Abstract
Background
Individuals with osteoarthritis (OA) often experience significant changes in their metabolic status and have a higher risk of mortality compared to the general population. However, no study has quantified the metabolic status of OA patients. Moreover, the association between metabolic status and risk of mortality among OA patients remains unclear.
Methods
The analysis included baseline populations with OA from the UK Biobank (UKBB) study and the National Health and Nutrition Examination Survey (NHANES) 1999–2018. Metabolic status was assessed using composite scores based on body mass index, waist-hip ratio, blood pressure, fasting plasma glucose, hemoglobin A1c, C-reactive protein, triacylglycerols, and lipoproteins. Restricted cubic splines were used to define healthy ranges for each indicator regarding mortality risk. Factors received 1 if within a healthy range, else 0. Total scores ranged from 0 to 7, with 3+ indicating good metabolic health. The traditional metabolic syndrome criteria (ATP III) and the Strict definition were used for comparison. Associations between metabolic unhealth and mortality were examined using Cox proportional hazards models. Integrated Discrimination Improvement (IDI) evaluated discriminatory capacity, and Net Reclassification Improvement (NRI) assessed reclassification performance in predicting ten-year mortality.
Results
Among 44,302 UKBB OA participants, 36.0 % were metabolically unhealthy (scores: 0–2), while in 5233 NHANES OA participants, 39.1 % were metabolically unhealthy based on our newly-established definition. After adjustment for covariates, metabolically unhealthy individuals had significantly higher all-cause mortality risk (HR: 1.36; 95 % CI: 1.26–1.47 in UKBB; HR: 1.16; 95 % CI: 1.03–1.30 in NHANES) compared to metabolically healthy individuals. In UKBB, compared to ATP III definition, the changes in IDI and NRI for our newly-established definition were 0.10 % (0.01 %, 0.20 %) and 10.60 % (1.8 %, 13.60 %), respectively. When the comparison was made with the Strict definition, the newly-established definition showed changes in IDI and NRI of 0.18 % (0.10 %, 0.20 %) and 12.10 % (10.30 %, 13.80 %), respectively. results.
Conclusions
A new definition for quantifying metabolic unhealth in OA was proposed. It identified a significant mortality risk with poorer metabolic status and outperformed general definitions in predictive accuracy. conclusion.
The translational potential of this article
OA is the leading joint disease worldwide, carrying a heavy health burden. Our newly established definition, evaluating metabolic health through waist-to-hip ratio, BMI, blood pressure, glucose, lipids, and C-reactive protein, identified 36.0 % OA patients in UK Biobank and 39.1 % OA patients in NHANES as having poor metabolic status, which is applicable in clinical and research settings. This study also showed that poor metabolic status, as classified by these criteria, is significantly correlated with an elevated risk of all-cause mortality in OA patients. It further highlights the importance of maintaining a holistic approach to care for OA patients, emphasizing the need to address both joint and metabolic health.
Keywords: Osteoarthritis, Mortality, Metabolic health
Graphical abstract
1. Introduction
As the most prevalent joint disease worldwide, osteoarthritis (OA) is estimated to affect approximately 10 % of men and 18 % of women aged 60 years and older[[1], [2], [3]]. A cohort study conducted in the South West of England has revealed that individuals with OA exhibit a heightened mortality risk when compared to the general population [4]. However, the reasons for the high mortality rate in OA patients remain poorly understood.
Metabolic disorders significantly impact health and increase mortality risks. Emerging research suggests a potential association between OA and metabolic disorders, as evidenced by significant metabolic alterations observed in OA patients [5,6]. A cohort study involved 7714 participants reported that metabolic syndrome is prevalent in 59 % of patients with OA [7]. Furthermore, OA is recognized as a "metabolic disorder" itself, with its pathogenesis closely tied to obesity, diabetes, dyslipidemia, hypertension, and insulin resistance, all of which contribute to the disease's initiation and progression [8]. Previous studies have primarily focused on the association between metabolic health status and all-cause mortality, cardiovascular disease, diabetes, and cancer in the general population[[9], [10], [11], [12]]. These studies utilized a metabolic health measure derived from Adult Treatment Panel-III criteria (ATPIII), which is based on outcomes related to metabolic syndrome [13]. It is worth noting that metabolic alterations in OA patients were significantly greater than in the general population [6]. However, there is currently a lack of research investigating the association between metabolic health status and all-cause mortality in individuals with OA, and it remains unclear which metabolic criteria are closely linked to mortality in this population.
Therefore, the objective of this study was to propose a metabolic health criterion for individuals with OA and to examine the associations between metabolic health status and all-cause mortality risk among participants with OA using data from the UK Biobank and the US National Health and Nutrition Examination Survey (NHANE, 1999–2018).
2. Methods
2.1. Study populations
The UK Biobank is a population-based cohort study that involved over 500,000 participants between the ages of 37 and 73. These individuals attended 1 of the 22 assessment centers located across the United Kingdom (UK) from 2006 to 2010 [14]. The study received approval from the North West Multi-center research ethics committee (REC reference 11/NW/0382), and participants provided written informed consent prior to their involvement [15]. In addition, the baseline measures collected during this period were linked to mortality records. The UK Biobank study was further approved by the National Health Service (NHS) National Research Ethics Service (16/NW/0274) for conducting sub-studies within it [16]. In the current study, a total of 44,302 individuals with OA at baseline were included after excluding 1298 participants who had missing values in their data. The National Health and Nutrition Examination Survey (NHANES) is a comprehensive study conducted to evaluate the health and nutritional status of the noninstitutionalized civilian population in the United States [17], aiming to provide a representative snapshot of the nation's overall health and well-being. For our current analysis, we made use of data from ten cycles of NHANES study, which spanned from 1999 to 2018. The sample from NHANES consisted of 101,316 adults aged 20 years or older. After excluding 96,083 participants who did not have OA at the beginning of the study, our analysis focused on a total of 5233 participants who were diagnosed with OA. Subsequently, interpolation was performed using the random forest method (Fig. 1).
Fig. 1.
Flow chart of UK Biobank and NHANES participants screening.
2.2. Assessment of metabolic health
In our study, we utilized three definitions to identify metabolic factors: our newly proposed definition, the traditional definition as outlined by ATPIII, which form the basis of the traditional definition of metabolic syndrome, and the Strict definition (Table 1) [13,18].
Table 1.
Three definitions of metabolic health.
