The Clinical Outcomes of Preclinical and Clinical Obesity Across Multiple Cohorts

ABSTRACT

Objective

The Lancet Diabetes & Endocrinology Commission proposed a new definition for clinical obesity. However, its prognostic value remains unclear.

Methods

Clinical obesity was defined using the Commission's criteria in the China Health and Retirement Longitudinal Study (CHARLS), the English Longitudinal Study of Aging (ELSA), and the National Health and Nutrition Examination Survey (NHANES). Prevalence estimation was conducted across four categories based on obesity and organ dysfunction status. Cox proportional hazards models were used to assess associations with clinical outcomes.

Results

The prevalence of clinical obesity was 2.50% (CHARLS), 6.18% (ELSA), and 9.08% (NHANES), respectively. Overall, older adults and men had lower prevalence, while older men in NHANES showed higher prevalence. Compared with non‐obesity without dysfunctions, clinical obesity and non‐obesity with dysfunctions showed higher risks of cardiovascular outcomes. Compared with preclinical obesity, clinical obesity elevated cardiovascular risk in CHARLS (HR = 2.49, 95% CI: 1.32–4.70) and ELSA (HR = 1.80, 95% CI: 1.02–3.17) and increased all‐cause (HR = 1.88, 95% CI: 1.23–2.87) and cardiovascular (HR = 3.11, 95% CI: 1.49–6.50) mortality risk in NHANES.

Conclusions

The prevalence of clinical obesity varied by age, sex, and country. Overall, the prevalence was lower among older adults and men. Clinical obesity was associated with increased cardiovascular event risk across cohorts and increased all‐cause mortality risk in NHANES, highlighting its value for risk stratification.

1. Introduction

Obesity has become a major global public health concern, with rising prevalence worldwide [1]. Recent epidemiological data indicate that ~40% of adults in the United States [2], 19% in the United Kingdom [3], and 16% in China [4] are affected by obesity. As a multifactorial condition, obesity substantially increases the risk of numerous adverse health outcomes, including premature all‐cause mortality, cardiovascular diseases (CVDs), type 2 diabetes, and various types of cancer, thereby imposing a considerable burden on individuals, healthcare systems, and society [5].

Body mass index (BMI) has long been used as the primary criterion for defining obesity owing to its simplicity. However, BMI does not account for body fat distribution or differentiate between adipose and lean mass, which may result in misclassification of obesity status [6, 7]. Recently, the Lancet Diabetes & Endocrinology Commission proposed a new definition of clinical obesity, emphasizing precise identification through multiple measures of adiposity distribution—such as body composition assessment, waist circumference, waist to hip ratio, and waist to height ratio—as well as evaluation of organ or tissue dysfunctions and limitations in daily activities resulting from excess fat accumulation. Within this framework, preclinical obesity refers to excess body fat accumulation without evidence of organ dysfunctions or limitations in daily activities. This new definition aims to reframe obesity as a chronic clinical disease and to improve the identification and management of obesity‐related complications [8].

The new definition provides important insights for improving the precision management of obesity, prompting growing interest in its potential clinical application. Recently, a study using data from the All of Us Research Program investigated the prevalence of clinical obesity based on this framework [9]. Another analysis of NHANES 2017–2020 data proposed a novel approach to adjust for organ dysfunctions by considering the proportion of the same dysfunctions among individuals without obesity [10]. Several analyses in UK Biobank (UKB) evaluated the transitions from preclinical obesity to obesity‐induced dysfunction and mortality [11], as well as associations of clinical obesity with autoimmune diseases [12], cancer [13], and all‐cause mortality [14]. Despite these preliminary efforts, the prevalence of clinical obesity, as well as its prognostic implications for adverse outcomes such as mortality and cardiovascular events across different populations, remains inadequately characterized.

This study utilizes longitudinal data from multiple cohorts to characterize the prevalence and clinical outcomes of clinical and preclinical obesity and employs different reference subgroups to compare the risks of adverse outcomes between these two conditions.

