Interventions to increase physical activity level in patients with whole spectrum chronic kidney disease: a systematic review and meta-analysis

Abstract

Background: Little is known about effective interventions to increase physical activity levels in this population. This systematic review and meta-analysis evaluated the effectiveness of different interventions for physical activity levels in whole-spectrum CKD patients.

Methods: In this systematic review and meta-analysis, PubMed, Embase, and Web of Science were searched from inception to December 2022, with an update in March 2023. Studies of clinical interventions based on assessing physical activity outcomes (e.g., daily steps, time spent doing physical activity) in patients with whole-spectrum CKD were included. The pooled effect size was calculated using a restricted maximum likelihood method and reported as a standardized mean difference (SMD) with 95% confidence interval (95% CI).

Results: The systematic review included 2,156 participants (59.9 ± 8.7 years) from 35 studies. Interventions aimed at modifying physical activity were associated with significantly higher physical activity levels in patients with CKD (SMD = 0.22; 95% CI: 0.08 to 0.36; I² = 55%). Exercise-based interventions and real-time step feedback increased physical activity by 0.26 (95% CI: 0.07 to 0.45; I² = 59%) and 0.36 (95% CI: 0.12 to 0.60; I² = 0%) standard deviations, respectively. Effect sizes did not vary by disease stage or study duration; however, there was evidence of small study or publication bias for the primary analysis.

Conclusion: In this systematic review and meta-analysis, intervention strategies aimed at modifying physical activity were associated with significantly increased physical activity levels in patients with whole-spectrum CKD.

1. Introduction

Chronic kidney disease (CKD) is a common, chronic disease with a persistent but irreversible decline in kidney function that relies on renal replacement therapy (RRT) to sustain life at the end stage [1]. The prevalence of CKD was estimated at 10% to 12% of the global population [2]. A recent survey showed that the incidence of CKD in the Chinese population is about 8.2% [3]. It is evident that kidney disease has evolved into a global health problem [4].

Physical inactivity is considered an important modifiable risk factor for frailty progression, poor quality of life, increased risk of worsening disease severity, and increased mortality risk in CKD patients, and it is observed across the disease spectrum [5–7]. Our previous systematic review and meta-analysis found that daily steps in patients with CKD were below the recommended values in the early stage of the disease [8]. CKD patients also spent significantly less time participating in moderate to vigorous physical activity than the general population [9,10]. Qualitative studies suggested that fear of participating in physical activity that triggers fear of injury or aggravates their condition may partially explain these findings [11], as individuals with increased disease severity are more likely to believe that exercise should be avoided [12]. However, it has also been shown that patients cite participation in physical activity as one of the most critical outcomes achieved after treatment for CKD-related symptoms [13,14].

Increasing awareness of the challenges posed by low physical activity in patients with CKD at various stages of the disease course has highlighted the need for interventions to increase physical activity participation in this population [13,15,16]. The proliferation of evidence demonstrating the promising role of physical activity as a modifiable lifestyle in complementing existing renal rehabilitation [17,18], coupled with the increasing number of wearable monitoring devices that can objectively measure participation [19], has led to a dramatic increase in the number and variety of studies aimed at increasing physical activity levels in CKD patients. In this context, it is urgent to address effective ‘exit strategies’ to free CKD patients from physical inactivity [20].

The present systematic review and meta-analysis aimed to assess the effectiveness of different interventions for physical activity outcomes in the whole-spectrum CKD.

2. Methods

2.1. Protocol and registration

This review was guided by the Cochrane Handbook for Systematic Reviews of Interventions [21] and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement (PRISMA 2020) [22] (Supporting Information Table S1). The protocol was registered on the PROSPERO website (CRD42022385441).

The current study methodology is similar to the previously described protocol with a few modifications. First, we added meta-analysis to quantify mean effect estimates for similar interventions. Second, the Methodological Index for Non-Randomized Studies (MINORS) scale previously used to assess non-RCTs was changed to the Risk Of Bias In Non-Randomized Studies-of Interventions (ROBIN-I). Third, subgroup analyses were performed for non-dialysis patients and patients with RRT-dependent CKD rather than different disease spectrums; and subgroups were revised for the study duration.