| Components of the three definitions |
Low risk range |
|
|---|---|---|
| Female | Male | |
| New definition | ||
| Waist-hip ratio | ≤0.95 | ≤0.95 |
| Body mass index | 28–32 kg/m2 | 28–31 kg/m2 |
| Blood pressure | SBP≤135 mmHg, DBP≤85 mmHg, no self-reported hypertension, no antihypertensive medications |
SBP≤150 mmHg, DBP≤85 mmHg, no self-reported hypertension, no antihypertensive medications |
| Blood glucose | fasting plasma glucose≤ 5.5 mmol/L, HbA1C ≤ 38 mmol/mol, no self-reported diabetes, no antidiabetic medications |
fasting plasma glucose≤ 5.5 mmol/L, HbA1C ≤ 38 mmol/mol, no self-reported diabetes, no antidiabetic medications |
| C-reactive protein | ≤1.7 mg/L | ≤1.7 mg/L |
| Triglyceride | ≤150 mg/dL, no self-reported hyperlipemia, no lipid-lowering medications |
≤150 mg/dL, no self-reported hyperlipemia, no lipid-lowering medications |
| Blood lipoprotein | HDL≥1.3 mmol/L, LDL ≥3.7 mmol/L | HDL≤1.5 mmol/L, LDL ≥3.7 mmol/L |
| ATPIII definition | ||
| Waist circumference | ≤88 cm | ≤102 cm |
| Blood pressure | SBP<135 mmHg, DBP<85 mmHg, no antihypertensive medications |
SBP<135 mmHg, DBP<85 mmHg, no antihypertensive medications |
| Fasting plasma glucose | <6.11 mmol/L, no self-reported diabetes |
<6.11 mmol/L, no self-reported diabetes |
| High-density lipoprotein | >1.3 mmol/L | >1.04 mmol/L |
| Triglyceride | <1.70 mmol/L, no lipid-lowering medications |
<1.70 mmol/L, no lipid-lowering medications |
| Strict definition | ||
| Blood Glucose | fasting plasma glucose< 5.56 mmol/L, HbA1C < 39 mmol/mol |
fasting plasma glucose< 5.56 mmol/L, HbA1C < 39 mmol/mol |
| Blood pressure | SBP<130 mmHg, DBP<85 mmHg, no antihypertensive medications |
SBP<130 mmHg, DBP<85 mmHg, no antihypertensive medications |
| Blood lipoprotein | total cholesterol <240 mg/dL, HDL >1.04 mmol/L, triglyceride <1.70 mmol/L, no lipid-lowering medications |
total cholesterol <240 mg/dL, HDL >1.04 mmol/L, triglyceride <1.70 mmol/L, no lipid-lowering medications |
The new definition is considered metabolically healthy if it achieves a score of 3 or higher on a 7-point scale, indicating an improved level of metabolic health. In comparison, the ATPIII definition considers a score of 3 or higher on a 5-point scale as indicative of metabolic health. Notably, Strict definition attained a perfect score of 3 out of 3 to meet the criteria for being classified as metabolically healthy.
Abbreviation: SBP, systolic blood pressure; DBP, diastolic blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein.
To establish our new definition, we analyzed 10 factors previously associated with metabolic syndrome using Restricted cubic splines (RCS): body mass index (BMI), waist-hip ratio (WHR), systolic blood pressure (SBP), diastolic blood pressure (DBP), blood glucose (fasting plasma glucose and hemoglobin A1c), C-reactive protein (CRP), triacylglycerols (TG), as well as lipoproteins such as high-density lipoprotein (HDL) and low-density lipoprotein (LDL) [[19], [20], [21]]. HDL and LDL levels were collectively considered.
2.3. Disease diagnosis
In UK biobank, the presence of OA was determined by analyzing hospital inpatient records, which contained admission and diagnosis information sourced from the Hospital Episode Statistics for England, Scottish Morbidity Record data for Scotland, and the Patient Episode Database for Wales. Diagnoses were coded using the International Classification of Diseases (ICD) system, with participants identified through ICD-9 codes (7151, 7152, 7153, 7158, 7159) and ICD-10 codes (M15 - M19). Additionally, patient self-reports were considered (Data-Field 20002). In NHANES study, participants were identified as having arthritis if they responded "yes" to the question " Has a doctor or other health professional ever told you that you had arthritis?". The specific type of arthritis was determined by their response to the question "Which type of arthritis was it?", with "Osteoarthritis" being one of the available options (Supplementary Table 1).
2.4. Outcome assessment
The outcome of interest was all-cause mortality, which was assessed based on follow-up data. Participants were followed until death, loss to follow-up, or the end of the study period, whichever occurred first. For UK Biobank study, follow-up data was available until 30 September 2021 for England and Wales and 31 October 2021 for Scotland. Cause of death was determined using International Classification of Diseases, 10th revision (ICD-10) codes. Survival time was calculated from baseline until the earliest occurrence of death or the end of follow-up (December 31, 2021). For NHANES study, mortality data was provided by the National Center for Health Statistics and linked to the National Death Index through 31 December 2019 [22]. The underlying cause of death was coded using ICD-10 codes. The duration of follow-up was calculated as the interval in months from the interview date to the date of death or through 31 December 2019 for participants who did not die during the study period.
2.5. Assessment of covariates
In the analysis of the UK Biobank and NHANES datasets, the covariates considered included age, sex, race, body mass index, Townsend deprivation index (poverty income ratio in NHANES dataset), education, smoking status, alcohol intake, regular physical activity, sedentary time, sleep duration, healthy diet and glucosamine use [23]. The covariates were derived from baseline questionnaire assessments.
2.6. Statistical analysis
To examine the non-linear associations between BMI, WHR, SBP, DBP, fasting plasma glucose and hemoglobin A1c, CRP, TG, HDL, LDL and all-cause mortality, we utilized RCS models within Cox proportional hazards models [24]. We recruited 44,302 participants with OA at baseline from UK Biobank study and stratified them by sex to apply RCS for each of the 10 aforementioned indicators. These RCS models were fitted with three knots and adjusted for age, sex, race, body mass index, Townsend deprivation index, education, smoking status, alcohol intake, regular physical activity, sedentary time, sleep duration, healthy diet and glucosamine use. Participants with self-reported diabetes or those taking hypoglycemic medications were excluded prior to the prediction of the association between blood sugar levels and all-cause mortality. Similarly, blood pressure, triglyceride levels, and cholesterol levels were handled in a comparable manner. For each metabolic factor, we assigned 1 point for a healthy level, otherwise 0 point. The metabolic score was constructed as the sum of all factors, ranging from 0 to 7, with a higher score indicating better metabolic status. A score of three or greater was considered metabolically healthy on a scale of 7.
Prior to utilizing Cox proportional hazard regression models, we first verify the proportionality assumptions of covariates associated with the outcomes (Supplementary Figs. 1 and 2). Using Cox proportional hazard regression models, we estimated the hazard ratios (HR) and 95 % confidence intervals (CI) to determine the association between metabolic status and all-cause mortality. Using Kaplan–Meier (KM) curves, we explored the relationships between metabolic score, metabolic health, and the survival rate of participants [25]. Additionally, we performed interaction analyses based on age, sex and OA site to explore the associations between the metabolic score and all-cause mortality among adults in different subgroups [26]. Considering that the age of 45 is commonly used as an inclusion criterion in OA cohorts and 60 years is a widely accepted reference point for studying overall mortality among older adults, this study utilized these ages as important cutoffs for categorizing the population in our subgroup analyses.
The Integrated Discrimination Improvement (IDI) was employed to assess the discriminatory capacity of three different metabolic health definitions (Our new definition, ATPIII definition, Strict definition) in predicting all-cause mortality over a ten-year duration [26]. Furthermore, the Net Reclassification Improvement (NRI) was utilized to evaluate the reclassification performance of these definitions for predicting all-cause mortality within the same time frame. Additionally, we opted to measure the goodness-of-fit between the observed data and the three definitions using Concordance values of the models [27].
Several sensitivity analyses were carried out to test the robustness of our findings. First, participants below the age of 45 at baseline were excluded. Second, individuals who died within the first two years of follow-up were excluded. Then, we excluded participants who had been diagnosed with cardiovascular disease or cancer before the start of the study. Considering that commonly used non-steroidal anti-inflammatory drugs and tramadol may alter mortality rates, individuals using such medications were also excluded [28]. Similarly, the population that has undergone joint replacement surgery was also excluded [29]. In constructing the weighted metabolic score, we employed a Cox proportional hazards model to estimate the β coefficients for each metabolic factor, with all-cause mortality as the outcome. The model was adjusted for potential confounders: these including age, sex, race, body mass index, Townsend deprivation index (poverty income ratio in NHANES dataset), education, smoking status, alcohol intake, regular physical activity, sedentary time, sleep duration, healthy diet and glucosamine use. The weighted metabolic score was derived as follows: binary metabolic variables were each multiplied by their corresponding β coefficients, the results summed, normalized by the total sum of β coefficients, and then multiplied by 100 [30]. This score was subsequently categorized into eight ordinal groups (0–7), based on the distribution of a simple additive metabolic score. A score of three or greater was considered metabolically healthy on a scale of 7.