2. Methods

2.1. Study Population

The China Health and Retirement Longitudinal Study (CHARLS) is a longitudinal study of Chinese participants aged 45 and above initiated in 2011, with follow‐ups every 2–3 years until 2020 [15]. The CHARLS protocols were approved by the Peking University Institutional Review Board (PU IRB) (IRB00001052‐11014, IRB00001052‐11015) in accordance with the Declaration of Helsinki. The English Longitudinal Study of Aging (ELSA) is a nationally representative cohort of participants aged 50 and above in England with regular follow‐ups every 2 years [16]. The ELSA study was approved by the London Multicentre Research Ethics Committee (MREC/01/2/91).

The US National Health and Nutrition Examination Survey (NHANES) is a representative cross‐sectional study of residents in the United States with a national, multistage, stratified, clustered probability sampling design (https://www.cdc.gov/nchs/nhanes/) [17]. The protocols of NHANES were approved by the Research Ethics Review Board of the National Center for Health Statistics (NCHS). We selected participants from NHANES 1999–2006 as baseline data.

Participants with missing data on age, sex, BMI, waist circumference (WC), body composition measurements, or follow‐up were excluded. In NHANES, participants aged < 18 years were additionally excluded (Figure S1A–C). For analysis of new‐onset CVDs in CHARLS and ELSA, we excluded participants with CVDs at baseline. In the analysis evaluating the association between obesity status and clinical outcomes across the three cohorts, participants with missing data on covariates (including race [in ELSA and NHANES], education level, smoking status, alcohol drinking status) were excluded.

2.2. Data Acquisition

Information on demographic characteristics (including age, sex, race [in ELSA and NHANES], education level [> 12 years of education, yes/no], smoking status [current smoking, yes/no], alcohol consumption status [current alcohol consumption, yes/no], and health status) was collected through standardized questionnaires in CHARLS, ELSA, and NHANES. Blood pressure, comprising systolic blood pressure (SBP) and diastolic blood pressure (DBP), was measured three times in the mobile examination center (MEC) or during home examinations using a mercury sphygmomanometer, and the mean value was used. Blood specimens, including blood glucose, glycated hemoglobin (HbA1c), C‐reactive protein (CRP), total cholesterol, total triglycerides, and HDL‐cholesterol (HDL‐C), were collected following standardized procedures. Details of examination protocols have been described on the website of CHARLS, ELSA and NHANES. (https://charls.pku.edu.cn/, https://www.elsa‐project.ac.uk/data‐and‐documentation, https://www.cdc.gov/nchs/nhanes/).

2.3. Definitions

Clinical obesity and preclinical obesity were defined according to the criteria from Lancet Diabetes & Endocrinology Commission. The Commission defined clinical obesity as a condition of illness resulting from organ dysfunctions directly attributable to excess adiposity, with excess adiposity defined according to the relevant national guidelines for each cohort and clinical obesity classified accordingly within each cohort [8]. In CHARLS, participants with BMI ≥ 28 kg/m² and WC ≥ 90 cm (for men) or ≥ 85 cm (for women) were classified as having excess fat accumulation. In ELSA and NHANES, participants with BMI ≥ 30 kg/m² and WC ≥ 102 cm (for men) or ≥ 88 cm (for women) were classified as having excessive fat. As NHANES provided data on body composition, we additionally defined excess adiposity in the sensitivity analysis as a body fat percentage > 25% in men and > 35% in women measured by dual‐energy X‐ray absorptiometry (DXA). In all cohorts, individuals with BMI > 40 kg/m² were directly classified as having excessive fat accumulation. Considering differences in data availability across cohorts, organ dysfunctions were assessed by questionnaires covering the following conditions: central nervous system disorders (in NHANES and CHARLS), respiratory disorders (in NHANES), cardiovascular disorders (in NHANES and ELSA), raised blood pressure, metabolism disorders, renal and urinal disorders (in NHANES and CHARLS), and musculoskeletal system disorders. Because ascertainment of heart failure differed across cohorts, history of heart failure was included as an organ dysfunction criterion in NHANES and ELSA, but not in CHARLS, where consistently ascertained baseline heart failure data were unavailable. Limitations in daily activities, including dressing, bathing, eating, walking, climbing stairs, picking up a coin, stooping, kneeling, crouching, and reaching arms, were assessed using self‐administered questionnaires (Table S1) [8].

