Weight-adjusted waist index and cognitive impairment in older adults: A nationwide longitudinal cohort study Authors: Hongli Nie, Xuexue Huang, Xiaohong Yu, Wenzhen Yang, Peiwen Li Published: 2026 ABSTRACT The association between adiposity and cognitive impairment remains inconclusive. This study aimed to evaluate the relationship between weight-adjusted waist index (WWI), a newly proposed indicator of adiposity, and cognitive impairment in a Chinese middle-aged and elderly population. Data from the 2011–2018 China Health and Retirement Longitudinal Study were used in this study. The WWI was calculated as waist circumference divided by the square root of body weight. Multivariate Cox regression models were employed to evaluate the association between WWI and cognitive impairment, while restricted cubic spline analysis was conducted to explore potential nonlinear associations between WWI and cognitive impairment. Model performance was assessed using Harrell’s C-index, continuous net reclassification improvement, and integrated discrimination improvement to evaluate the incremental predictive value of WWI. The study included 5333 participants aged ≥45 years, comprising 2928 males (54.9%) and 2405 females (45.1%), with a mean age of 56.9 years (standard deviation = 8.2). During the 7-year follow-up, 1334 participants (25.0%) developed cognitive impairment. After adjusting for potential confounders, the WWI was positively associated with cognitive impairment (hazard ratio [HR] = 1.13, 95% confidence interval [CI]: 1.05–1.21, P < .001). Compared with the lowest quartile (Q1), participants in higher WWI quartiles exhibited an increased risk of cognitive impairment: Q2 (HR = 1.05, 95% CI: 0.89–1.24, P = .559), Q3 (HR = 1.13, 95% CI: 0.96–1.34, P = .144), and Q4 (HR = 1.26, 95% CI: 1.06–1.49, P = .009). Restricted cubic spline regression analysis demonstrated a positive linear relationship between WWI and cognitive impairment risk (P for nonlinearity = .678). Furthermore, the continuous NRI and integrated discrimination improvement for WWI demonstrated statistically significant improvements in predicting cognitive impairment compared with traditional adiposity metrics (both P < .05). The findings were further validated through a series of sensitivity analyses. This study demonstrates a significant association between higher WWI levels and an increased risk of cognitive impairment among middle-aged and older Chinese adults, suggesting that maintaining lower WWI levels may be associated with a lower risk of cognitive impairment. 1. Introduction Cognitive impairment has emerged as a major public health concern.[1] It is characterized by deterioration in one or more cognitive domains (including memory, attention, language, executive function, visuospatial ability, comprehension, judgment, and calculation),[2] Global dementia cases are projected to rise from 57 million in 2019 to 153 million by 2050.[3] Similarly, dementia cases among Chinese adults aged ≥50 years are estimated to reach 66.3 million by 2050.[4] This rising prevalence substantially threatens older adults’ quality of life and imposes severe burdens on healthcare systems and society. Therefore, the early identification of risk factors for cognitive impairment is crucial for implementing timely interventions and mitigating disease progression. Obesity, with its rising prevalence among middle-aged and elderly populations,[5,6] is not only associated with metabolic diseases such as cardiovascular disease and diabetes,[7,8] but also adversely affects the central nervous system and cognitive function.[9,10] While several studies report a positive association between obesity and cognitive impairment or dementia in older adults,[11,12] recent evidence suggests that obesity does not consistently elevate dementia risk,[13] Some studies have even indicated a potential protective association.[14,15] These inconsistencies may stem from overreliance on body mass index (BMI), which cannot differentiate adipose tissue from muscle mass or assess central adiposity.[16] Therefore, developing more precise metrics to characterize obesity heterogeneity and its cognitive implications is critically essential. The weight-adjusted waist index (WWI), a novel anthropometric indicator reflecting both fat and muscle mass, was calculated as waist circumference (WC) divided by the square root of body weight.[17] Existing studies indicate that WWI outperforms both WC and BMI in predicting risks of specific diseases (e.g., hypertension and cardiovascular diseases),[18,19] and adverse clinical outcomes, including mortality.