Visual Abstract
Keywords: CKD, chronic kidney failure, ESKD, GFR, kidney dysfunction, kidney failure, mortality, mortality risk, renal function
Abstract
Key Points
Increased eGFR variability over 3 years independently predicts a higher risk of kidney outcomes and all-cause mortality.
This association remained consistent across subgroups and sensitivity analyses.
Routine eGFR variability assessment may enable identification of high-risk patients and provide an opportunity for the initiation of interventions.
Background
eGFR variability may predict adverse outcomes, such as cardiovascular events and mortality, yet its influence on kidney impairment progression in routine clinical practice is not well described.
Methods
This retrospective cohort study used longitudinal eGFR data from MedicineInsight, a comprehensive primary care database. We included adults (18 years or older) with at least three eGFR measurements over 3 years between January 1, 2011, and December 31, 2018. eGFR variability between visits was assessed using the coefficient of variation and categorized into groups by quintiles. A kidney composite end point, comprising a sustained 40% decline in eGFR from baseline, a sustained eGFR of <15 ml/min per 1.73 m2, and all-cause mortality, was tracked over a 3-year follow-up. Cox proportional hazards models quantified the association between eGFR variability and outcomes.
Results
Among 754,306 patients, with a mean age of 59.1 years and 58.0% female, higher eGFR variability was associated with an increased risk of the kidney composite end point (hazard ratio, 2.17;95% confidence interval, 2.03 to 2.32) for the highest versus lowest fifth after adjusting for mean eGFR and eGFR slope, as well as other cardiovascular disease risk factors and medication use. Similar trends were observed for components of the primary outcome and across all subgroups including age, sex, hypertension, diabetes, and baseline eGFR.
Conclusions
Increased eGFR variability independently predicts adverse kidney outcomes, underscoring its potential as a clinical biomarker for identifying high-risk patients. Including eGFR variability in routine kidney assessments may improve risk stratification, enabling timely interventions and potentially enhancing patient outcomes in primary care.
Introduction
CKD, marked by reduced kidney function or lasting damage, poses a major public health challenge and economic burden, particularly when it progresses to kidney failure which requires costly treatment.1–4 Lower eGFR is strongly linked to increased risks of cardiovascular disease and mortality.5,6 Moreover, decline in eGFR over time is a significant indicator of increased risk of adverse outcomes.7–9 Therefore, monitoring eGFR is essential for assessing disease diagnosis and progression.10,11
Fluctuations in eGFR measurements (i.e., eGFR variability) are commonly observed in routine clinical practice, occurring over short (days or weeks) or long-term (months or years) periods. They are influenced by a complex interplay of physiologic factors, underlying health conditions, and external influences such as medications, diet, and hydration. Although isolated eGFR measurements have traditionally been used in research studies evaluating kidney health, emerging evidence suggests that eGFR variability may offer additional insights into risks for adverse outcomes. For example, higher eGFR variability has been linked to an elevated risk of cardiovascular events, mortality, and worsening nephropathy.12–15 However, much of the evidence base to date has concentrated on specific populations and settings with restricted inclusion criteria (e.g., diabetes, trial participants,12,14,15 and military veterans13). Consequently, data on the implications of eGFR variability across broader populations in routine clinical settings remain limited. Using a large, national general practice data source encompassing individuals with a broad range of patient profiles, we sought to investigate the relationship between eGFR variability and adverse kidney outcomes.
Methods
Study Design and Population
We conducted a retrospective cohort study using data from MedicineInsight, a national primary care database in Australia. In brief, MedicineInsight data comprise deidentified longitudinal patient records, including demographics, comorbidities, laboratory results, clinical measurements, and prescriptions from general practices across Australia. Description of MedicineInsight dataset are reported elsewhere.16–18 The base population included adults aged 18 years or older who visited a general practice participating in the MedicineInsight program and had at least one serum creatinine measurement between January 1, 2011, and December 31, 2021 (2,717,966 patients across 392 practices).