The present study utilized UK Biobank for analysis, with validation analysis conducted using NHANES database. The baseline characteristics of various metabolic states were described, and differences between groups were assessed using the Student's t-test for continuous variables and categorical variable chi-square (χ2) tests. A significance level of 0.05 was employed to establish statistical significance in both directions. R software version 4.2.1 was utilized for conducting statistical analyses and visualizing the data.
3. Results
3.1. Baseline characteristics of participants
The baseline characteristics of the study population are presented in Table 2. Among the 44,302 UK Biobank participants with OA at baseline (mean age 60.6 years, 60.4 % women), 36.0 % were metabolically unhealthy (scores: 0–2). In the UK Biobank cohort, those who were metabolically unhealthy exhibited a higher likelihood of having lower socioeconomic status, lower educational attainment, a history of tobacco use, less alcohol consumption, less regular physical activity, more sedentary time, more sleep duration, a more unhealthy diet, lesser utilization of glucosamine supplements, more frequent utilization of medication, a history of joint replacement, a higher prevalence of knee OA, and higher prevalence of cardiovascular disease and cancer, and higher all - cause mortality rates. Of the 5233 participants from NHANES (mean age 64.7 years; 64.4 % women), 39.1 % showed metabolically unhealthy. In the NHANES cohort, metabolically unhealthy individuals tended to have lower incomes, lower educational levels, less tobacco use, reduced alcohol intake, less physical activity, more sedentary behavior, a history of joint replacement, a higher prevalence of cardiovascular diseases and cancer, and higher all-cause mortality rates. Supplementary Figs. 1 and 2 evaluated the proportional hazards assumption for UK Biobank and NHANES covariates, respectively. Graphical analysis showed parallel lines, indicating no violation of the assumption [31].
Table 2.
Baseline characteristics of participants from UK Biobank and NHANES according to level of metabolic health using our new definition.
| Characteristic |
UK Biobank |
NHANES |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Overall | Metabolically healthy | Metabolically unhealthy | P | Overall | Metabolically healthy | Metabolically unhealthy | P | |||
| N | 44,302 | 28,346 | 15,956 | 5233 | 3189 | 2044 | ||||
| Age at recruitment (years) | 60.60 (6.46) | 59.98 (6.65) | 61.70 (5.94) | <0.001 | 64.66 (13.65) | 63.08 (14.39) | 67.12 (11.99) | <0.001 | ||
| Sex (%) | Female | 26756 (60.40) | 19105 (67.40) | 7651 (48.0) | <0.001 | Female | 3368 (64.40) | 2215 (69.50) | 1153 (56.40) | <0.001 |
| Male | 17546 (39.60) | 9241 (32.60) | 8305 (52.00) | Male | 1865 (35.60) | 974 (30.50) | 891 (43.60) | |||
| Body mass index (kg/m2) | 29.19 (5.38) | 27.64 (4.42) | 31.93 (5.83) | <0.001 | 30.61 (7.48) | 29.39 (6.65) | 32.50 (8.28) | <0.001 | ||
| Waist hip ratio | 0.88 (0.09) | 0.85 (0.08) | 0.94 (0.09) | <0.001 | 0.95 (0.11) | 0.91 (0.10) | 1.00 (0.10) | <0.001 | ||
| Diastolic blood pressure (mmHg) | 142.72 (19.30) | 140.12 (19.19) | 147.35 (18.61) | <0.001 | 132.24 (21.08) | 128.47 (19.61) | 138.14 (21.94) | <0.001 | ||
| Systolic blood pressure (mmHg) | 82.52 (10.43) | 81.46 (10.24) | 84.40 (10.50) | <0.001 | 69.09 (14.44) | 68.86 (13.28) | 69.44 (16.08) | 0.152 | ||
| Hemoglobin A1c (mmol/mol) | 37.15 (7.01) | 35.19 (4.29) | 40.62 (9.21) | <0.001 | 40.82 (10.56) | 39.00 (9.30) | 43.66 (11.71) | <0.001 | ||
| Glucose (mmol/mol) | 5.22 (1.29) | 4.97 (0.78) | 5.67 (1.79) | <0.001 | 6.18 (2.02) | 5.86 (1.65) | 6.67 (2.41) | <0.001 | ||
| Triglycerides (mmol/mol) | 1.89 (1.04) | 1.67 (0.90) | 2.28 (1.16) | <0.001 | 1.53 (0.99) | 1.34 (0.85) | 1.82 (1.11) | <0.001 | ||
| LDL cholesterol (mmol/mol) | 3.55 (0.90) | 3.73 (0.85) | 3.24 (0.91) | <0.001 | 2.97 (0.97) | 3.13 (0.99) | 2.72 (0.88) | <0.001 | ||
| HDL cholesterol (mmol/mol) | 1.43 (0.38) | 1.52 (0.38) | 1.26 (0.32) | <0.001 | 1.44 (0.44) | 1.52 (0.45) | 1.32 (0.40) | <0.001 | ||
| Total cholesterol (mmol/mol) | 5.69 (1.20) | 5.95 (1.12) | 5.23 (1.20) | <0.001 | 5.13 (1.13) | 5.28 (1.12) | 4.89 (1.10) | <0.001 | ||
| C reactive protein (mg/L) | 3.49 (5.23) | 2.66 (4.46) | 4.96 (6.11) | <0.001 | 0.65 (1.10) | 0.50 (0.86) | 0.87 (1.36) | <0.001 | ||
| Townsend deprivation index | 2.58 (1.12) | 2.49 (1.11) | 2.74 (1.11) | <0.001 | — | — | — | — | ||
| Poverty income ratio | — | — | — | — | 2.70 (1.60) | 2.77 (1.62) | 2.59 (1.56) | <0.001 | ||
| Education (%) | College | 2260 (7.40) | 1544 (8.20) | 716 (6.10) | <0.001 | College | 1245 (23.80) | 798 (25.00) | 447 (21.90) | 0.010 |
| Other | 28264 (92.60) | 17196 (91.80) | 11068 (93.90) | Other | 3988 (76.20) | 2391 (75.00) | 1597 (78.10) | |||
| Smoking status (%) | Never | 21830 (49.30) | 15020 (53.00) | 6810 (42.70) | <0.001 | Never | 1897 (36.30) | 1066 (33.50) | 831 (40.70) | <0.001 |
| Previous | 17854 (40.30) | 10682 (37.70) | 7172 (44.90) | Former | 2494 (47.70) | 1580 (49.60) | 914 (44.70) | |||
| Current | 4618 (10.40) | 2644 (9.30) | 1974 (12.40) | Now | 838 (16.00) | 540 (16.90) | 298 (14.60) | |||
| Alcohol intake (%) | Never | 2368 (5.30) | 1323 (4.70) | 1045 (6.50) | <0.001 | Never | 1036 (23.20) | 551 (20.20) | 485 (27.80) | <0.001 |
| Previous | 2323 (5.20) | 1258 (4.40) | 1065 (6.70) | Former | 420 (9.40) | 290 (10.60) | 130 (7.50) | |||
| Current | 39611 (89.40) | 25765 (90.90) | 13846 (86.80) | Moderate | 1769 (39.60) | 1096 (40.20) | 673 (38.60) | |||
| — | — | — | — | — | Mild | 586 (13.1) | 403 (14.8) | 183 (10.5) | ||
| — | — | — | — | — | Heavy | 659 (14.7) | 387 (14.2) | 272 (15.6) | ||
| Regular physical activity (%) | No | 19153 (43.20) | 11281 (39.80) | 7872 (49.30) | <0.001 | No | 4103 (78.40) | 2427 (76.10) | 1676 (82.00) | <0.001 |
| Yes | 25149 (56.80) | 17065 (60.20) | 8084 (50.70) | Yes | 1130 (21.60) | 762 (23.90) | 368 (18.00) | |||
| Sedentary time (hour) | 4.32 (2.17) | 4.04 (2.02) | 4.76 (2.33) | <0.001 | 6.47 (3.42) | 6.25 (3.39) | 6.81 (3.43) | <0.001 | ||
| Sleep duration (hour) | 7.08 (1.32) | 7.06 (1.24) | 7.12 (1.43) | <0.001 | 7.12 (1.65) | 7.08 (1.63) | 7.17 (1.68) | 0.075 | ||
| Healthy diet (%) | No | 16097 (51.90) | 9339 (49.20) | 6758 (56.20) | <0.001 | No | 3140 (60.00) | 1885 (59.10) | 1255 (61.40) | 0.105 |
