Abstract
Background
Relative fat mass (RFM) is a novel indicator of central adiposity. Cognitive impairment (CI), obesity, and type 2 diabetes mellitus (T2DM) collectively pose substantial public health challenges, contributing to irreversible health consequences and imposing significant economic burdens on healthcare systems worldwide. This study aimed to evaluate the sex-specific association between RFM and CI in middle-aged and older hospitalized patients with T2DM.
Methods
This cross-sectional study analyzed data from 1328 patients (aged ≥45 years) admitted to the Department of Endocrinology, Tianjin Union Medical Center, between July 2018 and July 2022. CI was defined as a Montreal Cognitive Assessment score < 26. Logistic regression models, restricted cubic spline analyses, and sensitivity analyses were used to assess the association between RFM and CI across sex-specific quartiles.
Results
The overall prevalence of CI was 45.6% (605/1,328). The mean age of the study population was 62.4 ± 7.3 years, and 49.3% were female. Among the female participants, we identified a nonlinear inverse J-shaped relationship between RFM and CI (p for nonlinearity = 0.013), characterized by an increasing risk of CI up to an inflection point at an RFM of approximately 43.7, after which the risk plateaued. The third RFM quartile (40.02–42.86) exhibited the highest prevalence of CI (adjusted odds ratio: 2.54, 95% confidence interval: 1.54–4.16). In contrast, no significant association was observed between RFM and CI in males in either unadjusted or fully adjusted models. These findings were consistent across stratified and sensitivity analyses.
Conclusions
Among middle-aged and older patients with T2DM, RFM demonstrates a sex-specific, nonlinear association with CI in females. These results highlight the potential utility of RFM as a simple and practical indicator for identifying cognitive risk in female patients with T2DM, supplementing traditional adiposity measures such as body mass index.
Clinical trial number
Not applicable.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12888-026-07856-x.
Keywords: Relative fat mass, Cognitive impairment, Middle-aged and elderly, Type 2 diabetes, Sex difference
Introduction
Cognitive impairment (CI) and dementia represent a formidable and escalating public health challenge in the context of global population aging. In 2015, an estimated 46.8 million people worldwide were living with dementia, a number projected to rise to 130 million by 2050 [1]. Type 2 diabetes mellitus (T2DM) is a well-established risk factor for cognitive decline, with the prevalence of mild CI (MCI) reported to be as high as 45.0% among individuals with T2DM [2]. In China, the prevalence of CI among older patients with diabetes mellitus (DM) is approximately 48%, with higher rates observed in women, older adults, and individuals with lower educational attainment [3]. This association is particularly concerning because CI can significantly compromise a patient’s capacity to perform essential self-management tasks, such as glucose monitoring and dietary regulation, thereby creating a vicious cycle in which poor glycemic control accelerates cognitive decline [4].
Obesity, characterized by excessive adiposity that impairs health, has also reached epidemic proportions and is a frequent comorbidity in T2DM. The pathophysiological mechanisms linking T2DM, obesity, and CI are multifactorial and interconnected, involving chronic inflammation, insulin resistance, oxidative stress, and vascular dysfunction [5, 6]. Although body mass index (BMI) is the most widely used screening tool for obesity [7], it does not differentiate fat mass from lean mass and fails to accurately capture fat distribution [8, 9]. Waist circumference (WC) provides a more direct measure of abdominal adiposity [10]. More recently, the relative fat mass (RFM) index—calculated using WC and height—has been proposed as a more accurate estimator of whole-body fat percentage than BMI [11]. Derived and validated against dual-energy X-ray absorptiometry–assessed whole-body adiposity in a large, ethnically diverse cohort [11], RFM has since been linked to numerous cardiometabolic conditions, including hypertension [12], T2DM [13], depression [14], dyslipidemia [15], metabolic syndrome [15], cardiovascular disease [16], and all-cause mortality [17].
The relationship between adiposity and cognitive function is complex and appears to differ by sex. Prior studies have yielded inconsistent findings: some have identified higher BMI or WC as risk factors for CI [18–20], whereas others have reported null or even protective associations—an observation often referred to as the “obesity paradox” [21–23]. These inconsistencies may be partly attributable to the limitations of BMI. Notably, a growing body of literature suggests that the influence of adiposity on cognitive outcomes differs substantially between men and women. For example, central obesity has been identified as a risk factor for dementia in older women with T2DM, but not in men [24]. Recent longitudinal studies further highlight these sex-specific patterns, showing divergent cognitive trajectories among older adults with T2DM and obesity that appear to be shaped by sex [25, 26]. Biological factors such as menopause-related hormonal changes and differences in the management of cardiovascular risk factors may contribute to this sexual dimorphism [27, 28].
Despite the recognized value of RFM as a robust measure of adiposity and accumulating evidence of sex differences in cognitive outcomes, the association between RFM and CI in individuals with T2DM remains underexplored. Accordingly, this study analyzed data from hospitalized patients with T2DM to evaluate the sex-specific association between RFM and CI. We hypothesized that higher RFM would be associated with an increased risk of CI, and that this relationship would be more pronounced and potentially nonlinear in women compared with men. Clarifying this association may help establish a more precise, sex-specific anthropometric measure for identifying patients with T2DM who are at elevated risk for CI.
Research design and methods
This single-center, cross-sectional study included consecutive patients admitted to the Department of Endocrinology at Tianjin Union Medical Center between July 2018 and July 2022. We analyzed an original, de-identified electronic medical record (EMR) dataset. To obtain lifestyle information not captured in the EMR, participants completed a standardized questionnaire and participated in a brief face-to-face interview after their clinical encounter (Supplement 7). No additional examinations, tests, or imaging procedures were performed.