Participants with excessive fat accumulation were classified as having clinical obesity if they also exhibited evidence of one or more organ dysfunctions or limitations in daily activities attributable to obesity. Those without such dysfunctions or limitations were classified as having preclinical obesity. Participants who did not meet any excess adiposity criterion were classified as participants without obesity. Those who met only one excess adiposity criterion were categorized as “others” and excluded from subsequent outcome analyses (Figure S1).

2.4. Study Outcomes

The primary outcome across the three cohorts was all‐cause mortality. The secondary outcomes differed due to differences in follow‐up information among cohorts. In NHANES, it was cardiovascular mortality defined using ICD‐10 coding acquired from NCHS, whereas in CHARLS and ELSA, because cause‐specific mortality data were not available, the secondary outcome was new‐onset CVDs ascertained through self‐reported questionnaires.

In CHARLS, mortality during follow‐up (Waves 2013, 2015, 2018, and 2020) was ascertained through vital status (alive or dead) reported by participants' family members in structured interviews. The exact date of death was available only for 2013 and 2020. For deaths recorded in the 2015 and 2018 waves, survival time was estimated as the midpoint between the date of the last interview at which the participant was known to be alive and the date of the first wave interview recording the death [18]. New‐onset CVDs were evaluated by questions on the presence of stroke or any heart diseases (including heart attack, angina, coronary heart disease, heart failure, or other heart problems) in participants without CVDs at baseline (2011), and time for the first diagnosis was also collected.

In ELSA, the interview status data (alive or dead) were not available in Wave 7, 8, and 9, so the follow‐up data for all‐cause mortality was assessed until Wave 6 (2012–2013). The assessment of new‐onset CVDs in ELSA was similar to the procedures in CHARLS (including stroke or any heart diseases), and follow‐up continued through Wave 9 (2018–2019).

In NHANES, all‐cause mortality and cardiovascular mortality data as well as date of death up to December 31, 2019, were obtained by linking participant records from Public‐Use National Death Index (NDI) Linked Mortality data at NCHS [19].

2.5. Statistical Analysis

Given that a considerable proportion of individuals with normal BMI and WC also exhibited organ dysfunctions or limitations in daily activities, an adjustment was made to more accurately estimate the prevalence of preclinical and clinical obesity. Specifically, the prevalence of clinical obesity was adjusted by subtracting the proportion of participants with organ dysfunctions or activity limitations without excess adiposity from participants with clinical obesity, as described in a previous study [10]. To improve the accuracy of excess adiposity assessment, individuals who did not meet both predefined criteria of excess adiposity (e.g., BMI, WC, body fat percentage) were excluded, consistent with previous research [10]. Consequently, participants were categorized into four groups for subsequent analyses: (1) participants without obesity and without organ dysfunctions (reference group), (2) participants without obesity but with organ dysfunctions, (3) participants with obesity but without organ dysfunctions (preclinical obesity), and (4) participants with obesity and with organ dysfunctions (clinical obesity).

Cox proportional hazards models were used to examine the association between obesity status and clinical outcomes, adjusting for confounders including age, sex, race (in NHANES and ELSA), smoking status, alcohol consumption, education level, CRP, HDL‐C, SBP, and HbA1c. Risks were estimated for each subgroup using participants without obesity or organ dysfunctions as the reference group, and the hazard ratio (HR) for clinical obesity was further evaluated using preclinical obesity as the comparator. Similar analyses were performed using age stratification (65 years as the threshold) across the three cohorts, and interaction between terms for age group and obesity status was assessed by likelihood ratio tests. In NHANES, a competing risk analysis was conducted to examine the association between clinical obesity and cardiovascular mortality compared with competing events (mortality caused by other reasons) based on the Fine‐Gray model.