2.2. Data sources and search strategy

PubMed, Embase, and Web of Science databases were searched for relevant English language publications from inception to December 2022. We updated the search on 1 March 2023 to capture newly published studies. The search strategy covered all applicable terms related to ‘physical activity’, ‘daily steps’, ‘chronic kidney disease’, and ‘end-stage renal disease’ and was adjusted for each database (Supporting Information Table S2). Moreover, we manually investigated a relevant reference of systematic reviews to search for additional potential studies.

2.3. Eligibility and exclusion criteria

Eligible studies were included if they met the criteria: (1) Participants: Adults diagnosed with CKD according to established criteria [23]. Given the different mechanisms of physical activity promotion in children/adolescents and adults [24], only patients aged 18 years and older, including stages 1–5 nondialysis, peritoneal dialysis, hemodialysis, and kidney transplant recipients, were included in this review. (2) Interventions/Study design: Randomized controlled trials (RCTs), quasi-experiment studies, and single-arm trials (pre-post study designs) that incorporated any interventions. In this systematic review, we have summarized seven categories of strategies, with detailed operational definitions in Supporting Information Table S3. (3) Comparator/control: usual care or sham intervention. Studies comparing the effects of one intervention to another on physical activity were also included. (4) Outcome: Physical activity, refers to any bodily movement produced by skeletal muscles that requires energy expenditure. Studies must include variables reflecting objective measures or self-reported physical activity that can be measured with pedometers, accelerometers, activity monitors, or scales. Outcomes of interest include but are not limited to, daily steps, duration of physical activity (low-intensity physical activity, moderate-intensity physical activity, vigorous-intensity physical activity, total physical activity), and scale scores. The primary time points were baseline (before the start) and endpoint (at the completion of the intervention); wherever possible, we analyzed changes in physical activity from baseline. Higher physical activity was defined as an increase in physical activity from baseline, e.g., higher daily step counts, or higher scale score. We excluded conference abstracts, reviews, case reports, case series, non-English published articles, and published articles reporting duplicate data.

2.4. Study selection

Two independent authors screened the title and abstract of each citation based on the inclusion criteria. The reviewers’ decisions to include or exclude all retrieved articles were recorded in EndNote 20 software. Potentially eligible articles were highlighted for full-text evaluation, and reasons for excluding ineligible studies were recorded. We discussed disagreements and consulted a third review author if needed.

2.5. Data extraction

The following data were extracted and recorded by two independent authors from each included study: first author, year of publication, study design, sample size, patient characteristics (gender, age, stage of disease), intervention characteristic, study duration, physical activity measurement tool, and values of physical activity observed at baseline and completion of the intervention (endpoint). When two or more methods to assess outcomes were used in a study, the one with better validity and reliability was selected by discussion.

When the physical activity was measured more than twice, baseline and final follow-up outcomes were extracted. In studies comparing a particular intervention with usual care, when more than one intervention group of the same type was reported for a study (e.g., the intradialytic exercise vs. home-based exercise vs. control), these intervention groups were combined to avoid double counting of the control group.

2.6. Quality assessment

The RCTs were assessed by the Cochrane Collaborations’ second version of the risk of bias tool for randomized trials (RoB 2) [25] using the Excel macro tool (https://www.riskofbias.info/welcome). Quasi-experiment studies and single-arm trials were assessed using the ROBINS-I tool [26].

2.7. Handling continuous data

In the case of studies reporting data format of (1) median, first and third quartiles; (2) median, interquartile range; (3) median ± standard deviation; (4) mean, 95% confidence interval; and (5) mean, standard error, the established reference formula was used to estimate the mean and standard deviation (see Supporting Information Table S6 for details).

When data was missing or unclear [27–32], we emailed the original authors (Supporting Information Table S5). Studies without sufficient primary data (e.g., data reported in the literature for which means and standard deviations of physical activity could not be calculated) were excluded from the meta-analysis, but qualitative summaries were performed. Two authors responded with data from their studies [27,30], but Alahnoori et al. [27] reported categorical data and were excluded from the meta-analysis.

2.8. Qualitative review

We performed a qualitative review to present the results for articles that meet the eligibility criteria set by this systematic review and meta-analysis but for which meta-analysis is not possible (e.g., incomplete data).