Inclusion and exclusion criteria
T2DM was diagnosed according to the 1999 World Health Organization (WHO) criteria for DM [29], with reference to the WHO 2011 report on the use of glycated hemoglobin (HbA1c) in DM diagnosis [30] and the 2019 reaffirmation of diagnostic thresholds [31] (fasting plasma glucose ≥7.0 mmol/L, 2-h plasma glucose ≥11.1 mmol/L, or HbA1c ≥6.5%). Inclusion criteria were: (1) age ≥45 years with a diagnosis of T2DM; and (2) ability to independently complete neuropsychological assessments. Exclusion criteria were: (1) inability to complete neuropsychological tests due to speech disorders, hearing loss, or unwillingness to cooperate; (2) presence of conditions that may affect cognitive function, including head injury, cerebral ischemia, depression, anxiety, coma, drug addiction, or other mental and neurological disorders; and (3) concurrent anemia, severe pulmonary or renal disease, history of heart failure, malignant tumors, hypothyroidism, or hyperthyroidism.
This study was approved by the Medical Ethics Committee of Tianjin Union Medical Center and conducted in accordance with the Declaration of Helsinki (Approval Number: (2018) Fast-track Review No. C08). All participants were informed about the study and provided written consent prior to participation. Written consent was also obtained for any identifiable images or data included in this publication.
RFM
Professional nurses measured and recorded participants’ height, weight, and WC. RFM was calculated as 64 - [20*Height (m)/WC (m)] for men and 76 - [20*Height (m)/WC(m)] for women, following the formula 64 - (20*Height/WC) + (12*sex), where sex = 0 for men and 1 for women [11]. The Waist-to-Waist Index (WWI) was calculated by dividing WC (cm) by the square root of body weight (kg).
Screening evaluation for CI
The Montreal Cognitive Assessment (MoCA) is a brief, validated screening tool with 90% sensitivity and 87% specificity for detecting MCI at a cutoff score of < 26 [32]. To account for educational bias, one point was added for participants with ≤12 years of schooling [32]. Trained assessors administered the MoCA in a quiet, distraction-free environment following standardized procedures. For this analysis, participants were classified into a normal cognitive function group (MoCA ≥26) and a CI group (MoCA < 26).
Covariates
Demographic characteristics, lifestyle factors, and clinical data were collected through standardized questionnaires, face-to-face interviews, and medical record reviews. Participants self-reported their sex, age, ethnicity (Han or other), marital status (married, widowed, divorced, or never married), and education level (illiterate [0 years], primary [2–6 years], junior high [7–9 years], or high school and above [≥12 years]). Smoking status (current smoker or non-smoker) and drinking status (current drinker or non-drinker) were also recorded. Regular exercise was defined as ≥ 150 min of moderate-intensity or ≥ 75 min of vigorous-intensity aerobic activity per week, performed at least twice weekly. Adherence to a diabetic diet—emphasizing complex carbohydrates, high-fiber foods, portion control, regular meals to manage blood sugar, and restricted intake of simple sugars, sodium, and saturated fat—was documented.
Medical records provided data on DM duration and the use of antidiabetic medications (insulin or oral antidiabetic drugs). Disease history—including diabetic microangiopathy (nephropathy and/or retinopathy), hypoglycemic episodes within the past 3 months, lower-limb atherosclerosis, cerebrovascular disease (CVD; hemorrhagic and/or ischemic stroke), and hypertension (seated blood pressure ≥140/90 mmHg or current antihypertensive therapy)—was ascertained through self-report and verified via medical records. Dietary management for DM followed the recommendations of the “Expert Consensus on the Diagnosis and Treatment Measures for Older Patients with T2DM in China” [33].
Participants fasted for at least 8 h before undergoing serum biochemical testing. Total cholesterol, triglycerides (TG), and serum creatinine were measured using an automatic biochemical analyzer (TBA-120FR, Toshiba, Japan). HbA1c was assessed using a fully automated high-performance liquid chromatography-based glycohemoglobin analyzer (HA-8180, ARKRAY, Japan). Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation [34].
Statistical analysis
Continuous variables are presented as means (standard deviation) or medians (interquartile range), and categorical variables as frequencies and percentages. The chi-square test was used to compare categorical variables, while the Student’s t-test or the Mann–Whitney U test was applied to compare continuous variables between females and males. Cases with missing data (0–3.7%) were excluded from the analyses.
Logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (95% CIs) for the association between RFM and CI. Multivariate regression analyses were conducted separately for females and males. Model I adjusted for sociodemographic characteristics (age, marital status, and education level). Model II further adjusted for factors with p < 0.05 in univariate analysis (DM duration, diabetic dietary control, CVD, hypoglycemic events in the past 3 months, eGFR, and TG), in addition to Model I covariates. Model III additionally adjusted for variables that changed the matched OR by ≥10% when included (smoking status, drink status, and regular exercise), in addition to Model II covariates. Model IV further included hypertension to Model III covariates.
Linear trend tests were performed by treating categorical variables as continuous. Restricted cubic splines with four knots (at the 5th, 35th, 65th, and 95th percentiles) were used to flexibly model the association between RFM and CI after adjusting for Model IV covariates. A two-piecewise logistic regression model with smoothing was applied to examine threshold effects in the association between RFM and CI, adjusting for Model IV variables. Inflection points were identified using the likelihood-ratio tests and bootstrap resampling.
To assess the robustness of the relationship between RFM and CI across populations, interaction and subgroup analyses were conducted according to age (< 65 vs. ≥65 years), history of CVD (yes vs. no), and hypoglycemic events within the past 3 months (yes vs. no). Heterogeneity across subgroups and interaction effects were assessed using logistic regression models and likelihood-ratio testing, respectively. Since the sample size was determined by the available data, no a priori statistical power calculation was performed. All analyses were performed using R 4.2.1 (http://www.Rproject.org: The R Foundation for Statistical Computing, Vienna, Austria) and Free Statistics software (version 1.9.2; Beijing FreeClinical Medical Technology Co., Ltd., Beijing, China). A two-sided p < 0.05 was considered statistically significant. Data analysis was performed from June to July 2024.