Categorical variables were presented as count (percentage), and continuous variables with normal distribution were presented as mean (standard deviation [SD]). For parameters without normal distribution, median (interquartile range [IQR]) was displayed. For the comparison of baseline characteristics between groups, ANOVA was used for the continuous variables, and chi‐square test was used for categorical variables. We set the statistical significance at p < 0.05 (two‐sided).

Considering the complex, multistage probability sampling design of NHANES, we used primary sampling unit, pseudo‐strata, and sampling weights for analysis following NHANES analytic and reporting guidelines (https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx#analytic‐guidelines).

All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

2.6. Sensitivity Analysis

In NHANES, sensitivity analysis were conducted using body fat percentage by DXA as an indicator for excess adiposity evaluation. In CHARLS, sensitivity analysis was conducted by applying a common definition across all cohorts (BMI ≥ 30 kg/m² and WC ≥ 102 cm [for men] or ≥ 88 cm [for women]).

3. Results

3.1. Prevalence of Clinical Obesity in Different Cohorts

The baseline characteristics of participants from CHARLS, ELSA, and NHANES are summarized in Table S2. In CHARLS, during a median follow‐up of 8.18 years, 839 deaths (8.9% of 9453 participants) and 871 incident cardiovascular events (10.8% of 7511 participants) were recorded. In ELSA, over a median follow‐up of 9.25 years, 164 deaths (3.7% of 4395 participants) and 713 incident CVD cases (21.9% of 3261 participants) occurred. In NHANES, during a median follow‐up of 15.38 years, 3557 deaths (21.5%) and 923 cardiovascular‐related deaths (5.6%) were documented among 16,511 participants.

The prevalence of clinical obesity in participants from CHARLS, ELSA, and NHANES was presented in Table 1. In CHARLS, ELSA, and NHANES, the prevalence of clinical obesity was 2.50%, 6.18%, and 9.08%, respectively. In both CHARLS and ELSA, the prevalence was lower in men than in women and lower in participants aged ≥ 65 years than in those aged 45–64 years. In NHANES, the prevalence was highest among participants aged 45–64 years and lowest among those aged ≥ 65 years; it was lower in men than in women among participants aged < 65 years but higher in men among those aged ≥ 65 years.

TABLE 1.

Prevalence of clinical obesity in three cohorts using adjustment.

	Prevalence (%)	Prevalence in men (%)	Prevalence in women (%)
CHARLS (n = 13, 311)
All ages	2.50	1.58	3.19
45–64 years old	3.14	2.06	3.96
≥ 65 years old	1.68	0.99	1.85
ELSA (n = 7109)
All ages	6.18	5.17	6.75
45–64 years old	8.68	7.45	9.58
≥ 65 years old	3.93	3.47	3.82
NHANES 1999–2006 (n = 18,041)
All ages	9.08	6.82	10.68
18–44 years old	6.68	5.51	7.51
45–64 years old	9.10	5.79	11.33
≥ 65 years old	4.04	4.32	2.28

Note: The prevalence of clinical obesity was adjusted by subtracting the proportion of participants with organ dysfunctions or activity limitations without excess adiposity from those with dysfunctions and excess adiposity.

