Sex-specific differences in diabetes prevalence between normal-weight obesity and obesity among middle-aged adults in South Korea: a propensity score–matched study Authors: Youn-Soo Lee, Alex Soo In Park, Bo-Young Youn Published: 2026 ABSTRACT AbstractObjectiveTo examine sex differences in type 2 diabetes mellitus (T2DM) prevalence between individuals with normal-weight obesity (NWO) and individuals with obesity among middle-aged Korean adults using nationally representative data.MethodsThis cross-sectional study used Korea National Health and Nutrition Examination Survey data from 2019 to 2021. NWO was defined as a body mass index (BMI) of 18.5–24.9 kg/m2 with body fat ≥26% in men and ≥ 36% in women. Propensity score matching (1:1) between NWO and individuals with obesity was conducted separately by sex to balance demographic, socioeconomic, lifestyle, and clinical characteristics. Conditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs).ResultsA total of 684 participants were included after matching. Among men, T2DM prevalence was lower in the NWO group than in the obesity group (20.00% vs. 31.03%), with lower odds of T2DM (adjusted OR = 0.54; 95% CI 0.29, 0.93). Among women, no significant differences were observed between NWO and obesity groups (11.70% vs. 13.20%, p > 0.05).ConclusionsNWO showed sex-specific differences in association with T2DM prevalence; causal relationships cannot be inferred. Body composition assessment beyond BMI may identify metabolic risks, particularly in Asian populations.Keywords: Body composition, Adiposity, Cardiometabolic risk, Insulin resistance, Asian population, KNHANES Introduction The global prevalence of type 2 diabetes mellitus (T2DM) continues to increase, representing a major public health concern because of its association with cardiovascular disease, disability, and premature mortality (Saeedi et al., 2019). Obesity is a major modifiable risk factor for T2DM; however, conventional definitions based on body mass index (BMI) may not adequately capture metabolic risk, particularly among individuals with excess body fat despite having a normal BMI (Stefan et al., 2017; Blüher, 2019). Normal-weight obesity (NWO), typically defined as a normal BMI with elevated body fat percentage, has emerged as an important cardiometabolic risk phenotype (De Lorenzo et al., 2006; Oliveros and Somers, 2019). Individuals with NWO have been reported to exhibit insulin resistance, adverse lipid profiles, and chronic low-grade inflammation similar to those observed in individuals with obesity (Romero-Corral et al., 2010; Stefan et al., 2017). Previous studies have also suggested that NWO is associated with increased risk of T2DM and cardiovascular disease, highlighting the limitations of BMI-based risk assessment (Oliveros et al., 2014). Sex differences may further influence the relationship between adiposity and T2DM risk. Differences in fat distribution, hormonal regulation, and skeletal muscle mass contribute to distinct metabolic profiles between men and women (Kautzky-Willer et al., 2016; Tramunt et al., 2020). Men generally accumulate greater visceral adiposity, which is strongly associated with insulin resistance, whereas women tend to accumulate more subcutaneous fat, which may confer relative metabolic protection before menopause (Karastergiou et al., 2012). Lifestyle factors including smoking, alcohol consumption, and physical activity may also contribute to sex differences in T2DM risk (Logue et al., 2011). These relationships may be particularly important in Asian populations, where metabolic risk often occurs at lower BMI levels than in Western populations (WHO Expert Consultation, 2004). Studies in Korean populations have shown that body composition and central adiposity are associated with T2DM risk independent of BMI (Kim and Han, 2021; Lee and Kim, 2022). To date, most studies have compared NWO with metabolically healthy normal-weight individuals rather than directly with obesity phenotypes. Direct comparisons between NWO and obesity may provide important insights into relative metabolic risk across adiposity phenotypes. Therefore, this study examined sex differences in the association between NWO and T2DM prevalence among middle-aged Korean adults using nationally representative Korea National Health and Nutrition Examination Survey (KNHANES) data. Understanding how metabolic risk differs across body composition phenotypes may improve identification of