한국의 고령자에서 교육과 인지기능 관련성에 미치는 성별의 조절효과: 한국고령화연구패널조사자료의 분석

The Moderating Effect of Sex on the Association between Education and Cognitive Impairment in Older Koreans: Analysis of Korean Longitudinal Study of Ageing Data

Article information

J Health Info Stat. 2022;47(2):103-110
Publication date (electronic) : 2022 May 31
doi : https://doi.org/10.21032/jhis.2022.47.2.103
Professor, College of Nursing and Health, Kongju National University, Gongju, Korea
송인명orcid_icon
국립공주대학교 간호보건대학 교수
Corresponding author: Inmyung Song 56 Gongjudaehak-ro, Gongju 32588, Korea Tel: +82-41-850-0324, E-mail: inmyungs@gmail.com
No potential conflict of interest relevant to this article was reported.
Received 2022 February 10; Revised 2022 April 26; Accepted 2022 May 31.

Abstract

목적

전통적인 개발국의 연구결과에 따르면, 인지장애의 위험으로부터 보호하는 교육의 효과가 남녀 간 차이가 있었다. 이에 본 연구에서는 과거 교육기회에 대한 접근성에 있어 남녀 간 차별을 겪었던 고령 한국인들을 대상으로 인지기능과 교육수준 간의 관계에 미치는 성별의 조절효과를 조사하고자 한다.

방법

본 연구는 전국적인 조사자료인 2018년 고령화연구패널조사자료를 사용하였다. 인지장애는 Mini-Mental State Exam (MMSE) 점 수 24점 이하로 정의하였다. 단변량분석에서는 MMSE 점수와 각각의 사회경제학적 변수 또는 행위 변수들 간의 관계를 확인하기 위하 여 카이제곱검정을 사용하였다. 공변량을 통제한 상태에서 MMSE 점수와 교육수준 간의 관련성을 확인하기 위해서는 다변량로지스틱 회귀분석을 사용하였으며, 교차비(OR)와 95% 신뢰구간(CI)을 구하였다.

결과

분석에 포함된 총 5,793명의 대상자 중 31.53%은 교육을 전혀 받지 못하였거나 초등교육만 이수하였다. 이처럼 무교육/초등교육을 받은 대상자는 대학 이상의 교육을 받은 사람에 비해서 인지장애를 겪을 위험이 유의하게 높았다(OR=3.31, 95% CI=3.29-3.34, p<0.001). 낮은 교육수준과 인지장애의 관련성은 남성에 비해서 여성에게서 더 높았다. 교차비는 여성에서 4.58 (95% CI= 4.52-4.65) 남성에서 2.98 (95% CI=2.95-3.00)이었다.

결론

낮은 교육수준은 인지장애의 위험도 증가와 관련성이 있었으며 그 관련성이 남성보다는 여성에게서 더 높았다. 전통적으로 교육 기회에 대한 차별을 겪었던 여성고령자에게서 교육의 인지기능 보호효과가 클 가능성을 시사한다.

Trans Abstract

Objectives

In traditionally developed countries, the protective role of higher education on cognitive impairment differed between men and women. This study investigated the moderating impact of sex on the relationship between cognitive function and educational level in older Koreans, who may have experienced inequality in access to educational opportunities in the past.

Methods

This study used data from the 2018 Korean Longitudinal Study on Aging (KLoSA), a nationwide panel survey of community-dwelling older adults. Cognitive impairment was defined as a Mini-Mental State Exam (MMSE) score below 24. In univariate analyses, the χ2 test was used to examine the relationship between MMSE and each of sociodemographic and behavioral variables. Multiple logistic regression models were implemented to examine the association between MMSE score and educational level controlling for covariates. Multivariate-adjusted odds ratios (OR) and 95% confidence intervals (CI) were calculated.

Results

A total of 5,793 respondents were analyzed; 31.53% received no or only primary education. Individuals with up to primary education were more likely than those with college or higher education to have cognitive impairment (OR = 3.31, 95% CI=3.29-3.34, p <0.001). The association between lower educational level and cognitive impairment was stronger for women than for men; OR was 4.58 for women (95% CI=4.52-4.65) and 2.98 for men (95% CI=2.95-3.00).

Conclusion

Lower education was associated with increased risk of cognitive impairment and the protective role of education in cognitive function was stronger in women than in men.