2.9. Statistical analysis

We conducted three meta-analyses. First, random-effects models (restricted maximum likelihood, REML) were used to summarize the change from baseline to endpoint in physical activity in patients with CKD associated with the intervention strategies (including RCTs, quasi-experiment studies, and single-arm trials). Because the physical activity was assessed by different metrics, effect estimates were expressed as standardized mean differences (SMD) with 95% confidence intervals (CI). SMD was interpreted according to the rules of the Cochrane guidelines (i.e., small effect, 0.2; moderate effect, 0.5; and large effect, 0.8 [21]). Next, we calculated pooled between-study effect sizes (SMD and 95% CI) for RCTs and quasi-experiment studies covering any intervention compared with usual care using post-intervention values to assess the effect of interventions on modifying physical activity in comparison to the control group. According to the description in the Cochrane Handbook [21], it was assumed that comparing final measurements would yield the same estimates as the baseline changes. In addition, for outcomes of interest, we evaluated the effect of intradialytic exercise versus home-based exercise interventions on physical activity in patients with CKD.

Heterogeneity was examined using the Chi-squared test (p < .1 indicates significant heterogeneity) and the I² parameter (Heterogeneity explained by ≥25% means low heterogeneity, ≥50% means moderate heterogeneity, and ≥75% means high heterogeneity [33]). When the number of studies in the analysis exceeded 10, publication bias was assessed by the trim and fill method to estimate asymmetry in the funnel plot [34] and Egger’s regression test. We also calculated prediction intervals to determine the range of study effects that future trials might have [35]. A leave-one-out sensitivity analysis was performed by sequentially removing a single variant from the analysis.

Next, we conducted univariate and multivariate meta-regression analyses to examine the relationship between several study-level covariates and the primary outcome (physical activity). Based on available data, we selected six covariates: (1) male proportion (<50% vs. ≥50%); (2) mean age (<55 vs. ≥55 years); (3) assessment of physical activity (subjective vs. objective); (4) disease stage (non-RRT vs. RRT); (5) study duration (<24 vs. ≥24 weeks); (6) sample size (<30 vs. ≥30). Covariates meeting significance criteria (two-sided p < .15) were entered into the multivariate meta-regression model. A threshold of p < .15 was conservative to avoid premature discounting of potentially critical explanatory variables, and adjusted tests were used to control for type I error [36]. All analyses were performed on R software (version 4.2.0) using the meta [37], and metafor [38] packages, and p < .05 was considered statistically significant.

3. Results

The flowchart of the study is shown in Supporting InformationFigure S1. A total of 15,554 articles were retrieved, and five additional pieces of literature were obtained by reference. After removing duplicates from different database searches, 11,965 reports were excluded based on title and abstract screening, leaving 77 articles eligible for full-text analysis. After the final exclusion of 41 of these articles (The reasons for exclusion are provided in Supporting Information Table S4), 35 articles were included in this systematic review [27–32,39–67], five of which had incomplete reporting of physical activity results [27–29,31,32], were unsuccessful in contacting the original authors and could not be converted into corresponding means and standard deviations, and 30 articles were ultimately used for meta-analysis.

3.1. Characteristics of included studies

This systematic review included five single-arm trials, five quasi-experiment studies, and 25 RCTs with 2,156 participants (range 10–195), most of whom were hemodialysis-dependent CKD patients. All studies included adults with a mean age between 42.7 and 81.8 years (59.9 ± 8.7 years). The 35 included articles covered seven interventions for physical activity promotion in patients with whole-spectrum CKD: (1) exercise-based interventions, (2) real-time step feedback, (3) multicomponent interventions, (4) lifestyle modification, and (5) mobile health (mHealth) interventions, (6) different dialysis modalities, (7) group-based care. The study duration ranged from 7 weeks to 144 weeks. Instruments used to assess physical activity included accelerometers, pedometers, the Physical Activity Scale for the Elderly, and the International Physical Activity Questionnaire (Supporting Information Table S6). The quality assessment of the included literature is shown in Supporting Information Table S7 and Figure S2.

