpose of this study was twofold: (1) toinvestigate the social determinants of sleep problems intwo northeastern Asian countries, South Korea and Taiwan;and (2) to clarify whether social capital is associated withsleep problems.

Methods

Population and samples

Datawere collected from cross-sectional surveys conductedin South Korea and Taiwan in 2003 as part of “A Study ofStatistical Science on Health and Culture.” These data havebeen reported in part elsewhere [25], but here we investigate anew hypothesis using data that were not previously examined.The target populationswere adults (aged≥20years) residing ineach country. The survey involved a face-to-face interview bywell-trained interviewers using a semistructured questionnairethatwas specifically developed for the study and a standardizedanswer sheet to record responses. We used a multistageprobability sampling method, but, given the differences insurvey circumstances between the two countries, the sample-selection process varied slightly. First, primary sampling units(large administrative units), stratified by population size anddistrict based on official census data from 2000, weredetermined. Then, in South Korea, through several stages ofsampling-unit selection, a household (final sampling unit) wasrandomly selected and a respondent was selected from eachhousehold using the birthday rule. The birthday rule, whichselects a respondent based on the first birthday to occur within

the coming 12 months following the interview day, wasspecifically designed to reduce sampling bias. This processcontinued until the desired number of samples was obtained(quota sampling or random route sampling); thus, an exactresponse rate cannot be calculated for South Korea. In Taiwan,households (final sampling unit) were chosen from amongprimary sampling units by random sampling using a telephonecode book, which covers N90% of all households, and arespondent was selected from each household using the Kish[26] method. This method allows a random selection thatmaintains a representative sample in terms of age and genderand limits noncoverage error. All surveys were conducted afterthe subjects had given informed consent.

Measures

Self-reported sleep problemsRespondents were queried regarding three general

symptoms using the following questions: “Do you havedifficulty falling asleep at night?” [difficulty initiating sleep(DIS)], “Do you wake up during the night after you havegone to sleep?” [difficulty maintaining sleep (DMS)], and“Do you wake up too early in the morning and havedifficulty getting back to sleep?” [early morning waking(EMW)]. Self-reported sleep problems were noted when atleast one of the three symptoms was experienced. For eachitem, the respondents were asked to indicate whether thesymptoms occurred “on a regular, nightly basis experiencedfor more than two weeks,” “sometimes,” or “seldom ornever.” “On a regular, nightly basis” was taken as anaffirmative answer with regard to sleep problems.

Demographic characteristics and socioeconomic factors

Variables analyzed included gender and age as demo-graphic characteristics, and educational level and self-ratedhousehold income as socioeconomic factors. Educationallevel was divided into three categories: below high school(b12 years of education), high school (12–15 years), andcollege and above (≥ 16 years). Self-rated household incomewas divided into four categories: low, intermediate, high, anddo not know.

Social capital

We used a framework in which social capital wasmeasured based on cognitive or structural components.The cognitive dimension subsumes attitudinal manifesta-tions, such as sense of trust and the availability of persons toconsult in difficult situations, whereas the structuraldimension encompasses behavioral manifestations andparticipation in formal associations.

Sense of trust, a measure of cognitive social capital, wasassessed by two questions posed in the General SocialSurvey established by the National Opinions ResearchCenter (Chicago, IL, USA) [27]. Interpersonal trust was

437K. Nomura et al. / Journal of Psychosomatic Research 69 (2010) 435–440

assessed by asking, “Generally speaking, would you say thatmost people can be trusted, or that you cannot be too carefulin dealing with people?” The respondents then selected “canbe trusted” or “can't be too careful.” Norms of reciprocitywere assessed by asking, “Would you say that most of thetime people try to be helpful (i.e., the collective perception ofreciprocity), or are they mostly looking out for themselves?”The respondents then selected “try to be helpful” or “look outfor themselves.”