[20] Thus, WWI may serve as a more reliable metric for predicting cognitive impairment associated with adiposity. Nevertheless, existing evidence on the association between WWI and cognitive impairment is predominantly cross-sectional,[21–24] and the sole prospective investigation to date has been limited to elderly Chinese males.[25] To expand this limited longitudinal evidence and validate findings across broader populations, we conducted a prospective cohort analysis using nationally representative data from the China Health and Retirement Longitudinal Study (CHARLS, 2011-2018 waves). 2. Methods 2.1. Study design and population CHARLS is a nationally representative survey of middle-aged and older Chinese adults (aged ≥45 years), employing a stratified, multistage probability sampling design across 450 villages in 150 counties (28 provinces). Briefly, the baseline survey included 17,708 respondents and was conducted in 2011, with subsequent biennial follow-ups through 2020. Detailed information about CHARLS has been published elsewhere.[26] The CHARLS program was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-11015), and all participants provided written informed consent. This study used data from Wave 1 (2011–2012) through Wave 4 (2018), with Wave 1 as the baseline. Participants were excluded based on: age <45 years; missing WC, weight, or height data; WC outliers (<45 or>150 cm); missing demographic data; absent baseline cognitive assessment data; baseline brain injury/intellectual disability; baseline cognitive impairment; or were lost to follow-up. After exclusions, we enrolled 5333 participants in longitudinal analysis. The participant selection flowchart is shown in Figure 1. Flow chart shows the selection of the participants in this study. 2.2. Weight-adjusted waist index WWI was used as the independent variable and was calculated as the WC (cm) divided by the square root of body weight (kg).[17] Standardized physical examinations were performed by trained medical staff, with quality control by research personnel. WC was measured at the midpoint between the lowest rib and the iliac crest at the end of normal expiration using a nonelastic tape. Body weight was assessed using a calibrated digital scale (OmronTMHN-286, Yangzhou Korell Technology Co, Ltd.) with participants wearing light examination gowns and standing barefoot in an upright position. 2.3. Assessment of cognitive function Cognitive function in the CHARLS was assessed across 2 dimensions: episodic memory and mental status.[27,28] For episodic memory, participants were asked to immediately recall ten words read aloud by an interviewer (immediate recall). Approximately after a 5-minute interval, they recalled the same words again (delayed recall). Each correctly recalled word was scored as 1 point (total range: 0–20). For mental status, participants completed a series of tasks: serial subtraction of 7 from 100 for 5 times; answering questions regarding the current date, season, and day of the week; and redrawing a previously displayed figure. Each correct answer earned 1 point (maximum: 11 points). A global cognitive score (range: 0–31) was derived by summing the scores from both domains, with higher scores indicating better function. Cognitive impairment was defined as a total score <11.[29,30] 2.4. Potential covariates Potential covariates were selected through literature reviews, encompassing sociodemographic characteristics, lifestyle factors, and health status indicators. Sociodemographic characteristics included age, sex (male/female), educational level (junior high school and below/high school or above), marital status (married/other marital status), and residence (rural/urban). Lifestyle factors included smoking status (never/former/current), and drinking status (never/former/current). Health status indicators included self-reported physician-diagnosed hypertension and diabetes, as well as the use of corresponding medications (antihypertensive and glucose-lowering drugs, respectively). 