Association of Physical Activity Trajectories with Longitudinal Changes in Renal Function among Older Chinese Adults: A Prospective Cohort Study

doi:10.21203/rs.3.rs-3872026/v1

Download PDF

Research Article

Association of Physical Activity Trajectories with Longitudinal Changes in Renal Function among Older Chinese Adults: A Prospective Cohort Study

https://doi.org/10.21203/rs.3.rs-3872026/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

To investigate association between physical activity trajectories and longitudinal changes in renal function. We performed this study in Chinese adults with 60 years or older using data from China Health and Retirement Longitudinal Study (CHARLS) in 2011 and 2015.

Methods

Latent class trajectory modeling (LCTM) was applied to identify heterogeneous trajectories of physical activity scores (PAS) from 2011 to 2015. Generalize linear models were adopted to investigate the relationship between trajectories of PAS and the annualised rate of change in estimated glomerular filtration rate based on serum creatinine and cystatin C (slope of eGFRcr-cys). Serial mediation analysis was conducted to reveal significant mediating pathways from trajectories of PAS to the slope of eGFRcr-cys.

Results

A total number of 407 participants aged 60 years or older were divided into 5 classes based on different PAS trajectories. After adjustment for potential confounders (age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid), compared with class 1, class 2 was associated with an increase of 0.65 [95% confidence interval (CI), -0.39-1.68] in the slope of eGFRcr-cys; class 3 was associated with an increase of 1.24 (95%CI, 0.37–2.10) in the slope of eGFRcr-cys; class 4 was associated with an increase of 1.64 (95%CI, 0.65–2.63) in the slope of eGFRcr-cys; class 5 was associated with an increase of 2.37 (95%CI, 1.37–3.36) in the slope of eGFRcr-cys. Notably, a significant mediating pathway from trajectories of PAS to the slope of eGFRcr-cys through serial mediation of triglyceride-glucose (TyG) index and baseline eGFRcr-cys was found.

Conclusions

A slow-rise trajectory starting from a more than moderate level of PAS.(class 5) is preferable in terms of slowing renal function decline among older Chinese adults aged 60 years or more. For older people, this PA trajectory is recommended owing to its potential beneficial impact on renal function. A significant mediating pathway from trajectories of PAS to the slope of eGFRcr-cys is through serial mediation of TyG index and baseline eGFRcr-cys, which partially explains how PA trajectories are associated with longitudinal changes in renal function.

physical activity

renal function

slope of eGFRcr-cys

latent class trajectory modeling

serial mediation analysis

Physical activity (PA) exerts a wide range of beneficial effects for human health. Regular PA has showed significant cardioprotective effects across all age, gender and race groups regardless of its volume[1]. PA also show therapeutic promise for mental health disorders[2]. Intriguingly, for patients with cancer, PA has been shown to decrease cancer incidence and suppress tumor growth[3]. Given various favorable impacts of PA, lots of attention has been paid to this research field. A single blinded multi-center randomized controlled trial (RCT) conducted by Sobol et al.[4] demonstrated positive effects of aerobic exercise on physical performance in patients with mild Alzheimer's disease. Also by RCT, Zhou et al.[5] found that a six-month exercise intervention of mainly brisk walking was linked to improved quality of life in women with ovarian cancer.

In recent years, the relationship of PA and renal function has aroused researchers' interest. A previous cross-sectional study revealed that a higher level of PA was associated with a reduced odds ratio for lower estimated glomerular filtration rate (eGFR)[6]. In 2022, Shlipak et al.[7] showed that structured, moderate PA slowed the rate of decline in estimated glomerular filtration rate based on cystatin C (eGFRcys) among sedentary older adults with a mean age of 78.9 years in a large RCT with 1,199 participants. However, the effects of different PA trajectories on long-term changes in renal function remain unknown. To address this gap in knowledge, we performed the present study in Chinese adults with 60 years or older using two waves of data from China Health and Retirement Longitudinal Study (CHARLS) in 2011 and 2015.

Aim, design and setting of the study

To investigate the association between PA trajectories and longitudinal changes in renal function, we performed the present study with a prospective cohort. Data used in this study were obtained from the CHARLS, a nationally representative longitudinal survey conducted in Chinese population aged 45 years and above[8]. The baseline survey was launched in 2011 and then the participants were followed up biennially. Since serum creatinine and cystatin C levels were measured in 2011 and 2015, the present study used data from 2011 to 2015.