Person to consult was another measure of cognitive socialcapital. Participants were asked, “Which of the followingpeople would you first consult regarding your personalproblems and important matters?” “Problems” meanttroublesome or worrisome issues, such as health, money,interpersonal relations, etc. Response alternatives were “aspouse/partner,” “family members exclusive of a spouse/partner,” or “others/none.” “A spouse/partner” indicated the

Table 1Basic characteristics of subjects in relation to self-reported sleep symptoms

Variables

South Korea (n=1007)

Total(N)

Insomnia(%) a

Three sleep symptoms

DIS (%) DMS (%)

Sociodemographic characteristicsGenderFemale 507 9.1 5.8 6.3Male 500 7.8 5.1 3.8

Age50–100 260 15.8 10.5 10.435–49 440 4.6 3.2 2.720–34 307 7.8 4.3 3.9

Household income b

Low 278 17.3 12.2 11.9Middle 276 5.4 3.3 3.3High 418 4.8 2.4 1.9

Education b

Below high school 226 17.7 12.8 12.0High school 437 5.5 3.7 3.0College–university 343 6.1 2.6 3.2

Social capitalCognitive dimensionInterpersonal trust b

Careful with others 675 8.6 5.3 4.3Trust others 286 8.0 5.9 6.6Norms of reciprocity b

Think of myself only 422 8.8 6.4 4.7Think of others 521 8.1 4.8 5.0Person to consult b

Others/none 99 16.1 8.3 10.4Family membersexclusive of partner

314 8.0 5.5 4.9

A spouse/partner 594 7.4 4.9 4.4Structural dimensionCivic associationDoes not participate 330 10.0 5.8 6.4Participates 677 7.7 5.2 4.5

a Self-reported sleep problem is defined as having at least one of the three smaintaining sleep), and EMW (i.e., early morning waking).

b “Not known” (n); income: South Korea (35), Taiwan (43); education: South Ktrust: South Korea (46), Taiwan (13); person to consult: South Korea (153), Taiw

strongest social support, followed by “family membersexclusive of a spouse/partner,” and “others/none” indicatedpoor social support.

Membership in organizations, a measure of structuralsocial capital, was assessed using a question proposed byKawachi et al. [28] and was based on membership in a widevariety of voluntary associations, including sports groups,hobby groups, literature groups, alumni associations, churchgroups, and political organizations. Civic associations was adichotomous variable in the analysis (1=involvement in oneor more, 0=none).

Statistical analysis

The representative nature of the survey sample wasevaluated by statistical comparison with the national samplein each country. No significant difference was observed

Taiwan (n=785)

Total(N)

Insomnia(%) a

Three sleep symptoms

EMW (%) DIS (%) DMS (%) EMW (%)

4.7 453 11.0 8.4 4.6 4.94.0 332 6.9 4.5 2.1 2.4

12.3 230 17.8 13.0 7.8 7.80.9 324 6.5 4.9 2.2 2.52.6 231 4.8 3.0 1.3 1.7

12.6 313 13.1 10.2 5.8 6.40.0 346 6.7 4.3 2.6 2.32.2 83 8.4 4.8 1.2 2.4

12.4 343 13.4 10.5 5.8 5.82.3 259 7.3 5.0 2.3 2.71.8 183 4.4 2.2 1.1 1.6

4.7 646 8.4 5.7 2.9 3.33.9 126 13.5 11.1 6.4 6.4

5.2 453 8.8 6.0 2.7 3.33.8 292 10.3 8.6 5.1 4.8

7.3 141 8.5 5.7 1.4 3.64.2 267 12.7 9.0 5.6 4.1

4.1 377 7.2 5.6 2.9 3.7

4.7 580 9.1 6.2 3.6 3.64.3 205 9.8 8.3 3.4 4.4

leep symptoms of DIS (i.e., difficulty initiating sleep), DMS (i.e., difficult

orea (1); norms of reciprocity: South Korea (64), Taiwan (40); interpersonalan (24); neighborhood relationship: South Korea (3), Taiwan (1).

438 K. Nomura et al. / Journal of Psychosomatic Research 69 (2010) 435–440

between census and sample data with regard to gender or agedistribution. The reliability of the questionnaire used in thetwo countries was statistically acceptable; the translation andback-translation method was used [25].