3.2. Association Between Clinical Obesity and Outcomes

The associations with all‐cause mortality differed across the three cohorts. In CHARLS, compared with the reference group, participants without obesity but with organ dysfunctions or activity limitations showed a significantly higher risk of all‐cause mortality (HR = 1.32, 95% CI: 1.04–1.68). Neither preclinical obesity nor clinical obesity was significantly associated with all‐cause mortality, with HRs of 1.80 (95% CI: 0.75–4.34) and 1.23 (95% CI: 0.76–1.98), respectively. In ELSA, none of the obesity subgroups was significantly associated with all‐cause mortality. In NHANES, the participants without obesity but with dysfunctions subgroup and the clinical obesity subgroup had significantly elevated risks of all‐cause mortality relative to the reference group, with HRs of 1.56 (95% CI: 1.23–1.96) and 1.67 (95% CI: 1.29–2.17), respectively; no significant association was observed for preclinical obesity, with HR of 0.89 (95% CI: 0.63–1.27). In addition, compared with preclinical obesity, clinical obesity was associated with a higher risk of all‐cause mortality in NHANES (HR = 1.88, 95% CI: 1.25–2.84).

For CVD outcomes, the pattern was more consistent across cohorts. In CHARLS, ELSA, and NHANES, participants without obesity but with organ dysfunctions or activity limitations showed significantly higher risks than the reference group, with HRs of 1.54 (95% CI: 1.20–1.97), 1.40 (95% CI: 1.06–1.83), and 2.09 (95% CI: 1.33–3.26), respectively. Preclinical obesity was not significantly associated with CVD outcomes across the three cohorts, with HRs of 1.64 (95% CI: 0.84–3.19), 0.88 (95% CI: 0.47–1.64), and 1.60 (95% CI: 0.78–3.25), respectively. Clinical obesity was associated with significantly increased risks of CVD outcomes in all three cohorts, with HRs of 3.68 (95% CI: 2.58–5.24), 1.61 (95% CI: 1.17–2.22), and 2.92 (95% CI: 1.85–4.61), respectively. Compared with preclinical obesity, clinical obesity was associated with a higher risk of cardiovascular events in CHARLS and ELSA, with HRs of 2.49 (95% CI: 1.32–4.70) and 1.80 (95% CI: 1.02–3.17), respectively. In NHANES, clinical obesity was associated with a higher risk of cardiovascular mortality (HR = 3.11, 95% CI: 1.49–6.50).

In NHANES, the analysis of cardiovascular mortality using the competing risk method (Fine‐Gray model) showed similar results to those of the Cox proportional hazards models (Table S3).

In the age‐stratified analysis, participants aged < 65 years exhibited results that were statistically consistent with those of the overall cohort. No significant interaction was observed between age and subgroup classification for the outcomes (Table S4).

3.3. Sensitivity Analysis

In NHANES, when clinical obesity was defined using body fat percentage, its prevalence was slightly higher among participants aged 18–44 years than among those aged 45–64 years (9.17% vs. 8.84%) and was lowest among those aged ≥ 65 years (2.07%). Consistent with the primary analysis, men had lower prevalence than women before age 65 but higher prevalence at age ≥ 65 years (2.45% vs. 1.79%). For all‐cause mortality, clinical obesity was associated with a significantly higher risk than the reference group (adjusted HR = 1.67, 95% CI: 1.36–2.05), whereas the difference between clinical obesity and preclinical obesity was not statistically significant (adjusted HR = 1.12, 95% CI: 0.83–1.52). For cardiovascular mortality, clinical obesity was also associated with a significantly higher risk than the reference group (adjusted HR = 2.27, 95% CI: 1.50–3.44) and preclinical obesity (adjusted HR = 2.01, 95% CI: 1.04–3.87) (Table S5 and S6).

In CHARLS, when the same BMI and WC thresholds used in ELSA and NHANES were applied, the prevalence of clinical obesity remained lower than that in other cohorts and the risk ratios for clinical obesity were similar to the main analysis using the region‐specific definition (Table S7 and S8).

4. Discussion

In this study, we evaluated the prevalence of clinical obesity across diverse populations and examined its variations by age, sex, and country. Overall, the prevalence of clinical obesity was markedly lower among older adults and lower in men than in women. Notably, compared with participants without obesity and without organ dysfunctions, as well as those with preclinical obesity, clinical obesity was consistently associated with a higher risk of new‐onset cardiovascular diseases in CHARLS and ELSA, higher cardiovascular mortality in NHANES, and increased all‐cause mortality in NHANES.

