Introduction

Chronic kidney disease (CKD) is one of the leading causes of increased cardiovascular morbidity and mortality, [1] which is associated with higher healthcare costs and significantly burden. [2] Currently, CKD has become a growing public health problem. [3]

Frailty is a state of increased vulnerability to stressors due to decreased physiological reserves and thus leads to a poor health outcome, [4] and it is described as a spectrum from no frailty (i.e., robustness) to pre-frailty (i.e., early stage of frailty) and then physical frailty. [5] In the context of healthy aging, physical frailty was considered a public health priority, and its complexity requires specific management strategies. [6]

Physical frailty is common in patients with CKD, [7, 8] which can lead to a poor prognosis, as reflected by an increased risk of adverse events such as motor dysfunctions, limited mobility in daily life, and falls, leading to a reduced quality of life, and higher disability rates, ultimately an increased risk of mortality. [9, 10]

Between 2017 and 2021, seven meta-analyses assessing the prevalence of physical frailty in patients with CKD or the impact of frailty on survival were published. [11,12,13,14,15,16,17] However, meta-analyses on this topic have certain shortcomings, such as the inclusion of studies with overlapping populations (e.g., ACTIVE/ADIPOSE dialysis cohort [18], and CanFIT cohort [19]), missing some important relevant studies, and limited and/or lacking meta-regression analyses exploring sources of heterogeneity limit interpretation and conclusions.

Therefore, we conducted an updated, more comprehensive systematic review and meta-analysis with a larger sample size, and fewer overlapping cohorts, and its impact on survival. The primary goal of this study is to estimate the risk of mortality in patients with CKD who were affected by physical frailty. Our secondary objective is to assess the prevalence of physical frailty in patients with CKD, and to perform a detailed subgroup analysis to determine the distribution of frailty and the association of age and body mass index (BMI) with the risk of frailty-related mortality.

Methods

This study followed the newest Preferred Reporting Items for Systemic Reviews and Meta-Analyses (PRISMA 2020) [20] (Table S1) and the Meta-analyses of Observational Studies in Epidemiology (MOOSE) guidelines. [21] The systematic review protocol has been registered on the International Prospective Register of Systematic Reviews (PROSPERO) with a registration number CRD42022320312. The current study methodology is similar to the previously described protocol (https://www.crd.york.ac.uk/PROSPERO/) with a few modifications (Table S2).

Patient and public involvement

Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Search strategy

We searched PubMed, Embase, Web of Science, and two clinical trial registries-ClinicalTrials.gov and Cochrane Central Register of Controlled Trials (CENTRAL), from inception to May 8, 2022. The databases were searched independently by two authors (FZ and WH). We adopt a comprehensive retrieval strategy that combines medical subject heading (MeSH) terms, title/abstract, and synonyms to retrieve eligible studies fully. The search strategies for databases are available in Table S3. Moreover, we also reviewed conference abstracts from the International Society of Nephrology (https://www.theisn.org/), European Renal Association (https://www.era-online.org/en/), and American Society of Nephrology (https://www.asn-online.org/) for the past five years to look for other potential studies. Finally, we manually investigated a relevant reference of systematic reviews to search for additional potential studies. [11,12,13,14,15,16,17] Any disagreements were resolved by a third reviewer’s opinion (LYH).

Eligibility criteria

The inclusion criteria for studies were as follow: (1) adults (aged 18 years or older) confirmed CKD, including predialysis (i.e., stage 1-5 non-dialysis), peritoneal dialysis, hemodialysis, kidney transplant recipients; (2) reported prevalence of physical frailty according to a validated tool (e.g., Fried frailty phenotype; Frailty Index; Clinical Frailty Scale), or sufficient data to calculate it; (3) study reported data on physical frailty and mortality outcomes; (4) study design was limited to observational studies, including cohort study, cross-sectional study, case-control study, and longitudinal studies; (5) articles published in English.