2.5. Statistical analysis Continuous variables were presented as mean ± standard deviation or median with interquartile range, and categorical variables as frequencies (percentages). Intergroup differences across WWI quartiles were assessed using one-way ANOVA or the Kruskal-Wallis H test for continuous variables and χ2 tests for categorical variables. Cox proportional hazards regression models were used to evaluate associations between WWI and cognitive impairment. Kaplan–Meier curves were generated to determine differences in the occurrence of cognitive impairment across WWI quartiles. Restricted cubic spline regression was utilized to examine potential nonlinear relationships between WWI and cognitive impairment. Stratified analyses were conducted by age, sex, residence, education level, marital status, smoking status, drinking status, hypertension, diabetes, antihypertensive drugs, and glucose-lowering drugs. Several sensitivity analyses were performed to test the stability of the results by excluding participants who developed cognitive impairment <2 waves after baseline, or excluding participants with hypertension, or excluding participants with diabetes. To evaluate the incremental predictive value of WWI compared to conventional adiposity metrics, we assessed model discrimination using Harrell’s C-index, model reclassification using continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI).[31] All statistical analyses were performed using R software (version 4.4.3), with statistical significance defined as P < .05. 3. Results 3.1. Basic characteristics of participants The study included 5333 participants. The mean age of all participants was 56.9 ± 8.2 years, with 54.9% being males. Baseline characteristics stratified by WWI quartiles are shown in Table 1. Compared to the lowest WWI quartile, participants in the higher WWI were older, female, less educated, non-married, lived in urban areas, did not smoke or drink, had higher BMI, and had a higher prevalence of hypertension and diabetes. During the 7-year follow-up, 1334 participants developed cognitive impairment (incidence rate: 25.0%, 1334/5333). Table 1Baseline characteristics of participants categorized by WWI quartiles.VariablesOverall (n = 5333)Q1 (n = 1333)Q2 (n = 1334)Q3 (n = 1333)Q4 (n = 1333)P-valueAge, (y)56.9 ± 8.255.1 ± 7.655.8 ± 7.956.9 ± 8.159.9 ± 8.3<.001Gender, n (%)<.001Male2928 (54.9)977 (73.3)873 (65.4)697 (52.3)381 (28.6)Female2405 (45.1)356 (26.7)461 (34.6)636 (47.7)952 (71.4)Education, n (%)<.001Junior high school and below4483 (84.1)1045 (78.4)1091 (81.8)1124 (84.3)1223 (91.7)High school or above850 (15.9)288 (21.6)243 (18.2)209 (15.7)110 (8.3)Residence, n (%)<.001Urban2134 (40.0)452 (33.9)496 (37.2)597 (44.8)589 (44.2)Rural3199 (60.0)881 (66.1)838 (62.8)736 (55.2)744 (55.8)Marital status, n (%).044Married4645 (87.1)1174 (88.1)1171 (87.8)1169 (87.7)1131 (84.8)Other marital status688 (12.9)159 (11.9)163 (12.2)164 (12.3)202 (15.2)Drinking status, n (%)<.001Never2932 (55.0)616 (46.2)635 (47.6)742 (55.7)939 (70.4)Former432 (8.1)99 (7.4)117 (8.8)116 (8.7)100 (7.5)Current1969 (36.9)618 (46.4)582 (43.6)475 (35.6)294 (22.1)Smoking status, n (%)<.001Never3010 (56.4)574 (43.1)672 (50.4)793 (59.5)971 (72.8)Former500 (9.4)128 (9.6)142 (10.6)139 (10.4)91 (6.8)Current1823 (34.2)631 (47.3)520 (39.0)401 (30.1)271 (20.3)Hypertension, n (%)<.001No4039 (75.7)1126 (84.5)1069 (80.1)969 (72.7)875 (65.6)Yes1294 (24.3)207 (15.5)265 (19.9)364 (27.3)458 (34.4)Diabetes, n (%)<.001No4997 (93.7)1296 (97.2)1269 (95.1)1236 (92.7)1196 (89.7)Yes336 (6.3)37 (2.8)65 (4.9)97 (7.3)137 (10.3)antihypertensive drugs, n (%)<.001No4433 (83.1)1206 (90.5)1146 (85.9)1081 (81.1)1000 (75.0)Yes900 (16.9)127 (9.5)188 (14.1)252 (18.9)333 (25.0)glucose-lowering drugs, n (%)<.001No5144 (96.5)1315 (98.6)1298 (97.3)1275 (95.7)1256 (94.2)Yes188 (3.5)18 (1.4)36 (2.7)57 (4.3)77 (5.8)WWI11.0 ± 0.810.1 ± 0.510.8 ± 0.111.3 ± 0.212.1 ± 0.5<.001BMI, (kg/m2)23.8 ± 3.822.3 ± 4.123.4 ± 3.124.6 ± 3.525.0 ± 3.9<.001Duration of follow-up, (mo)75.7 ± 18.578.1 ± 15.976.9 ± 17.576.2 ± 18.071.8 ± 21.7<.001Cognitive impairment, n (%)1334 (25.0)261 (19.6)285 (21.4)324 (24.3)464 (34.8)<.001Continuous variables were presented as mean (SD) or median (IQR) according to their distribution and category variables were shown as frequencies (percentage).P values of continuous variables were calculated with one-way ANOVA or the Kruskal-Wallis H test, and Chi-square tests for category variables.BMI = body mass index, IQR = interquartile range, SD = standard deviation, WWI = weight-adjusted waist index. Baseline characteristics of participants categorized by WWI quartiles. Continuous variables were presented as mean (SD) or median (IQR) according to their distribution and category variables were shown as frequencies (percentage). P values of continuous variables were calculated with one-way ANOVA or the Kruskal-Wallis