3.2. Pre-post changes

The implementation of physical activity promotion strategies was associated with a significant increase in baseline values of physical activity levels in patients with whole-spectrum CKD (SMD = 0.22; 95% CI: 0.08 to 0.36; I² = 55%) and varied significantly (Q = 81.99; p < .001) across intervention types (Figure 1). The relatively wide prediction interval of 95% shows the uncertainty of the possible sensitivity and specificity values in future studies. Exercise-based interventions and real-time step feedback increased physical activity by 0.26 (95% CI: 0.07 to 0.45; I² = 59%) and 0.36 (95% CI: 0.12 to 0.60; I² = 0%) standard deviations, respectively, in this population; however, no significant effect of a multicomponent intervention, mHealth, lifestyle modification, different dialysis modalities, and group-based care was observed on changing baseline physical activity in whole-spectrum CKD patients (Figure 2 and Supporting Information Figure S3).

Figure 1.

Associations between interventions and pre- to post-intervention changes in physical activity. Abbreviations: IE, intradialytic exercise; HE, home-based exercise; AE, aerobic exercise; CKD, chronic kidney disease; KTR, kidney transplant recipient.

Figure 2.

Subgroup Forest plots of difference intervention strategies. Abbreviations: SMD, standard mean difference.

As shown in Supporting Information Figure S4, the trim-and-fill funnel plots indicate a possible publication bias (P_{for egger}: 0.026). In the sensitivity analysis, the meta-analysis was repeated consecutively after excluding each study, and the results showed a change in effect size between 0.18 (95% CI: 0.06 to 0.30) and 0.25 (95% CI: 0.11 to 0.39) (Supporting Information Figure S5).

The results of univariate and multivariate meta-regressions are given in Table 1. In univariate regression analyses, only mean age (β = −0.31; 95% CI, −0.60 to −0.02) was associated with modifying physical activity levels. In multivariate analysis, although heterogeneity persisted, the I² statistic decreased from 55% to 53%, suggesting that age and population may be potential factors for changes in physical activity (Table 1).

Table 1.

Analysis of moderators using meta-regression model.

Variable	Estimate	SE	95% CI	z	P	I2 (%)	R2 (%)
Univariate
Male proportion
<50% vs. ≥50%	−0.08	0.16	−0.40 to 0.24	−0.49	.624	54.16%	0.00%
Mean age
<55 y vs. ≥55 y	−0.31	0.15	−0.60 to −0.02	−2.12	.034	52.18%	3.70%
Sample sizea
<30 vs. ≥30	−0.18	0.13	−0.45 to 0.09	−1.31	.191	51.40%	7.07%
Duration
<24 w vs. ≥24 w	−0.07	0.14	−0.34 to 0.20	−0.51	.608	54.06%	0.00%
Disease of stage
Non-RRT vs. RRT	0.23	0.16	−0.08 to 0.54	1.45	.147	53.29%	5.04%
Outcome
Subjective vs. objective	0.09	0.14	−0.19 to 0.37	0.60	.546	54.40%	0.00%
Multivariate
Mean age	−0.27	0.15	−0.58 to 0.03	−1.76	.078	53.04%	0.00%
Non-RRT vs. RRT	0.15	0.17	−0.18 to 0.47	0.88	.377

^aSample size at the endpoint.

Abbreviations: RRT: renal replacement treatment; SE: standard error.

3.3. Comparison of physical activity promotion strategies with usual care

Twenty RCTs/quasi-experiment studies reported results of physical activity promotion strategies compared with usual care [28,31,41,42,45–52,54,57,59,60,62,64,65,67], with two studies [28,31] not included in the meta-analysis due to incomplete data reporting. Forest plot results showed a significant increase in physical activity with the addition of an intervention strategy to usual care compared to usual care alone (SMD = 0.29; 0.12 to 0.46; I² = 49%) (Figure 3). The prediction interval crossed the null line, suggesting that implementing interventions may have similar effects to usual care in future studies.

Figure 3.

Meta-analysis of trials examining the effect of interventions vs. usual care on physical activity.

The leave-one-out method demonstrated the robustness of this meta-analysis (Supporting Information Figure S6). No statistically significant sources of heterogeneity were found in the meta-regression analysis (Supporting Information Table S8). The results of this meta-analysis were less affected by individual study bias (P_{for egger} = 0.807) (Supporting Information Figure S8) but were limited by the imprecision of the wide confidence intervals between studies.

3.4. Intradialytic exercise versus home-based exercise program

Four RCTs reported the effect of intradialytic exercise versus home-based exercise on physical activity in hemodialysis-dependent CKD patients [43,52,58,61]. Although the effect size favored intradialytic exercise, there was no significant difference compared to the home-based exercise intervention (Figure 4). The analysis had no statistical heterogeneity, and the small number of studies and large confidence intervals limited the results.