Studies were excluded if any of the following exclusion criteria were met: (1) letter/comment/editorial, conference abstracts, and case reports; (2) studies of self-reported CKD; (3) reported other subtypes of frailty, like cognitive, psychosocial, or nutritional frailty. [22] For studies with overlapping cohorts or populations, we selected studies with more recent data, larger sample sizes, and/or richer information for inclusion. Disagreements were resolved by a third reviewer if necessary (LYH).

Methodological quality assessment

Risk of bias of included studies was assessed using the Newcastle-Ottawa Scale (NOS). [23] The scale assesses the risk of bias in three domains: selection, comparability, and exposure/outcome. In the selection and outcome sections, three of the nine items-“demonstration that outcome of interest was not present at start of study,“ “was follow-up long enough for outcomes to occur?“ and “adequacy of follow-up of cohorts-were removed since they do not apply to cross-sectional studies. [24]

Data extraction

Two independent authors (FZ and HW) extracted the data using an agreed form. We collected the following information from the included studies: (1) first authors; (2) publication year; (3) geographical location and country; (4) total sample size; (5) mean age and body mass index (BMI); (6) CKD stage; (7) the number of patients with physical frailty (where available, we also collected data on prefrail); (8) assessment tool used to define the presence of physical frailty; (9) the number of deaths; (10) follow-up time; 11) hazard ratio (HR) and 95% confidence interval (95% CI) of patients with frailty CKD corresponding to non-frail patients.

Where more than one study reported data from the same sample and outcome, only one study was selected based on a larger sample size or better suitability of data. When a study reported multiple instruments to assess physical frailty, we used only the results evaluated by the most used scales for inclusion in the primary meta-analysis to avoid the problem of non-independent data. If a study reported the prevalence of physical frailty at different stages of CKD, it was treated as two records included in the meta-analysis separately.

Where a study used hierarchical categorization, for example, mild, moderate, and severe frailty, for the primary analysis, we classified those in the moderate and severe category as frailty and mild as prefrail. If a study appeared potentially eligible but did not report the required data, we contacted the authors by email to obtain additional information and maximize existing studies’ availability. Studies were excluded when they did not report data available for extraction, and the authors could not be contacted. If a study reported longitudinal data on the prevalence of physical frailty in patients with CKD, only baseline data were extracted. Any disagreement about data extraction was resolved by consulting a third author (LYH).

Data synthesis

We performed a meta-analysis using meta [25] and the metafor package [26] in R software (version 4.2.0) to obtain weighted pooled estimates of physical frailty and prefrail for all studies. The prevalence was transformed using the Freeman-Tukey double arcsine transform to approximate a normal distribution and stabilize the variance. [27] Since study designs and assessment tools vary, we pooled data using a random-effects model based on the DerSimonian-Laird inverse variance method to provide conservative estimates of physical frailty prevalence and then reverse transformed the estimates and 95% CIs and expressed them as percentages to facilitate interpretation. [28] In addition, prediction intervals were generated to provide a range of predictions of true prevalence in any individual study. [29] Heterogeneity was quantified using the τ2 and I2 statistics, and I2 > 50% was defined as significant heterogeneity. [30]

We performed pre-planned subgroup analyses according to the CKD stages, different assessment tools, geographical location (i.e., the corresponding continent), and study design. Possible publication bias was tested by examining the funnel plot and an Egger regression. [31, 32]

To explain the significant heterogeneity observed, meta-regression analyses were conducted using the R metafor package to assess the effect of the following moderator variables on the heterogeneity of the prevalence of physical frailty: mean age, the proportion of males (percentage), mean BMI, geographical location, CKD stage, and assessment tool. Moreover, using a bubble plot, we visualize the relationship between mean age, BMI, and prevalence of physical frailty and the association between frailty and mortality risk.