| Yes | 14897 (48.10) | 9639 (50.80) | 5258 (43.80) | Yes | 2093 (40.00) | 1304 (40.90) | 789 (38.60) | |||
| Glucosamine use (%) | No | 28325 (63.90) | 17083 (60.30) | 11242 (70.50) | <0.001 | No | 5146 (98.30) | 3135 (98.30) | 2011 (98.40) | 0.915 |
| Yes | 15977 (36.10) | 11263 (39.70) | 4714 (29.50) | Yes | 87 (1.70) | 54 (1.70) | 33 (1.60) | |||
| Medication use (%) | No | 36629 (82.70) | 24027 (84.80) | 12602 (79.00) | <0.001 | No | 5037 (96.30) | 3068 (96.20) | 1969 (96.30) | 0.875 |
| Yes | 7673 (17.30) | 4319 (15.20) | 3354 (21.00) | Yes | 196 (3.70) | 121 (3.80) | 75 (3.70) | |||
| Joint replacement (%) | No | 30282 (68.40) | 20088 (70.90) | 10194 (63.90) | <0.001 | No | 683 (89.50) | 442 (92.10) | 241 (85.20) | 0.004 |
| Yes | 14020 (31.60) | 8258 (29.10) | 5762 (36.10) | Yes | 80 (10.50) | 38 (7.90) | 42 (14.80) | |||
| Cardiovascular disease (%) | No | 21167 (47.80) | 17007 (60.00) | 4160 (26.10) | <0.001 | No | 3975 (76.00) | 2590 (81.20) | 1385 (67.80) | <0.001 |
| Yes | 23135 (52.20) | 11339 (40.00) | 11796 (73.90) | Yes | 1258 (24.00) | 599 (18.80) | 659 (32.20) | |||
| Cancer (%) | No | 33962 (76.70) | 22186 (78.30) | 11776 (73.80) | <0.001 | No | 4174 (79.80) | 2597 (81.40) | 1577 (77.20) | <0.001 |
| Yes | 10340 (23.30) | 6160 (21.70) | 4180 (26.20) | Yes | 1059 (20.20) | 592 (18.60) | 467 (22.80) | |||
| All-cause mortality (%) | No | 39410 (89.00) | 26035 (91.80) | 13375 (83.80) | <0.001 | No | 3727 (71.20) | 2371 (74.30) | 1356 (66.30) | <0.001 |
| Yes | 4892 (11.00) | 2311 (8.20) | 2581 (16.20) | Yes | 1506 (28.80) | 818 (25.70) | 688 (33.70) | |||
| Race (%) | AFR | 419 (0.90) | 251 (0.90) | 168 (1.10) | <0.001 | MA | 441 (8.40) | 276 (8.70) | 165 (8.10) | 0.528 |
| EAS | 143 (0.30) | 77 (0.30) | 66 (0.40) | NHB | 777 (14.80) | 469 (14.70) | 308 (15.10) | |||
| EUR | 42640 (96.20) | 27441 (96.80) | 15199 (95.30) | NHW | 3376 (64.50) | 2037 (63.90) | 1339 (65.50) | |||
| Mixed | 173 (0.40) | 112 (0.40) | 61 (0.40) | OH | 308 (5.90) | 197 (6.20) | 111 (5.40) | |||
| SAS | 493 (1.10) | 229 (0.80) | 264 (1.70) | OHM | 331 (6.30) | 210 (6.60) | 121 (5.90) | |||
| UN | 434 (1.00) | 236 (0.80) | 198 (1.20) | — | — | — | ||||
| OA site (%) | Hip OA | 5717 (12.90) | 3749 (13.20) | 1968 (12.30) | <0.001 | All OA | 5233 (100.00) | 3189 (60.94) | 2044 (39.06) | <0.001 |
| Knee OA | 11564 (26.10) | 6766 (23.90) | 4798 (30.10) | — | — | — | ||||
| Hand OA | 1151 (2.60) | 806 (2.80) | 345 (2.20) | — | — | — | ||||
| Other OA | 22826 (51.50) | 15234 (53.70) | 7592 (47.60) | — | — | — | ||||
| Multi-site OA | 3044 (6.90) | 1791 (6.30) | 1253 (7.90) | — | — | — | ||||
The data were presented as mean (SD) for continuous variables or N (%) for categorical variables. P-values were calculated using a t-test and chi-square test for continuous and categorical variables, respectively.
Medication use refers to the usage of non-aspirin NSAIDs or tramadol.
Abbreviation: HDL, high-density lipoprotein; LDL, low-density lipoprotein; AFR, African; EAS, East Asian; Mixed, Mixed race; SAS, South Asian; UN, Unknow; MA, Mexican American; NHB, Non-Hispanic Black; NHW, Non-Hispanic White; OH, Other Hispanic; OHM, Other Race - Including Multi-racial.
3.2. Construction of metabolic health status
According to the results obtained from RCS curves, we established a WHR of less than or equal to 0.95, glucose levels of less than or equal to 5.5 mmol/L, HbA1c levels of less than or equal to 38 mmol/mol, CRP levels of less than or equal to 1.7 mg/L, LDL levels greater than or equal to 3.7 mmol/L as an appropriate threshold based for both sexes. Similarly, 28.5–30.5 kg/m2 for males (28–32 kg/m2 for females), systolic blood pressure of less than or equal to 150 mm/Hg for males, HDL levels less than or equal to 1.5 mmol/L for males (greater than or equal to 1.3 mmol/L for women), are classified as falling within the low-risk range (Supplementary Figs. 3 and 4). Furthermore, the application of RCS curves indicated no significant correlation between certain metabolic factors and all-cause mortality, leading to the adoption of ATPIII criteria to delineate low-risk thresholds: specifically, triglyceride levels ≤150 mg/dL for both sexes, systolic blood pressure ≤135 mmHg and diastolic blood pressure ≤85 mmHg for females. The final criteria for metabolic health are presented in Table 1.
In both the UK Biobank and NHANES datasets, the metabolic score demonstrated a linear negative correlation with all-cause mortality (Supplementary Fig. 5). Specifically, when the metabolic score was ≥3, it was associated with a lower HR value. Therefore, we defined a metabolic score of ≥3 as metabolically healthy, and a score of less than 3 as metabolically unhealthy. Survival analysis indicated that among patients with OA, the survival rate decreased as the metabolic score declined (Supplementary Fig. 6). Metabolically unhealthy individuals, defined with a metabolic score of less than 3, had a significantly lower survival rate, and the log - rank test showed statistical significance (p < 0.0001).
3.3. Association of metabolic score and metabolic health with all-cause mortality
The association between metabolic score, metabolic health status, and all-cause mortality are presented in Fig. 2. In the UK Biobank cohort, metabolically unhealthy individuals had a significantly higher risk of all-cause mortality compared to metabolically healthy participants: HR = 1.36 (95 % CI, 1.26–1.47) using our new definition, HR = 1.24 (1.15–1.35) using the ATPIII definition, and HR = 1.19 (0.95–1.49) using the Strict definition, after adjustment for age, sex, race, body mass index, Townsend deprivation index, education, smoking status, alcohol intake, regular physical activity, sedentary time, sleep duration, healthy diet and glucosamine use. As metabolic health scores improved, the risk of all-cause death decreased: compared to participants scoring 6–7 points, participants with 0 points had an HR = 1.87 (1.24–2.82) using our new definition and HR = 1.28 (0.86–1.91) using the ATPIII definition, while those with 0 points showed an HR = 1.32 (1.04–1.68) using the Strict definition (compared to participants scoring 3 points). We observed that a high metabolic score significantly associated with reduced risk of all-cause mortality for all three definitions of metabolic health.