Figure 4.

Meta-analysis of trials examining the effect of intradialytic exercise interventions vs. home-based exercise on physical activity.

3.5. Studies included in the qualitative review

Five studies [27–29,31,32] were included in the qualitative review, and their results agreed with the meta-analyses presented herein. The detailed study characteristics are shown in Supporting Information Table S6. The results of the study by Alahnoori et al. [27] showed that there was no significant difference in physical activity levels between the sumac fruit group and placebo group before and after the intervention. A randomized controlled trial by Chen et al. [28] found that strength training participants showed significant improvement in self-reported leisure-time physical activity compared to the control group at baseline. Dong et al. [29] implemented progressive resistance exercise for 12 weeks in hemodialysis-dependent CKD patients with sarcopenia, and at the end of the intervention, there was a significant difference in self-reported physical activity levels between the two groups. Another RCT that recruited hemodialysis-dependent CKD patients showed no significant increase in physical activity levels in the exercise group, but the decrease was lower than in the control group [31]. Turoń-Skrzypińska et al. [32] reported that the use of a pedometer was able to increase daily step counts in hemodialysis patients.

4. Discussion

Physical inactivity is prevalent in patients with whole-spectrum CKD [8,9] and is strongly associated with poor clinical prognosis and impaired quality of life [7,68,69]. Therefore, designating effective strategies to promote physical activity to induce long-term behavioral and lifestyle changes is a priority for this population [16]. In this comprehensive systematic review and meta-analysis, we confirm that interventions to modify physical activity, particularly exercise-based and real-time step feedback, can significantly increase physical activity levels that do not diminish over time.

In the past few years, literature has been widely published on the importance of physical activity in populations with chronic disease, including whole-spectrum CKD patients [5,70–73]. However, improvements in physical activity following any specific intervention have not been systematically demonstrated. Similar to previous reports that included patients with chronic obstructive pulmonary disease [74] and neuromuscular disease [75], this systematic review clarifies the diversity of interventions used in the current study.

To highlight the effects of interventions to increase physical activity level in patients with whole-spectrum CKD, we stratified the different strategies. The meta-analysis results further clarify that exercise-based interventions can significantly improve physical activity. The predictable decrease in physical activity in CKD patients due to anemia, inflammation, and malnutrition is accompanied by decreased cardiorespiratory adaptations and muscle function, and the resulting vicious cycle is worth pondering [7]. The exercise interventions interrupt this cycle, and although it does not restore the pre-diagnostic level of CKD, it improve physical function [76] and indirectly increase CKD patients’ confidence to participate in physical activity, i.e., exercise self-efficacy [77].

In recent years, wearable activity monitors have become increasingly popular in clinical practice to assess patients’ physical activity and supervise their behavior change [78]. The current meta-analysis suggests that improvements in physical activity occur when pedometers (i.e., continuous monitoring with real-time feedback) are added during an intervention, which is consistent with recent studies that have reported a strong association between the use of wearable devices and increased physical activity in patients with cardiometabolic disease [79], chronic obstructive pulmonary disease [80], and cancer survivors [81]. From a behavioral perspective, using wearable devices to modify physical activity supports real-time self-regulatory mechanisms (e.g., goal setting and self-monitoring) [82], which are recognized as established behavior change techniques that promote long-term health.

Our meta-regression showed that the effect was more prominent in participants younger than 55 years than those older than 55 years. One possible explanation for this is that older CKD patients have shallow levels of physical activity, and despite receiving the intervention, the magnitude of change is limited. Nevertheless, previous studies have shown that ‘even a small increase in physical activity can be beneficial’ [83]. In addition, we found that study duration and disease stage were not moderators of physical activity change, suggesting that interventions aimed at modifying physical activity in the whole-spectrum CKD population may have benefits regardless of time.

4.1. Clinical implications

Although nephrologists recognize the importance of physical activity, factors such as time constraints, inexperience, and lack of knowledge may limit the advice provided [84–86]. Evidence suggests that incorporating wearable devices to provide real-time feedback on daily steps during interventions for patients with CKD may benefit physical activity changes. Therefore, regarding the clinical application, healthcare providers can individually set different daily step goals and encourage using low-cost wearable devices (e.g., pedometers and accelerometers) and/or smartphones to record and track actual step counts. Electronic information and clinician encouragement could reduce healthcare use while ensuring that interventions to promote physical activity adhere to best practices and current guidelines.