Study population

Participants were excluded if they were lack of information on age or gender in 2011, missing information on PA in 2011 or 2015, or missing information on renal function including serum creatinine and cystatin C in 2011 or 2015, or under 60 years old. A total number of 407 participants were selected. The flow chart showing the selection of study population were presented in Fig. 1.

Physical activity assessment

The weekly volume of PA was reflected by the physical activity score (PAS). The variable “PAS” was constructed according to the International Physical Activity Questionnaire[9]. Specifically, variables used to calculate PAS in the “health status and functioning” section of CHARLS codebooks in 2011 and 2015 included da052_1_(days do vigorous activities at least 10 minutes continuously), da052_2_(days do moderate physical effort at least 10 minutes continuously), da052_3_(days walking at least 10 minutes continuously), da053_1_(2 hours everyday do vigorous activities at least 10 minutes continuously), da053_2_(2 hours everyday do moderate physical effort at least 10 minutes continuously), da053_3_(2 hours everyday walking at least 10 minutes continuously), da054_1_(minutes everyday do vigorous activities at least 10 minutes continuously), da054_2_(minutes everyday do moderate physical effort at least 10 minutes continuously), da054_3_(minutes everyday walking at least 10 minutes continuously), da055_1_(4 hours everyday do vigorous activities at least 10 minutes continuously), da055_2_(4 hours everyday do moderate physical effort at least 10 minutes continuously) and da055_3_(4 hours everyday walking at least 10 minutes continuously). Based on the responses of participants, the daily PA duration index was defined as follows: the duration of PA less than 30 minutes in one day was indexed as 1; the duration ranging from 30 minutes to 2 hours in one day was indexed as 2; the duration ranging from 2 hours to 4 hours in one day was indexed as 3; the duration more than 4 hours in one day was indexed as 4[10]. Then, the weekly PA duration index can be figured out by multiplying the daily PA duration index for each kind of activity and the number of days. Finally, the variable “PAS” was generated using metabolic equivalent multipliers as follows: PAS = 8.0 × weekly vigorous PA duration index + 4.0 × weekly moderate PA duration index + 3.3 × weekly walking duration index[10].

Evaluation of changes in renal function

In 2011 and 2015, the CHARLS collected venous blood samples to measure renal function including serum creatinine and cystatin C levels. We used the new equation without race developed in 2021 by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) to calculate estimated glomerular filtration rate based on serum creatinine and cystatin C (eGFRcr-cys)[11]. The follow-up years were calculated based on the time span between interview dates in 2015 and 2011. The annualised rate of changes in renal function over the four years from 2011 to 2015 (slope of eGFRcr-cys) equal to the quotient of difference between eGFRcr-cys in 2015 (eGFRcr-cys2015) and eGFRcr-cys in 2011 (eGFRcr-cys2011, also called baseline eGFRcr-cys) divided by follow-up years. The formula of changes in renal function is as follows: slope of eGFRcr-cys = (eGFRcr-cys2015 - eGFRcr-cys2011)/follow-up years.

Covariates

In this study, the following covariates were included: age, gender (male, female), area (urban, rural), education level (less than lower secondary, upper secondary & vocational training, tertiary), marital status (married, unmarried/divorced), current smoking (no, yes), drinking in last year (no, yes), body mass index (BMI) categorization (underweight: less than 18.5, normal weight: from 18.5 to 23.9, overweight: from 23 to 24.9, obesity: greater than 25), hypertension (no, yes), diabetes mellitus (no, yes), heart disease (no, yes), dyslipidemia (no, yes), liver disease (no, yes), kidney disease (no, yes), baseline eGFRcr-cys, baseline uric acid, baseline hypersensitive C-reactive protein (hsCRP), and Baseline triglyceride-glucose (TyG) index. The TyG index was calculated as ln[fasting triglyceride (mg/dL) × fasting plasma glucose (mg/dL)/2][12]. For missing of each categorical covariate, we set those missing values as another level of the categorical covariate. All continuous covariates were complete.

Statistical analysis

First, trajectories of PAS were identified by latent class trajectory modeling (LCTM), a method to identify heterogeneous subgroups of population in terms of longitudinal exposures or outcomes[13]. By this method, we can determine the optimal number of trajectory classes with the lowest Bayesian Information Criterion (BIC)[14]. The mean of posterior probabilities in each class less than 70% indicated a poor fit[14].

Second, to describe characteristics of study population classified by trajectories of PAS, normally distributed continuous variables were presented as mean ± standard deviation; non-normally distributed continuous variables were presented as median (1st quartile, 3rd quartile); categorical variables were presented as n (%). Student’s t-tests were used to estimate differences in normally distributed continuous variables among trajectories of PAS; Kruskal-Wallis tests were used to estimate differences in non-normally distributed continuous variables among trajectories of PAS; Chi-square tests were used to estimate differences in categorical variables among trajectories of PAS.