Results
Baseline characteristics
The study included 1328 adults, of whom 655 (49.3%) were female, with a mean age of 62.4 ± 7.3 years. As summarized in Table 1, significant sex-based differences were observed across several baseline characteristics. Female participants were significantly older than male participants (63.2 ± 7.3 vs. 61.7 ± 7.3 years, p < 0.001) and exhibited distinct sociodemographic profiles, including lower rates of marriage (84.1% vs. 93.3%) and fewer years of formal education (10.1 ± 2.8 vs. 11.1 ± 2.8 years; both p < 0.001).
Table 1.
Baseline characteristics of the study population by cognition status and sex
| Variables | Females | Males | ||||||
|---|---|---|---|---|---|---|---|---|
| Total(n = 655) | NCF(n = 341) | CI (n = 322) | p | Total(n = 673) | NCF (n = 398) | CI(n = 287) | p | |
| MOCA | 25.1 ± 2.7 | 27.2 ± 1.1 | 22.9 ± 2.2 | < 0.001* | 25.6 ± 2.5 | 25.6 ± 2.5 | 27.4 ± 1.1 | < 0.001* |
| Age(years) | 63.2 ± 7.3 | 61.9 ± 7.1 | 64.5 ± 7.2 | < 0.001* | 61.7 ± 7.3 | 61.7 ± 7.3 | 60.6 ± 7.2 | < 0.001* |
| Han, n (%) | 584 (89.2) | 299 (89.3) | 285 (89.1) | 0.994 | 610 (90.6) | 353 (91) | 257 (90.2) | 0.724 |
| Married, n (%) | 551 (84.1) | 302 (90.1) | 249 (77.8) | < 0.001* | 628 (93.3) | 373 (96.1) | 255 (89.5) | < 0.001* |
| Education level(years) | 10.1 ± 2.8 | 10.7 ± 2.5 | 9.5 ± 3.0 | < 0.001 | 11.1 ± 2.8 | 11.4 ± 2.9 | 10.5 ± 2.7 | < 0.001* |
| Course of diabetes(years) | 7.0 (1.0, 13.5) | 7.0 (1.0, 12.0) | 8.0 (2.0, 15.0) | 0.068 | 7.0 (1.0, 13.0) | 7.0 (1.0, 12.0) | 7.0 (1.0, 14.0) | 0.754 |
| Diabetic dietary control, n (%) | 209 (31.9) | 86 (25.7) | 123 (38.4) | < 0.001* | 247 (36.7) | 124 (32) | 123 (43.2) | 0.003* |
| Regular exercise, n (%) | 476 (72.7) | 246 (73.4) | 230 (71.9) | 0.655 | 463 (68.8) | 258 (66.5) | 205 (71.9) | 0.133 |
| Hypoglycemia n (%) | 171 (26.1) | 83 (24.8) | 88 (27.5) | 0.428 | 133 (19.8) | 66 (17) | 67 (23.5) | 0.032* |
| eGFR(mL/min/1.73 m2) | 121.7 ± 39.1 | 124.0 ± 41.8 | 119.3 ± 36.0 | 0.126 | 113.6 ± 27.5 | 115.9 ± 25.5 | 110.4 ± 29.8 | 0.009* |
| Smoke, n (%) | 59 (9.0) | 24 (7.2) | 35 (10.9) | 0.092 | 319 (47.4) | 182 (46.9) | 137 (48.1) | 0.765 |
| Drink, n (%) | 21 (3.2) | 11 (3.3) | 10 (3.1) | 0.908 | 308 (45.8) | 181 (46.6) | 127 (44.6) | 0.591 |
| Hypertension, n (%) | 460 (70.2) | 228 (68.1) | 232 (72.5) | 0.214 | 431 (64.0) | 248 (63.9) | 183 (64.2) | 0.938 |
| SBP(mmHg) | 133.8 ± 12.6 | 133.7 ± 12.3 | 134.0 ± 12.8 | 0.775 | 133.6 ± 21.4 | 134.1 ± 25.8 | 133.0 ± 13.3 | 0.503 |
| DBP(mmHg) | 80.0 ± 11.5 | 80.1 ± 7.8 | 79.9 ± 14.5 | 0.863 | 80.5 ± 8.1 | 80.5 ± 7.8 | 80.5 ± 8.5 | 0.895 |
| CVD, n (%) | 178 (27.2) | 62 (18.5) | 116 (36.2) | < 0.001* | 193 (28.7) | 93 (24) | 100 (35.1) | < 0.001* |
| LEAD, n (%) | 326 (49.8) | 164 (49) | 162 (50.6) | 0.669 | 451 (67.0) | 254 (65.5) | 197 (69.1) | 0.319 |
| Microangiopathy, n (%) | 257 (39.2) | 120 (35.8) | 137 (42.8) | 0.067 | 303 (45.0) | 172 (44.3) | 131 (46) | 0.674 |
| HbA1c(%) | 8.9 ± 2.1 | 8.9 ± 2.1 | 8.9 ± 2.1 | 0.634 | 9.0 ± 2.1 | 9.0 ± 2.0 | 9.1 ± 2.2 | 0.591 |
| TC(mmol/L) | 5.1 ± 1.2 | 5.1 ± 1.2 | 5.0 ± 1.2 | 0.265 | 4.7 ± 1.2 | 4.8 ± 1.2 | 4.6 ± 1.2 | 0.134 |
| TG(mmol/L) | 1.9 ± 1.2 | 2.0 ± 1.5 | 1.7 ± 0.8 | 0.008* | 2.2 ± 2.7 | 2.2 ± 2.0 | 2.1 ± 3.5 | 0.727 |
| Weight(kg) | 66.6 ± 11.1 | 8.8 ± 2.1 | 8.9 ± 2.1 | 0.526 | 76.8 ± 11.7 | 77.4 ± 11.2 | 76.0 ± 12.4 | 0.144 |
| Height(cm) | 159.7 ± 5.4 | 5.1 ± 1.2 | 5.0 ± 1.2 | 0.237 | 172.2 ± 5.6 | 172.3 ± 5.4 | 172.0 ± 5.7 | 0.391 |
| WC(cm) | 89.8 ± 10.0 | 2.0 ± 1.5 | 1.7 ± 0.8 | 0.006 | 92.9 ± 9.3 | 92.5 ± 8.9 | 93.4 ± 9.7 | 0.206 |
| BMI(m/kg2) | 26.1 ± 3.8 | 67.2 ± 11.3 | 66.0 ± 10.9 | 0.163 | 25.9 ± 3.4 | 26.0 ± 3.4 | 25.6 ± 3.4 | 0.152 |