INTRODUCTION

The burden of cognitive impairment in Korea is increasing [1,2]. Dementia and mild cognitive impairment, collectively referred to as cognitive impairment [3], registered the prevalence of 8.1% and 24.1%, respectively, among Korean adults aged 65 years and older in 2008 [1]. The prevalence of dementia alone was estimated to have increased 30 fold in just twelve years from 2003 to 2015 [2]. A number of studies have demonstrated that cognitive function is associated with sociodemographic variables, such as sex [46], age [7,8], and educational level [9], and behavioral characteristics, such as physical exercise [10] and social activity [11]. In particular, the protective role of higher education in the risk for dementia has been extensively studied [12,13]. The negative association between higher education and the risk of cognitive decline was widely documented across countries: in the United States [8,14], in Korea [5,6], and in Europe [7]. Nonetheless, the association between education and a dementia outcome was not uniformly confirmed, according to a systematic review [13], calling for more work in this subject area.

While most studies found cognitive function was independently associated with sex and education, only a few focused on the moderating impact of sex on the relationship between education and cognitive function [15,16]. While these studies found a significant effect of education on the risk for dementia in women but not in men, more research based in Korea is called for. First, the previous studies were based in the traditionally developed countries such as the Netherlands and the United Kingdom. In addition, they relied on constructing separate statistical models for men and women to examine the moderating role of sex in the relationship between cognitive function and education. In this context, it is worth examining whether the impact on education on the risk of cognitive function differs between men and women in the Korean population.

The Korean Longitudinal Study of Ageing (KLoSA) collects Mini-Mental State Exam (MMSE) scores in older Koreans and thus presents a natural opportunity to examine the intriguing role of gender in the relationship between higher education and cognitive function. Women in Korea may have experienced educational disadvantages regardless of their intellectual capacity, while growing up in the past when the country was still developing economically and under the influence of the prevalent patriarchical ideology [17]. The educationally disadvantaged women are very well represented in the panel of KLoSA [18]. Therefore, using the publicly available nationwide survey data, this study aims to investigate the moderating impact of sex on the relationship between MMSE and educational level in older Koreans. It is hypothesized that lower educational attainment is associated with increased risk of cognitive impairment as measured by MMSE and that there is a gender difference in the role of education in decreasing the risk.

METHODS

Data source

This study used data from the KLoSA. The KLoSA is a nationwide longitudinal panel survey of community-dwelling adults who were aged 45 and older at the time of the baseline interview in 2006 [18]. A total of 10,254 individuals were interviewed at the 2006 survey and they were followed-up every two years onward. This study used data from the 7th wave of the survey, which was conducted in 2018; 77.6% (n=6,136) of the baseline panel, who were then aged 57 years and older, responded to the survey. This analysis included individuals who rated their MMSE scores (n=5,793) and did not impute missing data

Outcome measure

Cognitive impairment has been screened by using a number of tools [19]. The MMSE is one of the most commonly used measures to screen cognitive function in the community and clinical settings, and was reported to be reliable [20,21]. The MMSE score ranges from 0 to 30, and lower scores represent greater cognitive declines. Considered a practical measure, MMSE was used to assess cognitive function in the KLoSA since the inception of the survey [22]. In this current study, cognitive impairment was defined as an MMSE score below 24, in accordance with the Cohort Studies of Memory in an International Consortium (COS-MIC) [23].

Explanatory variables

Variables of paramount interest in the present study are sex, education, and MMSE. Alongside these, this study also considered a number of potential covariates of cognitive impairment identified based on previous studies [10,11,24]. These covariates included sociodemographic characteristics, such as age, marital status, employment type, and household income, and behavioral and other characteristics, such as self-rated health, physical exercise, and social activity. Age was grouped into ≤64, 65-74, 75-84, and ≥85 years. Marital status was categorized into four groups: married, separated/divorced, widowed, and single. Respondents were categorized into four levels of education (up to primary education, middle school, high school, and college or higher). Employment type was categorized into three groups (salaried, self-employed, and economically inactive). Household income was grouped into <30 million won and ≥30 million won (per year) at which point all respondents in the sample were nearly equally divided. Self-rated health was measured as five categories (very healthy, healthy, neural, unhealthy, and very healthy) in the KLoSA but was recategorized into three groups (healthy, neutral, and unhealthy) due to the presence of too few responses in extreme categories [25]. Respondents were grouped based on physical exercise of ≥1 time per week. Social activity was defined based on social contact frequency and categorized into ≥1 per week, <1 per week, and none per year, according to the number of times respondents meet with close friends. The number of categories in this analysis was reduced from that in the KLoSA for the sake of ease of interpretation [25].