Assessment of eGFR Variability, Follow-Up, and Outcomes
From the study base population, we identified patients who had at least three eGFR measurements over a 3-year period (hereinafter referred to as the “exposure window”; Supplemental Figures 1 and 2). The first exposure window over the study period that met these criteria was defined as the exposure period for the study (i.e., the time between the date of the first eligible eGFR measurement and the date corresponding to the end of the 3-year period after the date of the first eligible eGFR measurement). Study outcomes were assessed in the 3-year period after the exposure window (hereinafter referred to as the “outcome window”). To allow a 3-year outcome window for all included patients, ascertainment of the exposure window was from January 1, 2011, to December 31, 2018. Individuals who had eGFR <15 ml/min per 1.73 m2, implausibly high eGFR (>200 ml/min per 1.73 m2), or experienced the study outcomes of interest during the exposure window were excluded. MedicineInsight includes outpatient eGFR values that are automatically calculated and routinely reported alongside serum creatinine results. To account for potential variability arising from differences in eGFR equations used during the study period, eGFR was recalculated using the 2009 CKD-Epidemiology Collaboration creatinine equation (without race).19
The primary outcome was a composite kidney outcome, defined as CKD progression (sustained decline in eGFR of ≥40% from baseline [i.e., the last eligible eGFR measurement during the exposure window]), kidney failure (sustained eGFR <15 ml/min per 1.73 m2), or all-cause mortality. Assessment of sustained decline in eGFR was based on at least two eGFR assessments with no eGFR measurements below (above for the outcome of sustained eGFR <15 ml/min per 1.73 m2) this threshold in the intervening period (>90 days). Secondary outcomes were the individual components of the primary outcome: (1) sustained decline in eGFR ≥40%, (2) kidney failure, and (3) all-cause mortality. Individuals were followed from the end of the exposure window until the date of the outcome or the end of the 3-year outcome window.
Covariates
Baseline covariates were assessed using information closest to the last date of the exposure window. Covariates included sociodemographic information (age, sex, and smoking status), laboratory and clinical measurements (mean eGFR, eGFR slope [both mean and slope assessed within the exposure window]), comorbid conditions (coronary heart disease, heart failure, hypertension, and type 2 diabetes mellitus), and prescription medications (statins, diuretics, renin-angiotensin-aldosterone system [RAAS] inhibitors, β-blockers [BBs], calcium channel blockers, metformin, sodium-glucose cotransporter 2 [SGLT2] inhibitors, glucagon-like peptide-1 receptor agonists, and insulin).
Statistical Analysis
Continuous variables are reported as mean (with SD) and categorical variables as number with percentages. Variability of eGFR was assessed using the coefficient of variation (CV; CVeGFR) defined as the percentage of the SD of eGFR to the mean eGFR. Individuals were categorized according to fifths of CVeGFR. Linear regression and logistic regression were used to analyze trends in baseline characteristics across the levels of eGFR variability. For each patient, eGFR slope during the exposure window was calculated using a linear regression model. Cumulative incidence for the primary outcome was illustrated by inverse Kaplan-Meier curves, where the log-rank trend test was used to compare incidence across the levels of eGFR variability. Multivariable Cox proportional hazards models were used to estimate hazard ratios (HRs) and their 95% confidence intervals (CIs) for the outcomes across the fifths of eGFR variability (using the lowest fifth as the reference category). Cox models were adjusted for age, sex, mean eGFR, eGFR slope during exposure window, smoking status, coronary heart disease, heart failure, hypertension, type 2 diabetes, and medication prescription (statin, dipeptidyl peptidase-4 inhibitors, metformin, SGLT2 inhibitors, glucagon-like peptide-1 receptor agonists, insulins, diuretics, RAAS inhibitors, BBs, and CCBs). Variability in eGFR was also assessed as a continuous variable using cubic spline regression models with knots placed at the 5th, 35th, 50th (reference), 65th, and 95th percentile points. We performed subgroup analysis to assess variations in the relationship between eGFR variability and outcomes across predefined groups according to age, sex, hypertension status, type 2 diabetes status, baseline eGFR levels, and baseline eGFR slope levels.