We performed three sensitivity analyses: (1) to exclude studies with a sample size of fewer than 100 participants to verify the possibility that the results overestimated the severity of the problem due to the small sample size; (2) considering the recent debate on the meta-analysis of proportions, [33, 34] we used generalized linear mixed models (i.e., random intercept logistic regression models) to pool the prevalence of physical frailty in CKD patients; [35] (3) using a “leave-one-out” approach, in which we removed all studies one by one to analyze their effects on the primary analysis. The grading recommendations assessment development and evaluation (GRADE) approach was used to assess the evidence on prognosis. [36]

Results

After removing duplicate articles from different databases, a total of 9,697 articles were included. These articles were first screened by title and abstract, and eventually, 369 articles and an additional seven articles were subjected to full-text review. A total of 187 articles were included after a rigorous screening process (The reasons for exclusion are listed in Table S4). Notably, some populations of included studies were duplicated (full details are listed in Table S5). One hundred and thirty-nine articles with 144 studies were ultimately included in the meta-analysis. The PRISMA flowchart is shown in Fig. 1.

Fig. 1
figure 1

The PRISMA flowchart of studies included in the meta-analysis. Legend: CENTRAL = Cochrane Central Register of Controlled Trials; CKD = Chronic Kidney Disease

Full size image

The characteristic of included studies

Table S6 summarizes the characteristics of 139 articles included in this systematic review and meta-analysis. There were 44 articles from Asia, 54 from the Americas (South and North America), 38 from Europe, and only three from Oceania. The included population was predominantly dialysis-dependent, with sample sizes ranging from 22 to 1,424,026 of the included studies. The mean age of the patients ranged from 31 to 83.7 years, and the mean BMI from 20.7 to 31.7 kg/m2. The scores for all studies are shown in Table S7.

Prevalence of physical frailty in CKD patients

The results of a meta-analysis based on 135 articles (142 studies in total) with a total of 1,672,650 patients showed an overall prevalence of physical frailty in CKD patients of 34.5% (95% CI 31.0 to 38.1%), with significant heterogeneity between studies as expected (Figure S1). The effect of individual studies on the combined estimates is very small by the leave-one-out analysis (Figure S2). The prevalence of physical frailty combined using the generalized linear mixed model was 33.0% (95% CI 29.2 to 36.59%) (Figure S3), similar to the Freeman-Tukey double arcsine transform. After excluding studies with sample sizes below 100, the pooled result was 33.6% (95% CI 29.7 to 37.7%) (Figure S4).

According to the different scales for judging pre-frailty, 60 articles (61 studies in total) reported data on prefrail in patients with CKD, with a combined prevalence of 39.4% (95% CI 35.3 to43.6%) and high heterogeneity between studies (Figure S5). Leave-one-out analysis showed that individual studies had minimal effect on the pool estimates (Figure S6). The prevalence of prefrail using the generalized linear mixed model was 38.8% (95% CI 34.7 to 43.1%), slightly lower than the Freeman-Tukey double arcsine transform (Figure S7).

Subgroup analysis for the prevalence of physical frailty in CKD patients

Figure 2 reports the results of subgroup analyses of physical frailty prevalence. Significant interactions were found based on the assessment tool, study design, and disease stage. A higher prevalence of physical frailty was found in the South American population and studies using the Chinese Frailty Score and Tilburg Frailty Indicator, while the lowest prevalence was found in studies using the Edmonton Frail Scale and FRAIL scales. In cross-sectional studies, lower proportions of CKD patients were frail. Regarding disease stage, kidney transplant recipients had the lowest prevalence of frailty. The heterogeneity found in most subgroup analyses was high. Based on subgroup analysis, Figure S8 shows the prevalence of frailty in 19 countries.

Fig. 2
figure 2

Subgroup analyses for the prevalence of physical frailty. Legend: KTR = Kidney Transplantation Recipient; CFS = Clinical Frailty Scale; TFI = Tilburg Frailty Index; EFS = Edmonton Frail Scale; FI = Frailty Index; CI = Confidence Interval

Full size image

Meta-regression for the prevalence of physical frailty in CKD patients

To explore potential heterogeneity sources in frailty prevalence, we performed univariate meta-regression analyses based on several study-level characteristics. The results identified disease stage and assessment tools as potential factors that could cause high heterogeneity (Table S8). Specifically, the prevalence of frailty was higher in dialysis-dependent CKD patients and lower in kidney transplant recipients compared to predialysis, consistent with the results of the subgroup analysis. Figure 3 shows the relationship between mean age, BMI, and prevalence of frailty.