at-risk individuals beyond BMI-based classification. It was hypothesized that NWO would be associated with lower T2DM prevalence compared with obesity, and that this association would differ by sex owing to sex-specific differences in fat distribution and metabolic regulation. Methods Study design and population This study used a cross-sectional research design, utilizing data from the KNHANES conducted between 2019 and 2021. The KNHANES is a nationally representative survey administered by the Korea Disease Control and Prevention Agency, aimed at comprehensively assessing the health and nutritional status of the Korean population (Kweon et al., 2014; Oh et al., 2021). Among the 6265 individuals aged over 40 and up to 60 years who participated in the KNHANES study from 2019 to 2021, several groups were excluded based on specific criteria: (1) participants with missing body composition data (n = 1384); (2) individuals lacking complete information on T2DM (n = 616); (3) underweight participants (n = 176); (4) normal-weight participants without obesity (n = 1112); and (5) those with missing data for covariates included in the study (n = 1268). A flowchart detailing the study population is provided in Fig. 1. Flow diagram of the study population of middle-aged Korean adults (aged 40–60 years), Korea National Health and Nutrition Examination Survey, 2019–2021. NWO = normal-weight obesity. OB = obesity. After applying exclusion criteria, 1709 participants remained eligible, including men with obesity (n = 734), men with NWO (n = 154), women with obesity (n = 623), and women with NWO (n = 198). After 1:1 propensity score matching (PSM) performed separately by sex, 684 participants remained (men: 145 pairs; women: 197 pairs). Measures NWO and obesity were defined using BMI and body fat percentage. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m2). Body fat percentage was measured using dual-energy X-ray absorptiometry according to KNHANES examination protocols (Jung et al., 2016). NWO was defined as BMI between 18.5 and 24.9 kg/m2 with elevated body fat percentage (≥ 26% in men and ≥ 36% in women); obesity was defined as BMI ≥ 25 kg/m2 based on Asia-Pacific obesity criteria (Kim et al., 2023; Parfenteva et al., 2024). T2DM was defined as fasting plasma glucose ≥126 mg/dL, glycated hemoglobin (HbA1c) ≥ 6.5%, self-reported physician diagnosis of T2DM, or current use of antidiabetic medication (Kim et al., 2017). Covariates were selected a priori based on prior evidence and potential associations with both obesity phenotype and T2DM to minimize confounding in the association between NWO and T2DM. Demographic variables included age, education level, marital status, employment status, household income, and private health insurance. Health behavior variables included smoking status, alcohol consumption, aerobic physical activity, perceived stress, self-rated health status, and meal frequency. Aerobic physical activity was included because it represents the standardized physical activity indicator available in the KNHANES dataset and is the modality most consistently associated with insulin sensitivity and T2DM risk in epidemiological research; it was assessed as a binary variable indicating whether participants met national guidelines (≥150 min/week of moderate-intensity or ≥ 75 min/week of vigorous-intensity activity), without capturing exercise intensity levels or total energy expenditure. Alcohol consumption was coded as a dichotomous variable (current drinker vs. non-drinker) based on self-report, without quantifying frequency or amount consumed. Meal frequency was categorized by weekly consumption. Clinical variables included systolic and diastolic blood pressure and parental history of T2DM. Statistical analysis Before PSM, participant characteristics were compared using independent t-tests for continuous variables and chi-square tests for categorical variables. Analyses were performed separately for men and women. PSM was conducted to reduce baseline differences between NWO and obesity groups. Propensity scores were estimated using logistic regression including demographic characteristics, health behaviors, and clinical variables. BMI and body fat percentage were excluded from the matching model because they define the exposure groups. Nearest-neighbor matching was performed at a 1:1 ratio without replacement using a caliper width of 0.2 standard deviations of the logit of the propensity score. Covariate balance was assessed using standardized mean differences, with values <0.1 indicating adequate balance using Love plots (eFigure 1). Logistic regression analyses were performed to examine the association between NWO and T2DM. Crude models were first estimated, followed by adjusted models including covariates. After matching, conditional logistic regression models were used to account for the matched design. BMI and body fat percentage were additionally adjusted to account for residual differences in body composition between exposure groups. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported. Statistical significance was defined as p < 0.05. All analyses were conducted using RStudio® version 2026.01.2 (Posit® Software, Boston, MA, USA). This study was based on a publicly available, anonymized database and thus met the data curator's guidelines for the protection of human subjects with respect to safety and privacy. The study was reviewed by the Institutional Review Board of Hwasung Medi-Science University, South Korea, and was granted an exemption (HSMUIRB-2024-05) because it used secondary data with anonymized and encrypted personal information; a statement of exemption is on file with the authors' institution. Results Baseline characteristics before and after PSM Baseline characteristics of participants according to obesity phenotype before and after PSM are presented in Table 1. Before matching, several demographic and clinical differences were observed between the NWO and obesity groups. Table 1Characteristics of middle-aged Korean adults (aged 40–60 years) with obesity or normal-weight obesity, before and after propensity score matching, Korea National Health and Nutrition Examination Survey, 2019–2021.Before propensity score matchingAfter propensity score matchingMenWomenMenWomenWith obesityWith normal-weight obesitypWith obesityWith normal-weight obesitypWith obesityWith normal-weight obesitypWith obesityWith normal-weight obesityp(n = 734)(n = 154)(n = 623)(n = 198)(n = 145)(n = 145)(n = 197)(n = 197)Matched variablesAge in years, mean ± SD50.7 ± 5.752 ± 5.80.0251.4 ± 5.750.6 ± 5.80.1750.7 ± 5.751.7 ± 5.70.1451.2 ± 5.750.9 ± 5.80.57Perceived health status, n (%)Very good52 (7.1%)7 (4.5%)24 (3.9%)6 (3%)5 (3.4%)7 (4.8%)5 (2.5%)6 (3.0%)Good208 (28.3%)37 (24.0%)120 (19.3%)44 (22.2%)33 (22.8%)36 (24.8%)44 (22.3%)43 (21.8%)Fair333 (45.4%)80 (51.9%)290 (46.5%)102 (51.5%)79 (54.5%)77 (53.1%)107 (54.3%)102 (51.8%)Bad130 (17.7%)23 (14.9%)154 (24.7%)38 (19.2%)21 (14.5%)22 (15.2%)35 (17.8%)38 (19.3%)Very bad11 (1.5%)7 (4.5%)35 (5.6%)8 (4%)7 (4.8%)3 (2.1%)6 (3.0%)8 (4.1%)Alcohol consumption, n (%)501 (68.3%)94 (61.0%)0.08209 (33.5%)65 (32.8%)0.85112 (77.2)106 (73.1)0.4272 (36.5)70 (35.5)0.83Current smoker, n (%)212 (28.9%)43 (28.7%)0.0631 (5.0%)7 (3.6%)0.8354 (37.2)54 (37.2)0.178 (4.1)7 (3.6)0.79Perceived stress, n (%)Very stressful37 (5.0%)3 (1.9%)29 (4.7%)8 (4.0%)3 (2.1%)3 (2.1%)7 (3.6%)8 (4.1%)Stressful147 (20.0%)15 (9.7%)143 (23%)40 (20.2%)18 (12.4%)15 (10.3%)41 (20.8%)40 (20.3%)Little stressful409 (55.7%)97 (63.0%)330 (53%)113 (57.1%)89 (61.4%)89 (61.4%)112 (56.9%)113 (57.4%)Not stressful at all141 (19.2%)39 (25.3%)121 (19.4%)37 (18.7%)35 (24.1%)33 (22.8%)37 (18.8%)36 (18.3%)Aerobic Physical Activity Rate, n (%)362 (49.3%)54 (35.1%)0.04261 (41.9%)85 (42.9%)0.6387 (60.0)71 (49.0)0.0694 (47.7)91 (46.2)0.76Family history of type 2 diabetes mellitus, n (%)Father96 (13.1%)16 (10.4%)0.1869 (11.1%)18 (9.1%)0.6636 (24.8)35 (24.1)0.8953 (26.9)50 (25.4)0.73Mother98 (13.4%)9 (5.8%)0.0283 (13.3%)20 (10.1%)0.6636 (24.8)35 (24.1)0.8953 (26.9)50 (25.4)0.73Systolic blood pressure (mmHg), mean ± SD125.4 ± 14.7121.1 ± 14.80.01124.9 ± 17.1120.3 ± 16.20.01124.8 ± 14.4121.6 ± 14.90.07124.2 ± 16.8121 ± 16.30.06Diastolic blood pressure (mmHg), mean ± SD83.7 ± 10.480.7 ± 9.80.0178.6 ± 10.376.1 ± 10.00.0283.4 ± 10.280.9 ± 9.90.0578.2 ± 10.276.6 ± 10.10.11Breakfast consumption, n (%)5–7 per week379 (51.6%)117 (76.0%)388 (62.3%)112 (56.6%)117 (80.7%)108 (74.5%)117 (59.4%)111 (56.3%)3–4 per week75 (10.2%)9 (5.8%)63 (10.1%)23 (11.6%)8 (5.5%)9 (6.2%)20 (10.2%)23 (11.7%)1–2 per week84 (11.4%)9 (5.8%)66 (10.6%)23 (11.6%)6 (4.1%)9 (6.2%)21 (10.7%)23 (11.7%)0 per week196 (26.7%)19 (12.3%)106 (17%)40 (20.2%)14 (9.7%)19 (13.1%)39 (19.8%)40 (20.3%)Lunch consumption, n (%)5–7 per week658 (89.6%)140 (90.9%)544 (87.3%)173 (87.4%)131 (90.3%)132 (91%)172 (87.3%)172 (87.3%)3–4 per week39 (5.3%)6 (3.9%)47 (7.5%)12 (6.1%)4 (2.8%)5 (3.4%)15 (7.6%)12 (6.1%)1–2 per week20 (2.7%)4 (2.6%)19 (3.0%)7 (3.5%)4 (2.8%)4 (2.8%)5 (2.5%)7 (3.6%)0 per week17 (2.3%)4 (2.6%)13 (2.1%)6 (3%)6 (4.1%)4 (2.8%)5 (2.5%)6 (3.0%)Dinner consumption, n (%)5–7 per week677 (92.2%)147 (95.5%)546 (87.6%)173 (87.4%)138 (95.2%)139 (95.9%)171 (86.8%)172 (87.3%)3–4 per week44 (6.0%)5 (3.2%)62 (10.0%)15 (7.6%)6 (4.1%)4 (2.8%)21 (10.7%)15 (7.6%)1–2 per week8 (1.1%)0 (0.0%)12 (1.9%)8 (4%)1 (0.7%)0 (0.0%)4 (2.0%)8 (4.1%)0 per week5 (0.7%)2 (1.3%)3 (0.5%)2 (1%)0 (0.0%)2 (1.4%)1 (0.5%)2 (1.0%)Unmatched VariablesBody mass index (kg/m2), mean ± SD27.9 ± 2.323.4 ± 1.1<0.0128.2 ± 2.723.1 ± 1.2<0.0127.8 ± 2.223.5 ± 1.1<0.0128.1 ± 2.623.2 ± 1.2<0.01Body fat (%), mean ± SD28.4 ± 3.829.5 ± 3.60.0139.7 ± 3.739.4 ± 3.40.3828.5 ± 3.829.4 ± 3.60.0439.6 ± 3.639.5 ± 3.40.74Type 2 diabetes mellitus status, n (%)0.680.000.030.27Normal289 (39.4%)66 (42.9%)253 (40.6%)107 (54.0%)43 (29.7%)63 (43.4%)90 (45.7%)106 (53.8%)Prediabetes292 (39.8%)56 (36.4%)258 (41.4%)68 (34.3%)57 (39.3%)53 (36.6%)81 (41.1%)68 (34.5%)Type 2 diabetes mellitus153 (20.8%)32 (20.8%)112 (18.0%)23 (11.6%)45 (31.0%)29 (20.0%)26 (13.2%)23 (11.7%)OB = obesity. NWO = normal-weight obesity. PSM = propensity score matching. SD = standard deviation. P-values were obtained from independent t-tests for continuous variables and chi-square tests for categorical variables before matching, and from paired analyses accounting for the matched design after propensity score matching. Characteristics of middle-aged Korean adults (aged 40–60 years) with obesity or normal-weight obesity, before and after propensity score matching, Korea National Health and Nutrition Examination Survey, 2019–2021. OB = obesity. NWO = normal-weight obesity. PSM = propensity score matching. SD = standard deviation. P-values were obtained from independent t-tests for continuous variables and chi-square tests for categorical variables before matching, and from paired analyses accounting for the matched design after propensity score matching. Among men, individuals with NWO were older than those with obesity (52.0 ± 5.8 vs. 50.4 ± 5.7 years, p = 0.02). Men with NWO also had lower systolic blood pressure (121.1 ± 14.8 vs. 125.4 ± 14.7 mmHg, p = 0.01) and diastolic blood pressure (80.7 ± 9.8 vs. 83.7 ± 10.4 mmHg, p = 0.01). Aerobic physical activity was less common among men with NWO compared with those with obesity (48.7% vs. 60.4%, p = 0.04). No significant differences were observed in education level, marital status, employment, smoking, alcohol consumption, parental history of T2DM, meal frequency, self-rated health, or perceived stress. Among women, systolic blood pressure (120.3 ± 16.2 vs. 124.9 ± 17.1 mmHg, p = 0.01) and diastolic blood pressure (76.1 ± 10.0 vs. 78.6 ± 10.3 mmHg, p = 0.02) were lower in the NWO group compared with the obesity group. No significant differences were observed in age, education, marital status, employment, smoking, alcohol consumption, aerobic physical activity, parental history of T2DM, meal frequency, self-rated health, or perceived stress. As expected, BMI was lower and body fat percentage was higher in the NWO group compared with the obesity group among men. Among women, BMI was lower in the NWO group, while body fat percentage did not differ significantly between groups as expected due to exposure definition. After PSM, baseline characteristics were generally balanced between NWO and obesity groups in both men and women. No significant differences remained after matching, whereas BMI and body fat percentage remained different between groups as defining characteristics of the exposure. Association between NWO and T2DM before and after PSM The prevalence of T2DM according to obesity phenotype is presented in Fig. 2, and the corresponding regression results are shown in Table 2. Adjusted Model 1 included demographic, socioeconomic, lifestyle, and clinical covariates, whereas Adjusted Model 2 additionally adjusted for BMI and body fat percentage after matching to account for residual body composition differences. Prevalence of type 2 diabetes mellitus among middle-aged Korean adults (aged 40–60 years) with obesity or normal-weight obesity, by sex, before and after propensity score matching, Korea National Health and Nutrition Examination Survey, 2019–2021. OB = obesity. NWO = normal-weight obesity. PSM = propensity score matching. Table 2Association between normal-weight obesity and type 2 diabetes mellitus