The relationships between self-reported sleep problemsand demographic characteristics, socioeconomic factors, andsocial capital were summarized for each country. We used alogistic regression model to compute both crude and adjustedodds ratios (ORs), with 95% confidence intervals (CIs), forself-perceived insomnia. The variables used in the multivar-iate models were gender, age, and those selected using astepwise method (inclusion and exclusion criteria=0.3 foreach). Age was treated as a continuous independent variablein the logistic regression model, as the outcome of self-perceived insomnia appeared to increase with age. ORsreflected an increase in the odds of self-perceived insomniaper 10-year increase in age. P values for income, educationallevel, and person to consult were calculated to determine thesignificance of each item. In addition, the relationshipbetween self-reported sleep problems and interactionsbetween gender or social capital and the remainingexplanatory variables was assessed. The analyses wereconducted using SAS software (version 8.12 for Windows).All tests were two sided, with a significance level of Pb.05,and Wald chi-square tests were used.

Table 2Odds ratios (95% confidence interval) of self-reported sleep problems a

Variables Reference

South Ko

Univariat

Sociodemographic characteristicsGenderFemale Male 1.2 (0.8–Age 1.5 (1.3–Household income c, d ⁎⁎⁎

Low High 4.2 (2.4–Middle 1.1 (0.6–Education c ⁎⁎⁎

Below high school College–university 3.3 (1.9–High school 0.9 (0.3–

Social capitalCognitive dimensionInterpersonal trust d

Careful with others Trust others 1.1 (0.7–Norms of reciprocity d

Think of myself only Think of others 1.2 (0.8–Person to consult c, d ⁎⁎

Others/none A spouse/partner 2.6 (1.3–Family members exclusive of partner A spouse/partner 1.1 (0.6–

Structural dimensionCivic associationDoes not participate Participates 1.5 (0.7–

†Pb.1, ⁎Pb.05, ⁎⁎Pb.01, ⁎⁎⁎Pb.001, based on the Wald chi-square.–, Indicates items excluded from the finals.

a Self-reported sleep problem is defined as having at least one of the three slb Based on stepwise logistic regression model.c P values in “income, ” “education,” and “person to consult” were the signifd “Not known” was excluded.

Results

The number of valid questionnaires returned was 1007 inSouth Korea (complete sample) and 785 in Taiwan (responserate=44%). Table 1 shows the basic characteristics ofsubjects in relation to self-reported sleep symptoms. Theprevalence of self-reported sleep problems was 8.4% (5.4%for DIS, 5.1% for DMS, and 4.4% for EMW) in SouthKorea, compared to 9.3% (6.8% for DIS, 3.6% for DMS, and3.8% for EMW) in Taiwan. Regardless of country, peoplewho were female, older and with lower income oreducational level were more likely to report sleep problems.In both countries, subjects who did not participate in civicassociations tended to report more sleep problems. Com-pared to subjects who relied on a spouse/partner when theyneeded someone to consult regarding personal problems andimportant matters, participants who indicated “others/none”tended to have a higher prevalence of sleep symptoms inSouth Korea, whereas participants who consulted otherfamily members tended to have a higher prevalence of sleepproblems in Taiwan.

Table 2 shows the ORs and 95% CIs for self-reportedsleep problems calculated by logistic regression models.Univariate models showed that significant variables associ-ated with sleep problems included increasing age, low

rea Taiwan

e Multivariate b Univariate Multivariate b

1.9) 1.2 (0.7–2.0) 1.7 (1.0–2.8)† 1.9 (1.1–3.3)⁎

1.8)⁎⁎⁎ 1.2 (1.0–1.5)⁎ 1.6 (1.4–1.9)⁎⁎⁎ 1.6 (1.4–1.9)⁎⁎⁎⁎⁎ ⁎

7.2) 3.2 (1.7–5.9) 1.6 (0.7–3.8) –2.3) 1.1 (0.6–2.3) 0.7 (0.3–1.9) –

⁎⁎

5.8) – 3.4 (1.6–7.3) –1.6) – 1.7 (0.7–4.1)

1.9) – 0.6 (0.3–1.1)† 0.6 (0.3–1.2)

2.0) – 0.9 (0.5–1.4) 1.3 (0.8–2.4)⁎ ⁎ ⁎

5.0) 2.3 (1.2–4.4) 1.3 (0.6–2.7) 1.3 (0.6–2.8)1.9) 1.1 (0.6–2.0) 2.0 (1.1–3.4) 2.3 (1.2–4.0)

2.9) – 1.1 (0.6–2.0) –

eep symptoms of DIS, DMS, and EMW.

icance of the item.