Third, to investigate the association between trajectories of PAS and the slope of eGFRcr-cys, univariate and multivariate logistic regression models were used. Three models were constructed: crude model with no adjusted covariates; model 1 with adjusted age and gender; model 2 with adjusted age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid. In model 2, variates selected for adjustment were determined by backward stepwise selection[15].

Fourth, subgroup analysis was performed as a sensitivity test to evaluate the robustness of our results. We tested potential interactions between trajectories of PAS and the slope of eGFRcr-cys in different subgroups of participants. P for interaction was estimated by the likelihood ratio test. The variate “trajectories of PAS” was treated as a continuous variable. Subgroups were stratified by drinking in last year (no, yes), diabetes mellitus (no, yes), heart disease (no, yes), kidney disease (no, yes), baseline eGFRcr-cys (< 60 mL/min/1.73m2, ≥ 60 mL/min/1.73m2), and baseline uric acid (< 6 mg/dL, ≥ 6 mg/dL).

Fifth, serial mediation analysis was conducted to reveal significant mediating pathways from trajectories of PAS to the slope of eGFRcr-cys. A direct effect is the association between predictor variable (X) and outcome variable (Y) after controlling for all mediators, whereas an indirect effect is the association between X and Y via particular mediators[16]. Trajectories of PAS was entered as X. The slope of eGFRcr-cys was entered as Y. Baseline TyG index was entered as mediator 1(M1). Baseline eGFRcr-cys was entered as mediator 2(M2). A 1000 sample bootstrapping technique was used to estimate the indirect effects.

All analyses and graphics programs were performed with R-4.2.2 (https://www.R-project.org, The R Foundation). A p value less than 0.05 (two-sided) was considered statistically significant.

Trajectory modeling of physical activity scores

Trajectory modeling of PAS was performed by LCTM. As shown in Table 1, the nadir of BIC was 9604.18 when the number of classes was 5. In this trajectory model, the mean of posterior probabilities in each class was above 70%, which indicated a good fit. A total number of 407 participants aged 60 years or older were included. As shown in Fig. 2, 166 (40.79%) individuals were identified as class 1 with a constant low trajectory of PAS; 49 (12.04%) respondents were classified as class 2 with a decline trajectory starting from a moderate level of PAS; 82 (20.15%) participants were divided into class 3 with a slow-rise trajectory starting from a less than moderate level of PAS; 54 (13.27%) individuals were sorted as class 4 with a steep decline trajectory starting from a relative high level of PAS; 56 (13.76%) respondents were defined as class 5 with a slow-rise trajectory starting from a more than moderate level of PAS.

Table 1

Parameters of latent class trajectory models
Number of classes	BIC	Proportion of each class	Mean of posterior probabilities in each class
1	10058.29	100%	100%
2	9701.14	40.79%/59.21%	92.06%/98.44%
3	9653.25	39.07%/12.29%/48.65%	90.49%/94.53%/95.03%
4	9638.67	40.79%/9.34%/11.55% /38.33%	95.12%/82.28%/95.76%/93.99%
5	9604.18	40.79%/12.04%/20.15%/13.27%/ 13.76%	95.46%/92.85%/86.91%/94.94%/95.88%
6	9659.67	41.77%/10.32%/3.69%/13.02%/ 14.99%/16.22%	90.67%/83.28%/70.20%/87.55%/92.27%/95.59%
BIC, Bayesian Information Criterion.

Descriptive characteristics

Characteristics of study population classified by trajectories of PAS were compared and shown in Table 2. The number of participants in each class was 166, 49, 82, 54 and 56, respectively. There was no significant difference in proportion of current smokers, prevalence of diabetes mellitus, dyslipidemia, liver disease and kidney disease, gender, education level, marital status, drinking in last year, baseline eGFRcr-cys and TyG index levels, and follow-up duration among the five classes. Participants in class 1 tended to be older, have higher baseline uric acid and hsCRP levels, have a higher proportion of overweight individuals and urban residents, and have a higher prevalence of hypertension and heart disease than people in other classes. On the contrary, participants in class 1 were more likely to have lower slope of eGFRcr-cys levels when compared with those in other classes.