Fig. 3
figure 3

Univariable meta-regression for the prevalence of physical frailty according to study-level characteristics. Legend: A = mean age; B = mean body mass index

Full size image

Association between physical frailty and mortality risk in CKD patients

From the 37 articles (38 studies in total) that provided analyzable data, physical frailty was associated with increased mortality risk in patients with CKD. The pooled adjusted HR was 1.941 (95% CI 1.586 to 2.375), but heterogeneity was high (Figure S9).

Eleven articles (12 studies in total) provided prefrail-related data. The pooled adjusted HR was 1.345 (95% CI 1.231 to 1.469), with moderate heterogeneity between studies (Figure S10). In addition, four studies reported data on continuous variables, i.e., each 1-unit increase in frailty score was associated with a 29.1% (95% CI 1.191 to 1.400) increased mortality risk (Figure S11).

Sensitivity analyses based on leave-one-out revealed no changes in the significance or direction of effects for these results (Figure S12-S13). The evidence level for these results was very low according to GRADE (Table S9).

Meta-regression for physical frailty and mortality risk in CKD patients

Meta-regression investigating potential sources of heterogeneity demonstrated that mean age, BMI, percentage of males, and CKD stage were not significantly associated with physical frailty and mortality risk, but geographical location explained between-study heterogeneity (Table S10). Furthermore, we visualized the association of mean age and BMI with physical frailty and mortality risk (Figure S14).

Publication bias

Funnel plots for all meta-analysis outcomes are shown in Figure S15 and were tested by Egger linear regression.

Discussion

Summary of main results

In this systematic review and meta-analysis of 1,672,650 CKD patients, approximately one-third (95% CI 31.0–38.1%) were physical frail, with a prediction interval pointing to a prevalence range of 3.2%–76.9%, which was a higher in older individuals and lower in those with higher BMI. Pre-frailty is defined as the transitional stage of transition from a normal clinical state to frailty. [5] The pooled results demonstrated a prevalence of pre-frailty in CKD patients of approximately 39.4% (95% CI 35.3 to 43.6%), which is slightly higher than that of frailty.

On the other hand, with significant heterogeneity, the risk of mortality in patients with frailty CKD was nearly two times higher than in patients without frailty (HR: 1.941; 95% CI 1.586 to 2.375), and the results of subgroup analysis at different stages of CKD supported this association. Further analysis revealed a positive association between physical frailty and mortality risk in those with older age and higher BMI.

Discussion of main findings

In the past 20 years, frailty has been introduced in geriatrics as the aging population has rapidly expanded, highlighting the significant impact on patients and health services, clinical care, and chronic non-communicable diseases. [5, 6] Studies of patients with early-stage CKD and ESKD treated with long-term dialysis have reported a prevalence of physical frailty ranging from 7% in CKD stages 1–4 to 73% in hemodialysis patients, [11] and the prediction interval results of this meta-analysis support this wider result, depending on the frailty assessment tool, the participants, [10] and different glomerular filtration rate calculation formulas. [37]

The view in recent years has been that the assessment of frailty provides additional insight into the various phenomena that affect the physiologic status of patients with chronic diseases, including CKD. [38, 39] Although there is a consensus on the theoretical concept of frailty, many tools have been used to assess it. [40] Among these, the frailty phenotype, which assesses remaining physiologic reserve based on the phenotypic expression of different signs and symptoms, is most used to screen individuals for frailty, [5] whereas the frailty index provides an overall assessment of health deficits. [41]