439K. Nomura et al. / Journal of Psychosomatic Research 69 (2010) 435–440

income, low educational level, and selecting “others/none”as persons one would consult in South Korea, and increasingage, low income, low educational level, and selecting“family members exclusive of partner” as persons onewould consult in Taiwan. Finally, stepwise multivariatelogistic regression models showed that, in South Korea,significant variables associated with sleep problems includedincreasing age, low income, and persons who had “others/none” to consult (compared to persons having “a spouse/partner” to consult). In Taiwan, female gender, increasingage, and persons who had “family members exclusive ofpartner” to consult (compared with persons having “aspouse/partner” to consult) were more likely to experiencesleep problems. No interaction was observed between genderor social capital and the remaining explanatory variables.

Discussion

This study sought to assess the social determinants ofself-reported sleep problems in South Korea and Taiwan.Sleep problems in these two northeastern Asian countrieswere associated with demographic characteristics, socio-economic factors, and social support. We discuss theresults of the present study in light of its strengths andlimitations below.

A meta-analysis showed that women were more likely toreport insomnia than were men [7,6], but the majority of theindividual studies included in the meta-analysis wereconducted in Western countries and only four wereconducted in Asia (i.e., two studies in Hong Kong, one inSouth Korea [14] and one in Japan). We reviewed this study[14] and one other study conducted in South Korea [12] thatwe retrieved by literature search and found that their results[12,14] were inconsistent with respect to the effect of genderon sleep problems. Ohayon and Hong [14] reported nodifference in the prevalence of sleep problems (i.e., DIS,EMA, and nonrestorative sleep) between men and women,whereas Cho et al. [12] reported a higher prevalence of sleepproblems in women than in men. A higher prevalence inwomen than in men was also observed in our study;however, after adjusting for potential confounders, femalegender was not found to be a significant risk factor for sleepproblems in South Korea. The discrepancy among thesethree studies, including ours, may be explained bydifferences in definitions: Ohayon and Hong [14] definedinsomnia as at least three nights per week of sleeplessness,whereas Cho et al. [12] defined insomnia in terms of thepresence of symptoms (i.e., yes or no). In fact, our definition(i.e., “on a regular, nightly basis experienced for more thantwo weeks.”) was stricter than that used in Ohayon andHong's [14] study. Indeed, our additional analyses definingsleep problems in terms of “sometimes” showed a significantgender effect in South Korea. Thus, the lack of a significantgender effect in South Korea in this study may be a result ofour strict definition of sleep problems.

Low household income and low educational level weresignificantly associated with sleep problems, a finding thatagrees with the results of previous health inequality studies[13,14]. However, the results of multivariate analysesshowed that household income was significant in SouthKorea, but not in Taiwan. A national study in Taiwanconducted in 2001 [9] failed to show a statisticallysignificant association between income and insomnia. Theauthors attributed the nonsignificance of the result to the higheconomic growth and relatively fair distribution of income inTaiwan. Thus, despite the limited number of previousreports, the effect of household income may need to becaptured in the context of the economic situation of thewhole country, requiring careful interpretation of our results.

Our study showed that the structural (i.e., civicassociation) and cognitive (i.e., sense of trust) componentsof social capital were not significantly related to insomnia,although social support showed a positive relationship.Thus, the effect of social support may not truly reflect“social capital,” as conceptualized in Western studies.Nevertheless, before making such an interpretation, wemust discuss two important points. First, some might arguethat the nonsignificance of social capital may be due to thecultural characteristics of the study samples. However, ourprevious study in five East Asian countries (Japan, SouthKorea, Singapore, five areas of mainland China, andTaiwan) [20], which used part of the same data setdescribed here, demonstrated that social capital wasgenerally positively associated with overall well-being.Second, although previous reports were limited to mentalhealth, most studies using aggregated data or multilevelmodels failed to find statistically significant associationsbetween social capital and common mental illness at higherlevels of analysis such as neighborhoods [29–31]. Hence,our results are consistent with previous results [29–31] and,therefore, do not indicate that social capital influences theprevalence of sleep problems.