prevalence among middle-aged Korean adults (aged 40–60 years), by sex, before and after propensity score matching, Korea National Health and Nutrition Examination Survey, 2019–2021.MenWomenWith obesity, OR (reference)With normal-weight obesity, OR (95% CI)With obesity, OR (reference)With normal-weight obesity, OR (95% CI)Before propensity score matchingCrude model1.00 (reference)1.01 (0.65, 1.52)1.00 (reference)0.60 (0.37, 0.97)Adjusted Model 1⁎1.00 (reference)0.52 (0.32, 0.83)1.00 (reference)0.54 (0.33, 0.90)After propensity score matchingCrude model1.00 (reference)0.56 (0.32, 0.95)1.00 (reference)1.00 (0.67, 1.78)Adjusted model 2⁎⁎1.00 (reference)0.54 (0.29, 0.93)1.00 (reference)0.96 (0.47, 1.94)OR = odds ratio; CI = confidence interval; BMI = body mass index; NWO = normal-weight obesity; OB = obesity; PSM = propensity score matching.⁎Adjusted Model 1: Multiple logistic regression adjusted for age, household income, marital status, private insurance status, self-rated health, education, employment status, alcohol consumption, perceived stress, smoking status, aerobic physical activity, parental history of T2DM, systolic blood pressure, diastolic blood pressure, and meal frequency.⁎⁎Adjusted Model 2: Conditional logistic regression analysis after PSM, additionally adjusted for BMI and body fat percentage to account for residual differences in body composition between exposure groups. Association between normal-weight obesity and type 2 diabetes mellitus prevalence among middle-aged Korean adults (aged 40–60 years), by sex, before and after propensity score matching, Korea National Health and Nutrition Examination Survey, 2019–2021. OR = odds ratio; CI = confidence interval; BMI = body mass index; NWO = normal-weight obesity; OB = obesity; PSM = propensity score matching. Adjusted Model 1: Multiple logistic regression adjusted for age, household income, marital status, private insurance status, self-rated health, education, employment status, alcohol consumption, perceived stress, smoking status, aerobic physical activity, parental history of T2DM, systolic blood pressure, diastolic blood pressure, and meal frequency. Adjusted Model 2: Conditional logistic regression analysis after PSM, additionally adjusted for BMI and body fat percentage to account for residual differences in body composition between exposure groups. Before PSM, the prevalence of T2DM did not significantly differ between the NWO and obesity groups among men (20.78% vs. 20.84%, p > 0.05). Among women, T2DM prevalence was significantly lower in the NWO group compared with the obesity group (11.62% vs. 17.98%, p = 0.04). In logistic regression analyses prior to matching, no significant association between NWO and T2DM was observed among men in the crude model (OR = 1.01, 95% CI 0.65, 1.52). However, after adjustment for demographic, socioeconomic, lifestyle, and clinical covariates (Adjusted Model 1), men with NWO showed significantly lower odds of T2DM compared with those with obesity (OR = 0.52, 95% CI 0.32, 0.83). Among women, NWO was associated with lower odds of T2DM in both the crude model (OR = 0.60, 95% CI 0.37, 0.97) and Adjusted Model 1 (OR = 0.54, 95% CI 0.33, 0.90). After PSM, T2DM prevalence among men was significantly lower in the NWO group compared with the obesity group (20.00% vs. 31.03%, p = 0.03). Conditional logistic regression analyses demonstrated that NWO remained associated with lower odds of T2DM among men in both the crude model (OR = 0.56, 95% CI 0.32, 0.95) and the adjusted model (Adjusted Model 2), which additionally controlled for residual differences in BMI and body fat percentage (OR = 0.54, 95% CI 0.29, 0.93). Among women, no significant differences in T2DM prevalence were observed between the NWO and obesity groups after matching (11.68% vs. 13.20%, p > 0.05). Similarly, no significant associations were observed in either the crude model (OR = 1.00, 95% CI 0.67, 1.78) or Adjusted Model 2 (OR = 0.96, 95% CI 0.47, 1.94). Overall, these findings indicate that NWO was associated with lower T2DM prevalence compared with obesity among men but not among women after balancing covariates through PSM. Discussion The present study identified sex-specific differences in the association between NWO and T2DM prevalence among middle-aged Korean adults. After balancing demographic, socioeconomic, lifestyle, and clinical factors using PSM, men with NWO demonstrated lower T2DM prevalence compared with men with obesity, whereas