H test, and Chi-square tests for category variables. BMI = body mass index, IQR = interquartile range, SD = standard deviation, WWI = weight-adjusted waist index. We compared baseline characteristics between participants included in the final analysis (n = 5333) and those excluded due to missing data on cognitive impairment or other key covariates (n = 11,598). This procedure aimed to assess the potential for selection bias and to confirm the generalizability of the study’s findings. As shown in Table S1, Supplemental Digital Contentsignificant differences were observed for several demographic and health-related variables (all P < .05). Compared with excluded individuals, the included cohort was younger, more likely to be male and married, and had a higher education level. Regarding health-related variables, the included group exhibited a higher prevalence of current smoking and alcohol consumption, but a lower prevalence of hypertension. No significant difference was observed for diabetes prevalence, rural residence, BMI, or WWI. 3.2. Association between WWI and cognitive impairment As presented in Table 2, in the fully adjusted model, WWI as a continuous variable was significantly associated with an increased risk of cognitive impairment (hazard ratio [HR] 1.13, 95% confidence interval [CI) 1.05–1.21, P < .001). When analyzing WWI as quartiles, compared to the lowest quartile (Q1), the risk progressively increased, with the highest quartile (Q4) demonstrating a statistically significant 26% higher risk (HR 1.26, 95% CI 1.06–1.49, P = .009), while Q2 and Q3 showed no significant associations. The overall trend across quartiles was significant (P for trend = .006). Table 2Adjusted HRs (95% CI) of cognitive impairment according to WWI quartiles.VariableEvents, n (%)Crude modelModel 1Model 2HR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueWWI1334 (25.0)1.39 (1.31-1.49)<.0011.10 (1.03–1.18).0051.13 (1.05–1.21)<.001Q1 (<10.53)261 (19.6)ReferenceReferenceReferenceQ2 (10.53–11.01)285 (21.4)1.10 (0.93–1.31).2481.04 (0.88–1.24).6111.05 (0.89–1.24).559Q3 (11.01–11.54)324 (24.3)1.29 (1.10–1.52).0021.08 (0.91–1.27).3671.13 (0.96–1.34).144Q4 (>11.54)464 (34.8)1.94 (1.67–2.26)<.0011.23 (1.04–1.46).0151.26 (1.06–1.49).009P for trend1.48 (1.36–1.61)<.0011.13 (1.03–1.24).0111.14 (1.04–1.25).006Crude model: unadjusted.Model 1: Cox proportional hazards regression models adjusted for age (continuous), gender (female/male).Model 2: Model 1 plus additionally adjusted for educational level (junior high school and below/high school or above), marital status (married/other marital status), residence (rural or urban), smoking status (never/former/current), drinking status (never/former/current), hypertension (yes/no), Diabetes (yes/no), antihypertensive drugs (yes/no) and glucose-lowering drugs (yes/no).CI = confidence interval, HR = hazard ratio, WWI = weight-adjusted waist index. Adjusted HRs (95% CI) of cognitive impairment according to WWI quartiles. Crude model: unadjusted. Model 1: Cox proportional hazards regression models adjusted for age (continuous), gender (female/male). Model 2: Model 1 plus additionally adjusted for educational level (junior high school and below/high school or above), marital status (married/other marital status), residence (rural or urban), smoking status (never/former/current), drinking status (never/former/current), hypertension (yes/no), Diabetes (yes/no), antihypertensive drugs (yes/no) and glucose-lowering drugs (yes/no). CI = confidence interval, HR = hazard ratio, WWI = weight-adjusted waist index. As shown in Figure 2, Kaplan–Meier curves revealed corresponding patterns, with Q4 exhibiting progressively lower cognitive impairment-free survival compared to Q1 over the follow-up period, while Q2 and Q3 showed trajectories comparable to the reference group (log-rank P < .001). The restricted cubic spline analysis further confirmed a positive linear relationship between WWI levels and cognitive impairment (P for nonlinearity = .678) (Fig. 3). The KM survival curves for cognitive impairment across WWI quartiles. KM = Kaplan–Meier, WWI = weight-adjusted waist index. Restricted cubic spline analysis with multivariate-adjusted associations between WWI and cognitive impairment. Adjusted for age, gender, educational level, marital status, residence, smoking status, drinking status, hypertension, diabetes, antihypertensive drugs, and glucose-lowering drugs. CI = confidence interval, HR = hazard ratio, WWI = weight-adjusted waist index. 