Article Text

Download PDFPDF
Habitual physical activity, renal function and chronic kidney disease: a cohort study of nearly 200 000 adults
  1. Cui Guo1,
  2. Tony Tam2,
  3. Yacong Bo1,3,
  4. Ly-yun Chang4,5,
  5. Xiang Qian Lao1,6,
  6. G Neil Thomas7
  1. 1 The Jockey Club School of Public Health and Primary Care, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong
  2. 2 Department of Sociology, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  3. 3 School of Public Health, Zhengzhou University, Zhengzhou, Henan, China
  4. 4 Institute of Sociology, Academia Sinica, Taipei, Taiwan
  5. 5 MJ Health Research Foundation, MJ Group, Taipei, Taiwan
  6. 6 Shenzhen Research Institute of the Chinese University of Hong Kong, Shenzhen, Guangdong, China
  7. 7 Public Health, Epidemiology and Biostatistics, University of Birmingham, Birmingham, UK
  1. Correspondence to Professor Xiang Qian Lao, The Jockey Club School of Public Health and Primary Care, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong; xqlao{at}cuhk.edu.hk

Abstract

Background There is limited information on the association between habitual physical activity (PA) and renal function.

Objective To report the longitudinal association between self-reported habitual PA and measures of renal function in a large cohort in Taiwan.

Methods A total of 199 421 participants (aged ≥20 years) were selected from a Taiwan cohort between 1996 and 2014. All participants underwent at least two standardised medical examinations between 1996 and 2014. Self-administrated questionnaires were used to collect information on habitual PA. We used a generalised linear mixed model to investigate the associations between habitual PA and yearly change in estimated glomerular filtration rate (eGFR). The Cox proportional hazard regression model was used to investigate the associations between habitual PA and incident chronic kidney disease (CKD).

Results Participants had a median follow-up duration of 4.2 years (0.2–18.9). The yearly mean (±SD) decrease in eGFR in participants with baseline very low-PA, low-PA, moderate-PA and high-PA was 0.46±1.01, 0.36±0.97, 0.30±0.94 and 0.27±0.91 mL/min/1.73 m2, respectively. Relative to the participants with very low-PA, the coefficients of yearly eGFR change were −43.93 (95% CI −79.18 to −8.68), 35.20 (95% CI −2.56 to 72.96) and 53.56 (95% CI 10.42 to 96.70) µL/min/1.73 m2, respectively, for the participants with low-PA, moderate-PA and high-PA, after controlling for a wide range of covariates. Relative to the very low-PA participants, those who had low-PA, moderate-PA and high-habitual PA had HRs of 0.93 (95% CI 0.88 to 0.98), 0.94 (95% CI 0.89 to 0.99) and 0.91 (95% CI 0.85 to 0.96) to develop CKD, respectively, after controlling for the covariates.

Conclusions A higher level of habitual PA is associated with a smaller decrease in the level of eGFR and a lower risk of developing CKD.

  • chronic
  • exercises
  • kidney
  • physical activity

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Chronic kidney disease (CKD) was the 12th leading cause of death and resulted in more than 2.5 million (2.17%) deaths worldwide in 2016.1 The global prevalence of CKD is estimated at around 13.4%.2 People with kidney disease often present with a range of comorbidities and have an increased risk of myocardial infarction and death.3–5The awareness of CKD is low partly due to its early asymptomatic characteristics.5 Many risk factors may contribute to CKD including unhealthy lifestyles, toxic environmental factors and cardiovascular disease (CVD).6–8

To date, there remains limited information describing habitual physical activity (PA), renal function and the onset of CKD. In epidemiological studies, participants who exercise regularly have a better estimated glomerular filtration rate (eGFR) than those with a sedentary behaviour.9 10 Three cross-sectional studies suggested an association between higher PA and a lower prevalence of CKD in men6 and in the general population.11 12 The few published cohort studies generally focused on select populations with relatively small sample sizes.13–23 Therefore, we conducted a large cohort study to investigate the effects of habitual PA on changes in eGFR and the development of CKD in 199 421 adults in Taiwan.

Methods

Study design and participants

The participants in this study were selected from an ongoing large longitudinal cohort in Taiwan. The cohort’s details have been described in our previous publications.8 24 Briefly, since 1996, MJ Health Management Institution has recruited more than half a million Taiwanese residents for paid membership in a standard medical screening programme that provides a series of medical examinations with standard protocols including anthropometric measurements and physical examinations such as spirometry and blood and urinary tests. We refer to this as the MJ cohort.25 Participants' demographic and socioeconomic information, including lifestyle and medical history, were collected from a standard self-administered questionnaire.25 Each participant provided written informed consent before participation. Ethical approval for this study was given by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee.