| Secretagogue, n (%) | 91 (13.9) | 51 (15.2) | 40 (12.5) | 0.314 | 103 (15.3) | 57 (14.7) | 46 (16.1) | 0.606 |
| Insulin, n (%) | 200 (30.5) | 95 (28.4) | 105 (32.8) | 0.216 | 231 (34.3) | 129 (33.2) | 102 (35.8) | 0.493 |
| MET, n (%) | 271 (41.4) | 146 (43.6) | 125 (39.1) | 0.24 | 256 (38.0) | 152 (39.2) | 104 (36.5) | 0.479 |
| SU, n (%) | 164 (25.0) | 88 (26.3) | 76 (23.8) | 0.457 | 112 (16.6) | 69 (17.8) | 43 (15.1) | 0.534 |
| Glinides, n (%) | 99 (15.1) | 46 (13.7) | 53 (16.6) | 0.312 | 93 (13.8) | 60 (15.5) | 33 (11.6) | 0.149 |
| TZD, n (%) | 12 (1.8) | 8 (2.4) | 4 (1.2) | 0.278 | 14 (2.1) | 9 (2.3) | 5 (1.8) | 0.612 |
| AGI, n (%) | 359 (54.8) | 182 (54.3) | 177 (55.3) | 0.8 | 329 (48.9) | 182 (46.9) | 147 (51.6) | 0.231 |
| DPP-4i, n (%) | 85 (13.0) | 47 (14) | 38 (11.9) | 0.412 | 97 (14.4) | 54 (13.9) | 43 (15.1) | 0.669 |
| SGLT2i, n (%) | 15 (2.3) | 7 (2.1) | 8 (2.5) | 0.726 | 18 (2.7) | 11 (2.8) | 7 (2.5) | 0.763 |
| GLP-1RA, n (%) | 6 (0.9) | 4 (1.2) | 2 (0.6) | 0.687 | 7 (1.0) | 4 (1) | 3 (1.1) | 1 |
| RFM | 40.0 ± 3.9 | 39.5 ± 4.2 | 40.6 ± 3.4 | < 0.001 | 26.6 ± 3.6 | 26.4 ± 3.5 | 26.8 ± 3.7 | 0.15 |
Continuous variables are expressed as mean (SD) or median (interquartile range), and categorical variables are expressed as n (%)
Abbreviations: MOCA, Montreal Cognitive Assessment; eGFR,estimated glomerular filtration rate; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; CVD,Cerebrovascular disease; LEAD, Lower Extremity Arterial Disease; HbA1c, Hemoglobin A1c; TC, Total Cholesterol; TG, Triglycerides; WC,waist circumference; BMI, Body Mass Index; MET, Metformin; SU,Sulfonylureas; TZD,Thiazolidinediones; AGI, Alpha-Glucosidase Inhibitors; DPP-4i, Dipeptidyl Peptidase-4 Inhibitors; SGLT2i, Sodium-Glucose Cotransporter 2 Inhibitors; GLP-1RA, Glucagon-Like Peptide-1 Receptor Agonists; RFM,relative fat mass
Several clinical and lifestyle differences were also noted. Females had higher eGFR (121.7 ± 39.1 vs. 113.6 ± 27.5 mL/min/1.73 m2, p < 0.001) but lower MoCA scores (p < 0.05). Consistent with established patterns, females reported markedly lower rates of alcohol and tobacco use. The prevalence of lower-limb atherosclerosis and diabetic microangiopathy was also lower in females (all p < 0.05). In contrast, females exhibited a significantly higher prevalence of CI (48.9% vs. 42.3%, p < 0.001) and hypertension (70.2% vs. 64.0%, p = 0.016).
Association between RFM and CI
Univariate analysis identified several factors associated with CI, including age, sex, marital status, educational level, DM duration, diabetic dietary control, CVD, hypoglycemic events in the past 3 months, and eGFR (Table S1).
The relationships between BMI, WC, WWI, RFM, and CI are presented in Table S2. After adjusting for all covariates (age, sex, marital status, education level, DM duration, dietary control, CVDs, hypoglycemic events in the past 3 months, eGFR, smoking status, drinking consumption, regular exercise, and hypertension), the ORs (95% CI) for CI across RFM quartiles were: 1.00 (reference), 1.20 (0.86–1.67), 1.50 (1.08–2.10), and 1.35 (0.96–1.90).
Among females, higher RFM was markedly associated with a greater prevalence of CI (OR, 1.06; 95% CI: 1.01–1.11; p = 0.013). When RFM was analyzed categorically, the third quartile (Q3: ≤40.02 to ≤42.86) showed an adjusted OR of 2.54 (95% CI: 1.54–4.16; p < 0.001) for CI in Model IV, compared with the first quartile (Q1: < 19.84 to ≤37.72). This association remained consistent across all adjusted models (Table 2), indicating that the relationship between RFM and CI in females is independent of the covariates included. In contrast, no significant association was observed between RFM and CI across all models in males (Table 3).
Table 2.
