Association Between Obesity and Education Level Among the Elderly in Taipei, Taiwan

Global population ageing is accelerating, with the UN and WHO defining ageing, aged, and super-aged societies by the proportion of people aged 65 years and older. Taiwan has rapidly progressed to an aged society, and Taipei reached this status by 2014. Concurrently, global adult obesity has nearly tripled since 1975, and in Taiwan adult obesity rose from 11.5% (1993–1996) to 22.8% (2013–2016). Among older adults in Taiwan, obesity prevalence increased from 18.9% (1993–1996) to 22.8% (2013–2016). Obesity in older adults is associated with elevated mortality risk and multiple chronic diseases including hypertension, dyslipidemia, type 2 diabetes, coronary heart disease, and cancer, imposing significant health burdens. While an inverse association between education level and obesity has been shown in many developed countries, geriatric populations have been less studied. Given the rising burden of geriatric obesity and its policy implications, this study aimed to evaluate the association between education level and obesity among elderly residents of Taipei, Taiwan, and to assess other demographic correlates including age, race, income, and smoking.

Prior reports in the United States show differences in obesity prevalence by sex, age, race, household income, education, smoking, and urbanization, with lower obesity among college graduates and residents of large metropolitan areas. OECD reports indicate similar patterns in other developed countries, with obesity lower among those with higher education. In South Korea, education has been positively associated with obesity among men but inversely among women, with patterns differing by model specification and interaction terms. In Taiwan, women with less education have higher rates of overweight/obesity, and morbid obesity has been linked to low socioeconomic status. Few studies have focused on elderly populations; one study comparing Japan and the US reported a 5–9% reduction in obesity likelihood per additional year of education, though limited by self-reported data and low obesity prevalence in Japanese older adults. Multiple pathways link socioeconomic status (education, income, occupation) to health through differential exposure to chronic stress and related biological effects. Establishing whether education relates to geriatric obesity in Taipei can inform targeted interventions.

Design and data source: Cross-sectional study using data from the Taipei elderly health examination programme (2013–2015), an annual, government-funded program. Data were de-identified; ethics approval was obtained (TCHIRB-10805022-W) with informed consent waived. Participants: Taipei citizens aged ≥65 years (aboriginal identity ≥55 years). If a participant attended multiple years, only the latest year’s data were used. Exclusion criteria included missing/wrong data, extreme or implausible values: age >110 years; body weight >120 kg or <20 kg; height >200 cm or <120 cm; BMI >50 kg/m² or <10 kg/m². Measures: Height and weight were measured at contracted hospitals using standardized, calibrated devices with shoes removed. Obesity defined as BMI ≥27 kg/m² per Taiwan’s Ministry of Health and Welfare (overweight 24 to <27 kg/m²; normal 18.5 to <24 kg/m²; underweight <18.5 kg/m²). Education level originally categorized (illiterate to graduate institute, with completeness), was converted to years and collapsed into five categories: ≤6, 7–9, 10–12, 13–15, and ≥16 years. Race: aborigine vs non-aborigine per official records. Income status: normal, middle-income, middle/low-income, and low-income; for analysis, middle-income, middle/low-income, and low-income were collapsed into lower-income households. Smoking status categorized as non-smoker (never), occasional smoker (only while socialising/by invitation), and current smoker (every day or only after meals). Statistical analysis: Multivariate logistic regression assessed associations between education level and obesity separately for men and women, controlling for age, race, income status, and smoking status. Reference categories: education ≥16 years, non-aborigine, normal income, and non-smoker. Results included prevalence estimates and adjusted odds ratios (OR) with 95% Wald confidence intervals. Subgroup analyses: by data year (2013, 2014, 2015) and by age groups (≤70, 71–80, >80 years). Analyses were conducted using SAS 9.4.