Because organ dysfunctions and activity limitations were also present among individuals without obesity, we adjusted the prevalence of clinical obesity by subtracting this background proportion from that observed among individuals with excess adiposity [10]. Consequently, the estimated prevalence of clinical obesity in all three cohorts was below 10%, substantially lower than estimates based solely on BMI or waist circumference. Obesity‐related comorbidities, such as cardiovascular disorders, cancer, chronic kidney disease, and sarcopenia, are also highly associated with aging [20, 21]. Moreover, aging is accompanied by metabolic abnormalities that resemble those observed in obesity, such as insulin resistance, dyslipidemia, and systemic inflammation [22, 23]. Thus, the lower prevalence of clinical obesity in older adults may partly reflect the higher background burden of comorbidities among older adults without obesity. Our findings suggest that diagnosing and interpreting clinical obesity in older adults are inherently more complex. The definition and management of obesity in this population should therefore focus more on its impact on health outcomes and on implementing appropriate, individualized interventions.

We found that the prevalence of clinical obesity was lower in men than in women, particularly among participants aged 45–64 years. In general, body fat distribution had a significant difference between men and women. Women tended to have more subcutaneous adipose tissue (SAT) while men accumulated more visceral adipose tissue (VAT), which contributed to a higher risk of metabolic disorders. This protective effect in females was partially attributed to estrogen [24]. Across different regions of the world, the sex distribution of obesity shows marked heterogeneity [25]. In China, BMI‐defined obesity is more prevalent in men [26, 27], yet compound obesity—defined by simultaneous elevations in BMI and waist circumference—is more common in women, indicating a greater burden of central adiposity among women with obesity [28]. This pattern likely explains the higher prevalence of clinical obesity among women observed in CHARLS. In contrast, data from NHANES 2017–2020 showed similar rates between men and women [10]. In our analysis of NHANES 1999–2006, the overall prevalence was comparable, but the sex difference was more pronounced. Among adults aged 65 years and older, the sex difference narrowed, and clinical obesity became more prevalent in men, possibly reflecting age‐ and sex‐specific patterns of adiposity distribution and the higher comorbidity burden in older adults (Table S9).

Because this definition has implications for clinical prioritization and resource allocation, it is essential that participants with clinical obesity demonstrate worse prognosis. To reduce confounding from organ dysfunctions unrelated to excess adiposity, we compared clinical obesity not only with the reference group but also with preclinical obesity. Across all three cohorts, clinical obesity was consistently associated with elevated cardiovascular risk, and in NHANES, it was also associated with higher all‐cause mortality and cardiovascular mortality (also after applying competing risk analysis to reduce potential biases arising from deaths due to other causes). In contrast, preclinical obesity was not significantly associated with all‐cause mortality or cardiovascular events compared with the reference group, which suggested that the prognostic signal was stronger when excess adiposity was accompanied by organ dysfunctions. When compared with participants without excess adiposity (rather than with the reference group), preclinical obesity in NHANES was associated with significantly lower hazard ratios for all‐cause and cardiovascular mortality, and a similar trend was observed in CHARLS and ELSA (Table S10). This observation may reflect the so‐called “obesity paradox.” Previous studies have reported a J‐shaped relationship between BMI and mortality, where individuals with overweight or mild obesity exhibit lower mortality than those with normal or low BMI [29]. This pattern is thought to be partly driven by the higher prevalence of chronic conditions and frailty among individuals with low body weight [30]. Indeed, in our analysis, participants without obesity but with comorbidities also experienced a higher risk of adverse outcomes than those without comorbidities (Figures 1 and 2). Therefore, preclinical obesity should not be misinterpreted as protective even when its association with adverse outcomes does not reach statistical significance.