A series of sensitivity analyses were conducted to test for robustness of the findings: (1) using SD of eGFR as an alternative index for eGFR variability; (2) using different definitions for CKD progression (advanced CKD [sustained eGFR of <30 ml/min per 1.73 m2] and sustained decline in eGFR of ≥30% from the baseline); (3) adjusting for additional covariates with multiple-imputation, systolic BP, diastolic BP, body mass index, serum low-density lipoprotein, and blood glycated hemoglobin; (4) adjusting for urinary albumin-to-creatinine ratio; (5) adjusting for number of eGFR measurements during the exposure window; (6) landmark analysis excluding study events during the first year of the outcome window to account for potential reverse causality (e.g., eGFR variability attributable to early acute illness events in the follow-up period); and (7) using Fine and Gray model with deaths as a competing risk. For the analysis adjusting for additional covariates with multiple imputation, since values for these covariates were predominantly missing at baseline, missing values were imputed using Fully Conditional Specification multiple imputation.20 To do this, we fitted a regression model using basic demographic variables and relevant variables (i.e., systolic BP, diastolic BP, body mass index, glycated hemoglobin, and low-density lipoprotein cholesterol) as covariates, using the standard error and parameter estimates of the covariates to estimate missing values, refit the regression model without estimated missing values, repeating this process a total of 25 times to refine our estimation of the missing values. P values of <0.05 were considered statistically significant. Statistical analyses were performed with SAS version 9.4.
Ethics Approval
This study was approved by the MedicineInsight Data Governance Committee (2020-004) and Research Ethics Review Committee of the Sydney Local Health District, New South Wales, Australia (X21-0428, 2020/ETH00963).
Results
Baseline Characteristics
We studied 754,306 patients as shown in Supplemental Figure 2. Table 1 presents the baseline characteristics of the identified individuals across fifths of CVeGFR (F1: <3.37%; F2: 3.37%–5.59%; F3: 5.60%–8.09%; F4: 8.10%–11.49%; and F5: >11.49%). The mean age of the cohort was 59.1 (SD: 17.1) years, and 58.0% were female. Individuals who experienced greater eGFR variability were older and more likely to be female, with a higher prevalence of comorbidities and medication use.
Table 1.
Baseline characteristics across the fifths of the coefficient of variance of eGFR
| Variables | Overall | Fifths of CVeGFR (%) | ||||
|---|---|---|---|---|---|---|
| F1 | F2 | F3 | F4 | F5 | ||
| (<3.37) | (3.37–5.59) | (5.60–8.09) | (8.10–11.49) | (>11.49) | ||
| Number of Participants | 754,306 | 150,843 | 150,578 | 151,304 | 150,704 | 150,877 |
| Sociodemographic information | ||||||
| Age (yr), mean (SD) | 59.1 (17.1) | 56.0 (15.9) | 57.2 (16.5) | 58.7 (16.7) | 60.3 (17.3) | 63.4 (18.2) |
| Male sex, n (%) | 316,924 (42.0) | 66,424 (44.0) | 62,224 (41.3) | 65,140 (43.1) | 63,192 (41.9) | 59,962 (39.7) |