Fig. 2.
Multivariable Cox regression analysis of metabolic score and metabolic health in relation to all-cause mortality.
Model1: crude model, non-adjusted; Model 2: adjustment was made for age, sex, race, body mass index, Townsend deprivation index (in the UK Biobank, equivalent to the Poverty income ratio in the NHANES dataset), education, smoking status, alcohol intake, regular physical activity, sedentary time, sleep duration, healthy diet and glucosamine use.
Abbreviation: HR, hazard ratio; CI, confidence interval.
In the NHANES dataset, similar results were found where individuals with metabolic unhealthiness exhibited a significantly elevated risk of all-cause mortality compared to metabolically healthy participants: our new definition yielded an HR = 1.16 (1.03–1.30), while the ATPIII definition resulted in an HR = 1.08 (0.96–1.22), and Strict definition showed an HR = 0.99 (0.73–1.34). Compared to participants scoring 6–7 points, participants with 1 point had an HR = 12.61 (2.45–64.99) using our new definition, participants with 0 point had an HR = 1.11 (0.58–2.13) using the ATPIII definition, while those with 0 points showed an HR = 1.02 (0.74–1.41) using the Strict definition (compared to participants scoring 3 points). Our new definition demonstrated that a higher metabolic score (healthy metabolic status) was associated with a significant reduction in the risk of all-cause mortality.
3.4. Subgroup analysis
The sex subgroup analyses of the UK Biobank, presented in Supplementary Fig. 7, indicated that the association between metabolically healthy status and all-cause mortality was more pronounced in males than in females (P for interaction = 0.0204). In the NHANES database, across all three metabolic definitions, the risk was found to be higher in males, yet no significant interaction effects were observed.
Across all three metabolic definitions, in the UK Biobank dataset, the risk was notably higher among individuals aged 45 and above. In contrast, within the NHANES dataset, the risk was more pronounced among those aged 60 and above. Nevertheless, no significant interaction effects were detected (P for interaction >0.05, Supplementary Fig. 8).
Both the new definition and the ATP III definition indicated that, in the UK Biobank dataset, the association between metabolically healthy status and all - cause mortality was more pronounced in hip OA and other - site OA. Statistical analysis showed no significant interaction effects (P for interaction >0.05, Supplementary Fig. 9).
Metabolic health was associated with a reduced risk of all-cause mortality, consistent with the findings reported in Fig. 2. Moreover, compared with the other two definitions, our new one showed a far more robust link to all-cause mortality. In multivariable-adjusted Cox regression analysis, the HR value for our new definition was the highest among the three, indicating its stronger predictive power for all-cause mortality. Across many subgroups in both datasets, the associations defined by our new approach remained statistically significant. This consistency highlights the reliability and stability of our new definition across different sub-populations. (In the UK Biobank dataset, among individuals aged over 60 years, our new definition yielded a HR of 1.43 (1.31–1.55), the ATPIII definition resulted in an HR of 1.20 (1.10–1.31), and the Strict definition showed an HR of 1.35 (1.01–1.81). In the NHANES dataset, among individuals aged over 60 years, our new definition yielded a HR of 1.26 (1.12–1.43), the ATPIII definition resulted in an HR of 1.15 (1.02–1.31), and the Strict definition showed an HR of 1.35 (0.96–1.90). Supplementary Figs. 7, 8, 9).
3.5. Sensitivity analysis
Sensitivity analyses in both databases, after excluding those below 45 years old at baseline, individuals who died within the first two years of follow-up, participants with a history of cardiovascular disease and cancer, those taking NSAIDs or tramadol, and those with a history of joint replacement, did not significantly alter the results (Supplementary Fig. 10). In both study, adherence to our newly constructed healthy weighted metabolic score was significantly linked to a reduced risk of all-cause mortality, with HRs remaining robustly indicative of this association after multivariable adjustment and in comparison to other two definitions (Supplementary Fig. 11).
3.6. Comparison of the three definitions for predicting 10-year all-cause mortality using IDI and NRI
In the UK Biobank, relative to the ATP III definition, the changes in IDI and NRI for our new definition were 0.1 (95 % CI: 0.01–0.20) and 10.60 (95 % CI: 1.80–13.60), respectively. Meanwhile, in comparison with the Strict definition, the corresponding changes in IDI and NRI for our new definition were 0.18 (95 % CI: 0.10–0.20) and 12.10 (95 % CI: 10.30–13.80), respectively (Supplementary Figs. 12 and 13). These findings suggest that our new definition exhibits a superior capacity for discriminating and reclassifying all-cause mortality risk over a decade. No significant discrepancies were observed among the three definitions of IDI and NRI in NHANES dataset.
3.7. Comparison of the three definitions using concordance
In the multivariate adjusted models, the Concordance values for the new definition, ATP-III definition, and Strict definition in the UK Biobank were 0.704, 0.704, and 0.702, respectively. In the NHANES dataset, after multivariate adjustment, the Concordance values for all three definitions were consistently 0.781. These results suggest that in both datasets, there were no significant differences in Concordance values among the new definition, ATP-III definition, and Strict definition (Supplementary Fig. 14).
4. Discussion
This systematic evaluation of mortality data and various anthropometric and metabolic factors led to the development of a new definition which classifies OA participants as either metabolically healthy or unhealthy. According to the new definition, 36.0 % of OA individuals in UK Biobank were categorized as metabolically unhealthy. In the representative sample of American adults from NHANES, 39.1 % fell into the metabolically unhealthy category. The escalating prevalence of metabolic syndrome among various sociodemographic groups in the United States in recent years is well-documented, with more than a third of American classified as having metabolic syndrome according to the conventional criteria [27]. This rate is approximately three times higher than that observed in South Korea [32]. Furthermore, when compared to their British counterparts, American Whites present with increased body weight and waist circumference, particularly among males, which is indicative of a higher risk for metabolic dysregulation [33]. These observations are consistent with our findings and provide a coherent explanation for the trends observed, aligning with the broader scientific understanding of metabolic health disparities. Metabolically unhealthy individuals exhibited an elevated risk of all-cause mortality in both cohorts compared to their metabolically healthy counterparts. The robustness of these findings was confirmed after adjusting for multiple variables. Furthermore, our new definition demonstrated a more pronounced HR than both ATP-III and Strict definitions in both cohorts (HR, 1.36 in UK Biobank and 1.16 in NHANES), suggesting a stronger predictive value for identifying increased mortality risk among OA participants with altered metabolic status. The combination of seven metabolic factors implies that maintaining a healthier metabolic status may be linked to a greater reduction in risk of all-cause mortality, indicating the importance of preserving multiple healthy metabolic factors rather than solely focusing on one as another adverse metabolic factor may counteract its benefits.