Association between RFM and CI in multiple regression model in females
| RFM | Crude Model | Model I | Model II | Model III | Model IV | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Continuous | 1.08 (1.04~1.13) | < 0.001 | 1.06 (1.02~1.11) | 0.006 | 1.06 (1.01~1.11) | 0.016 | 1.06 (1.01~1.11) | 0.017 | 1.06 (1.01~1.11) | 0.013 |
| Categorical | ||||||||||
| Q1(19.84~37.72) | Reference | Reference | Reference | Reference | Reference | |||||
| Q2(37.72~40.02) | 1.77 (1.13~2.75) | 0.012 | 1.76 (1.11~2.8) | 0.017 | 1.74 (1.08~2.82) | 0.023 | 1.74 (1.07~2.82) | 0.025 | 1.75 (1.08~2.83) | 0.024 |
| Q3(40.02~42.86) | 2.72 (1.74~4.25) | < 0.001 | 2.53 (1.58~4.05) | < 0.001 | 2.5 (1.53~4.08) | < 0.001 | 2.51 (1.53~4.11) | < 0.001 | 2.54 (1.54~4.16) | < 0.001 |
| Q4(42.86~52.26) | 2.12 (1.36~3.31) | 0.001 | 1.8 (1.13~2.86) | 0.014 | 1.75 (1.07~2.84) | 0.025 | 1.74 (1.07~2.84) | 0.026 | 1.78 (1.08~2.91) | 0.023 |
| P for trend | < 0.001 | 0.006 | 0.012 | 0.013 | 0.011 | |||||
Model I:Adjust for age, marital status, education level.Model II:Adjust for diabetes course, diabetic dietary pattern, CVD, hypoglycemic events in last 3 months, eGFR and TG.Model III:Adjust for variables that, when added to this model, changed the matched odds ratio by at least 10%, including smoking status, drink status and regular exercise plus Model II;Model IV:Adjust for hypertension plus Model III.Abbreviations: OR,odd ratio; CI, confidence interval., RFM,relative fat ma
Table 3.
Association between RFM and CI in multiple regression model in males
| RFM | Crude Model | Model I | Model II | Model III | Model IV | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | OR (95% CI) | P-value | |
| Continuous | 1.03 (0.99~1.08) | 0.15 | 1.01 (0.97~1.06) | 0.577 | 1 (0.96~1.05) | 0.93 | 1.01 (0.96~1.05) | 0.794 | 1.01 (0.96~1.06) | 0.678 |
| Categorical | ||||||||||
| Q1(13.71~24.24) | Reference | Reference | Reference | Reference | Reference | |||||
| Q2(24.24~26.64) | 0.98 (0.63~1.51) | 0.911 | 0.89 (0.57~1.4) | 0.618 | 0.83 (0.53~1.32) | 0.437 | 0.86 (0.54~1.37) | 0.532 | 0.87 (0.54~1.38) | 0.542 |
| Q3(26.64~29.09) | 1.02 (0.66~1.58) | 0.912 | 0.97 (0.62~1.53) | 0.903 | 0.91 (0.57~1.44) | 0.69 | 0.93 (0.59~1.49) | 0.776 | 0.96 (0.6~1.52) | 0.849 |
| Q4(29.09~36.48) | 1.35 (0.88~2.08) | 0.169 | 1.11 (0.71~1.74) | 0.651 | 1 (0.63~1.58) | 0.986 | 1.03 (0.65~1.64) | 0.905 | 1.07 (0.66~1.71) | 0.791 |
| P for trend | 0.166 | 0.579 | 0.909 | 0.822 | 0.707 | |||||
Model I:Adjust for age, marital status, education level.Model II:Adjust for diabetes course, diabetic dietary pattern, CVD, hypoglycemic events in last 3 months, eGFR and TG.Model III:Adjust for variables that, when added to this model, changed the matched odds ratio by at least 10%, including smoking status, drink status and regular exercise plus Model II;Model IV:Adjust for hypertension plus Model III.Abbreviations: OR,odd ratio; CI, confidence interval., RFM,relative fat mass
Restricted cubic spline analysis further demonstrated a sex-specific pattern, with a significant nonlinear relationship in females (p for nonlinearity = 0.006; Figure S1A) but not in males (p for nonlinearity = 0.414; Figure S1B) (Fig. 1).
Fig. 1.
Association between RFM and cognitive impairment. Adjusted odds ratio of cognitive impairment from restricted cubic spline (RCS) logistic regression models with knots at the 5th, 35th, 65th, and 95th percentiles. Adjusted for all covariates (Adjust for age, marital status, education level, diabetes course, diabetic dietary pattern, cerebrovascular diseases, hypoglycemic events in last 3 months, eGFR, TG, smoking status, drink status, regular exercise and hypertension). The solid lines and bands represent the ORs and corresponding 95% CIs
In threshold analysis among females, for RFM values < 43.693, the OR for CI was 1.16 (95% CI: 1.09–1.23, p < 0.001), corresponding to a 16% increase in CI risk per one-unit increase in RFM. No significant association was observed when RFM was ≥43.693, suggesting that CI risk plateaus beyond this threshold (Table 4).
Table 4.
Association between RFM and CI using two-piecewise regression models in females
| RFM | Crude model | Adjusted model* | ||
|---|---|---|---|---|
| OR (95%CI) | P-value | OR (95%CI) | P-value | |
| < 43.693 | 1.16 (1.09~1.23) | < 0.001 | 1.14 (1.07~1.23) | < 0.001 |
| ≥43.693 | 1.04 (0.82~1.31) | 0.74 | 1.00 (0.77~1.30) | 0.997 |
| Nonlinear test | 0.006 | |||
RFM,relative fat mass
*Adjust for age, marital status, education level, diabetes course, diabetic dietary pattern, cerebrovascular diseases, Hypoglycemic events in last 3 months, eGFR, TG, smoking status，drink status, regular exercise and hypertension
Stratified analyses
Figure 2 (females) and Fig. 3 (males) show the results of subgroup analyses. No significant interactions were observed in any subgroups stratified by age, history of CVD, or hypoglycemic events in the past 3 months.
Fig. 2.