| Current smoker, n (%) | 78,778 (10.4) | 17,563 (11.6) | 16,234 (10.8) | 15,229 (10.1) | 14,587 (9.7) | 15,165 (10.1) |
| eGFR | ||||||
| Mean eGFR (ml/min per 1.73 m2), mean (SD) | 87.7 (19.3) | 99.6 (15.3) | 93.6 (17.4) | 87.2 (17.8) | 82.3 (18.0) | 75.9 (18.7) |
| eGFR slope (ml/min per 1.73 m2/yr), mean (SD)a | −0.4 (53.6) | −0.3 (14.0) | −0.4 (25.5) | −0.4 (32.2) | −0.6 (35.5) | −0.4 (105.9) |
| Numbers of eGFR measurements six or more, n (%) | 186,170 (24.7) | 18,631 (12.4) | 29,383 (19.5) | 38,986 (25.8) | 46,139 (30.6) | 53,031 (35.1) |
| Baseline comorbid conditions | ||||||
| Cardiovascular disease, n (%) | 77,725 (10.3) | 10,381 (6.9) | 12,145 (8.1) | 14,670 (9.7) | 17,360 (11.5) | 23,169 (15.4) |
| Heart failure, n (%) | 25,673 (3.4) | 1943 (1.3) | 2650 (1.8) | 3654 (2.4) | 5523 (3.7) | 11,903 (7.9) |
| Hypertension, n (%) | 316,377 (41.9) | 54,341 (36.0) | 57,042 (37.9) | 62,083 (41.0) | 66,729 (44.3) | 76,182 (50.5) |
| Diabetes mellitus, n (%) | 103,804 (13.8) | 17,369 (11.5) | 18,388 (12.2) | 20,300 (13.4) | 21,947 (14.6) | 25,800 (17.1) |
| Baseline medication use | ||||||
| Statin, n (%) | 270,271 (35.8) | 46,080 (30.5) | 49,651 (33.0) | 54,444 (36.0) | 57,886 (38.4) | 62,210 (41.2) |
| Diuretics, n (%) | 114,105 (15.1) | 14,905 (9.9) | 17,375 (11.5) | 20,069 (13.3) | 24,519 (16.3) | 37,237 (24.7) |
| RAAS inhibitors, n (%) | 289,336 (38.4) | 47,930 (31.8) | 51,283 (34.1) | 56,427 (37.3) | 61,996 (41.1) | 71,700 (47.5) |
| β-blockers, n (%) | 122,257 (16.2) | 16,415 (10.9) | 19,154 (12.7) | 22,632 (15.0) | 27,133 (18.0) | 36,923 (24.5) |
| CCB, n (%) | 112,427 (14.9) | 17,162 (11.4) | 18,830 (12.5) | 21,170 (14.0) | 24,259 (16.1) | 31,006 (20.6) |
| DPP4 inhibitors, n (%) | 21,896 (2.9) | 3577 (2.4) | 3862 (2.6) | 4096 (2.7) | 4640 (3.1) | 5721 (3.8) |
| Metformin, n (%) | 88,061 (11.7) | 15,291 (10.1) | 15,960 (10.6) | 17,154 (11.3) | 18,325 (12.2) | 21,331 (14.1) |
| SGLT2 inhibitors, n (%) | 6286 (0.8) | 1132 (0.8) | 1177 (0.8) | 1174 (0.8) | 1295 (0.9) | 1508 (1.0) |
| GLP1-RAs, n (%) | 5465 (0.7) | 952 (0.6) | 993 (0.7) | 1044 (0.7) | 1139 (0.8) | 1337 (0.9) |
| Insulins, n (%) | 26,152 (3.5) | 3426 (2.3) | 3902 (2.6) | 4581 (3.0) | 5681 (3.8) | 8561 (5.7) |
| Additional baseline characteristics including missing data b | ||||||
| Systolic BP (mm Hg), mean (SD) | 129.6 (17.4) | 129.6 (17.2) | 129.4 (17.1) | 129.8 (17.0) | 130.0 (17.2) | 130.0 (18.2) |
| Diastolic BP (mm Hg), mean (SD) | 77.1 (10.9) | 77.8 (10.4) | 77.5 (10.4) | 77.4 (10.9) | 77.0 (11.1) | 76.1 (11.3) |
| BMI (kg/m2), mean (SD) | 29.5 (7.9) | 29.3 (7.9) | 29.3 (7.8) | 29.5 (7.7) | 29.6 (8.0) | 29.6 (8.1) |
| Serum LDL (mmol/L), mean (SD) | 2.89 (1.18) | 2.96 (0.95) | 2.93 (0.98) | 2.90 (0.97) | 2.86 (1.76) | 2.76 (1.00) |
| Blood glycated hemoglobin (%), mean (SD) | 7.79 (1.77) | 7.65 (1.29) | 7.63 (1.29) | 7.59 (1.28) | 7.62 (1.30) | 7.77 (2.68) |
BMI, body mass index; CCB, calcium channel blockers; CVeGFR, coefficient of variation; DPP4, dipeptidyl peptidase-4; GLP1-RA, glucagon-like peptide-1 receptor agonist; RAAS, renin-angiotensin-aldosterone system; SGLT2, sodium-glucose cotransporter 2.