Statistical analysis

Respondents were described in terms of sociodemographic and behavioral characteristics. The number of respondents was calculated from the sample and population estimates were obtained by adjusting for sampling weights, because weighting is regarded as a standard procedure in handling data from the complex survey design, such as KLoSA, which is exposed to the problem of unequal selection probability [26]. The frequency and percentage of respondents with an MMSE score of less than 24 were calculated by age group, marital status, educational level, employment type, household income, self-rated health, physical activity status, and social contact frequency for men and women. In univariate analyses, the χ2 (chi-square) test was used to examine the relationship between MMSE and each of sociodemographic and behavioral variables.

Multiple logistic regression models were implemented to examine the association between MMSE score and educational level controlling for covariates. Respondents with college or higher education were treated as the reference group. To examine if sex moderates the association between education level and cognitive impairment, the model included the interaction term of sex and educational level. Multivariate-adjusted odds ratios (OR) and 95% confidence interval (CI) were obtained for each independent variable. All statistical analyses were performed by using SAS (SAS Institute Inc., Cary, NC, USA). This study was approved by the Kongju National University Institutional Review Board (IRB No.: KNU_IRB_ 2021-12) and the need for consent was waived by the ethics committee.

RESULTS

A total of 5,793 respondents rated MMSE scores in the 2018 KLoSA; 54.75% were females and 51.85% were aged 65 years and older (Table 1). 31.53% received no or only primary education. The majority of respondents were married (77.13%) and economically inactive (59.60%). Only 36.15% reported they did exercise at least once a week. More than half of the respondent said they met with close friends at least once per week.

Sociodemographic characteristics of subjects

The proportion of respondents with cognitive impairment was greater for women than for men (35.85% vs. 23.95%) (Table 2). The proportion of respondents with cognitive impairment increased with advancing age and decreased with more educational attainment. 36.43% and 53.94% of the widowed men and women, respectively, were cognitively impaired. The proportion of respondents with cognitive impairment was highest in the economically inactive group and in respondents with no social contact. The proportion was higher in respondents with household income of less than 30 million won per year, who rated themselves unhealthy, and doing no exercise than their counterparts. All variables were statistically significant (p <0.001).

Distribution of respondents with cognitive impairment

The results of the multiple logistic regression model without the interaction term show that individuals with up to primary education were more than three times likely to have cognitive impairment than those with at least college education (OR=3.31, 95% CI=3.29-3.34, p <0.001) (Table 3). The odds of having cognitive impairment were higher in the single group than the married (OR=2.54, 95% CI=2.51-2.57, p <0.001) and among economically inactive people than among salaried workers (OR=1.52, 95% CI=1.51-1.53, p <0.001). Having no physical exercise or social contact (per year) was associated with increased risk for cognitive impairment (p <0.001). The multiple regression model with the interaction term shows that individuals with up to primary education were more likely than those with college or higher education to have cognitive impairment and that the association differed between men and women (Table 4). OR was 4.58 for women (95% CI=4.52-4.65) and 2.98 for men (95% CI=2.95-3.00).

Multiple logistic regression model without interaction term

Multiple logistic regression model with interaction term

DISCUSSION

Using nationwide survey data, this present study provides yet more evidence that lower education was associated with increased likelihood of cognitive impairment in both men and women. This general finding is in support of the cognitive reserve hypothesis that higher levels of education build cognitive reserve or brain reserve capacity and attenuate cognitive decline [27]. Added value of this present study lies in the finding of a significant interaction effect between sex and educational level on cognitive function in older Koreans who may have experienced unequal education opportunities between two sexes [17]. The significant interaction term (sex∗educational level) indicates that sex moderates the relationship between educational level and cognitive performance measured by the MMSE scores. In particular, the deleterious impact of lower education on the risk for cognitive impairment was stronger in women than in men, suggesting the possibility that cognitive performance can be enhanced by higher education, especially amongst women. While this possibility should be further examined based on an experimental design, there appears to be a considerable potential for harnessing the protective role of education in cognitive performance in older Koreans, given that a sizable proportion of the study population (31.53%) received no or only primary education.