Stratified analyses of the association between RFM and cognitive impairment according to baseline characteristics in female group. NFM was divided into four levels by quartile(19.84 < Q1≤37.72; 37.72 < Q2≤40.02;40.02 < Q3≤42.86;42.86 < Q4≤52.26). Note: the p value for interaction represents the likelihood of interaction between the variable and the RFM. Abbreviations: OR,odd ratio; CI, confidence interval., CVD,Cerebrovascular diseases, hypoglycemia, hypoglycemic events in last 3 months
Fig. 3.
Stratified analyses of the association between RFM and cognitive impairment according to baseline characteristics in male group. NFM was divided into four levels by quartile (13.71 < Q1≤24.24; 24.24 < Q2≤26.64;26.64 < Q3≤29.09;29.09 < Q4≤36.48)
Discussion
In this single-center, cross-sectional study of 1328 middle-aged and older inpatients with T2DM, we examined the sex-specific association between RFM and CI. Our primary finding was a nonlinear, inverse J-shaped relationship between RFM and CI in female patients, which remained significant after extensive adjustment for demographic, clinical, and metabolic confounders. In contrast, no significant association was observed in males across any of the models. This sex-specific difference was further supported by restricted cubic spline and threshold analyses, which identified an inflection point at an RFM of approximately 43.7 in women, beyond which the risk of CI plateaued.
The sex-specific nature of our findings provides an important context for interpreting the often-inconsistent literature on obesity and cognitive health. Our null findings in males are consistent with several prior studies. For example, a large study in an older Chinese population reported no evidence that general obesity increased the risk of CI [23]. In contrast, research from the ELSI-Brazil study found that the relationship between glycemic control and cognitive decline was more pronounced in women [26]. More directly relevant to our analysis, West et al. demonstrated that central obesity, measured by WC, was associated with poorer cognition in older women with T2DM but not in men [24]. The biological mechanisms underlying this sex disparity are likely multifactorial. Estrogen is known to exert neuroprotective effects, including reducing oxidative stress, promoting neuronal survival, and enhancing neurotransmitter function [35]. These actions help preserve cognitive performance in women. However, women are predisposed to accumulating central adiposity, particularly after menopause, and the decline in ovarian hormones has been linked to an increased risk of cardiovascular diseases [36] and potentially to altered neuropathology. For instance, associations between parity and Alzheimer’s disease pathology have been observed exclusively in women [36]. Additionally, women with T2DM often have poorer control over cardiovascular risk factors such as cholesterol and blood pressure [37], which may further exacerbate the adverse effects of central fat on cerebrovascular health and cognitive function.
Our primary finding was an inverse J-shaped association between RFM and CI in female patients, which persisted after adjustment for covariates. This observation indicates that the relationship between RFM and CI is not simply linear: CI risk increased with RFM up to an inflection point around 43.7, beyond which the risk plateaued. First, the steep left portion of the curve (RFM < 43.7) shows that each one-unit increase in RFM was associated with a 16% increase in the odds of CI (OR 1.16, 95% CI: 1.09–1.23), suggesting that rising visceral fat may contribute to insulin resistance [38], chronic low-grade inflammation [5], and microvascular endothelial injury [39]—all of which are established drivers of cognitive decline. Second, above this threshold, additional adiposity did not further increase CI risk in our cross-sectional sample. This finding may reflect survival bias, saturation of inflammatory pathways, or reduced discriminative capacity of RFM at higher levels of adiposity. Such a pattern may partly explain the “obesity paradox” frequently reported in the literature [21, 22] and likely reflects the plateauing of neuroinflammatory processes, underscoring the need for prospective studies with extended follow-up. Moreover, the sex-specific nature of this association reinforces the potential utility of RFM as a simple tool for cognitive risk stratification in women with T2DM, supplementing more traditional measures such as BMI.
Our stratified analyses further revealed that the association between RFM and CI in women was particularly pronounced among those without a history of CVD and among those who had experienced recent hypoglycemic events. This pattern suggests that the adverse cognitive effects of central adiposity may be amplified in the presence of additional metabolic or vascular stressors, or in relatively younger cohorts for whom competing mortality risks are lower. However, the absence of statistically significant interaction effects indicates that, although the strength of the association may vary across subgroups, the underlying relationship is consistently observed, reinforcing the robustness of our primary finding.
This study has some limitations. First, due to its cross-sectional design, temporality and causality could not be established between RFM and CI; it remains unclear whether higher RFM precedes and contributes to CI, or whether CI leads to behavioral or lifestyle changes that elevate RFM. Future longitudinal cohort studies are needed to clarify the directionality of these associations. Second, because this investigation was conducted at a single center and included hospitalized patients, the findings may not be fully generalizable to community-dwelling individuals with T2DM. Hospitalized patients typically have more advanced disease and multiple comorbidities, which may influence both RFM and cognitive status. Third, despite the extensive adjustment for potential confounders and the consistency of results across various analytical approaches, residual confounding from unmeasured factors (e.g., apolipoprotein E genotype, sex hormone levels) cannot be excluded. Lastly, the low use of newer glucose-lowering medications—such as Sodium-Glucose Co-transporter 2 inhibitors and Glucagon-Like Peptide-1 receptor agonists—in our cohort reflects clinical practice during the study period (2018–2022); given emerging evidence that these agents may influence both metabolic and cognitive outcomes, this context should be considered when interpreting our findings.
Despite these limitations, this study has notable strengths, including the use of a validated cognitive assessment tool (MoCA) and robust statistical methods that accounted for nonlinearity and examined stratified associations. From a clinical perspective, our findings suggest that RFM—a simple and low-cost anthropometric measure—probably serve as a valuable sex-specific screening tool for identifying middle-aged and older women with T2DM who are at elevated risk of CI. This could facilitate earlier and more targeted interventions, such as intensified management of cardiovascular risk factors and body composition. From a public health standpoint, integrating RFM assessment into the routine care of patients with DM may help in stratifying the future burden of dementia and allocating resources more effectively.