P for trend=0.26. All the other P for trends <0.001.
481,335 for systolic BP, 481,351 for diastolic BP, 274,751 for body mass index, 545,863 for serum LDL, and 658,721 for blood glycated hemoglobin are missing.
eGFR Variability and Outcomes
During the 3-year outcome window, a total of 14,239 patients (1.89%) experienced the kidney composite end point. There were 4636 sustained declines in eGFR ≥40% events, 189 kidney failure events, and 9940 deaths. Increments by fifths of CVeGFR were associated with a graded increased risk of the primary outcome (P for trend <0.001; Figure 1). This association remained consistent after covariate adjustment in multivariable Cox regression (HR, 2.17 [95% CI, 2.03 to 2.32] for F5 versus F1; Figure 2). When individual components of the kidney composite outcome were assessed separately, the results for sustained decline in eGFR ≥40% and all-cause mortality were similar, while no significant differences in the risk of kidney failure were observed across the CVeGFR categories, although confidence intervals were wide (Figure 2). Assessment of the association between continuous CVeGFR and the risk of outcomes (primary and secondary) showed a linear association between the levels of CVeGFR and all study outcomes except kidney failure (Figure 3).
Figure 1.
Cumulative incidence of the composite kidney outcome according to fifths of the coefficient of variance of eGFR. The numbers presented below figure mean the number of patients as risk in each time point. P for trend was calculated using the log-rank trend test.
Figure 2.
The association between eGFR variability CV and the risk of study outcomes. Each HR was calculated by a Cox proportional hazards model with F1 of CV of eGFR as reference. *All P for trends <0.001. CI, confidence interval; CV, coefficient of variance; HR, hazard ratio.
Figure 3.
Cubic spline curve of multivariable adjusted HRs for the development of the composite kidney end point and its components as per the fifths of the CV of eGFR. The cubic spline curves of HRs for each outcome were illustrated using the natural cubic spline without intercept and with knots of the 5th, 35th, 65th, and 95th percentile points and reference of the median of CV of eGFR.
Although significant heterogeneity was observed in subgroups by age and sex, the positive log-linear association between the CVeGFR and the primary outcome was evident across all subgroups assessed (P for trend for all subgroups <0.001; Figure 4), where the HRs for Q5 across all subgroups ranged from 1.81 to 2.45. Furthermore, a significant association between CVeGFR and the primary outcome was observed across all subgroups defined by kidney function (i.e., eGFR levels and eGFR slope levels), with a P for trend <0.001 in all subgroups (Figure 5). Overall results remained consistent in all sensitivity analyses (Supplemental Figure 3).
Figure 4.
The association between eGFR variability CV and the risk of the composite kidney outcome across patient subgroups. Each HR was calculated by a Cox proportional hazards model with the lowest fifth of CV of eGFR as reference. P for interactions were calculated by incorporating interaction term between each subgroup variable and CV of eGFR in the relevant statistical models. *All P for trends <0.001.
Figure 5.
Forest plots for a subgroup analysis in accordance with baseline eGFR slope levels.*All P for trends <0.001. The cutoff values for eGFR slope in exposure window were based on its tertile.