no significant differences were observed among women. These findings suggest that the metabolic implications of excess adiposity within a normal BMI range may differ according to sex and body composition phenotype. These findings differ from several previous studies reporting increased T2DM risk among individuals with NWO compared with normal-weight individuals without excess adiposity (Jia et al., 2018; Kapoor et al., 2020). However, those studies primarily compared NWO with metabolically healthy normal-weight groups rather than directly with obesity. Because obesity represents a higher adiposity state, comparisons between NWO and obesity may produce different patterns of association. The present findings therefore extend prior work by suggesting that T2DM risk in NWO may be intermediate between normal weight without excess adiposity and obesity, particularly among men. Nevertheless, a lower T2DM risk relative to the obesity group should not be interpreted as evidence that NWO is a metabolically benign condition, as this comparison does not establish absolute metabolic risk. The observed lower T2DM prevalence among men with NWO compared with men with obesity is consistent with the established relationship between increasing adiposity and T2DM risk. Previous large cohort studies have shown that T2DM risk increases progressively across adiposity categories and that men often develop T2DM at lower BMI thresholds than women (Gray et al., 2015; Kautzky-Willer et al., 2023). Differences in fat distribution may also contribute to these findings. Visceral adiposity is strongly associated with insulin resistance and cardiometabolic risk (Tao et al., 2018; Lönnqvist et al., 1997), whereas individuals with NWO may have lower visceral fat accumulation than those with generalized obesity despite elevated total body fat (Porter et al., 2009; Bigaard et al., 2004). This difference in fat distribution may partially explain the observed differences between groups. Among women, the significant difference observed before matching was no longer present after adjustment, suggesting that baseline demographic and lifestyle differences may have contributed to the crude association. This pattern is consistent with previous studies showing attenuation of metabolic associations after adjustment for adiposity distribution and behavioral factors (Jia et al., 2018). These findings highlight the importance of accounting for confounding when examining obesity phenotypes in observational data. Menopausal status was not included as a covariate; given the study's age range of 40–64 years, menopause-related shifts in fat distribution toward visceral adiposity and the associated increase in metabolic risk may represent an unmeasured source of confounding that contributed to the attenuation of differences observed among women. The absence of significant differences between NWO and obesity among women should not be interpreted as indicating low metabolic risk in NWO. Prior Korean studies have demonstrated that women with NWO remain at increased risk of metabolic syndrome compared with normal-weight individuals without excess adiposity (Kim et al., 2023). Additional studies have similarly shown that excess body fat within a normal BMI may be associated with insulin resistance and adverse metabolic profiles (Ruderman et al., 1998; Madeira et al., 2013). These findings support the interpretation that NWO may represent a metabolically at-risk phenotype even when differences relative to obesity are not observed. Differences in lean mass may also contribute to the observed patterns. Skeletal muscle is a major determinant of glucose disposal, and lower lean mass relative to body weight has been associated with higher T2DM risk (Ruderman et al., 1998; Srikanthan and Karlamangla, 2011; Hong et al., 2017). Individuals with NWO may retain relatively greater lean mass compared with those with obesity, which may contribute to differences in metabolic risk despite similar or elevated body fat percentage. This observation supports the importance of evaluating body composition rather than relying solely on BMI. These findings have implications for metabolic risk assessment in Asian populations, where T2DM and metabolic disorders frequently occur at lower BMI levels than in Western populations (Kapoor et al., 2019; Kim et al., 2023). The results suggest that BMI alone may not adequately capture metabolic heterogeneity and that incorporation