After conducting a more comprehensive search and excluding overlapping populations, our current study provides a relatively reliable estimate of the prevalence of physical frailty in patients with CKD, documenting that approximately 30 of every 100 patients with CKD are frail and 39.4% are prefrail. This result suggests that up to 73.5% of CKD patients have some degree of physical frailty, compared to 63% in the general population. [42] It is important to note that the prevalence of physical frailty varies widely across CKD stages and should be more targeted for screening and intervention. [43] The highest risk is in the dialysis population, while the lowest is in kidney transplant recipients, with a prevalence of physical frailty of 40.3% and 20.6%. It is well known that CKD is a dynamic process. [44] In the early stages of CKD, frailty is triggered by increased catabolism due to anemia, insulin resistance, and vitamin D deficiency and progresses with loss of renal function. [45] In dialysis-dependent CKD patients, this frailty cycle is further exacerbated by metabolic acidosis, inflammation, malnutrition, and dialysis. [46] However, frailty can improve within a few months after kidney transplantation. [47] Nevertheless, it has been suggested that physical frailty may or may not be reversible and that patients may develop severe after transplantation, thus necessitating concern for post-transplant frailty. [48]

On a national level, some countries reported sparse data, with India reporting a relatively higher incidence of frailty, as shown in Figure S8. The population of India is reported to be undergoing a marked shift and is now aging at an accelerated rate. [49] It is well known that advanced age is closely associated with frailty development. [4] As a comparison, western countries such as Spain, Germany, and Australia have lower rates of frailty in CKD patients, which may be attributed to two factors: (1) outcome bias caused by a smaller included population; and (2) national emphasis on CKD management, for example, the United Stated government released the “Advancing American Kidney Health” initiative in 2019, which aims to improve the lives of Americans suffering from CKD. [50]

In the regression analysis, we observed an interesting phenomenon. The reality that the prevalence of physical frailty increases with age is unquestionable; after all, frailty is an age-related clinical condition. [51] However, the relationship between frailty and BMI is negative, although not statistically significant. According to the explanation of Lowrie et al., [52] during inflammatory states or malnutrition, body protein stores are diverted to defend against inflammation and repair damage; therefore, the increased body weight of overweight CKD patients protects against inflammation and infection, which are vital factors that exacerbate frailty development. [53] This theory may explain the lower prevalence of physical frailty in CKD patients with higher BMI.

The pooled adjusted HR associated with frailty for categorical and continuous indicators ranged from 1.291 to 1.941. In this case, the pooled adjusted HR was lower in kidney transplant recipients at 1.450 (95% CI 1.163 to 1.806) and higher in dialysis patients at 1.936 (95% CI 1.626 to 2.306). For dialysis patients, factors such as reduced physical activity and inadequate intake contribute to the high prevalence of frailty. [18, 54] In turn, frailty can cause decreased immune function and frequent infections, while inflammatory mediators released by infections cause increased protein metabolism, decreased muscle and fat volume, and worsened frailty, [55] which aggravates the development and progression of atherosclerosis through inflammatory mechanisms. [56] The three form a vicious cycle that increases the incidence of cardiovascular disease and mortality in dialysis patients. However, frailty-induced mortality is higher across the CKD spectrum, which emphasizes the need for screening and intervention for frailty in patients with all stages of CKD to prevent premature death in frail patients. This finding also highlights the need to incorporate physical frailty scores into CKD prognostic models, particularly in the dialysis population.

In patients with CKD, the “obesity paradox” has been reported, implying that higher BMI may be associated with better survival. [57] However, our meta-regression found that, although no statistical significance was observed, patients with higher BMI had a higher risk of frailty-induced death. This result is contrary to the results of Chang et al. [58]. The reason for this contradiction with the “obesity paradox” may be that defining obesity solely in terms of BMI does not quantify the body fat ratio and distribution. [59]

An unavoidable and challenging issue is the inclusion of studies using a variety of frailty assessment scales. Anderson et al. reported that the major frailty scales (Frailty Phenotype, Frailty Index, Edmonton Frailty Scale, and Clinical Frailty Scale) were similar in assessing the prevalence of frailty in dialysis-dependent CKD patients. [60] However, the agreement in frailty status between instruments was weak at best, which may explain the reason for the high heterogeneity of this study. Nevertheless, Imamura K et al. reported that the Fried Frailty Phenotype, Study of Osteoporotic Fractures Index, Short Physical Performance Battery, and Clinical Frailty Scale were similar about predicting clinical events in outpatient hemodialysis patients; also, Imamura K et al. suggested that caution should be considered when using questionnaire-based frailty assessments. [61] These findings suggest that it may be feasible and informative to pool studies for meta-analysis for the specific purpose of this study, based on the support of available evidence in the literature.