Table 2

Characteristics of study population classified by trajectories of physical activity score
	Trajectories of physical activity score
	Class 1	Class 2	Class 3	Class 4	Class 5	P-value
Number of participants	166	49	82	54	56
Age, years	67.62 ± 5.24	67.02 ± 5.21	66.12 ± 5.27	66.20 ± 4.67	65.52 ± 4.84	0.034*
Gender, n (%)						0.102
Male	76 (45.78)	28 (57.14)	41 (50.00)	32 (59.26)	36 (64.29)
Female	90 (54.22)	21 (42.86)	41 (50.00)	22 (40.74)	20 (35.71)
Area, n (%)						< 0.001*
Urban	78 (46.99)	15 (30.61)	19 (23.17)	13 (24.07)	8 (14.29)
Rural	88 (53.01)	34 (69.39)	63 (76.83)	41 (75.93)	48 (85.71)
Education level, n (%)						0.215
Less than lower secondary	152 (91.57)	49 (100.00)	77 (93.90)	53 (98.15)	54 (96.43)
Upper secondary & vocational training	10 (6.02)	0 (0.00)	5 (6.10)	1 (1.85)	2 (3.57)
Tertiary	4 (2.41)	0 (0.00)	0 (0.00)	0 (0.00)	0 (0.00)
Marital Status, n (%)						0.275
Married	129 (77.71)	41 (83.67)	67 (81.71)	47 (87.04)	50 (89.29)
Unmarried/divorced	37 (22.29)	8 (16.33)	15 (18.29)	7 (12.96)	6 (10.71)
Current smoking, n (%)						0.389
No	120 (72.29)	33 (67.35)	54 (65.85)	32 (59.26)	35 (62.50)
Yes	46 (27.71)	16 (32.65)	28 (34.15)	22 (40.74)	21 (37.50)
Drinking in last year, n (%)						0.094
No	123 (74.10)	28 (57.14)	54 (65.85)	32 (59.26)	35 (62.50)
Yes	46 (27.71)	16 (32.65)	28 (34.15)	22 (40.74)	21 (37.50)
BMI categorization, n (%)						0.005*
Underweight (less than 18.5)	8 (4.82)	4 (8.16)	7 (8.54)	4 (7.41)	1 (1.79)
Normal weight (from 18.5 to 23.9)	55 (33.13)	22 (44.90)	32 (39.02)	26 (48.15)	37 (66.07)
Overweight (from 23 to 24.9)	38 (22.89)	4 (8.16)	17 (20.73)	12 (22.22)	8 (14.29)
Obesity (greater than 25)	57 (34.34)	17 (34.69)	22 (26.83)	7 (12.96)	9 (16.07)
Hypertension, n (%)						0.015*
No	92 (55.42)	32 (65.31)	50 (60.98)	37 (68.52)	45 (80.36)
Yes	74 (44.58)	17 (34.69)	32 (39.02)	17 (31.48)	11 (19.64)
Diabetes mellitus, n (%)						0.447
No	150 (90.36)	48 (97.96)	77 (93.90)	51 (94.44)	52 (92.86)
Yes	15 (9.04)	0 (0.00)	4 (4.88)	3 (5.56)	4 (7.14)
Heart disease, n (%)						0.008*
No	120 (72.29)	44 (89.80)	64 (78.05)	46 (85.19)	54 (96.43)
Yes	45 (27.11)	5 (10.20)	18 (21.95)	8 (14.81)	2 (3.57)
Dyslipidemia, n (%)						0.070
No	139 (83.73)	43 (87.76)	71 (86.59)	51 (94.44)	55 (98.21)
Yes	26 (15.66)	5 (10.20)	9 (10.98)	2 (3.70)	1 (1.79)
Liver disease, n (%)						0.086
No	162 (97.59)	49 (100.00)	80 (97.56)	50 (92.59)	56 (100.00)
Yes	4 (2.41)	0 (0.00)	2 (2.44)	4 (7.41)	0 (0.00)
Kidney disease, n (%)						0.498
No	153 (92.17)	49 (100.00)	76 (92.68)	51 (94.44)	55 (98.21)
Yes	12 (7.23)	0 (0.00)	5 (6.10)	3 (5.56)	1 (1.79)
Baseline eGFRcr-cys, mL/min/1.73m2	79.75 ± 17.76	82.12 ± 16.50	84.00 ± 13.89	81.51 ± 15.78	86.43 ± 13.66	0.050
Slope of eGFRcr-cys, mL/min/1.73m2/year	0.92 (-1.16-3.25)	1.88 (-0.01-3.29)	1.83 (-0.22-3.65)	1.61 (-0.13-4.43)	3.25 (0.30–5.19)	0.002*
Baseline uric acid, mg/dL	4.68 ± 1.33	4.60 ± 0.97	4.06 ± 1.05	4.50 ± 1.12	4.51 ± 1.16	0.005*
Baseline hsCRP, mg/L	1.39 (0.75–2.69)	1.02 (0.56–2.93)	0.86 (0.61–1.86)	0.79 (0.56–1.31)	0.72 (0.50–1.78)	< 0.001*
Baseline TyG index	8.75 ± 0.64	8.73 ± 0.62	8.64 ± 0.61	8.57 ± 0.52	8.52 ± 0.58	0.096
Follow-up duration, years	3.98 ± 0.10	3.98 ± 0.14	4.00 ± 0.07	3.98 ± 0.08	4.00 ± 0.10	0.560
Normally distributed continuous variables were presented as mean ± standard deviation (Student’s t-test); non-normally distributed continuous variables were presented as median (1st quartile- 3rd quartile) (Kruskal-Wallis test); categorical variables were presented as n (%) (Chi-square test).
BMI, body mass index. eGFRcr-cys, estimated glomerular filtration rate based on serum creatinine and cystatin C. hsCRP, hypersensitive C-reactive protein. TyG, triglyceride-glucose.
*P -value < 0.05.