Conclusion
In conclusion, this study identified a significant sex-specific difference in the association between RFM and CI among hospitalized patients with T2DM. In female patients, a nonlinear, inverse J-shaped relationship was observed: CI risk increased with rising RFM up to approximately 43.7, beyond which the risk plateaued. No significant association was detected in male patients. These findings suggest that RFM may serve as a useful sex-specific indicator for identifying women with T2DM who are at increased risk for cognitive decline. However, the cross-sectional design limits causal inference, underscoring the need for future prospective studies to further clarify this association, elucidate its biological mechanisms, and determine whether interventions aimed at modifying RFM can improve cognitive outcomes in at-risk populations.
Electronic Supplementary Material
Below is the link to the electronic supplementary material.
Acknowledgements
We thank all participants and the Department of Endocrinology at Tianjin Union Medical Center for their support.
Abbreviations
- BMI
Body mass index
- CI
Cognitive impairment
- CVD
Cerebrovascular disease
- DM
Diabetes mellitus
- eGFR
Estimated glomerular filtration rate
- EMR
Electronic medical record
- HbA1c
Glycated hemoglobin
- MCI
Mild cognitive impairment
- MoCA
Montreal Cognitive Assessment
- RFM
Relative fat mass
- TG
Triglycerides
- T2DM
Type 2 diabetes mellitus
- WC
Waist circumference
- WHO
World Health Organization
- WWI
Waist-to-Waist Index; 95% CI: 95% confidence interval
Author contributions
YTL contributed to conception, design, and write the manuscript. YLL contributed to conception, design, and revise the manuscript. HNQ Collected and managed data, performed statistical analysis. JNL and MYZ contributed to conception, study design, supervision and review and approved the final manuscript. All authors reviewed the manuscript.
Funding
This study was supported by the Natural Science Foundation of Tianjin City (18ZXDBSY00120) and the Science and Technology Project of Tianjin Municipal Health Commission (ZD20006) and Natural Science Foundation of Tianjin (25JCYBJC01200).
Data availability
All data from this study are included in the article and supplementary files.
Declarations
Ethics approval and consent to participate
The study was approved by the Institutional Review Board of Tianjin Union Medical Center, Nankai University, following the principles of the Declaration of Helsinki (Approval Number: (2018) Fast-track Review No. C08). Written informed consent was obtained from all participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yanting Liu and Yanlan Liu contributed equally to this work.
Contributor Information
Meiyun Zhang, Email: zmy22202@aliyun.com.
Jingna Lin, Email: 13207628978@163.com.
References
- 1.Realdon O, Rossetto F, Nalin M, Baroni I, Cabinio M, Fioravanti R, et al. Technology-enhanced multi-domain at home continuum of care program with respect to usual care for people with cognitive impairment: the ability-TelerehABILITation study protocol for a randomized controlled trial. BMC Psychiatry. 2016;16(1):425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.You Y, Liu Z, Chen Y, Xu Y, Qin J, Guo S, et al. The prevalence of mild cognitive impairment in type 2 diabetes mellitus patients: a systematic review and meta-analysis. Acta Diabetol. 2021;58(6):671–85. [DOI] [PubMed] [Google Scholar]
- 3.Chen JF, Zhang YP, Han JX, Wang YD, Fu GF. Systematic evaluation of the prevalence of cognitive impairment in elderly patients with diabetes in China. Clin Neurol Neurosur. 2023;225:107557. [DOI] [PubMed] [Google Scholar]
- 4.Tomlin A, Sinclair A. The influence of cognition on self-management of type 2 diabetes in older people. Psychol Res Behav Ma. 2016;9:7–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Miller AA, Spencer SJ. Obesity and neuroinflammation: a pathway to cognitive impairment. Brain Behav Immun. 2014;42:10–21. [DOI] [PubMed] [Google Scholar]
- 6.Guillemot-Legris O, Muccioli GG. Obesity-induced neuroinflammation: beyond the hypothalamus. Trends Neurosci. 2017;40(4):237–53. [DOI] [PubMed]
- 7.Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19.2 million participants. Lancet. 2016;387(10026):1377–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Antonopoulos AS, Oikonomou EK, Antoniades C, Tousoulis D. From the BMI paradox to the obesity paradox: the obesity-mortality association in coronary heart disease. Obes Rev. 2016;17(10):989–1000. [DOI] [PubMed] [Google Scholar]
- 9.Piche ME, Poirier P, Lemieux I, Despres JP. Overview of Epidemiology and contribution of obesity and body fat distribution to cardiovascular disease: an update. Prog Cardiovasc Dis. 2018;61(2):103–13. [DOI] [PubMed] [Google Scholar]
- 10.Ross R, Neeland IJ, Yamashita S, Shai I, Seidell J, Magni P, et al. Waist circumference as a vital sign in clinical practice: a Consensus statement from the IAS and ICCR working group on visceral obesity. Nat Rev Endocrinol. 2020;16(3):177–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Woolcott OO, Bergman RN. Relative fat mass (RFM) as a new estimator of whole-body fat percentage horizontal line a cross-sectional study in American adult individuals. Sci Rep-Uk. 2018;8(1):10980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yu P, Huang T, Hu S, Yu X. Predictive value of relative fat mass algorithm for incident hypertension: a 6-year prospective study in Chinese population. BMJ Open. 2020;10(10):e38420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Xiao B, Cao C, Han Y, Hu H, He Y. Non-linear relationship between relative fat mass and diabetes risk in Japanese adults: a retrospective cohort study. Sci Rep-Uk. 2024;14(1):23496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhu X, Yue Y, Li L, Zhu L, Cai Y, Shu Y. The relationship between depression and relative fat mass (RFM): a population-based study. J Affect Disord. 2024;356:323–28. [DOI] [PubMed] [Google Scholar]