Three subgroup analyses were conducted to investigate the association between metabolic health and all-cause mortality. The results revealed that, across all three definitions of metabolic health, men had a higher incidence of all - cause mortality than women. Significantly, in the UK Biobank dataset, only our new definition detected a sex - metabolic health interaction. This finding emphasizes the importance of paying particular attention to the metabolic status of men, suggesting that there may be underlying factors or mechanisms specific to men that increase their susceptibility to adverse health outcomes related to metabolic dysfunction. Understanding these sex differences can significantly impact public health interventions and strategies aimed at improving overall population well-being. By recognizing the high risk faced by men, healthcare professionals can adjust prevention measures and treatments accordingly. Additionally, it is imperative to underscore the significance of metabolic health in individuals over 45 years old (especially over 60 years old) with OA. When comparing metabolically unhealthy individuals to their healthy counterparts, in the UK Biobank, the group aged 45–60 had the highest all - cause mortality rates. In NHANES, the group of individuals over 60 years old exhibited the highest all-cause mortality rates. Although there was no interaction between age and metabolic health, we contend that closely monitoring one's metabolic status is particularly crucial for those over 60 years old as it may potentially mitigate adverse health outcomes. In the UK Biobank, during the subgroup analysis of OA sites, the new definition revealed that among patients, excluding those with hand OA, metabolic unhealthiness was more strongly associated with an elevated risk of all - cause mortality. The ATPIII definition demonstrated that in patients with hip OA and other OA (encompassing OA at different sites such as the spine, shoulder, and sacroiliac joints, excluding hip, knee, and hand OA), metabolic unhealthiness was more strongly correlated with an increased risk of all - cause mortality. Overall, when evaluating the health risks of OA patients, it is crucial to comprehensively consider the specificity of joint sites and the intricate role that metabolic factors play. Particular attention should be paid to the management of the metabolic status of patients with hip OA and those with other types of OA (including OA at different sites like the spine, shoulder, and sacroiliac joints).
Emerging evidence delineates a pathological continuum linking metabolic dysregulation to OA progression. Central to this process, obesity-induced adipokines (e.g., leptin) activate JAK/STAT3 signaling in chondrocytes, triggering cartilage matrix degradation through MMP-13 overexpression [34]. These inflammatory cascades elevate systemic CRP levels (HR = 1.22 for all-cause mortality), creating a pro-osteoclastic microenvironment [35]. Crucially, lipid overload disrupts ubiquitin-proteasome homeostasis, as evidenced by palmitoylated protein accumulation impairing TRAF6 degradation – a mechanism perpetuating NF-κB activation and synovial M1 macrophage polarization via USP7-mediated histone modifications [36]. Therapeutic interventions like semaglutide could break this vicious cycle through dual actions: substantial weight reduction (−13.7 % vs placebo) and NLRP3 inflammasome suppression in joint-resident macrophages [37]. The convergence of metabolic stress and ubiquitination imbalance explains the heightened mortality risk (HR 1.36) in hip OA patients, thus underscoring the necessity for mechanism-targeted management strategies.
Our definition significantly deviates from the conventional metabolic syndrome definition. Instead of using waist circumference as an indicator of abdominal obesity, we have substituted it with waist-to-hip ratio, which offers the advantage of considering hip circumference as well. Waist circumference serves as an estimation for visceral fat, a risk factor for metabolic disorders, while hip circumference acts as a proxy for lower body fat, potentially exerting a protective effect on metabolism[[38], [39], [40]]. Hence, we have opted for a waist-to-hip ratio that comprehensively encompasses body fat assessment. Furthermore, we incorporated BMI into our definition as an indicator linked to the development of OA and all-cause mortality[[41], [42], [43]]. The selection of other metabolic factors aligns with the conventional definition of metabolic syndrome. Although measures of blood pressure and triglycerides were not significantly associated with all-cause mortality in women in our study, our new definition still included health criteria for blood pressure and triglycerides, given the well-known association between dyslipidemia and increased risk of death and cardiovascular disease [44,45]. We observed that the association between individual metabolic factors and all-cause mortality risk was nonlinear, and employed adjusted RCS to identify optimal health ranges for these factors. It is noteworthy that our study revealed an inverse relationship between LDL levels and the risk of all-cause mortality in both sexes, contradicting the common association of elevated LDL with increased risks of all-cause and cardiovascular mortality [46]. Moreover, a U-shaped relationship appears to exist in heart failure patients, where both very low and high LDL levels are associated with adverse outcomes [43]. The optimal range of LDL levels may vary depending on the patient population and underlying health conditions. Therefore, we advocate that for individuals with OA, a LDL level above 3.7 mmol/L falls within a healthy range. In addition, our study revealed a positive relationship between HDL levels and the risk of all-cause mortality in men. A previous study has suggested that extremely high HDL levels are associated with an increased risk of all-cause mortality [47]. Therefore, we advocate that for men with OA, a HDL level below 1.5 mmol/L falls within a healthy range.
In UK Biobank and NHANES datasets, we compared the ATP III definition and Strict definition with our novel definition in predicting 10-year all-cause mortality. We assessed the discriminatory ability of the new model compared to the old models using IDI and NRI. IDI evaluates whether the new model better distinguishes between high-risk and low-risk samples, while NRI assesses whether it improves sample reclassification into correct risk groups [48]. In the UK Biobank dataset, our new definition exhibited improved performance compared to both the ATP III definition and the Strict definition, as indicated by a significant reduction in IDI and NRI values. These findings highlight that our new definition demonstrates a robust capability to identify metabolic-related all-cause mortality risk specifically within OA population. Additionally, we utilized concordance to evaluate the correlation between the likelihood of all-cause mortality and the model constructed from three metabolic health definitions [49]. Our findings demonstrate that the new definition exhibits a higher concordance value compared to the other two definitions in both UK Biobank and NHANES datasets, indicating that our new definition has superior fitting accuracy with observed data. Our new definition offers a more comprehensive explanation for predicting all-cause mortality than existing criteria by incorporating additional factors or enhancing current standards to better capture how metabolic health impacts overall mortality risk. The consistency of our results across populations in both UK Biobank and NHANES datasets further strengthens our conclusions, suggesting that metabolically unhealthy individuals identified by our new definition are at greater risk for adverse health outcomes.
The key strengths of current study lie in its large sample sizes from two well-established national-wide cohorts in the United Kingdom and the United States. Additionally, we developed a metabolically healthy score specifically for individuals with OA to evaluate the complex relationship between metabolic status and risk of mortality, which yielded better predicted effect sizes than conventional metabolic scores. Furthermore, sensitivity analysis was conducted to demonstrate the robustness of results. However, it should be noted that there are also limitations. First, the assumption that all metabolically healthy factors have an equal impact on survival outcomes may not hold true, as the biological relevance and risk associations of individual factors can vary significantly. Second, exclusion of participants due to lack of data on blood biochemistry could potentially introduce bias into the survival analysis. Third, since only baseline metabolic status was considered in this study, changes in metabolic health during follow-up cannot be estimated. Fourth, although some covariates were adjusted in our models, residual confounding cannot be completely ruled out. Last, these findings are based on UK Biobank and NHANES datasets and may not necessarily generalize to other populations or countries. Future studies should aim to replicate these results across diverse cohorts globally for a more comprehensive understanding of the specific relationship between metabolic health and risk of mortality.
5. Conclusion
In this study, we have developed a new definition for identifying metabolic status in individuals with OA, based on waist-to-hip ratio, BMI, blood pressure, blood glucose levels, lipid profiles, and C-reactive protein levels. This comprehensive new definition can be effectively utilized in both clinical and research settings. Our findings demonstrate that the classification of having poor metabolic health according to this new definition is significantly associated with an elevated risk of all-cause mortality among patients with OA. Moreover, our improved criteria for defining metabolic health outperform previous standards and exhibit predictive capabilities for the risk of mortality. These results underscore the significance of providing holistic care to individuals with OA by addressing both joint health and overall metabolic well-being.
Author contributions
HY, MZ and ZZ designed the study. HY, MZ, TF, CH managed and analyzed the data. HY wrote the first draft of the article. All authors contributed to the data interpretation and report preparation, providing intellectual content to the manuscript, and granting approval for submission.
Data Availability
The dataset supporting the conclusions of this article is available in the UK Biobank repository in https://www.ukbiobank.ac.uk/(Application ID: 67654). Data from NHANES are publicly available at https://www.cdc.gov/nchs/nhanes/index.htm.
Ethics
UK Biobank has received ethical approval from the UK National Health Service's National Research Ethics Service (16/NW/0274). NHANES has received ethical approval from the NCHS Research Ethics Review Board.