Discussion
In this cohort of over 750,000 individuals aged 18 years or older with diverse patient characteristics, we found that increased eGFR variability over 3 years was strongly associated with a higher risk of experiencing adverse kidney outcomes, independent of mean eGFR and eGFR slope, and patient complexity. Overall results were consistently observed across various subgroups and a series of sensitivity analyses which demonstrate the robustness of our findings. Our findings indicate that eGFR variability, which can be easily calculated from routine eGFR measurements, may be a valuable prognostic factor for future kidney outcomes and mortality in clinical settings.
Studies to date on the association between eGFR variability and adverse kidney outcomes have reported mixed results. Tseng et al. reported a significant link between higher eGFR variability and increased risk of CKD progression, even after adjusting for mean and slope of eGFR.21 Another study of ADVANCE trial participants found that increased eGFR variability was linked to a greater risk of incident or worsening diabetic nephropathy12 among individuals with type 2 diabetes. Notably, a recent study by Nishiwaki et al. (n=4224), which used a time-updated model of eGFR variability among patients with CKD, showed that each SD increase in eGFR was associated with a 3.34-fold increase in the risk of cardiovascular disease, a 5.97-fold increase in the risk of ESKD, and a 1.52-fold increase in the risk of all-cause mortality—suggesting that longitudinal variability in eGFR predicts adverse outcomes.22 Our findings align with those reported in these studies, collectively suggesting that eGFR variability serves as a predictor of adverse outcomes.12–15 Conversely, some studies reported no significant association between eGFR variability and kidney-related outcomes.23,24 The inconsistencies across studies may be due in part to limited power to detect significant associations attributable to small sample size. Indeed, this has been acknowledged as a limitation in previous studies. To this end, our findings, based on a large cohort of over 750,000 individuals, provide clinically important data on the prognostic impact of eGFR variability and a range of kidney outcomes across individuals with varying patient profiles.
The mechanisms underlying the association between eGFR variability and adverse outcomes are not yet fully elucidated. Variability in eGFR observed in clinical practice may indicate reduced kidney reserve or structural changes such as renal artery atherosclerosis, as seen with aging kidneys.14,25 It may also reflect impaired renal autoregulation or dynamic fluctuations in renal perfusion due to diuretic use, heart failure, endothelial dysfunction, or renovascular disease. These physiologic disturbances could underlie or exacerbate instability in kidney function.14 It is thus possible that variability in eGFR may be a marker of vascular or metabolic instability which in turn increases the risk of adverse kidney events (rather than variability itself serving as a direct cause). Of note, we observed that eGFR variability remained significantly associated with poor kidney outcomes across groups with differing changes in baseline eGFR levels (increasing, decreasing, or oscillating pattern), suggesting that eGFR variability predicts poor outcomes independent of eGFR slope. Moreover, significant associations were observed across various subgroups defined by age, sex, hypertension, and type 2 diabetes status. There was a significant, but slightly weaker, log-linear association between eGFR variability and the kidney composite end point in patients aged 65 years or older and male patients (with evidence of significant interaction between the subgroups). However, given that the overall estimates suggest a greater risk of adverse kidney outcomes with increasing CVeGFR (e.g., F5 compared with F1), these significant interactions are likely to be due to chance. Taken together, our findings suggest that fluctuation in eGFR, independent of eGFR decline and patient complexity, serves as an important prognostic marker. Importantly, eGFR variability can be easily calculated from routine eGFR measurements, which would support the feasibility of integrating its use in routine clinical settings. This could facilitate earlier identification of those at heightened risk of kidney dysfunction, enabling more timely, targeted interventions that can improve patient outcomes.