Before these results can be generalized, several limita-tions of our study should be considered. First, because thestudy was cross-sectional, we could not determine any causalrelationship between self-reported sleep problems andidentified social factors. Second, the range of variables thatcould be analyzed was limited. For example, information onbaseline depression, anxiety, and use of psychiatric drugswas not available. Third, sampling methods differed betweenthe two countries, which may have created bias. We usedrandom sampling methods, but people who were not residingin the household at the time of the investigation may havebeen excluded. The differences in the response rates betweenthe two countries may have also contributed to bias.Nevertheless, we confirmed that our samples in the twocountries did not differ statistically from each country'snational data. Fourth, because this study was based on aquestionnaire incorporated into a face-to-face interview,translation accuracy could have affected the responses. Tominimize potential errors, we validated our questionnaire by

440 K. Nomura et al. / Journal of Psychosomatic Research 69 (2010) 435–440

back translation and confirmed the stability of the data.Finally, recent research has demonstrated that sleep durationand quality, which have consequences for health, arestrongly associated with race, gender, and socioeconomicstatus [32]. In addition, because many studies have suggestedthat sleep duration is statistically associated with morbidityas well as mortality [33,34], future studies should includeboth sleep quality and sleep quantity as variables.

In conclusion, despite limited evidence, self-reportedsleep problems in two northeast Asian countries wereassociated with demographic characteristics and socioeco-nomic factors, which is consistent with previous results inWestern countries. In addition, the results of this studysuggested that sleep problems may also be associated withsocial support (i.e., the availability of persons to consultregarding personal problems and important matters).

References

[1] Soldatos CR, Allaert FA, Ohta T, Dikeos DG. How do individualssleep around the world? Results from a single-day survey in tencountries. Sleep Med 2005;6:5–13.

[2] Roth T, Roehrs T. Insomnia: Epidemiology, characteristics, andconsequences. Clin Cornerstone 2003;5:5–15.

[3] Nieto FJ, Young TB, Lind BK, Shahar E, Samet JM, Redline S, et al.Association of sleep-disordered breathing, sleep apnea, and hyperten-sion in a large community-based study: Sleep Heart Health Study.JAMA 2000;283:1829–36.

[4] Tishler PV, Larkin EK, Schluchter MD, Redline S. Incidence of sleep-disordered breathing in an urban adult population: The relativeimportance of risk factors in the development of sleep-disorderedbreathing. JAMA 2003;289:2230–7.

[5] Choi JB, Nelesen R, Loredo JS, Mills PJ, Ancoli-Israel S, Ziegler MG,et al. Sleepiness in obstructive sleep apnea: A harbinger of impairedcardiac function? Sleep 2006;29:1531–6.

[6] Sanford SD, Lichstein KL, Durrence HH, Riedel BW, Taylor DJ, BushAJ. The influence of age, gender, ethnicity, and insomnia on EpworthSleepiness Scores: A normative US population. Sleep Med 2006;7:319–26.

[7] Zhang B, Wing YK. Sex differences in insomnia: A meta-analysis.Sleep 2006;29:85–93.

[8] OhayonMM, Zulley J, Guilleminault C, Smirne S, Priest RG. How ageand daytime activities are related to insomnia in the general population:Consequences for older people. J Am Geriatr Soc 2001;49:360–6.

[9] Chen YY, Kawachi I, Subramanian SV, Acevedo-Garcia D, Lee YJ.Can social factors explain sex differences in insomnia? Findings from anational survey in Taiwan. J Epidemiol Community Health 2005;59:488–94.

[10] Roberts RE, Shema SJ, Kaplan GA, StrawbridgeWJ. Sleep complaintsand depression in an aging cohort: A prospective perspective. Am JPsychiatry 2000;157:81–8.

[11] Su TP, Huang SR, Chou P. Prevalence and risk factors of insomnia incommunity-dwelling Chinese elderly: A Taiwanese urban area survey.Aust N Z J Psychiatry 2004;38:706–13.

[12] Cho YW, Shin WC, Yun CH, Hong SB, Kim J, Earley CJ.Epidemiology of insomnia in Korean adults: Prevalence and associatedfactors. J Clin Neurol 2009;5:20–3.