of body composition measures may improve identification of individuals at metabolic risk. The observed sex differences further indicate that relationships between adiposity and metabolic risk may not be uniform across populations. Several limitations should be considered. First, the cross-sectional design prevents causal inference regarding the relationship between NWO and T2DM. Prospective studies are needed to clarify temporal relationships and potential mechanisms. Second, the study population consisted of Korean adults, which may limit generalizability to other populations. Third, residual confounding from unmeasured factors such as dietary intake, menopausal status, genetic susceptibility, and psychosocial factors cannot be excluded despite adjustment using PSM. Fourth, the study compared NWO exclusively with individuals with obesity and did not include a normal-weight, non-obese reference group; therefore, the findings reflect relative differences in T2DM prevalence within the elevated-adiposity spectrum and cannot be used to draw conclusions about the absolute metabolic risk conferred by NWO. Fifth, lifestyle variables including physical activity, meal frequency, and alcohol consumption were measured using proxy indicators derived from self-reported survey items that do not capture exercise intensity, total caloric intake, or dietary quality, which may result in residual confounding that cannot be fully addressed through statistical adjustment. Lastly, the exclusion of over 1200 participants due to missing body composition and T2DM data may have introduced selection bias and could limit the representativeness of the study sample, as individuals with missing data may differ systematically from those included in the analysis. Conclusions This cross-sectional study observed sex-specific differences in the co-occurrence of NWO and T2DM prevalence among middle-aged Korean adults. After PSM, men with NWO showed lower T2DM prevalence compared with men with obesity, while no significant differences were observed among women. As the study design is cross-sectional, these findings reflect differences in the co-occurrence of body composition phenotypes and T2DM status and do not allow inference about the direction of association or metabolic progression over time. These findings suggest that metabolic risk associated with body composition phenotypes may differ by sex and that BMI alone may not fully capture metabolic heterogeneity. Incorporating body composition measures may improve identification of metabolic risk, particularly in Asian populations. However, the sample size after PSM was relatively small, particularly among men (145 pairs), which may have limited statistical power and warrants cautious interpretation of the observed associations. Longitudinal studies are needed to clarify causal relationships and underlying mechanisms. The following are the supplementary data related to this article. Love plot of balance following PSM CRediT authorship contribution statement Youn-Soo Lee: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Alex Soo In Park: Writing – original draft, Investigation, Formal analysis. Bo-Young Youn: Writing – review & editing, Supervision, Investigation, Formal analysis. Funding Source This research received no specific grant from any funding agency in the public, commercial and not-for-profit sectors. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability The datasets analyzed in this study are publicly available from the Korea Disease Control and Prevention Agency (KDCA) through the KNHANES website (https://knhanes.kdca.go.kr/knhanes/postSendPage.do?url=/eng/intr/dataIntr.do&postparam=%7B%22menuId%22:%224003%22%7D). Researchers can access the data after completing a user registration process. All analytical code and preprocessing scripts used in this study are available from the corresponding author upon reasonable request. References Further reading Associated Data Data Availability Statement The datasets analyzed in this study are publicly available from the Korea Disease Control and Prevention Agency (KDCA) through the KNHANES website (https://knhanes.kdca.go.kr/knhanes/postSendPage.do?url=/eng/intr/dataIntr.do&postparam=%7B%22menuId%22:%224003%22%7D). Researchers can access the data after completing a user registration process. All analytical code and preprocessing scripts used in this study are available from the corresponding author upon reasonable request.