Increased risk of cognitive impairment associated with low education in Korean women relative to men is somewhat consistent with the findings in other traditionally developed countries; low education was associated with increased risk of dementia in women but not in men [15,16]. However, compared to the previous findings, this current study showed that lower education increased the the risk of cognitive impairment not only in women but also in men; only the magnitude of the impact was stronger in women than in men. The differential impact of low education on risk for cognitive impairment between sexes may reflect the past social environment where educational priority in the household with limited means was given to male children, thereby leaving girls without even primary education [17]. Other than education, the current statistical models did not capture other characteristics that could have influenced gender inequality. Nourishment, occupational development, and social/familial support are other potential areas for inequal access between men and women in economically difficult times; however, this is only speculative and therefore needs to be explored further in the future.

A systematic literature review suggests that additional years of education do not uniformly attenuate the risk for dementia [13]. In contrast to the finding, this current study points to the continually decreased risk for cognitive impairment associated with each additional level of education. In other words, more education always is associated with improved cognitive function. Nonetheless, this current study indicates that the risk for cognitive impairment, as represented by OR, can be lowered by a huge margin by transitioning from no or primary education to middle school education.

The moderating effect of sex on the relationship observed in this current study is consistent with the finding of a study in the United States [23]. However, the U.S. study reports that it is only up to middle school education that improved cognitive function associated with education was stronger for women than for men. These findings suggest that women are likely to reap greater benefits in terms of decreased risk for cognitive impairement than men by providing even the most basic levels of education, since older women in Korea were traditionally at most disadvantages educationally. This statement, however plausible it may sound, needs to be closely examined by using a experimental design in the future to justify a scalable intervention in the population.

In addition to gender differences in the association between education and cognitive impairment, this current study found that a range of sociodemographic and behaviorable variables were associated with cognitive function. Being single or widowed, low household income, economical inactivity, being unhealthy, and lack of physical exercise and social contact all were tied to higher risks for cognitive impairement. These findings are consistent with the existing knowledge base [10,11,24]. While addition of these variables in statistical models does not create new knowledge, it was done so as to decipher the relationship between education and cognitive function by sex with more clarity.

While this current study can expand the existing knowledge by examining the moderating impact of sex on the relationship between education and cognitive impairment based on nationwide survey data in Korean men and women, caution should be exercised in interpreting the results. First, there is a limitation associated with MMSE itself [28]. Altough a widely used screening measure of cognitive impairment, it has a low sensitivity in people with high educational status and a low specificity in people with low socioeconomic status [29,30]. Second, the present study used a cross-sectional design and therefore precludes drawing the causal relationship between independent variables and cognitive impairment. A future study should exploit the longitudinal nature of KLoSA data. Last but not the least, respondents in the KLoSA self-reported on variables such as physical exercise and social contact frequency and therefore the information collated may not reflect actual statuses.

CONCLUSION

This study revealed the potential moderating impact of sex on the relationship between educational attainment and cognitive function in older adults in Korea. Lower education was associated with increased risk of cognitive impairment and the relationship differed between men and women. In particular, the protective role of educational level in decreased risk of cognitive impairment was stronger in women than in men. It is plausile that providing even the most elementary levels of education to older Koreans who may have experienced educational disadvantage in the past could offer benefits in terms of cognitive function. However, this plausibility should be examined by using an experimental design.