[13] Sutton DA, Moldofsky H, Badley EM. Insomnia and health problemsin Canadians. Sleep 2001;24:665–70.

[14] Ohayon MM, Hong SC. Prevalence of insomnia and associated factorsin South Korea. J Psychosom Res 2002;53:593–600.

[15] Moore PJ, Adler NE, Williams DR, Jackson JS. Socioeconomic statusand health: The role of sleep. Psychosom Med 2002;64:337–44.

[16] Van Cauter E, Spiegel K. In: Adler NE, et al, editor. Sleep as amediator of the relationship between socioeconomic status and health:A hypothesis. Socioeconomic status and health in industrial nations:Social, psychological, and biological pathways. New York: New YorkAcademy of Sciences, 1999.

[17] Kaplan GA, Pamuk ER, Lynch JW, Cohen RD, Balfour JL. Inequalityin income and mortality in the United States: Analysis of mortality andpotential pathways. BMJ 1996;312:999–1003.

[18] Kawachi I, Kim D, Coutts A, Subramanian SV. Commentary:Reconciling the three accounts of social capital. Int J Epidemiol2004;33:682–90 discussion 700-4.

[19] Bain K, Hicks N. Building social capital and reaching out to excludedgroups: The challenge of partnerships. Paper presented at the CELAMmeeting on The Struggle Against Poverty Toward the Turn of theMillennium, Washington, DC, 1998.

[20] Yamaoka K. Social capital and health and well-being in East Asia: Apopulation-based study. Soc Sci Med 2007;66:885–99.

[21] Yip W, Subramanian SV, Mitchell AD, Lee DT, Wang J, Kawachi I.Does social capital enhance health and well-being? Evidence fromrural China. Soc Sci Med 2007;64:35–49.

[22] Fone D, Dunstan F, Lloyd K, Williams G, Watkins J, Palmer S. Doessocial cohesion modify the association between area incomedeprivation and mental health? A multilevel analysis. Int J Epidemiol2007;36:338–45.

[23] Berkman LF, Glass T. Social integration, social networks, socialsupport, and health. In: Berkman LF, Kawachi I, editors. Socialepidemiology. New York: Oxford University Press, 2000. p. 137–73.

[24] Putnam RD. Making democracy work: Civic traditions in modernItaly. Princeton: Princeton University Press, 1993.

[25] Nomura K, Yamaoka K, Nakao M, Yano E. Impact of insomnia onindividual health dissatisfaction in Japan, South Korea, and Taiwan.Sleep 2005;28:1328–32.

[26] Kish L. Survey sampling. New York: Wiley, 1965.[27] Survey Cumulative File, Ann Arbor, Michigan: University of

Michigan, Interuniversity Consortium for Political and SocialResearch. 1972 through 1993.

[28] Kawachi I, Kennedy BP, Lochner K, Prothrow-Stith D. Socialcapital, income inequality, and mortality. Am J Public Health 1997;87:1491–8.

[29] Cutrona CE, Russell DW, Hessling RM, Brown PA, Murry V. Directand moderating effects of community context on the psychologicalwell-being of African American women. J Pers Soc Psychol 2000;79:1088–101.

[30] Skapinakis P, Lewis G, Araya R, Jones K, Williams G. Mental healthinequalities in Wales, UK: Multi-level investigation of the effect ofarea deprivation. Br J Psychiatry 2005;186:417–22.

[31] Araya R, Dunstan F, Playle R, Thomas H, Palmer S, Lewis G.Perceptions of social capital and the built environment and mentalhealth. Soc Sci Med 2006;62:3072–83.

[32] Lauderdale DS, Knutson KL, Yan LL, Rathouz PJ, Hulley SB, Sidney S,et al. Objectively measured sleep characteristics among early-middle-aged adults: The CARDIA study. Am J Epidemiol 2006;164:5–16 Epub2006 Jun 1.

[33] Tamakoshi A, Ohno Y. Self-reported sleep duration as a predictor ofall-cause mortality: Results from the JACC study, Japan. Sleep 2004;27:51–4.

[34] Ayas NT,White DP,Manson JE, StampferMJ, Speizer FE,Malhotra A,et al. A prospective study of sleep duration and coronary heart disease inwomen. Arch Intern Med 2003;163:205–9.