This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient-relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

The details of participant selection are shown in online supplementary eFigure 1. The cohort included 567 904 participants at least 20 years of age with eGFR measurements from 1996 to 2014. We excluded 99 247 participants due to incomplete information (31 326 for data on habitual PA and 67 921 for data on covariates). Another 837 participants were excluded: 49 with an eGFR of 200 mL/min/1.73 m2 or greater, or less than 2 mL/min/1.73 m2 (because the values suggest that the measurements were likely incorrect due to occasional technical errors)8 and 788 with a urinary protein level of at least 2.0 g/L (because urinary protein is also an important syndrome for CKD).

Supplemental material

To investigate the association between habitual PA and eGFR change, we included a total of 199 421 (42.6%) participants who had at least two medical visits for data analysis (ie, 268 399 participants were excluded because they had undergone only one medical examination). We further excluded 9347 participants with CKD at baseline (defined as an eGFR of less than 60 mL/min/1.73 m2 or self-report of a physician diagnosis of CKD) to investigate the association between habitual PA and the development of CKD. Compared with the 277 746 excluded participants, the 190 074 included non-CKD participants were of younger age (mean: 38.5 vs 41.4 years), similar percentage of males (51.6% vs 48.7%), higher education level (high school or lower: 38.1% vs 47.2%), similar physical labour at work (sedentary: 88.7% vs 86.6%) and similar volume of metabolic equivalent (MET) (mean: 6.3 vs 6.7 MET-hours (MET-h)).

Habitual PA

Details of the information collected for habitual PA have been described in previous studies.26–28 In brief, the MJ cohort used a self-administered questionnaire to collect information on weekly PA during the month before each medical examination. Activity was classified into four types according to intensity by asking the question “Which types of physical activities did you usually take in the previous month?” with several examples given under each type: light (eg, walking), moderate (eg, brisk walking), medium (eg, jogging) or high-vigorous (eg, running). Each type was assigned a corresponding MET (1 MET=1 kilocalorie per hour per kilogram of bodyweight); light, moderate, medium-vigorous and high-vigorous intensity were assigned METs of 2.5, 4.5, 6.5 and 8.5, respectively.29 The weekly total time spent on the PA was obtained by asking the question “How many hours did you spend on the PA weekly in previous month?” before 2009. Since 2009, we have used the two questions to obtain the weekly total time by asking “How often did you usually do the PA weekly in the previous month?” and “How many hours did you spend on the PA each time?”.

The weekly total time spent on the PA was calculated by multiplying the hours and frequency. The participants who had more than one activity intensity type were assigned a MET value that was weighted by the weekly total time spent in each type. We further calculated the habitual PA (MET-h) using the product of intensity (MET) and weekly total time (hour). The habitual PA was then classified as very low-PA (<3.75 MET-h), low-PA (3.75–7.49 MET-h), moderate-PA (7.50–16.49 MET-h) or high-PA (≥16.50 MET-h) roughly according to the current PA guidelines for Americans.30

Outcome measurements

The yearly change in eGFR and incident CKD were the outcomes of this study. An overnight fasting blood sample was collected from each participant during each medical examination for measurement of the serum creatinine level using an automatic biochemical analyser (Hitachi 7150 (Tokyo) before 2005 or Toshiba C8000 (Tokyo) since 2005). The eGFR was estimated using the equation given in the Modification of Diet in Renal Disease study31:

Embedded Image

The yearly change in eGFR (μL/min/1.73 m2 per year) was calculated for each participant as the difference in eGFR divided by the duration (years) between the value at follow-up and the corresponding value at the preceding visit.

Regarding incident CKD, we followed up all the 190 074 participants who were non-CKD at baseline assessment (ie, the first visit). The incident CKD was identified by medical assessment (defined as an eGFR of less than 60 mL/min/1.73 m2 or reported a physician diagnosis of CKD) in the subsequent visits.32 Because follow-up frequency and interval of medical visits among the participants varied, we used the first occurrence of CKD in the follow-up visits as the end point for those participants who developed CKD. For those who did not develop CKD in the study period, we used the last visit as the censoring date.

Covariates

The technical report by MJ Health Research Foundation and our previous publications described the measurement and quality control in detail.24 33 In brief, height and weight were measured with light clothing and without shoes. Seated blood pressure was measured with an auto-sphygmomanometer (Citizen CH-5000). The plasma glucose level was measured with an automatic biochemical analyser (Hitachi 7150 (Tokyo) before 2005 or Toshiba C8000 (Tokyo) since 2005). We further measured the urinary protein level with a semi-automated computer-assisted urinalysis system (Roche Miditron or Roche Cobas U411).