FIGURE 1.

Hazard ratios (95% CI) for all‐cause mortality according to obesity status and dysfunctions in CHARLS, ELSA, and NHANES 1999–2006. Cox regression was used to assess the hazard ratio of preclinical obesity and clinical obesity, adjusting for confounders including age, sex, race (in NHANES and ELSA), smoking, drinking, education level, CRP, HDL‐C, SBP, and HbA1c. Model 1: Adjusted for confounders including age, sex, race, smoking, drinking, education level. Model 2: Adjusted for confounders including age, sex, race, smoking, drinking, education level, CRP, HDL‐C, SBP, HbA1c. #Obesity with dysfunctions (clinical obesity) was compared with obesity without dysfunctions (preclinical obesity). *p < 0.05; **p < 0.01; ***p < 0.001. HR, hazard ratio.

FIGURE 2.

Hazard ratios (95% CI) for cardiovascular (CVD) outcomes according to obesity status and dysfunctions in CHARLS, ELSA, and NHANES 1999–2006. Cox regression was used to assess the hazard ratio of preclinical obesity and clinical obesity, adjusting for confounders including age, sex, race (in NHANES and ELSA), smoking, drinking, education level, CRP, HDL‐C, SBP, and HbA1c. Cardiovascular mortality was analyzed in the NHANES cohort, while cardiovascular events were assessed in the CHARLS and ELSA cohorts. Participants with pre‐existing cardiovascular diseases or missing follow‐up data were excluded from the cardiovascular event analysis. Model 1: Adjusted for confounders including age, sex, race, smoking, drinking, education level. Model 2: Adjusted for confounders including age, sex, race, smoking, drinking, education level, CRP, HDL‐C, SBP, HbA1c. #Obesity with dysfunctions (clinical obesity) was compared with obesity without dysfunctions (preclinical obesity). *p < 0.05; **p < 0.01; ***p < 0.001. HR, hazard ratio.

Given the age‐related differences in prevalence, we performed an age‐stratified analysis. Among participants younger than 65 years, the hazard ratios and their statistical significance were largely consistent with those in the overall cohort, while among those aged 65 years and older, the risk of adverse outcomes did not reach statistical significance in most subgroups. However, tests for interaction yielded nonsignificant p values across all subgroups, indicating no strong evidence that age modified these associations. The attenuated associations observed in older participants may reflect the higher burden of comorbidities among older participants without obesity, as well as the lower prevalence of clinical obesity and smaller sample size in this age group. We also observed that in CHARLS, both the non‐obesity with organ dysfunctions and obesity with organ dysfunctions subgroups were associated with significantly higher risks of adverse outcomes among participants younger than 65 years, whereas similar associations were not observed in those aged 65 years or older. In ELSA, the subgroup with obesity and organ dysfunction showed a comparable pattern. Overall, these age‐stratified findings were broadly consistent with the main analysis and further highlighted the prognostic importance of comorbidities and dysfunctions.

The main strength of this study is the use of nationally representative data from different countries to analyze the commonalities in the prevalence rates of clinical obesity and the risks of adverse outcomes across various nations. However, certain limitations should be acknowledged. Firstly, many organ dysfunctions included in the new definition of clinical obesity are also associated with aging. Consequently, among older adults, a relatively large number of individuals without obesity still have organ dysfunctions or daily activity limitations, making it difficult to clearly elucidate the causal relationship between excess adiposity and organ dysfunction. Secondly, although it is necessary to consider organ dysfunctions when evaluating clinical obesity in populations with or without obesity, distinguishing obesity‐related dysfunctions at the individual level remains challenging, especially in older adults. Further research is needed to develop precise, individual‐level assessment strategies. Thirdly, some of the data on organ dysfunctions or limitations in daily activities were self‐reported, which may introduce recall bias and require further verification by precise investigations. Fourthly, direct body composition measurements were not available in CHARLS and ELSA, which partially limited cross‐cohort comparability. Additionally, owing to differences in data availability across the three cohorts, not all information regarding organ dysfunctions and limitations in daily activities could be uniformly ascertained, leading to estimation biases and potential differences among the three cohorts. Specifically, heart failure was included in the definition of clinical obesity in NHANES and ELSA but not in CHARLS because baseline data were unavailable. In CHARLS, incident cardiovascular diseases may still have included new‐onset heart failure due to broad outcome definitions in this cohort. Because heart failure could not be uniformly separated from other heart diseases across cohorts, a fully harmonized exclusion analysis was not feasible and may have introduced some circularity.

