Introduction

Sedentary behavior is commonly defined as any awake activity performed while sitting, reclining, or lying down, with an energy expenditure ≤ 1.5 metabolic equivalents1. Because of the widespread use of electronic devices, changes in transportation patterns, growth of sedentary occupations, sedentary behavior has become common in modern society2. Recent studies conclude that sedentary behavior is a risk factor for various diseases, such as metabolic syndrome, cardiovascular disease, Parkinson’s disease, increase of body fat, and decrease of bone density2,3,4,5. Therefore, sedentary behavior is regarded as a modifiable risk factor for several diseases. As recommended by the World Health Organization (WHO) 2020 Global Guidelines on Physical Activity and Sedentary Behavior, adults should limit their sedentary time and replace it with physical activities.

Aging represents the deterioration of biological functions and dysregulation of multiple physiological systems6. Unquestionably, aging is an inevitable process, while the rate of aging is heterogeneous. Accelerated aging increases the susceptibility to death and diseases7. At the same time, it’s possible to avoid or postpone the onset of age-related diseases by slowing the rate of aging8. Phenotypic age (PhenoAge) serves as an emerging indicator of an individual’s biological age9 and has been shown to outperform in capturing health risk relative to telomere length10.

Recently, limited studies conclude that prolonged sedentary behavior may accelerate the biological aging process11,12. Sedentary behavior is related to a higher BMI, which has been shown to be linked to a biological aging13. Unlike molecular markers, BMI is routinely measured in clinical practice and effectively modified through lifestyle changes. Considering the limitations of previous studies, we aim to examine the association of sitting time and PhenoAge among adults in the U.S. population and we hypothesize that sedentary activity accelerates biological aging and BMI plays a mediating role.

Method

Data source

The National Health and Nutrition Examination Survey (NHANES) dataset is a public, cross-sectional, countrywide, two-year cycled survey which collects demographic, socioeconomic, health, and nutritional information from the US population. Conducted by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC), NHANES employs a complex, multistage, stratified, clustered probabilistic design and written informed consent are obtained from all participants14.

The data utilized in the present study were obtained from five successive NHANES cycles, spanning from 2007 to 2016 (n = 50,588). We excluded participants with missing data on sitting time (n = 15,299), who were unavailable for calculating PhenoAge (n = 19,733), with missing data on other covariates (n = 1,744), or with tumor (n = 1308). Consequently, 12,504 eligible participants aged more than 20 years old were included in the final analytic sample. The complete data integration process was depicted in Fig. 1.

Fig. 1
figure 1

Participant sample flowchart.

Full size image

Outcome measurements

PhenoAge was calculated using chronological age and nine biomarkers: albumin, creatinine, glucose, log [C-reactive protein (CRP)], lymphocyte percentage, mean cell volume, red blood cell distribution width, alkaline phosphatase, and white blood cell count7,15. We trained the PhenoAge algorithm with NHANES III data and calculated PhenoAge in NHANES IV cycles from 2007 to 2016 using the “BioAge” R package (https://github.com/dayoonkwon/BioAge)16. The final equation for calculating Phenotypic Age is provided below7,15:

$$\:Phenotypic\:Age=141.5+\frac{\text{l}\text{n}[-0.00553\times\:\text{l}\text{n}(1-\text{M}\left)\right]}{0.09165}$$

where

$$\:M=1-exp\left(\frac{-1.51714\times\:\text{exp}\left(xb\right)}{0.0076927}\right)$$

.

xb = − 19.907 − 0.0336 × albumin + 0.0095 × creatinine + 0.1953 × glucose + 0.0954 × ln(CRP) − 0.0120 × lymphocyte percent + 0.0268 × mean cell volume + 0.3306 × red cell distribution width + 0.00188 × alkaline phosphatase + 0.0554 × white blood cell count + 0.0804 × chronological age.

Besides, phenotypic age acceleration (PhenoAgeAccel) or deceleration/stasis was estimated from the difference between the PhenoAge and chronological age (determined by the date of birth)17.

Exposure measurements

In NHANES, the physical activity questionnaire was based on the Global Physical Activity Questionnaire (GPAQ) which includes questions related to sedentary activities18. Daily sitting time was measured by the following question: “On a typical day, how much time do you usually spend sitting at school, at home, getting to and from places, or with friends, including time spent sitting at a desk, traveling in a car or bus, reading, playing cards, watching television, or using a computer?”. Consistent with recent research19,20we classified daily sitting time into four categories (< 4, 4–6, 6–8, and ≥ 8 h/day), labeled as Q1, Q2, Q3, and Q4, respectively.