Covariates were mainly selected based on previous literature.8 26 Data on the following covariates were obtained at each medical visit: age (years); sex; education (less than high school (<10 years), high school (10–12 years), college or university (13–16 years) or postgraduate (>16 years)); smoking status (never, former or current); alcohol consumption (seldom (less than once per week), occasional (one to two times per week) or regular (at least three times per week)); physical labour at work (mostly sedentary (eg, clerk), sedentary with occasional walking (eg, seamstress), mostly standing or walking (eg, retail salesmen) or hard labour (eg, porter)); vegetable and fruit intake (seldom (<1 serving/day), moderate (1–2 servings/day) or frequent (>2 servings per day)); body mass index (BMI; calculated as weight (in kilograms) divided by height (in metres) squared); hypertension (defined as systolic blood pressure of at least 140 mmHg, diastolic blood pressure of at least 90 mmHg or self-reported hypertension); dyslipidaemia (defined as a total cholesterol level of at least 240 mg/dL (13.3 mmol/L), triglyceride level of at least 200 mg/dL (11.1 mmol/L) or high-density lipoprotein cholesterol level of lower than 40 mg/dL (2.2 mmol/L)); self-reported CVD (yes or no); diabetes (defined as a fasting blood glucose level of at least 126 mg/dL (7 mmol/L) or self-report of a physician diagnosis of diabetes); self-report of a physician diagnosis of cancer (yes or no); baseline eGFR (mL/min/1.73 m2); urinary protein (negative or normal (<0.1 g/L), trace (0.1~0.2 g/L), 1 plus (0.2~1.0 g/L) and 2 plus (1.0~2.0 g/L)); calendar season34 (spring, March to May; summer, June to August; autumn, September to November; winter, December to February); and calendar year.

Statistical analysis

To investigate the effects of habitual PA on the change in eGFR we used a generalised linear mixed model (GLMM) for data analysis, which is a typical statistical approach used in a longitudinal study to deal with repeated measurements. Because eGFR followed normal (or Gaussian) distribution in this study, we applied an identity link function in the model.

The variable 'habitual PA' and all covariates (except for sex and baseline eGFR) were treated as time-dependent variables. We present data from four hierarchical models. Model 1 was adjusted for demographic factors (age, sex and education) and baseline eGFR. Model 2 was further adjusted for lifestyle factors (physical labour at work, smoking status, alcohol consumption and vegetable and fruit intake), calendar season and year. Model 3 was further adjusted for potential mediating factors BMI, hypertension, diabetes, dyslipidaemia and self-report of a physician diagnosis of CVD and cancer. Model 4 was further adjusted for the baseline urinary protein level. A trend test was conducted by treating the habitual PA category as an ordinary variable. Coefficients with a 95% CI were estimated.

To examine the effects of habitual PA on the development of CKD we used a Cox proportional hazard regression model for data analysis. Hazard ratios (HRs) with 95% CIs were calculated. We examined the proportional hypothesis by plotting the Kaplan-Meier survival curves for categorical covariates and the Schoenfeld residuals for continuous covariates, respectively. All covariates agreed with the proportional hypothesis except for education and physical labour at work. We therefore further performed subgroup analyses stratified by education (<high school vs high school or above) and physical labour at work (mostly sedentary or sedentary with occasional walking vs mostly standing or walking or hard labour).

We also conducted three sensitivity analyses. (1) We excluded participants with fewer than 2 years of follow-up because the development of CKD is a chronic process. (2) We excluded participants enrolled after 2005 to avoid potential bias by changing the measurement instrument. (3) We excluded incident CKD identified by a self-report of a physician diagnosis of CKD to avoid potential bias from differences in the diagnostic criteria.

The software package R 3.4.0 (R Core Team, Vienna, Austria) was used to conduct all statistical analyses. The effects of habitual PA were regarded as statistically significant at a two-tailed 0.05 level.

Results

Table 1 shows the characteristics of the participants included in this study; 199 421 participants with 557 379 observations were included to assess the associations between habitual PA and eGFR change. Overall, the eGFR level decreased by 0.39 mL/min/1.73 m2 per year. The median number of medical visits was 4.0 (range 2–28), at a median visit interval of 1.4 (IQR 1.1–2.3) years. The follow-up rate is 42.6%. A total of 190 074 participants without CKD at baseline were included in the analysis of associations between habitual PA and incident CKD. We identified 10 596 incident cases of CKD (5.6%) in this study. The duration of follow-up ranged from 0.2 to 18.9 years (median (IQR) 4.2 (2.1–8.2) years). Slightly more than half of the participants were male, and they were generally well-educated. More than 60% of the participants were mostly sedentary during work, and 50.2% of the participants had a very low habitual PA level.