Role of the Funding source
This work was supported by the National Natural Science Foundation of China (Grant No. 32000925, 82372428, 81974342), and the Guangzhou Science and Technology Program (202002030481). The funders played no role in the design of the study, collection and analysis of data, determination to publish, or preparation of the manuscript.
Conflicts of interest
The authors declare no competing financial interests or personal relationships that could influence this work.
Acknowledgments
This study makes use of data from UK Biobank (Project ID: 67654) and we thank the UK Biobank participants and the UK Biobank team for generating an important research resource. The graphical abstract was created in BioRender. Yang, H. (2025) https://BioRender.com/n21d456.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jot.2025.02.004.
Contributor Information
Hao Yang, Email: yhaodoc@163.com.
Muhui Zeng, Email: 13mhzeng@stu.edu.cn.
Tianxiang Fan, Email: tianxiang.fan@connect.polyu.hk.
Haowei Chen, Email: smuchw@163.com.
Xiaofeng Fang, Email: 726364191@qq.com.
Zhong Alan Li, Email: alanli@cuhk.edu.hk.
Xiaoshuai Wang, Email: drwangxs2019@126.com.
David J. Hunter, Email: david.hunter@sydney.edu.au.
Changhai Ding, Email: changhai.ding@utas.edu.au.
Zhaohua Zhu, Email: zhaohua.zhu@utas.edu.au.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
References
- 1.Dj H., S B.-Z. Osteoarthritis. Lancet (London, England) 2019;393 doi: 10.1016/S0140-6736(19)30417-9. [DOI] [Google Scholar]
- 2.V D., Wm O., C D., Ag C., Dj H. Evaluation and treatment of knee pain: a review. JAMA. 2023;330 doi: 10.1001/jama.2023.19675. [DOI] [PubMed] [Google Scholar]
- 3.GBD 2021 Osteoarthritis Collaborators Global, regional, and national burden of osteoarthritis, 1990-2020 and projections to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Rheumatol. 2023;5:e508–e522. doi: 10.1016/S2665-9913(23)00163-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Nüesch E., Dieppe P., Reichenbach S., Williams S., Iff S., Jüni P. All cause and disease specific mortality in patients with knee or hip osteoarthritis: population based cohort study. BMJ. 2011;342 doi: 10.1136/bmj.d1165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Mobasheri A., Rayman M.P., Gualillo O., Sellam J., van der Kraan P., Fearon U. The role of metabolism in the pathogenesis of osteoarthritis. Nat Rev Rheumatol. 2017;13:302–311. doi: 10.1038/nrrheum.2017.50. [DOI] [PubMed] [Google Scholar]
- 6.Li H., George D.M., Jaarsma R.L., Mao X. Metabolic syndrome and components exacerbate osteoarthritis symptoms of pain, depression and reduced knee function. Ann Transl Med. 2016;4:133. doi: 10.21037/atm.2016.03.48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yoshimura N., Muraki S., Oka H., Kawaguchi H., Nakamura K., Akune T. Association of knee osteoarthritis with the accumulation of metabolic risk factors such as overweight, hypertension, dyslipidemia, and impaired glucose tolerance in Japanese men and women: the ROAD study. J Rheumatol. 2011;38:921–930. doi: 10.3899/jrheum.100569. [DOI] [PubMed] [Google Scholar]
- 8.Zhuo Q., Yang W., Chen J., Wang Y. Metabolic syndrome meets osteoarthritis. Nat Rev Rheumatol. 2012;8:729–737. doi: 10.1038/nrrheum.2012.135. [DOI] [PubMed] [Google Scholar]
- 9.Zhou Z., Macpherson J., Gray S.R., Gill J.M.R., Welsh P., Celis-Morales C., et al. Are people with metabolically healthy obesity really healthy? A prospective cohort study of 381,363 UK Biobank participants. Diabetologia. 2021;64:1963–1972. doi: 10.1007/s00125-021-05484-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gao M., Lv J., Yu C., Guo Y., Bian Z., Yang R., et al. Metabolically healthy obesity, transition to unhealthy metabolic status, and vascular disease in Chinese adults: a cohort study. PLoS Med. 2020;17 doi: 10.1371/journal.pmed.1003351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Shao F., Chen Y., Xu H., Chen X., Zhou J., Wu Y., et al. Metabolic obesity phenotypes and risk of lung cancer: a prospective cohort study of 450,482 UK biobank participants. Nutrients. 2022;14:3370. doi: 10.3390/nu14163370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Song Z., Gao M., Lv J., Yu C., Guo Y., Bian Z., et al. Metabolically healthy obesity, transition to unhealthy phenotypes, and type 2 diabetes in 0.5 million Chinese adults: the China Kadoorie Biobank. Eur J Endocrinol. 2022;186:233–244. doi: 10.1530/EJE-21-0743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Grundy S.M., Brewer H.B., Cleeman J.I., Smith S.C., Lenfant C., et al. Definition of metabolic syndrome: report of the national heart, lung, and blood institute/American heart association conference on scientific issues related to definition. Circulation. 2004;109:433–438. doi: 10.1161/01.CIR.0000111245.75752.C6. American Heart Association. [DOI] [PubMed] [Google Scholar]
- 14.Sudlow C., Gallacher J., Allen N., Beral V., Burton P., Danesh J., et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12 doi: 10.1371/journal.pmed.1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ahmed N.N., Sherman S.J., Vanwyck D. Frailty in Parkinson's disease and its clinical implications. Parkinsonism Relat Disord. 2008;14:334–337. doi: 10.1016/j.parkreldis.2007.10.004. [DOI] [PubMed] [Google Scholar]
- 16.UK BIOBANK ETHICS AND GOVERNANCE FRAMEWORK Version 3.0. October 2007. https://www.ukbiobank.ac.uk/media/0xsbmfmw/egf.pdf [Google Scholar]
- 17.Wan Z., Guo J., Pan A., Chen C., Liu L., Liu G. Association of serum 25-hydroxyvitamin D concentrations with all-cause and cause-specific mortality among individuals with diabetes. Diabetes Care. 2021;44:350–357. doi: 10.2337/dc20-1485. [DOI] [PubMed] [Google Scholar]
- 18.Zembic A., Eckel N., Stefan N., Baudry J., Schulze M.B. An empirically derived definition of metabolically healthy obesity based on risk of cardiovascular and total mortality. JAMA Netw Open. 2021;4 doi: 10.1001/jamanetworkopen.2021.8505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee D.H., Keum N., Hu F.B., Orav E.J., Rimm E.B., Willett W.C., et al. Predicted lean body mass, fat mass, and all cause and cause specific mortality in men: prospective US cohort study. BMJ. 2018;362 doi: 10.1136/bmj.k2575. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xie H., Zhang H., Ruan G., Wei L., Ge Y., Lin S., et al. Individualized threshold of the involuntary weight loss in prognostic assessment of cancer. J Cachexia Sarcopenia Muscle. 2023;14:2948–2958. doi: 10.1002/jcsm.13368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hou X.-Z., Lv Y.-F., Li Y.-S., Wu Q., Lv Q.-Y., Yang Y.-T., et al. Association between different insulin resistance surrogates and all-cause mortality in patients with coronary heart disease and hypertension: NHANES longitudinal cohort study. Cardiovasc Diabetol. 2024;23:86. doi: 10.1186/s12933-024-02173-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhang Y.-B., Chen C., Pan X.-F., Guo J., Li Y., Franco O.H., et al. Associations of healthy lifestyle and socioeconomic status with mortality and incident cardiovascular disease: two prospective cohort studies. BMJ. 2021;373 doi: 10.1136/bmj.n604. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Seeman T., Merkin S.S., Crimmins E., Koretz B., Charette S., Karlamangla A. Education, income and ethnic differences in cumulative biological risk profiles in a national sample of US adults: NHANES III (1988-1994) Soc Sci Med. 2008;66:72–87. doi: 10.1016/j.socscimed.2007.08.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Johannesen C.D.L., Langsted A., Mortensen M.B., Nordestgaard B.G. Association between low density lipoprotein and all cause and cause specific mortality in Denmark: prospective cohort study. BMJ. 2020;371 doi: 10.1136/bmj.m4266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Valgimigli M., Hong S.-J., Gragnano F., Chalkou K., Franzone A., da Costa B.R., et al. De-escalation to ticagrelor monotherapy versus 12 months of dual antiplatelet therapy in patients with and without acute coronary syndromes: a systematic review and individual patient-level meta-analysis of randomised trials. Lancet. 2024;404:937–948. doi: 10.1016/S0140-6736(24)01616-7. [DOI] [PubMed] [Google Scholar]