In conclusion, this study presented the clinical obesity prevalence rates among representative populations in different countries and provided convergent evidence from distinct populations on their associations with clinical outcomes. It emphasizes the value of the new definition of clinical obesity in assessing adverse outcomes and its significance for medical resource allocation and personalized diagnosis and treatment. Nevertheless, given the complexity of incorporating multiple indicators of organ dysfunction and functional limitation, further research is needed to determine how this definition can be most effectively implemented in clinical practice.

Author Contributions

Qi Huang, Xiantong Zou, and Linong Ji: conceptualization, supervision. Sitong Li, Qi Huang, and Xiantong Zou: resources, data curation. Sitong Li, and Qi Huang: methodology, formal analysis, visualization. Sitong Li: writing – original draft. Qi Huang, Yingning Liu, Song Wang, Yingying Luo, Xueyao Han, Xiantong Zou, and Linong Ji: writing – review and editing. Xiantong Zou, and Linong Ji: project administration, funding acquisition. All authors reviewed and approved the final version of the manuscript.

Funding

This work was supported by the Special Funds of the National Natural Science Foundation of China (T2341011 to Xiantong Zou), Beijing Nova Cross Program of Science and Technology (20250484806, to Xiantong Zou), and Peking University People's Hospital Research and Development Funds (grant numbers RZ2024‐03 to Xiantong Zou).

Disclosure

The authors employed ChatGPT (OpenAI, San Francisco, CA, USA) to assist with grammar corrections during manuscript preparation. All intellectual and analytical content was generated by the authors, and the final text was thoroughly reviewed and approved by all authors prior to submission.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: The items of diagnostic criteria for clinical obesity in three cohorts.

Table S2: Characteristics of participants in CHARLS, ELSA, and NHANES.

Table S3: The association of clinical obesity with mortality and cardiovascular outcomes in NHANES using Fine–Gray competing risk models.

Table S4: Hazard ratios (95% CI) for all‐cause mortality and cardiovascular events according to obesity and dysfunction status in CHARLS, ELSA, and NHANES (1999–2006), stratified by age.

Table S5: Prevalence of clinical obesity in NHANES (evaluated by body fat percentage) using adjustment.

Table S6: The association of clinical obesity with mortality and cardiovascular outcomes in NHANES (evaluated by body fat percentage).

Table S7: The association of clinical obesity with mortality and cardiovascular outcomes in CHARLS (using thresholds of BMI and waist circumference consistent with ELSA and NHANES).

Table S8: Adjusted prevalence of clinical obesity in CHARLS using BMI and waist circumference thresholds consistent with those in ELSA and NHANES.

Table S9: Numbers of participants in subgroups of clinical obesity in three cohort.

Table S10: Prevalence of preclinical obesity and clinical obesity and their association with mortality and cardiovascular outcomes in three cohorts.

Figure S1: (A) Flowchart of participant selection process in CHARLS. (B) Flowchart of participant selection process in ELSA. (C) Flowchart of participant selection process in NHANES 1999–2006.

Data Availability Statement

Data are available in a public, open access repository. CHARLS, ELSA, and NHANES are all publicly accessible databases at https://charls.pku.edu.cn/, https://www.elsa‐project.ac.uk/data‐and‐documentation, https://www.cdc.gov/nchs/nhanes/, respectively.