Association between trajectories of physical activity scores and the slope of eGFRcr-cys

To examine the association between trajectories of PAS and the slope of eGFRcr-cys, three logistic regression models were established (shown in Table 3). In the crude model without adjustment, compared with class 1, class 2 was associated with an increase of 0.68 [95% confidence interval (CI), -0.47-1.84] in the slope of eGFRcr-cys; class 3 was associated with an increase of 1.20 (95%CI, 0.24–2.16) in the slope of eGFRcr-cys; class 4 was associated with an increase of 1.71 (95%CI, 0.60–2.83) in the slope of eGFRcr-cys; class 5 was associated with an increase of 1.79 (95%CI, 0.69–2.88) in the slope of eGFRcr-cys. In the model 1 with adjustment of age and gender, compared with class 1, class 2 was associated with an increase of 0.69 (95%CI, -0.47-1.85) in the slope of eGFRcr-cys; class 3 was associated with an increase of 1.21 (95%CI, 0.24–2.18) in the slope of eGFRcr-cys; class 4 was associated with an increase of 1.73 (95%CI, 0.60–2.86) in the slope of eGFRcr-cys; class 5 was associated with an increase of 1.81 (95%CI, 0.69–2.94) in the slope of eGFRcr-cys. In the model 2 with adjustment of potential confounders (age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid), compared with class 1, class 2 was associated with an increase of 0.65 (95%CI, -0.39-1.68) in the slope of eGFRcr-cys; class 3 was associated with an increase of 1.24 (95%CI, 0.37–2.10) in the slope of eGFRcr-cys; class 4 was associated with an increase of 1.64 (95%CI, 0.65–2.63) in the slope of eGFRcr-cys; class 5 was associated with an increase of 2.37 (95%CI, 1.37–3.36) in the slope of eGFRcr-cys. In addition, a linear trend relationship between trajectories of PAS and the slope of eGFRcr-cys was found.

Table 3

Association between trajectories of physical activity score and the slope of eGFRcr-cys
	Model	Trajectories of physical activity score	Beta	95%CI	P-value	P for trend
Slope of eGFRcr-cys	Crude model ^a	Class 1	—	—	—	<0.001*
		Class 2	0.68	-0.47,1.84	0.248
		Class 3	1.20	0.24,2.16	0.015*
		Class 4	1.71	0.60,2.83	0.003*
		Class 5	1.79	0.69,2.88	0.002*
	Model 1 ^b	Class 1	—	—	—	<0.001*
		Class 2	0.69	-0.47,1.85	0.245
		Class 3	1.21	0.24,2.18	0.015*
		Class 4	1.73	0.60,2.86	0.003*
		Class 5	1.81	0.69,2.94	0.002*
	Model 2 ^c	Class 1	—	—	—	<0.001*
		Class 2	0.65	-0.39,1.68	0.223
		Class 3	1.24	0.37,2.10	0.005*
		Class 4	1.64	0.65,2.63	0.001*
		Class 5	2.37	1.37,3.36	< 0.001*
^a In the crude model, we did not adjust any covariates.
^b In model 1, we adjusted age and gender.
^c In model 2, we adjusted age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid.
eGFRcr-cys, estimated glomerular filtration rate based on serum creatinine and cystatin C. CI, confidence interval.
*P -value < 0.05.