- 15.Kobo O, Leiba R, Avizohar O, Karban A. Relative fat mass is a better predictor of dyslipidemia and metabolic syndrome than body mass index. Cardiovasc Endocr Me. 2019;8(3):77–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zwartkruis VW, Suthahar N, Idema DL, Mahmoud B, van Deutekom C, Rutten FH, et al. Relative fat mass and prediction of incident atrial fibrillation, heart failure and coronary artery disease in the general population. Int J Obes. 2023;47(12):1256–62. [DOI] [PubMed] [Google Scholar]
- 17.Woolcott OO, Bergman RN. Defining cutoffs to diagnose obesity using the relative fat mass (RFM): association with mortality in NHANES 1999-2014. Int J Obes. 2020;44(6):1301–10. [DOI] [PubMed] [Google Scholar]
- 18.Whitmer RA, Gustafson DR, Barrett-Connor E, Haan MN, Gunderson EP, Yaffe K. Central obesity and increased risk of dementia more than three decades later. Neurology. 2008;71(14):1057–64. [DOI] [PubMed] [Google Scholar]
- 19.Qizilbash N, Gregson J, Johnson ME, Pearce N, Douglas I, Wing K, et al. BMI and risk of dementia in two million people over two decades: a retrospective cohort study. Lancet Diabetes Endo. 2015;3(6):431–36. [DOI] [PubMed] [Google Scholar]
- 20.Tang X, Zhao W, Lu M, Zhang X, Zhang P, Xin Z, et al. Relationship between central obesity and the incidence of cognitive impairment and dementia from cohort studies involving 5, 060, 687 participants. NEUROSCI BIOBEHAV R. 2021;130:301–13. [DOI] [PubMed] [Google Scholar]
- 21.Chen J, Zhang J, Wang H, Zhu H, Fu J, Li C, et al. Obesity paradox in cognitive function: a longitudinal study of BMI and cognitive impairment in older adult Chinese population. Front Aging Neurosci. 2025;17:1543501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Dong W, Kan L, Zhang X, Li M, Wang M, Cao Y. Association between body mass index and cognitive impairment in Chinese older adults. Front Public Health. 2023;11:1255101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ren Z, Li Y, Li X, Shi H, Zhao H, He M, et al. Associations of body mass index, waist circumference and waist-to-height ratio with cognitive impairment among Chinese older adults: based on the CLHLS. J Affect Disord. 2021;295:463–70. [DOI] [PubMed] [Google Scholar]
- 24.West RK, Ravona-Springer R, Heymann A, Schmeidler J, Leroith D, Koifman K, et al. Waist circumference is correlated with poorer cognition in elderly type 2 diabetes women. Alzheimers Dement. 2016;12(8):925–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Espeland MA, Yassine H, Hayden KD, Hugenschmidt C, Bennett WL, Chao A, et al. Sex-related differences in cognitive trajectories in older individuals with type 2 diabetes and overweight or obesity. Alzh Dement-Trci. 2021;7(1):e12160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Cochar-Soares N, de Oliveira DC, Luiz MM, Aliberti M, Suemoto CK, Steptoe A, et al. Sex differences in the trajectories of cognitive decline and affected cognitive domains among older adults with controlled and uncontrolled glycemia. J Gerontol A-Biol. 2024;79(7). [DOI] [PMC free article] [PubMed]
- 27.Rettberg JR, Yao J, Brinton RD. Estrogen: a master regulator of bioenergetic systems in the brain and body. Front Neuroendocrin. 2014;35(1):8–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gerdts E, Regitz-Zagrosek V. Sex differences in cardiometabolic disorders. Nat Med. 2019;25(11):1657–66. [DOI] [PubMed] [Google Scholar]
- 29.Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part, 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med. 1998;15(7):539–53. [DOI] [PubMed] [Google Scholar]
- 30.World Health Organization. Use of Glycated haemoglobin (HbA1c) in the diagnosis of diabetes mellitus: abbreviated report of a WHO consultation. In: WHO guidelines approved by the guidelines review committee. Geneva: World Health Organization; 2011. [PubMed] [Google Scholar]
- 31.World Health Organization: wHO classification of diabetes mellitus. Geneva: World Health Organization; 2019.
- 32.Nasreddine ZS, Phillips NA, Bédirian V, Charbonneau S, Whitehead V, Collin I, et al. The Montreal cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53(4):695–99. [DOI] [PubMed] [Google Scholar]
- 33.Jia W, Weng J, Zhu D, Ji L, Lu J, Zhou Z, et al. Standards of medical care for type 2 diabetes in China 2019. Diabetes-METAB Res. 2019;35(6):e 3158. [DOI] [PubMed] [Google Scholar]
- 34.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AR, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Arevalo MA, Azcoitia I, Garcia-Segura LM. The neuroprotective actions of oestradiol and oestrogen receptors. Nat Rev Neurosci. 2015;16(1):17–29. [DOI] [PubMed] [Google Scholar]
- 36.Beeri MS, Rapp M, Schmeidler J, Reichenberg A, Purohit DP, Perl DP, et al. Number of children is associated with neuropathology of Alzheimer’s disease in women. Neurobiol Aging. 2009;30(8):1184–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Franzini L, Ardigo D, Cavalot F, Miccoli R, Rivellese AA, Trovati M, et al. Women show worse control of type 2 diabetes and cardiovascular disease risk factors than men: results from the MIND.IT study group of the Italian society of Diabetology. Nutr Metab Cardiovas. 2013;23(3):235–41. [DOI] [PubMed] [Google Scholar]
- 38.Naomi R, Teoh SH, Embong H, Balan SS, Othman F, Bahari H, et al. The role of oxidative stress and inflammation in obesity and its impact on cognitive impairments-A narrative Review. Antioxidants-Basel. 2023;12(5). [DOI] [PMC free article] [PubMed]
- 39.Gorelick PB, Scuteri A, Black SE, Decarli C, Greenberg SM, Iadecola C, et al. Vascular contributions to cognitive impairment and dementia: a statement for healthcare professionals from the American heart association/American stroke association. Stroke. 2011;42(9):2672–713. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data from this study are included in the article and supplementary files.