Clinical and public health implications

The results of our study have important clinical and public health implications. From a clinical perspective, assessment of frailty and consequent individualized care can improve physical condition before dialysis and awaiting kidney transplantation by assigning intervention strategies appropriate to the individual. It has been argued that the degree of frailty in ESKD is one of the critical decision factors for receiving renal replacement therapy. [62, 63] Nephrologists are less likely to recommend dialysis treatment for patients with established frailty. [64] Furthermore, Haugen et al. reported that among kidney transplant candidates, the enrollment rate of frail patients was 38% lower (HR: 0.62; 95% CI 0.56 to 0.69) compared to non-frail patients. [65] Various interventions have been proposed to influence the frailty state. Some interventions such as prehabilitation [66], aerobic exercise and resistance training, and strict blood pressure/glycemic control have an established role in managing frailty in CKD. [43] For the reversal of frailty in dialysis patients, the current study proposes a strategy of exercise training combined with oral nutritional supplementation, which has gained some benefits. [67, 68]

The developmental benefits of reversing frailty are not limited to the patient level but may also positively impact public health. Based on our findings on the positive correlation between the prevalence of physical frailty and age in CKD and the progression of frailty with declining renal function, early detection, and intervention in predialysis stages and the provision of timely, accessible, and adequate affordable support for CKD patients through effective referral mechanisms to prolong the functional decline trajectory of older patients as much as possible, which is the concept of “healthy aging” advocated by the World Health Organization. [69]

Strength and limitation

This is the largest meta-analysis to date focusing on studies of the prevalence of frailty in CKD patients and its impact on the mortality risk in this population. We acknowledge the following limitations. First, statistical heterogeneity between studies remained high even after subgroup analysis of disease stage, geographic location, study design, and assessment tools. This suggests differences in the demographic characteristics of the included study populations that were not explained in this study (e.g., comorbidity, ethnicity). It may have partially contributed to the differences in reported clinical outcomes between studies. Nonetheless, high heterogeneity is a common problem in epidemiological meta-analysis and has been gradually accepted. [70] Second, small sample studies may compromise the statistical efficacy of the study. Therefore, we performed the additional manipulation of excluding studies with sample sizes below 100 to observe changes in effect sizes. The results were similar to those of the primary analysis. Third, although frailty is a dynamic process, variables including changes in frailty over time after diagnosis of CKD, dialysis initiation, and kidney transplantation were unavailable due to the nature of the meta-analysis and require further investigation. Fourth, despite our rigorous review, there may be partial overlap in the included studies. Fifth, different included studies adjusted for different confounders in the multivariate Cox regression model, which may have contributed to the high heterogeneity of the pooled adjusted HR. A better approach would be to use individual patient meta-analysis to further assess the relationship between frailty and survival after adjusting for key confounding variables. Sixth, the funnel plot asymmetry and the significance of the Egger regression test suggest a possible publication bias in this study. Seventh, because the included studies were designed as observational studies, the level of evidence for frailty and mortality risk in CKD was rated as very low in terms of GRADE. Given these limitations, the results of this systematic review and meta-analysis should be interpreted cautiously.

Conclusion

In conclusion, this systematic review and meta-analysis suggest that frailty affects approximately one-third of patients with CKD. The current study also showed that frailty was associated with increased mortality risk among patients at different stages of CKD. Considering this, (1) frailty should be an essential consideration in decision-making, prognosis, and advance care planning for patients with CKD; (2) all patients with CKD, regardless of their degree of renal dysfunction, should be regularly assessed for frailty status with a standardized tool; and (3) more studies are needed to incorporate frailty into the prognostic scores of patients with CKD.