Subgroup analysis

To assess the robustness of our results, subgroup analysis of association between trajectories of PAS and the slope of eGFRcr-cys was carried out as a sensitivity test. The variate “trajectories of PAS” was treated as a continuous variable. Beta values were obtained after adjusting age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid, except for the subgroup variable itself. As shown in Table 4, all beta values were positive, which indicated that results of subgroup analysis were in consistent with the primary results. No significant interaction effect between trajectories of PAS and subgroup variables for the slope of eGFRcr-cys was detected.

Table 4

Subgroup analysis of association between trajectories of physical activity score and the slope of eGFRcr-cys
	Subgroup	Trajectories of physical activity score ^a
	Subgroup	Number of participants	Beta ^b	95%CI	P-value	P for interaction
Slope of eGFRcr-cys	Drinking in last year
	No	271	0.63	0.36,0.90	< 0.001*	0.576
	Yes	135	0.50	0.12,0.87	0.010*	0.576
	Diabetes mellitus
	No	378	0.51	0.29,0.73	< 0.001*	0.063
	Yes	26	1.37	0.03,2.71	0.061	0.063
	Heart disease
	No	328	0.48	0.25,0.71	< 0.001*	0.096
	Yes	78	1.02	0.37,1.66	0.003*	0.096
	Kidney disease
	No	384	0.59	0.37,0.81	< 0.001*	0.834
	Yes	21	0.31	-0.96,1.57	0.645	0.834
	Baseline eGFRcr-cys, mL/min/1.73m²
	< 60.00	38	1.27	0.11,2.44	0.040*	0.723
	≥ 60.00	369	0.47	0.22,0.71	< 0.001*	0.723
	Baseline uric acid, mg/dL
	< 6.00	360	0.53	0.30,0.76	< 0.001*	0.548
	≥ 6.00	47	1.24	0.49,2.00	0.003*	0.548
^a The variate “trajectories of PAS” was treated as a continuous variable.
^b Beta values were obtained after adjusting age, drinking in last year, diabetes mellitus, heart disease, kidney disease, baseline eGFRcr-cys and baseline uric acid, except for the subgroup variable itself.
eGFRcr-cys, estimated glomerular filtration rate based on serum creatinine and cystatin C. CI, confidence interval.
*P -value < 0.05.

Serial mediation analysis

To reveal significant mediating pathways from trajectories of PAS to the slope of eGFRcr-cys, serial mediation analysis was conducted. Beta values were obtained after adjusting age, drinking in last year, diabetes mellitus, heart disease, kidney disease and baseline uric acid. As shown in Table 5, trajectories of PAS had a significant impact on the slope of eGFRcr-cys through the serial mediating path of “TyG index–baseline eGFRcr-cys” with a mediated proportion of 6.28% (ratio of indirect to total effect). These results indicated that TyG index and baseline eGFRcr-cys play serial mediating roles between trajectories of PAS and the slope of eGFRcr-cys.

Table 5

Serial mediating effect analysis
Path	Effect	^a Beta	^b Boost 95%CI	P-value
Total indirect effect	-0.022	-0.009	-0.144, 0.092	0.715
X→M1→Y	0.019	0.008	-0.022, 0.063	0.401
X→M2→Y	-0.074	-0.030	-0.191, 0.029	0.178
X→M1→M2→Y	0.034	0.014	0.009, 0.067	0.027*
Direct effect	0.562	0.224	0.366, 0.788	< 0.001*
Total effect	0.540	0.215	0.304, 0.800	< 0.001*
^a Beta values were obtained after adjusting age, drinking in last year, diabetes mellitus, heart disease, kidney disease and baseline uric acid.
^b 1000 bootstrap samples were used.
X, trajectories of physical activity scores. M1, baseline TyG index. M2, baseline eGFRcr-cys. Y, slope of eGFRcr-cys.
CI, confidence interval. TyG, triglyceride-glucose. eGFRcr-cys, estimated glomerular filtration rate based on serum creatinine and cystatin C.
*P -value < 0.05.

This is the first study to examine the effects of different PA trajectories on long-term changes in renal function. Our study implicates beneficial effects of specific PA trajectories on the slope of eGFRcr-cys among older Chinese population. The significance of this study lies in identifying a preferable PA trajectory (class 5) in terms of slowing the rate of renal function decline. Also, it denotes the worst PA trajectory (class 1). These findings may provide evidence for PA guidance to older people aged 60 years or more. The disclosure of a significant mediating pathway from trajectories of PAS to the slope of eGFRcr-cys further deepens our understanding on how PA trajectories are associated with longitudinal changes in renal function.