References

1. . Kim KW, Park JH, Kim MH, Kim MD, Kim BJ, Kim SK, et al. A nationwide survey on the orevalence of dementia and mild cognitive impairment in South Korea. J Alzheimers Dis 2011;23:281–291. DOI: 10.3233/JAD-2010-101221.
2. . Jang JW, Park JH, Kim S, Lee SH, Lee SH, Kim YJ. Prevalence and incidence of dementia in South Korea: a Nationwide Analysis of the National Health Insurance Service Senior Cohort. J Clin Neurol 2021;17(2):249–256. DOI: 10.3988/jcn.2021.17.2.249.
3. . US Preventive Services Task Force. Screening for cognitive impairment in older adults: US Preventive Services Task Force Recommendation Statement. JAMA 2020;323(8):757–763. DOI: 10.1001/jama.2020.0435.
4. . Juva K, Sulkava R, Verkkoniemi A, Niinistö L. Sex differences in cognitive performance among the very old-Mini-Mental State Examination in a population aged 85 years and over. J Clin Geropsychol 2001;7:39–45. DOI: 10.1023/A:1026412701660.
5. . Han JW, Kim TH, Jhoo JH, Park JH, Kim JL, Ryu SH, et al. A normative study of the Mini-Mental State Examination for Dementia Screening (MMSE-DS) and its short form (SMMSE-DS) in the Korean elderly. J Korean Geriatr Psychiatry 2013;14(1):27–37. (Korean).
6. . Lee DY, Lee KU, Lee JH, Kim KW, Jhoo JH, Youn JC, et al. A normative study of the Mini-Mental State Examination in the Korean elderly. J Korean Neuropsychiatr Assoc 2002;41(3):508–525. (Korean).
7. . Magni E, Binetti A, Rozzini R, Trabucchi M. Mini-Mental State Examination: a normative study in Italian elderly population. Eur J Neurol 1996;3(3):198–202. DOI: 10.1111/j.1468-1331.1996.tb00423.x.
8. . Crum RM, Anthony JC, Bassett SS, Folstein MF. Population-based norms for the Mini-Mental State Examination by age and educational level. JAMA 1993;269(18):2386–2391. DOI: 10.1001/jama.1993. 03500180078038.
9. . Brucki SMD, Nitrini R, Caramelli P, Bertolucci PHF, Okamoto IH. Suggestions for utilization of the Mini-Mental State Examination in Brazil. Arq Neuropsiquiatr 2003;61(3B):777–781. DOI: 10.1590/s0004-282x2003000500014.
10. . Colcombe S, Kramer AF. Fitness effects on the cognitive function of older adults: a meta-analytic study. Psychol Sci 2003;14(2):125–130. DOI: 10.1111/1467-9280.t01-1-01430.
11. . Kim D, Arai H, Kim S. Social activities are associated with cognitive decline in older Koreans. Geriatr Gerontol Int 2017;17(8):1191–1196. DOI: 10.1111/ggi.12861.
12. . Meng X, D' Arcy C. Education and dementia in the context of the cognitive reserve hypothesis: a systematic review with meta-analyses and qualitative analyses. PLoS One 2012;7(6):e38268. DOI: 10.1371/journal.pone.0038268.
13. . Sharp ES, Gatz M. The relationship between education and dementia an updated systematic review. Alzheimer Dis Assoc Disord 2011;25(4):289–304. DOI: 10.1097/WAD.0b013e318211c83c.
14. . Clouston SAP, Smith DM, Mukherjee S, Zhang Y, Hou W, Link BG, et al. Education and cognitive decline: an integrative analysis of global longitudinal studies of cognitive aging. J Gerontol B Psychol Sci Soc Sci 2020;75(7):e151–e160. DOI: 10.1093/geronb/gbz053.
15. . Ott A, van Rossum CT, van Harskamp F, van de Mheen H, Hofman A, Breteler MM. Education and the incidence of dementia in a large population-based study: the Rotterdam Study. Neurology 1999;52(3):663–666. DOI: 10.1212/wnl.52.3.663.