3.3. Subgroup and sensitivity analyses As shown in Figure 4, the subgroup analysis indicated no significant interactions among subgroup variables after adjusting for confounding factors. This implies that the impact of WWI on cognitive impairment was generally stable. Forest plot of the relationship between WWI and cognitive impairment in different subgroups. Adjusted for age, gender, educational level, marital status, residence, smoking status, drinking status, hypertension, diabetes, antihypertensive drugs, and glucose-lowering drugs, except for the stratified factor. BMI = body mass index, CI = confidence interval, HR = hazard ratio, WWI = weight-adjusted waist index. Repeating the primary analyses in sensitivity analyses, by excluding participants who developed cognitive impairment <2 waves after baseline, or those with hypertension, or diabetes at baseline, yielded similar results (Table S2,S3 and S4, Supplemental Digital Content). 3.4. Incremental predictive value of WWI over conventional adiposity metrics As shown in Table 3, the addition of WWI or conventional adiposity metrics modestly improved the C-index for cognitive impairment (all P < .001). Furthermore, WWI uniquely demonstrated significant enhancement across both reclassification indices, with continuous NRI improved by 0.106, and IDI improved by 0.002 (both P = .004). BMI showed a modest reclassification benefit (NRI = 0.069, P = .046) but a nonsignificant IDI, while WC did not reach statistical significance for either reclassification metric. These results suggest that WWI provides more consistent and clinically relevant incremental predictive information beyond conventional adiposity metrics in this aging population. Table 3The incremental predictive value of WWI and conventional obesity metrics for cognitive impairment.C-index (95% CI)P valueContinuousNRI, % (95% CI)P valueIDI (95% CI)P valueBasic model0.728 (0.715–0.742)ReferenceReferenceBasic model + WWI0.730 (0.716–0.743)<0.0010.106 (0.033–0.169).0040.002 (0.001–0.004)0.004Basic model + BMI0.731 (0.717–0.744)<0.0010.069 (0.005–0.151).0460.001 (-0.0001–0.003)0.080Basic model + WC0.730 (0.716–0.743)<0.0010.068 (-0.005–0.144).0740.001 (−0.0004–0.002)0.192The basic model included all variables of fully adjusted Model 2 except for WWI.BMI = body mass index, CI = confidence interval, HR = hazard ratio, IDI = integrated discrimination improvement, NRI = net reclassification improvement, WC = waist circumference, WWI = weight-adjusted waist index. The incremental predictive value of WWI and conventional obesity metrics for cognitive impairment. The basic model included all variables of fully adjusted Model 2 except for WWI. BMI = body mass index, CI = confidence interval, HR = hazard ratio, IDI = integrated discrimination improvement, NRI = net reclassification improvement, WC = waist circumference, WWI = weight-adjusted waist index. 4. Discussion To our knowledge, this is among the few longitudinal studies to explore the association between WWI and cognitive function in aging Chinese populations. After comprehensive adjustment for covariates, each 1-unit increase in WWI was associated with a 13% higher risk of cognitive impairment (HR = 1.13, 95% CI: 1.05–1.21). This association remained consistent across all subgroups, confirming its robustness. WWI demonstrated statistically significant improvements in both continuous NRI and IDI, indicating that WWI provides incremental predictive value beyond traditional adiposity metrics by more accurately reclassifying individuals into appropriate risk categories. The relationship between obesity and cognitive impairment has been extensively studied. Some studies have found a positive association between high BMI and cognitive impairment.[32–36] However, Liang et al reported that BMI-defined overweight and obesity reduced the risks of cognitive impairment, whereas underweight status increased risk.[37] These discrepancies may reflect the limited sensitivity of BMI for diagnosing obesity in elderly populations.[38] Furthermore, Tang et al found that high WC was associated with a greater risk of cognitive impairment and dementia.[39] Conversely, Liang et al found that higher WC was associated with slower cognitive decline.[40] In conclusion, despite extensive research on these anthropometric indicators of adiposity and cognitive impairment, their diagnostic utility remains controversial. As an alternative, the WWI, a metric incorporating WC with reduced BMI dependence, may offer superior diagnostic and prognostic utility for conditions related to adiposity. Li et al found that elevated WWI levels were associated with higher risks of cognitive decline.