Covariates

Based on prior researches21,22covariates included age, gender (male, female), race (Mexican American, non-Hispanic Black, non-Hispanic White, other Hispanic, and other races), marital status (divorced, married, never married, and others), education level (lower than 12th grade, high school grade, and college grade and above), body mass index (BMI, kg/m2) (≤ 20, 20–25, 25–30, and > 30), poverty income ratio (PIR) (< 1.3, 1.3–3.5, and ≥ 3.5), smoke, physical activity (lightly, moderate, and vigorous), self-reported history of diabetes, coronary heart disease (CHD), stroke, congestive heart failure (CHF), and hypertension (HTN). Smoking was defined as having smoked at least 100 cigarettes in life.

Statistical analyses

Referring to previous NHANES literatures23,24we followed NHANES analytical guidelines and applied appropriate sample weights for the enrollment of participants. Continuous variables were expressed as weighted mean ± standard error (SE) and analyzed via the Student’s t-test for normally distributed data or the Mann-Whitney U-test for skewed distribution data, while categorical variables were expressed as weighted percentages and analyzed using the Rao-Scott Chi-square. The baseline characteristics of the participants were compared according to PhenoAge deceleration/stasis and PhenoAgeAccel. Weighted multivariate logistic regression was employed to calculate the odd ratios (OR) and 95% confidence intervals for the association of sitting time and PhenoAge. Three models were constructed: Model 1 was unadjusted, Model 2 was adjusted for age, gender, race, education level, PIR, marital status, and Model 3 incorporated further adjustments for BMI, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN. Restricted cubic spline regression was used to explore potential non-linear relationships between sitting time and PhenoAge. Referring to previous studies, RCS is particularly suitable for analyzing dose-response relationships with non-linear characteristics25,26. Additionally, RCS models with three knots positioned at the (10th, 50th, and 90th) percentiles were applied to further explore the dose-effect relationship. Stratified analyses by potential effect modifiers and tests for multiplicative interactions were performed. Furthermore, we adopted a nonparametric bootstrap method resampled 5000 times to evaluate the indirect impact of sitting time on PhenoAgeAccel mediated through BMI. The proportion mediated was calculated as the ratio of the indirect effect to the total effect, and statistical significance was determined using bootstrapped 95% confidence intervals. This robust method, widely applied in epidemiological studies27,28avoids assumptions of normality and provides reliable estimates of mediation effects.

All statistical analyses were performed using R software (version 4.2.2) with a 2-sided P < 0.05 considered statistically significant. All regression models were fitted using the svyglm function from the survey package in R.

Results

Population characteristics

Table 1 showed the characteristics of the participants. A total of 12,504 eligible participants (weighted population, 100,939,130) with an average age of 45.3 ± 16.0 years were enrolled in the analysis, among whom 48.4% were males. Individuals with PhenoAgeAccel were usually older, more likely to be male, had less education level and longer sitting time, exhibited a higher BMI and lower PIR, more likely to smoke. Additionally, they were more likely to suffer from diabetes, CHD, stroke, CHF, and HTN.

Table 1 Baseline characteristics of participants.
Full size table

Association of sitting time and phenoageaccel

The results of weighted univariable and multivariable logistic regression models were summarized in Table 2. In crude model (Model 1), adults who sat longer significantly showed greater increases in PhenoAgeAccel (Q2: OR = 1.27, 95%CI = 1.07, 1.51, p = 0.007; Q3: OR = 1.28, 95%CI = 1.06, 1.55, p = 0.011; Q4: OR = 1.45, 95%CI = 1.25, 1.69, p < 0.001) compared to the reference (Q1). Similar association was observed in Model 3. After fully adjusting for age, gender, race, education level, PIR, marital status, BMI, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN, longer sitting time (Q2: OR = 1.30, 95%CI = 1.06, 1.58, p = 0.013; Q3: OR = 1.25, 95%CI = 1.01, 1.55, p = 0.038; Q4: OR = 1.58, 95%CI = 1.33, 1.88, p < 0.001) statistically significantly corresponded to an increase in PhenoAgeAccel in comparison to the sitting time < 4 h/d. Moreover, Fig. 2 illustrated an approximately positively linear relationship (p for nonlinear = 0.1842) between sitting time and PhenoAgeAccel in a fully adjusted model (Model 3).

Table 2 Associations between sitting time and PhenoAgeAccel.
Full size table
Fig. 2
figure 2

The association of sitting time with PhenoAgeAccel visualized by restricted cubic spline. Risk ratios were adjusted for age, gender, race, education level, PIR, marital status, BMI, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN.

Full size image

Stratified analyses

We conducted stratified analyses to explored the relationship between sitting time and PhenoAgeAccel based on BMI, gender, race, marital status, education, PIR, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN (Fig. 3). The findings indicated that race, education level, CHF and HNT could serve as moderating factors in the relationship between sitting time and PhenoAgeAccel (P for interaction < 0.05).

Fig. 3
figure 3

Subgroup analysis for the association between sitting time and PhenoAgeAccel. HRs were adjusted for age, gender, race, education level, PIR, marital status, BMI, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN.