- 26.Guven G., Brankovic M., Constantinescu A.A., Brugts J.J., Hesselink D.A., Akin S., et al. Preoperative right heart hemodynamics predict postoperative acute kidney injury after heart transplantation. Intensive Care Med. 2018;44:588–597. doi: 10.1007/s00134-018-5159-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Rawshani A., Rawshani A., Franzén S., Sattar N., Eliasson B., Svensson A.-M., et al. Risk factors, mortality, and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2018;379:633–644. doi: 10.1056/NEJMoa1800256. [DOI] [PubMed] [Google Scholar]
- 28.Zeng C., Dubreuil M., LaRochelle M.R., Lu N., Wei J., Choi H.K., et al. Association of tramadol with all-cause mortality among patients with osteoarthritis. JAMA. 2019;321:969–982. doi: 10.1001/jama.2019.1347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hunt L.P., Ben-Shlomo Y., Clark E.M., Dieppe P., Judge A., MacGregor A.J., et al. 45-day mortality after 467,779 knee replacements for osteoarthritis from the National Joint Registry for England and Wales: an observational study. Lancet. 2014;384:1429–1436. doi: 10.1016/S0140-6736(14)60540-7. [DOI] [PubMed] [Google Scholar]
- 30.Lourida I., Hannon E., Littlejohns T.J., Langa K.M., Hyppönen E., Kuzma E., et al. Association of lifestyle and genetic risk with incidence of dementia. JAMA. 2019;322:430–437. doi: 10.1001/jama.2019.9879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cox model assumptions - easy guides - wiki - STHDA n.d. http://www.sthda.com/english/wiki/cox-model-assumptions
- 32.Park J., Mendoza J.A., O'Neil C.E., Hilmers D.C., Liu Y., Nicklas T.A. A comparison of the prevalence of the metabolic syndrome in the United States (US) and Korea in young adults aged 20 to 39 years. Asia Pac J Clin Nutr. 2008;17:471–482. [PubMed] [Google Scholar]
- 33.Wells J.C.K., Cole T.J., Bruner D., Treleaven P. Body shape in American and British adults: between-country and inter-ethnic comparisons. Int J Obes. 2008;32:152–159. doi: 10.1038/sj.ijo.0803685. [DOI] [PubMed] [Google Scholar]
- 34.Sampath S.J.P., Venkatesan V., Ghosh S., Kotikalapudi N. Obesity, metabolic syndrome, and osteoarthritis-an updated review. Curr Obes Rep. 2023;12:308–331. doi: 10.1007/s13679-023-00520-5. [DOI] [PubMed] [Google Scholar]
- 35.Fu K., Cai Q., Jin X., Chen L., Oo W.M., Duong V., et al. Association of serum calcium, vitamin D, and C-reactive protein with all-cause and cause-specific mortality in an osteoarthritis population in the UK: a prospective cohort study. BMC Public Health. 2024;24:2286. doi: 10.1186/s12889-024-19825-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Su S., Tian R., Jiao Y., Zheng S., Liang S., Liu T., et al. Ubiquitination and deubiquitination: implications for the pathogenesis and treatment of osteoarthritis. Journal of Orthopaedic Translation. 2024;49:156–166. doi: 10.1016/j.jot.2024.09.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Bliddal H., Bays H., Czernichow S., Uddén Hemmingsson J., Hjelmesæth J., Hoffmann Morville T., et al. Once-weekly semaglutide in persons with obesity and knee osteoarthritis. N Engl J Med. 2024;391:1573–1583. doi: 10.1056/NEJMoa2403664. [DOI] [PubMed] [Google Scholar]
- 38.Stefan N. Causes, consequences, and treatment of metabolically unhealthy fat distribution. Lancet Diabetes Endocrinol. 2020;8:616–627. doi: 10.1016/S2213-8587(20)30110-8. [DOI] [PubMed] [Google Scholar]
- 39.Lotta L.A., Wittemans L.B.L., Zuber V., Stewart I.D., Sharp S.J., Luan J., et al. Association of genetic variants related to gluteofemoral vs abdominal fat distribution with type 2 diabetes, coronary disease, and cardiovascular risk factors. JAMA. 2018;320:2553–2563. doi: 10.1001/jama.2018.19329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Neeland I.J., Poirier P., Després J.-P. Cardiovascular and metabolic heterogeneity of obesity: clinical challenges and implications for management. Circulation. 2018;137:1391–1406. doi: 10.1161/CIRCULATIONAHA.117.029617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Salis Z., Gallego B., Nguyen T.V., Sainsbury A. Association of decrease in body mass index with reduced incidence and progression of the structural defects of knee osteoarthritis: a prospective multi-cohort study. Arthritis Rheumatol. 2023;75:533–543. doi: 10.1002/art.42307. [DOI] [PubMed] [Google Scholar]
- 42.Lohmander L.S., Gerhardsson de Verdier M., Rollof J., Nilsson P.M., Engström G. Incidence of severe knee and hip osteoarthritis in relation to different measures of body mass: a population-based prospective cohort study. Ann Rheum Dis. 2009;68:490–496. doi: 10.1136/ard.2008.089748. [DOI] [PubMed] [Google Scholar]
- 43.Di Angelantonio E., Bhupathiraju S., Wormser D., Gao P., Kaptoge S., et al. Body-mass index and all-cause mortality: individual-participant-data meta-analysis of 239 prospective studies in four continents. Lancet. 2016;388:776–786. doi: 10.1016/S0140-6736(16)30175-1. Global BMI Mortality Collaboration null. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kopin L., Lowenstein C. Dyslipidemia. Ann Intern Med. 2017;167:ITC81–96. doi: 10.7326/AITC201712050. [DOI] [PubMed] [Google Scholar]
- 45.Yang W.-Y., Melgarejo J.D., Thijs L., Zhang Z.-Y., Boggia J., Wei F.-F., et al. Association of office and ambulatory blood pressure with mortality and cardiovascular outcomes. JAMA. 2019;322:409–420. doi: 10.1001/jama.2019.9811. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Johannesen C.D.L., Langsted A., Mortensen M.B., Nordestgaard B.G. Association between low density lipoprotein and all cause and cause specific mortality in Denmark: prospective cohort study. BMJ. 2020;371 doi: 10.1136/bmj.m4266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.von Eckardstein A., Nordestgaard B.G., Remaley A.T., Catapano A.L. High-density lipoprotein revisited: biological functions and clinical relevance. Eur Heart J. 2023;44:1394–1407. doi: 10.1093/eurheartj/ehac605. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Everett B.M., Cook N.R., Conen D., Chasman D.I., Ridker P.M., Albert C.M. Novel genetic markers improve measures of atrial fibrillation risk prediction. Eur Heart J. 2013;34:2243–2251. doi: 10.1093/eurheartj/eht033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Heller G. A measure of explained risk in the proportional hazards model. Biostatistics. 2012;13:315–325. doi: 10.1093/biostatistics/kxr047. [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
The dataset supporting the conclusions of this article is available in the UK Biobank repository in https://www.ukbiobank.ac.uk/(Application ID: 67654). Data from NHANES are publicly available at https://www.cdc.gov/nchs/nhanes/index.htm.