Table 1

Particpant characteristics

The eGFR level decreased over the follow-up period at various levels of baseline habitual PA (figure 1). The yearly adjusted mean decreases in eGFR in the participants with baseline very low-PA, low-PA, moderate-PA and high-PA were 0.46±1.01, 0.36±0.97, 0.30±0.94 and 0.27±0.91 mL/min/1.73 m2, respectively.

Figure 1

Adjusted mean of yearly change of estimated glomerular filtration rate (eGFR) over time for each participant classified by baseline habitual physical activity (PA). Columns A–D represent the four categories of baseline habitual PA – very low-PA, low-PA, moderate-PA and high-PA – respectively. Circles indicate the adjusted mean of the eGFR change for each participant. The mean was calculated with the estimated eGFR change based on Model 4, adjusted for age, sex, education, baseline eGFR, physical labour at work, smoking status, alcohol consumption, vegetable and fruit intake, calendar season, calendar year, body mass index, hypertension, diabetes, urinary protein level, dyslipidaemia, self-report of a physician diagnosis of cardiovascular disease and self-report of a physician diagnosis of cancer. The yearly adjusted means of decreased eGFR for the categories of baseline very low-PA, low-PA, moderate-PA and high-PA were 0.46±1.01, 0.36±0.97, 0.30±0.94 and 0.27±0.91 µL/min/1.73 m2, respectively.

Table 2 shows the association between habitual PA and the change in eGFR. The results were generally consistent across the four models. Relative to the participants with very low-PA, the coefficients of yearly eGFR change were −43.93 (95% CI −79.18 to −8.68), 35.20 (95% CI −2.56 to 72.96) and 53.56 (95% CI 10.42 to 96.70) µL/min/1.73 m2, respectively, for the participants with low-PA, moderate-PA and high-PA, after controlling for a wide range of covariates. A significant trend test was also observed, and every category increase in habitual PA was associated with a coefficient of 18.07 (95% CI 5.00 to 31.14) µL/min/1.73 m2/year.

Table 2

Associations of yearly change in estimated glomerular filtration rate with habitual physical activity in Taiwanese adults

Table 3 shows the health effects of habitual PA on incident CKD. Similar to eGFR, the results were generally consistent across the four models. Relative to the very low-PA participants, those who had low-PA, moderate-PA and high-habitual PA had HRs of 0.93 (95% CI 0.88 to 0.98), 0.94 (95% CI 0.89 to 0.99) and 0.91 (95% CI 0.85 to 0.96), respectively, after controlling for the covariates. The trend test indicated that the risk of incident CKD generally decreased as the habitual PA increased (HR 0.97, 95% CI 0.95 to 0.99). Table 4 shows the results of subgroup analyses. The beneficial effects of habitual PA were higher in the participants who had higher education or sedentary physical labour. Table 5 presents the results of the sensitivity analyses, which generally yielded similar results.

Table 3

Associations of incident chronic kidney disease with habitual physical activity in Taiwanese adults

Table 4

Associations of incident chronic kidney disease with habitual physical activity in subgroup analyses in Taiwanese adults

Table 5

Sensitivity analyses for associations of habitual physical activity with yearly change in estimated glomerular filtration rate and incident chronic kidney disease

Discussion

In this established large Asian adult cohort, a higher habitual PA was associated with a smaller decrease in eGFR over time and with a decreased risk of incident CKD.

A study from the Netherlands found no statistically significant effects of PA or a 5-year change in PA on eGFR in adults.15 White et al 13 reported that physically inactive participants had a higher prevalence of a low eGFR in Australian adults, although the association was not statistically significant. Our results differ from those and extend the findings of five other studies,18 19 21–23 that reported significant benefits of PA in mitigating age-related reduction in eGFR. These five studies focused on specific groups of participants, such as men,18 19 women,23 older people (≥65 years)22 or patients with CKD.21

We observed that habitual PA was associated with a lower risk of the onset of CKD and this extends findings in previous studies.16–20 35 36 Among those, two studies were conducted in patients with type 2 diabetes,16 20 four in men only,17–19 one in middle-aged subjects35 and one in older subjects.36 In contrast, some cohort studies did not find significant associations between PA and CKD development.14 37 38 However, two studies reported that active male participants39 and female divers40 had a higher risk of CKD than inactive males and female non-divers, respectively. Many factors might contribute to the inconsistency, including study design, information bias and the targeted population. We believe our longitudinal study design based on a general population with comprehensive PA information collected may provide more robust evidence.