Full size image

Mediation analyses

Mediation analyses revealed the mediating effect of BMI on the correlation between sitting time and PhenoAgeAccel. The indirect mediated effect through BMI was 0.004 (95%CI: 0.002–0.004). The results indicated that BMI mediated 21.0% variance of the total effect and was a significant mediator of the association between sitting time and PhenoAgeAccel. Figure 4 displayed the mediation pathway model.

Fig. 4
figure 4

The mediating effect of BMI on the relationship between sitting time and PhenoAgeAccel. Adjusted for age, gender, race, education level, PIR, marital status, BMI, smoking, physical activity, diabetes, CHD, stroke, CHF, and HTN.

Full size image

Discussion

Using the 12,504 nationally representative U.S. adult participants in the NHANES, we found a strong positive correlation between sitting time and PhenoAgeAccel. The dose–response analysis showed a linear association of sitting time and PhenoAgeAccel and the mediation analyses suggested the role of BMI as a mediator in this relationship.

With the changes in lifestyle attributed to technological advances, sitting time was prolonged and the negative health effects of sedentary behaviors attracted attention. In 2023, You et al.11 enrolled 6,439 representative adults from NHANES to conduct a cross-sectional study and concluded that prolonged sedentary time was associated with elevated PhenoAge. Lately, Lu et al., using data from the Guangzhou Biobank Cohort Study, triangulate observational and mendelian randomization studies and find that sedentary behavior had a detrimental impact on PhenoAgeAccel12. These results were consistent with ours. Nevertheless, our study adjusted for more confounding variables, explored the dose-response relationship between sitting time and PhenoAgeAccel, and the role of BMI in this relationship.

There were various potential pathways through which sedentary behavior might affect aging. A randomized controlled trial found a negative correlation between sitting time and telomere length29. However, it should be noted that only 49 older people were included in the trial. Hence, the results could simply be a chance finding due to the inadequate sample size30. Moreover, it had been investigated that nutrient sensing pathways could modulate longevity31. Meanwhile, sedentary behavior was able to blunt the nutrient signaling pathways responsible for muscle growth such as insulin/insulin-like growth factor 1 (IGF-1) signaling (IIS) pathways, mechanistic target of rapamycin (mTOR), growth hormone (GH), and testosterone signaling. Furthermore, sedentary behavior was possibly associated with genomic instability, epigenetic alterations, loss of proteostasis, mitochondrial dysfunction, which aggravates the aging process32.

In our mediation analysis, we found that sedentary behavior accelerated aging in part by increasing BMI, which suggested a potential underlying mechanism. Consistent with previous research, sedentary behavior, characterized by low muscle activity, high energy intake, and reduced physical exercise, was related with higher BMI33,34,35,36,37. This increased BMI could induce a highly inflammatory systemic milieu with perturbations of white blood cell counts38,39. An observational twin study conducted by Sara et al.40 demonstrated a significant association between BMI and accelerated epigenetic aging, independent of genetic factors, potentially mediated by insulin resistance. Furthermore, Sangmi et al.41 found that BMI might accelerate aging by shortening telomere length. Moreover, aging of the immune system, reactive oxygen species and redox homeostasis were also confirmed to potentially play roles in obesity-induced aging13.

Subgroup analyses revealed significant effect modifications by race, education level, CHF, and HNT status. The association between sedentary time and PhenoAgeAccel was particularly pronounced in Non-Hispanic Black individuals, those with low education levels, those with HTN, and those without CHF, suggesting potential socioeconomic and cardiometabolic disparities in susceptibility to sedentary-related aging. While these findings underscore high-risk populations for prioritized intervention, residual confounding and small subgroup samples warrant cautious interpretation. Prospective studies with repeated measures are needed to verify these interactions.

It should be acknowledged that this study inevitably had several shortcomings. First, in our research, daily sitting time was assessed by the self-report, which may introduce recall bias. Second, the cross-sectional study design was unavailable to determine causality, which require prospective studies to ascertain. Third, some baseline information, including sitting time and history of diseases, was based on recollection and self-report, which might lead to recall bias. Fourth, although we had tried to take into account as many variables as possible, there were still some unaccounted variables (like energy intake, occupation, and psychological factors) that could influence the relationship between sitting time and aging. Fifth, we didn’t further distinguish work sitting time and leisure sitting time, so we failed to separately explore the impacts of these contexts in these PhenoAgeAccel. Finally, while our stringent inclusion criteria ensured data quality for the mediation analysis, only 24% (12,504/50,588) of the original NHANES participants met all eligibility requirements. The limited sample size may affect the generalizability of our findings to populations with different characteristics .

Conclusion

Using a representative nationwide population, the study determined the association between sitting time and emerging PhenoAge index. Our findings revealed a strong significant and independent relationship between sedentary behavior and PhenoAgeAccel, which exhibited a proximate linear dose-response relationship. Furthermore, BMI served as a mediator of the correlation between sitting time and PhenoAgeAccel.