16. . Russ TC, Stamatakis E, Hamer M, Starr JM, Kivimäki M, Batty GD. Socioeconomic status as a risk factor for dementia death: individual participant meta-analysis of 86 508 men and women from the UK. Br J Psychiatry 2013;203(1):10–17. DOI: 10.1192/bjp.bp.112.119479.
17. . Chung JS. Women's unequal access to education in South Korea. Comp Educ Rev 1994;38(4):487–505. DOI: 10.1086/447272.
18. . Kim H, Park SM, Jang SN, Kwon S. Depressive symptoms, chronic medical illness, and health care utilization: findings from the Korean Longitudinal Study of Ageing (KLoSA). Int Psychogeriatr 2011;23(8):1285–1293. DOI: 10.1017/S1041610211000123.
19. . Ismail Z, Rajji TK, Shulman KI. Brief cognitive screening instruments: an update. Int J Geriatr Psychiatry 2010;25(2):111–120. DOI: 10.1002/gps.2306.
20. . Arevalo-Rodriguez I, Smailagic N, Roqué i Figuls M, Ciapponi A, Sanchez-Perez E, Giannakou A, et al. Mini-Mental State Examination (MMSE) for the detection of Alzheimer's disease and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2015;2015(3):CD010783. DOI: 10.1002/14651858.CD010783. pub2.
21. . Cockrell JR, Folstein MF. Mini-Mental State Examination 2nd ed.th ed. West Sussex: John Wiley & Sons, Ltd.; 2002.
22. . Folstein MF, Folstein SE, McHugh PR. “ Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12(3):189–198. DOI: 10.1016/0022-3956(75) 90026-6.
23. . Makkar SR, Lipnicki DM, Crawford JD, Kochan NA, Castro-Costa E, Lima-Costa MF, et al. Education and the moderating roles of age, sex, ethnicity and apolipoprotein epsilon 4 on the risk of cognitive impairment. Arch Gerontol Geriatr 2020;91:104112. DOI: 10.1016/j.archger. 2020.104112.
24. . Subramaniapillai S, Almey A, Natasha Rajah M, Einstein G. Sex and gender differences in cognitive and brain reserve: Implications for Alzheimer's disease in women. Front Neuroendocrinol 2021;60:100879. DOI: 10.1016/j.yfrne.2020.100879.
25. . DiStefano C, Shi D, Morgan GB. Collapsing categories is often more advantageous than modeling sparse data: Investigations in the CFA Framework. Struct Equ Modeling 2021;28(2):237–249. DOI: 10.1080/10705511.2020.1803073.
26. . DL Hahs-Vaughn, CM McWayne, RJ Bulotsky-Shearer, Wen X, Faria AM. Methodological considerations in using complex survey data: an applied example with the head start family and child experiences survey. Eval Rev 2011;35(3):269–303. DOI: 10.1177/0193841X11412071.
27. . Fratiglioni L, Wang HX. Brain reserve hypothesis in dementia. J Alzheimers Dis 2007;12(1):11–22. DOI: 10.3233/jad-2007-12103.
28. . Kang IW, Beom IG, Cho JY, Son HR. Accuracy of Korean-Mini-Mental Status Examination based on Seoul Neuro-Psychological Screening Battery II Results. Korean J Fam Med 2016;37(3):177–181. Korean. DOI: 10.4082/kjfm.2016.37.3.177.
29. . Boo YY, Jutila OE, Cupp MA, Manikam L, Cho SI. The identification of established modifiable mid-life risk factors for cardiovascular disease which contribute to cognitive decline: Korean Longitudinal Study of Aging (KLoSA). Aging Clin Exp Res 2021;33(9):2573–2586. DOI: 10.1007/s40520-020-01783-x.
30. . Nelson A, Fogel BS, Faust D.. Bedside cognitive screening instruments. A critical assessment. J Nerv Ment Dis 1986;174(2):73–83. DOI: 10. 1097/00005053-198602000-00002.