Potential mechanisms

Several pathways have been hypothesised as to the potential biological mechanism of the association between PA and renal health. First, physical activity can improve cardiovascular endothelial function and improve insulin sensitivity.41 A similar effect on the kidney vasculature would improve renal function. Second, low habitual PA contributes to insulin resistance, which may directly damage renal vasculature (including angiogenesis, mesangial expansion, glomerular hyperfiltration) and also a detrimental effect on the kidney by intensifying insulin-responsive signalling.42 A reduction in adiposity or in adipocytokines could explain the beneficial effects of exercise on renal health. Visceral adipocytes can increase angiotensinogen and damage kidney endothelium.43

Limitations of the study

This study has some limitations. First, habitual PA was assessed with a self-administered questionnaire. We used the available questionnaire data, whose content validity and reliability have been reported previously.4 Second, the diagnosis of CKD was based mainly on a single measurement of eGFR lower than 60 mL/min/1.73 m2. In a clinical setting, the diagnosis of CKD should be based on two measurements taken at least 90 days apart. Our method may have included some acute CKD cases and diluted the associations. Third, Mansournia et al 44 showed that use of a standard statistical method such as GLMM may produce biassed estimates due to prior exposure. However, the correlations of the confounders with habitual PA and previous PA in our study were weak (ρspearman ranged from −0.09 to 0.24 for habitual PA and ρspearman ranged from −0.10 to 0.24 for previous PA, respectively) in this study. Thus, prior exposure should not affect our conclusions. Finally, the follow-up rate of this cohort was 42.6%. Thus, we only include data from approximately 35% of adults in the cohort to investigate the association of eGFR with PA and data from approximately 33% of adults in the cohort to investigate the association of CKD with PA, respectively. Potential selection bias cannot be ruled out and an alternative approach such as the inverse probability weighting technique45 46 would have been used to correct such bias, if any. However, we have adjusted for a wide range of covariates and the results were consistent across the four models (tables 2 and 3). We conducted a series of sensitivity and subgroup analyses (tables 4 and 5) which show that the associations are robust. Both adjusting for covariates associated with selection and sensitivity analysis are effective methods to control for selection bias.46

Strengths of the study

This study has several strengths. First, we report a prospective association between habitual PA and renal function. Second, the large sample size resulted in more stable and precise estimates and provided sufficient power to conduct a series of sensitivity analyses. Third, the potential effects of a wide range of covariates were considered. Fourth, the information collected for PA was comprehensive and included intensity, frequency and duration. We also collected information on physical labour at work and considered it in our data analysis. Finally, data were routinely collected from a standardised medical screening programme, which may minimise the likelihood of investigator bias.

In conclusion, we found that habitual PA was associated with an attenuated decrease in eGFR and development of CKD in this large Asian adult population. This is an important finding given that PA is a behaviour that can be changed and because of the accelerating global burden of kidney disease.

What are the new findings?

  • We found that a higher habitual physical activity was associated with a lower annual decrease in estimated glomerular filtration rate (eGFR) in Chinese adults based on a large longitudinal cohort. A higher habitual physical activity was associated with a lower risk of chronic kidney disease in the same population.

How might it impact on clinical practice in the future?

  • This study extends proof of the health benefits of physical activity to include mitigating age-related deterioration in kidney function.

Acknowledgments

The authors appreciate the MJ Health Research Foundation for authorising the use of its health data (authorisation code: MJHR2015002A). Any interpretation or conclusion related to this article does not represent the views of MJ Health Research Foundation. Cui Guo and Yacong Bo are supported by PhD Studentships of the Chinese University of Hong Kong. The authors are grateful to the anonymous reviewers and the journal editors for their valuable comments.

References

Footnotes

  • Contributors XQL and GNT conceived and designed the study. LC, TT and XQL acquired the data. CG and YB searched the literature. CG cleaned and analysed the data. CG, TT, XQL and GNT interpreted the results. CG and XQL drafted the manuscript. XQL, CG, GNT, YB, TT and LC revised the manuscript. All authors contributed to the content and critical revision of the manuscript and approved the final version. XQL obtained the funding. XQL is the guarantor of this study.

  • Funding This work was supported in part by the Environmental Health Research Fund of the Chinese University of Hong Kong (7104946).

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Approved by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request. The data are deidentified participant data, which are available from MJ Health Institute upon reasonable request.