Several observational studies have reported decreased physical activity in patients with CKD from the early disease stage to a ‘nadir’ during the hemodialysis stage [7,8]. This study showed that an exercise-based intervention strategy could significantly increase physical activity in whole-spectrum CKD patients. In a meta-analysis, there was no difference in the change in physical activity between intradialytic and home-based exercise in hemodialysis-dependent CKD patients. This result suggests an inspiration for future clinical practice. For hemodialysis patients with relatively stable physical status, clinical healthcare professionals can provide appropriate guidance and encouragement to support them in performing home exercise training; those with poor compliance can perform intradialytic exercise under the supervision of health providers. The latter can not only improve dialysis adequacy but also gain a higher exercise self-efficacy through communication with patients [87].

4.2. Strength and limitation

Different types of clinical studies were included in this study to describe the role of intervention strategies in increasing physical activity levels in whole-spectrum CKD patients in terms of pre-post change and postintervention comparison, respectively. Second, the analysis used univariate and multivariate meta-regression to explore the association between specific moderators and physical activity change.

However, there are several limitations of this systematic review that deserve consideration. First, this study included highly biased quasi-experiment studies and single-arm trials, and the meta-analysis results should be interpreted with caution. Second, physical activity may be a secondary outcome in some studies; however, finding these studies was difficult, as the titles and abstracts do not necessarily describe physical activity assessment in the articles. Despite a rigorous screening process, we cannot completely exclude the possibility that a few studies were not identified. Third, most studies were centered on hemodialysis-dependent CKD patients, which may limit the generalizability of findings. Fourth, changes in physical activity are influenced by the objective environment (e.g., weather and traffic [13]), and we could not account for these factors in our analysis. Fifth, keeping participants uninformed about exercise as a type of intervention is inherently difficult, and the risk of bias due to deviations from the intended intervention and patient-reported outcome measures is of concern. The fact that some physical activity outcomes are self-reported means that it is subject to recall bias [88]. Sixth, in this systematic review and meta-analysis, we pooled subjectively and objectively measured physical activity outcomes, which may have been the source of the higher heterogeneity (I² = 55%); additionally, subjectively reported levels of physical activity may have inflated the results.

4.3. Future consideration

In this systematic review, only one high-quality study assessed physical activity beyond 12 months. Although a 3-year lifestyle intervention increased the proportion of participants meeting physical activity guidelines, this RCT used a questionnaire, which had the disadvantage of a greater effect of chance on outcomes [89]. Future studies should introduce wearable devices to observe the role of long-term interventions in changing physical activity. In addition, person-centered medicine is increasingly being emphasized. Although exercise programs provide a valuable opportunity for patients with CKD to improve physical fitness and other health conditions, a checklist to support a range of physical activities should include not only increased behavioral interventions but also the removal of innate barriers that allow patients to increase their habitual physical activity as much as possible [90]. Therefore, future research needs a menu of adaptable interventions to encourage physical activity for whole-spectrum CKD patients.

5. Conclusion

In conclusion, our findings suggest that interventions aimed at increasing physical activity, particularly those based on exercise and step feedback, can significantly improve physical activity in whole-spectrum CKD patients. Although we were unable to draw definitive conclusions regarding the optimal physical activity promotion intervention, what is clear from this systematic review is that implementing interventions targeting physical activity in whole-spectrum CKD patients may be able to obtain physical activity increases in the short term. This suggests that future focus should be on sustained maintenance of physical activity, which could be combined with wearable devices based on goal-setting theory.

Funding Statement

This study was funded by Budgeted projects in Shanghai University of Traditional Chinese Medicine (Grant number: 2021WK118), Shanghai Municipal Health Commission (20224Y0166).

Authors’ contributions

Fan Zhang had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Fan Zhang, Yan Bai. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Fan Zhang, Hui Wang. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Fan Zhang, Hui Wang. Obtained funding: Fan Zhang. Administrative, technical, or material support: Liuyan Huang, Huachun Zhang. Supervision: Huachun Zhang.

Disclosures statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data sets covered in this article are listed in the manuscript and the supplementary material.