[21] Zhou et al found that WWI was independently associated with dementia risk.[24] A prospective cohort study of 1457 elderly Chinese men found a significant positive association between WWI and cognitive decline, with particularly pronounced effects above a WWI threshold of 11.52 cm/√kg.[25] In our study, higher WWI values were also associated with increased cognitive impairment risk. Participants in the highest WWI quartile (Q4 > 11.54 cm/√kg) exhibited a 26% significantly increased risk of cognitive impairment (HR 1.26, 95% CI: 1.06–1.49) compared to Q1. These findings solidify WWI as an independent risk factor for cognitive impairment. While the precise mechanisms linking higher WWI to cognitive impairment remain to be fully elucidated, several interrelated pathways may explain this association. First, adipose tissue is metabolically active, and its accumulation is closely linked to increased oxidative stress and mitochondrial dysfunction,[41,42] characterized by elevated production of reactive oxygen species. Both oxidative stress and mitochondrial dysfunction have been implicated in the pathogenesis of chronic low-grade inflammation, which is a well-established risk factor for cognitive decline and neurodegeneration.[43] Second, beyond systemic effects, excess adiposity may directly influence muscle health. Fat infiltration into skeletal muscle can impair muscle function and contribute to the development of sarcopenic obesity.[44] Skeletal muscle functions as an endocrine organ, secreting myokines such as brain-derived neurotrophic factor and insulin-like growth factor-1, which exert neuroprotective effects.[45] Reduced muscle mass may diminish this protective signaling, thereby accelerating neurodegeneration.[46] This combined effect of increased adiposity and reduced muscle mass, captured by the WWI, may synergistically contribute to cognitive decline through the pathways described above. Our study has several advantages. First, we incorporated a total of 5333 individuals drawn from a nationally representative sample. The substantial sample size supported comprehensive subgroup analyses, enabling rigorous evaluation of the WWI-cognitive impairment association across diverse demographic strata. This analytical approach significantly strengthens the statistical robustness and credibility of our findings. Moreover, this prospective study with a 7-year follow-up enabled assessment of cognitive impairment incidence and evaluation of potential causal associations between WWI and cognitive impairment. However, our study is not without its limitations. First, despite adjusting for numerous potential confounders, the possibility of residual confounding exists, especially considering unmeasured variables like genetic factors, dietary habits, and physical activity levels, which could potentially impact the incidence of cognitive impairment. Second, this study did not account for changes in body weight or WC during the 7-year follow-up period. In aging populations, such fluctuations are common and may reflect underlying health transitions that could influence cognitive performance. Therefore, future studies with repeated anthropometric measurements are warranted to validate our findings. Third, since this study was restricted to middle-aged and older Chinese adults, it is not known whether the results show heterogeneity across other populations. Thus, future studies should aim to address these limitations to validate our findings and explore their generalizability. 5. Conclusion Among Chinese older adults, elevated WWI levels demonstrated a significant association with increased risk of cognitive impairment. This relationship remained robust across diverse subgroups, supporting WWI’s validity as a consistent and generalizable biomarker for cognitive risk assessment. Overall, these findings suggest that WWI may represent a modifiable target for strategies aimed at preventing cognitive impairment in aging populations. Acknowledgments We thank all the participants for their contribution to the CHARLS. Author contributions Conceptualization: Hongli Nie, Xuexue Huang. Data curation: Hongli Nie. Formal analysis: Hongli Nie. Methodology: Hongli Nie. Writing – original draft: Hongli Nie, Xuexue Huang. Writing – review & editing: Hongli Nie, Xuexue Huang, Xiaohong Yu, Wenzhen Yang, Peiwen Li.