Our results showed that certain trajectories of PAS (classes 3 to 5) were associated with an increased slope of eGFRcr-cys when compared with the constant low trajectory of PAS (class 1). The class 1 trajectory represents an inactive lifestyle with persistently low levels of PA. Numerous studies have demonstrated harms of physical inactivity (PI). For instance, Katzmarzyk et al.[17] provided evidence that PI was the major risk factor for non-communicable diseases including stroke, hypertension, coronary heart disease, type 2 diabetes mellitus, dementia, depression and cancers. Besides, Kari et al.[18] found that PI increased average individual-level productivity costs. These findings along with our results indicate detrimental roles of PI in human health.

We also found a linear trend relationship between trajectories of PAS and the slope of eGFRcr-cys, indicating that the class 5 trajectory is preferable in terms of slowing the rate of renal function decline. This trajectory reflects an active lifestyle with more than moderate levels of PA. Many studies have reported that PA has beneficial impacts on renal function. For example, a RCT carried out by Shlipak et al.[7] showed that structured, moderate PA mitigated the rate of decrease in eGFRcys among sedentary older adults. The Mendelian randomization study performed by Park et al.[19] demonstrated that moderate-to-vigorous PA causally associated with a higher eGFR in the general population. These publications along with our findings support favorable effects of PA on renal function.

The results of serial mediation analysis denoted that TyG index and baseline eGFRcr-cys play serial mediating roles between trajectories of PAS and the slope of eGFRcr-cys. This mediating pathway has not been reported yet. As an insulin resistance indicator, TyG index was reported to be a risk factor for rapid decline of renal function and new-onset kidney disease[20]. However, the impact of PA on TyG index has not been well studied. A relatively small mediated proportion suggested that PA trajectories are associated with the slope of eGFRcr-cys mainly by direct effect.

Strengths of this study include a relatively long follow-up time with a mean duration of 3.99 years and the prospective study design. However, there are several limitations to be acknowledged. First, the study population were older Chinese aged 60 years or more with a mean age of 66.77 years from the CHARLS. Therefore, the results may not be generalizable to other demographics. Second, since answers to PA related questions were self-reported, potential information bias could not be eliminated. Third, selection bias may exist in that participants were excluded when lack of necessary information. Fourth, residual confounding may exist owing to unmeasured variables.

BIC, Bayesian Information Criterion

BMI, body mass index

CHARLS, China Health and Retirement Longitudinal Study

CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration

eGFR, estimated glomerular filtration rate

eGFRcys, estimated glomerular filtration rate based on cystatin C

eGFRcr-cys, estimated glomerular filtration rate based on serum creatinine and cystatin C

hsCRP, hypersensitive C-reactive protein

LCTM, latent class trajectory modeling

PA, Physical activity

PAS, physical activity score

PI, physical inactivity

RCT, randomized controlled trial

TyG, triglyceride-glucose

Ethics approval and consent to participate

The CHARLS has received ethical approval from the institutional review board (IRB) at Peking University. The IRB approval number for the main household survey and biomarker collection was IRB00001052–11015 and IRB00001052-11014, respectively. All participants gave written informed consent before they were interviewed.

Consent for publication

Not applicable.

Availability of data and materials

The data that support the findings of this study are available from the CHARLS (https://charls.charlsdata.com/pages/data/111/zh-cn.html) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the CHARLS.

Competing interests

The authors declare that they have no competing interests.

Funding

This study was supported by the National Natural Science Foundation of China (grant number 82200813 and 82000662) and the Scientifc Research Launch Project for new employees of the Second Xiangya Hospital of Central South University. The funders had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Authors' contributions

MZ designed the study. MZ extracted data from the CHARLS. DY, LZ, XC and MZ analyzed and interpreted data. DY and MZ drafted the manuscript. All authors revised the manuscript. All authors read and approved the final manuscript.

Acknowledgments

We thank all staffs and participants of the CHARLS.