Sample: After exclusions, 28,092 men (mean age 76.8±7.80) and 31,835 women (mean age 74.2±6.89) were included. Overall BMI distribution: Men—obesity 16.4%, overweight 33.2%, normal 46.6%, underweight 3.8%. Women—obesity 18.3%, overweight 27.5%, normal 49.2%, underweight 5.0%. Education distribution: Men—≤6 years 19.3%, 7–9 years 10.0%, 10–12 years 22.5%, 13–15 years 11.8%, ≥16 years 36.4%. Women—≤6 years 36.2%, 7–9 years 16.2%, 10–12 years 23.8%, 13–15 years 9.0%, ≥16 years 14.8%. Adjusted associations with obesity (reference: ≥16 years education): - Men: prevalence by education—≤6: 19.2%; 7–9: 18.5%; 10–12: 17.1%; 13–15: 14.2%; ≥16: 14.7%. ORs—≤6: 1.436 (95% CI 1.314–1.570; p<0.001); 7–9: 1.363 (1.219–1.523; p<0.001); 10–12: 1.213 (1.113–1.322; p<0.001); 13–15: 0.972 (0.868–1.087; p=0.614). - Women: prevalence by education—≤6: 24.1%; 7–9: 19.0%; 10–12: 15.1%; 13–15: 13.0%; ≥16: 12.1%. ORs—≤6: 2.278 (2.062–2.517; p<0.001); 7–9: 1.680 (1.500–1.881; p<0.001); 10–12: 1.278 (1.146–1.424; p<0.001); 13–15: 1.087 (0.945–1.251; p=0.241). Other covariates: - Race: Aborigines had markedly higher obesity—men: prevalence 43.7%, OR 2.936 (2.316–3.722; p<0.001); women: prevalence 38.3%, OR 2.616 (2.184–3.133; p<0.001) vs non-aborigines (men 16.1%; women 18.0%). - Income: No significant differences—men: lower-income OR 0.907 (0.746–1.102; p=0.325); women: OR 1.061 (0.868–1.297; p=0.566). - Smoking: Men—current smokers OR 0.861 (0.758–0.978; p=0.022); occasional smokers OR 1.079 (0.895–1.300; p=0.426) vs non-smokers. Women—no significant differences for current or occasional smokers. - Age: Men—per year OR 0.977 (0.972–0.981; p<0.001); Women—OR 0.999 (0.995–1.003; p=0.644). Subgroup analyses (by year and by age groups) consistently showed an inverse association between education level and obesity. Younger elderly cohorts exhibited higher average education and higher obesity prevalence, suggesting possible cohort effects.

The study demonstrates an inverse association between education level and obesity among elderly residents of Taipei, aligning with patterns observed in developed countries. The gradient is stronger among women, consistent with literature showing more pronounced education-obesity disparities in females. Despite adjusting for age, race, income, and smoking, aboriginal status remained strongly associated with higher odds of obesity, indicating unmeasured socioeconomic or contextual factors (e.g., wealth, debt, environmental exposures) may contribute. Lower household income was not associated with obesity in this cohort, likely influenced by the small proportion classified as lower-income and limited granularity of income categories. Age showed a modest inverse association with obesity in men but not women; however, cross-sectional design and cohort effects likely explain apparent age patterns, as younger elderly cohorts had higher obesity prevalence. The observed lower odds of obesity among older male current smokers accord with known weight-suppressive effects of nicotine through increased energy expenditure and appetite suppression, but smoking’s adverse cardiometabolic consequences outweigh any potential weight benefit. Participation patterns suggest more educated individuals may have been more likely to attend the health examinations, potentially influencing observed distributions. Overall, the findings support the role of educational attainment as a key social determinant of obesity in late life and highlight subgroups (women with lower education, aborigines) for targeted interventions.

Among elderly Taipei residents (2013–2015), higher educational attainment was associated with lower prevalence and odds of obesity, with a stronger effect in women. Taiwanese aborigines had substantially higher odds of obesity independent of age, income, and smoking. Male current smokers exhibited slightly lower odds of obesity, though this does not imply health benefit. Future longitudinal and quasi-experimental studies collecting education measures prior to obesity onset are needed to clarify causality and potential reverse causation. Public health programs should prioritize tailored empowerment, nutrition, and physical activity interventions for disadvantaged groups, particularly women with lower education and aboriginal populations, to reduce obesity-related morbidity and mortality in the elderly.

- Cross-sectional design precludes causal inference and is susceptible to cohort effects. - Potential selection bias: only Taipei citizens who received information and enrolled in the health examination were included; more educated individuals may have been more likely to participate. - Income categorization was coarse; most participants were in the normal-income group, limiting detection of income-obesity gradients and precluding finer stratification within normal income. - Exclusion of extreme anthropometric/age values likely had minimal impact due to low frequency but may slightly affect generalizability. - Findings may not generalize to all older adults in Taipei or Taiwan. - Low prevalence of female smokers reduced power to detect associations in women.