Article information Continued

Table 1.

Sociodemographic characteristics of subjects

Variables Category Respondents (n) % (weighte
Sex Male 2,409 45.25
Female 3,384 54.75
Age (y) ≤64 1,734 48.15
65-74 1,932 31.05
75-84 1,648 17.16
≥85 479 3.64
Marital status Married 4,285 77.13
Separated & Divorced 164 4.17
Widowed 1,308 17.51
Single 36 1.19
Education level Up to primary education 2,481 31.53
Middle school 1,031 18.03
High school 1,695 36.59
College or higher 586 13.86
Employment type Salaried 945 23.35
Self-employed 807 17.05
Economically inactive 4,041 59.60
Household income (million won) <30 3,581 50.77
≥30 2,212 49.23
Self-rated health Healthy 1,525 32.07
Neutral 2,640 45.48
Unhealthy 1,628 22.45
Exercise (≥once per week) Yes 1,920 36.15
No 3,873 63.85
Social contact frequency (per week) ≥1 3,328 54.81
<1 1,903 37.19
None per year 562 8.00

n is the sample frequency and % is the population estimate.

Table 2.

Distribution of respondents with cognitive impairment

Variables Category Men (n) Impaired (n) % p-value Women (n) Impaired (n) % p-value
Total 2,409 577 23.95 <0.001 3,384 1,213 35.84 <0.001
Age (y) ≤64 714 73 10.22 <0.001 1,020 111 10.88 <0.001
65-74 854 146 17.10 1,078 313 29.04
75-84 681 265 38.91 967 538 55.64
≥85 160 93 58.13 319 251 78.68
Marital status Married 2,165 491 22.68 <0.001 2,120 558 26.32 <0.001
Separated & Divorced 82 24 29.27 82 20 24.39
Widowed 140 51 36.43 1,168 630 53.94
Single 22 11 50.00 14 5 35.71
Education level Up to primary school 642 279 43.46 <0.001 1,839 976 53.07 <0.001
Middle school 442 110 24.89 589 130 22.07
High school 897 142 15.83 798 95 11.90
College or higher 428 46 10.75 158 12 7.59
Employment type Salaried 508 66 12.99 <0.001 437 63 14.42 <0.001
Self-employed 585 86 14.70 222 49 22.07
Economically inactive 1,316 425 32.29 2,725 1,101 40.40
Household income (million won) <30 1,377 426 30.94 <0.001 2,204 946 42.92 <0.001
≥30 1,032 151 14.63 1,180 267 22.63
Self-rated health Healthy 730 81 11.10 <0.001 795 133 16.73 <0.001
Neutral 1,134 239 21.08 1,506 432 28.69
Unhealthy 545 257 47.16 1,083 648 59.83
Exercise (≥once per week) Yes 909 148 16.28 <0.001 1,011 227 22.45 <0.001
No 1,500 429 28.60 2,373 986 41.55
Social contact frequency (per week) ≥1 1,258 260 20.67 <0.001 2,070 690 33.33 <0.001
<1 910 179 19.67 993 309 31.12
None per year 241 138 57.26 321 214 66.67

p-values were obtained by using the χ2 test.

Table 3.

Multiple logistic regression model without interaction term

Variables Category OR 95% CI p-value
Lower limit Upper limit
Sex (ref.: male) Female 1.05 1.04 1.05 <0.001
Age (y) (ref.: ≤64) 65-74 1.58 1.57 1.58 <0.001
75-84 3.38 3.36 3.39 <0.001
≥85 6.36 6.31 6.42 <0.001
Marital status (ref.: married) Separated & Divorced 1.06 1.05 1.07 <0.001
Widowed 1.17 1.17 1.18 <0.001
Single 2.54 2.51 2.57 <0.001
Education level (ref.: ≥college) Up to primary school 3.31 3.29 3.34 <0.001
Middle school 1.92 1.91 1.93 <0.001
High school 1.24 1.24 1.25 <0.001
Employment type (ref.: salaried) Self-employed 1.00 0.99 1.00 0.095
Economically inactive 1.52 1.51 1.53 <0.001
Household income (million won) (ref.: ≥30) <30 1.23 1.23 1.24 <0.001
Self-rated health (ref.: healthy) Neutral 1.41 1.40 1.41 <0.001
Unhealthy 2.75 2.74 2.76 <0.001
Physical exercise (ref.: ≥once per week) No 1.59 1.59 1.60 <0.001
Social contact (per week) (ref.: ≥1) <1 1.39 1.39 1.40 <0.001
None 2.29 2.28 2.30 <0.001
Observations 5,793
−2 Log L 10,294,627

ref, reference; OR, odds ratio; CI, confidence interval.

Table 4.

Multiple logistic regression model with interaction term

Variables Category OR 95% CI p-value
Lower limit Upper limit
Age (y) (ref.: 57-64) 65-74 1.58 1.57 1.58 <0.001
75-84 3.38 3.36 3.39 <0.001
≥85 6.36 6.31 6.41 <0.001
Marital status (ref.: married) Separated & Divorced 1.06 1.05 1.07 <0.001
Widowed 1.17 1.16 1.17 <0.001
Single 2.57 2.54 2.60 <0.001
Employment type (ref.: salaried) Self-employed 1.00 0.99 1.00 0.3263
Economically inactive 1.52 1.51 1.52 <0.001
Household income (million won) (ref.:≥30) <30 1.23 1.23 1.24 <0.001
Self-rated health (ref.: healthy) Neutral 1.40 1.40 1.41 <0.001
Unhealthy 2.75 2.73 2.76 <0.001
Physical exercise (per week) (ref.: ≥once) No 1.59 1.59 1.60 <0.001
Social contact frequency (per week) (ref.: ≥1) <1 1.39 1.39 1.40 <0.001
None 2.29 2.28 2.30 <0.001
Education∗sex (ref.: ≥college) ≤Primary school∗male 2.98 2.95 3.00 <0.001
Middle school∗male 1.71 1.70 1.73 <0.001
High school∗male 1.13 1.12 1.14 <0.001
≤Primary school∗female 4.58 4.52 4.65 <0.001
Middle school∗female 2.68 2.65 2.72 <0.001
High school∗female 1.72 1.70 1.75 <0.001
Observations 5,793
−2 Log L 10,291,136

ref, reference; OR, odds ratio; CI, confidence interval.