Lavie CJ, Ozemek C, Carbone S, Katzmarzyk PT, Blair SN. Sedentary Behavior, Exercise, and Cardiovascular Health. Circ Res. 2019;124(5):799–815.
Smith PJ, Merwin RM. The Role of Exercise in Management of Mental Health Disorders: An Integrative Review. Annu Rev Med. 2021;72:45–62.
Hojman P, Gehl J, Christensen JF, Pedersen BK. Molecular Mechanisms Linking Exercise to Cancer Prevention and Treatment. Cell Metab. 2018;27(1):10–21.
Sobol NA, Hoffmann K, Frederiksen KS, Vogel A, Vestergaard K, Braendgaard H, et al. Effect of aerobic exercise on physical performance in patients with Alzheimer's disease. Alzheimers Dement. 2016;12(12):1207–15.
Zhou Y, Cartmel B, Gottlieb L, Ercolano EA, Li F, Harrigan M et al. Randomized Trial of Exercise on Quality of Life in Women With Ovarian Cancer: Women's Activity and Lifestyle Study in Connecticut (WALC). J Natl Cancer Inst. 2017;109(12).
Parsons TJ, Sartini C, Ash S, Lennon LT, Wannamethee SG, Lee IM, et al. Objectively measured physical activity and kidney function in older men; a cross-sectional population-based study. Age Ageing. 2017;46(6):1010–14.
Shlipak MG, Sheshadri A, Hsu FC, Chen SH, Jotwani V, Tranah G, et al. Effect of Structured, Moderate Exercise on Kidney Function Decline in Sedentary Older Adults: An Ancillary Analysis of the LIFE Study Randomized Clinical Trial. JAMA Intern Med. 2022;182(6):650–59.
Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8.
Craig CL, Marshall AL, Sjostrom M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35(8):1381–95.
Bai A, Tao L, Huang J, Tao J, Liu J. Effects of physical activity on cognitive function among patients with diabetes in China: a nationally longitudinal study. BMC Public Health. 2021;21(1):481.
Inker LA, Eneanya ND, Coresh J, Tighiouart H, Wang D, Sang Y, et al. New Creatinine- and Cystatin C-Based Equations to Estimate GFR without Race. N Engl J Med. 2021;385(19):1737–49.
Huang R, Wang Z, Chen J, Bao X, Xu N, Guo S, et al. Prognostic value of triglyceride glucose (TyG) index in patients with acute decompensated heart failure. Cardiovasc Diabetol. 2022;21(1):88.
Watson C, Geifman N, Renehan AG. Latent class trajectory modelling: impact of changes in model specification. Am J Transl Res. 2022;14(10):7593–606.
Lennon H, Kelly S, Sperrin M, Buchan I, Cross AJ, Leitzmann M, et al. Framework to construct and interpret latent class trajectory modelling. BMJ open. 2018;8(7):e020683.
Sanchez-Pinto LN, Venable LR, Fahrenbach J, Churpek MM. Comparison of variable selection methods for clinical predictive modeling. Int J Med Inform. 2018;116:10–7.
Chen T, Luo H, Feng Q, Li G. Effect of Technology Acceptance on Blended Learning Satisfaction: The Serial Mediation of Emotional Experience, Social Belonging, and Higher-Order Thinking. Int J Environ Res Public Health. 2023;20(5).
Katzmarzyk PT, Friedenreich C, Shiroma EJ, Lee IM. Physical inactivity and non-communicable disease burden in low-income, middle-income and high-income countries. Br J Sports Med. 2022;56(2):101–06.
Kari JT, Nerg I, Huikari S, Leinonen AM, Nurkkala M, Farrahi V, et al. The Individual-Level Productivity Costs of Physical Inactivity. Med Sci Sports Exerc. 2023;55(2):255–63.
Park S, Lee S, Kim Y, Lee Y, Kang MW, Kim K, et al. Causal effects of physical activity or sedentary behaviors on kidney function: an integrated population-scale observational analysis and Mendelian randomization study. Nephrology, dialysis, transplantation: official publication of the European Dialysis and Transplant Association -. Eur Ren Association. 2022;37(6):1059–68.
Xiong J, Zheng X, Luo J, Luo X, Zhang Y, Xu L, et al. A follow-up study to explore factors associated with rapid kidney function decline and new-onset kidney disease among Chinese elderly population. Geriatr Gerontol Int. 2022;22(11):968–75.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Association of Physical Activity Trajectories with Longitudinal Changes in Renal Function among Older Chinese Adults: A Prospective Cohort Study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Background

Methods

Aim, design and setting of the study

Study population

Physical activity assessment

Evaluation of changes in renal function

Covariates

Statistical analysis

Results

Trajectory modeling of physical activity scores

Descriptive characteristics

Association between trajectories of physical activity scores and the slope of eGFRcr-cys

Subgroup analysis

Serial mediation analysis

Discussion

Conclusions

Abbreviations

Declarations

Ethics approval and consent to participate

Consent for publication

Availability of data and materials

Competing interests

Funding

Authors' contributions

Acknowledgments

References

Additional Declarations

Status:

Version 1