The strengths of our study include a large cohort size, sex balance of the present cohort, and broad inclusion criteria, inclusive of various patient profiles, which increase the generalizability of findings. In addition, the use of variability based on eGFR measurements routinely collected in the general practitioner setting suggests its applicability as a prognostic marker in clinical practice. However, our study has some limitations. First, although the study is based on a large national database of general practices across Australia (inclusive of all states and territories), we do not have information from general practices not participating in the MedicineInsight program. However, patients included in the MedicineInsight database are generally representative of the Australian population regarding age and sex.16 Second, we were not able to assess receipt of KRT and cause-specific mortality, especially kidney-related death, as this information is not systematically recorded in MedicineInsight. Nevertheless, we observed consistent results across other adverse kidney-related outcomes supporting the robustness of our findings. Third, it is possible that there may be other factors that influence the association between eGFR variability and outcomes that could not be fully accounted for in the analysis including other laboratory test results at baseline (e.g., albuminuria) and predictable changes in eGFR associated with the initiation or escalation of renoprotective agents (e.g., RAAS inhibitors, SGLT2 inhibitors). Of note, our analysis adjusted for baseline medication prescription (including RAAS inhibitors and SGLT2 inhibitors) and sensitivity analysis accounting for additional clinical and laboratory measurements showed that overall results remained unchanged.
In conclusion, in a large cohort of adults with broad-ranging patient characteristics, greater eGFR variability over 3 years was strongly associated with a higher risk of adverse kidney outcomes and all-cause mortality. Our findings suggest that routine monitoring of eGFR variability, independent of single measurements or other measures of eGFR change such as slope, may provide useful prognostic information for identifying patients at high risk of experiencing adverse kidney outcomes.
Supplementary Material
Acknowledgments
This study is based on data from MedicineInsight (project 2020-004), a national general practice data program developed by NPS MedicineWise and transitioned to the Australian Commission on Safety and Quality in Health Care in 2023. MedicineInsight extracts and collates longitudinal, deidentified, patient health data from the clinical information systems of consenting general practices across Australia. This study was supported by the Renal Division of The George Institute for Global Health, which is supported by the University of New South Wales Scientia Program and a sponsorship provided by Boehringer Ingelheim and Eli Lilly Alliance.
Disclosures
Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/KN9/B323.
Author Contributions
Conceptualization: Min Jun, Takaya Sasaki.
Data curation: Luke Buizen, Takaya Sasaki.
Formal analysis: Takaya Sasaki.
Funding acquisition: Min Jun.
Investigation: Jeffrey T. Ha, Min Jun, Daniel B. Ketema, Takaya Sasaki, Hannah Wallace.
Methodology: Luke Buizen, Katie Harris, Min Jun, Takaya Sasaki, Mark Woodward.
Project administration: Min Jun.
Supervision: Sunil V. Badve, Martin Gallagher, Katie Harris, Meg J. Jardine, Min Jun, Sradha S. Kotwal, Brendon L. Neuen, Paul E. Ronksley, Mark Woodward, Takashi Yokoo.
Validation: Luke Buizen.
Visualization: John Chalmers, Takaya Sasaki.
Writing – original draft: Min Jun, Takaya Sasaki.
Writing – review & editing: Sunil V. Badve, Luke Buizen, John Chalmers, Martin Gallagher, Jeffrey T. Ha, Katie Harris, Meg J. Jardine, Min Jun, Daniel B. Ketema, Sradha S. Kotwal, Brendon L. Neuen, Paul E. Ronksley, Takaya Sasaki, Hannah Wallace, Mark Woodward, Takashi Yokoo.
Funding
M. Jun: University of New South Wales Scientia Program. This work was supported by Boehringer Ingelheim and Eli Lilly Alliance.
Declarative Statements
This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. The need to obtain informed patient consent was waived.
Data Availability Statements
Data belong to a third party, and authors are not authorized to share the data. Identity of Third Party: The current study is based on data from MedicineInsight, a national general practice data collection developed by NPS MedicineWise and managed by the Australian Commission on Safety and Quality in Health Care from 2023. Data from MedicineInsight used in the current study are not publicly available, but researchers may express an interest in MedicineInsight data to the Australian Commission on Safety and Quality in Health Care.
Supplemental Material
This article contains the following supplemental material online at http://links.lww.com/KN9/B324.
Supplemental Figure 1. Scheme of this study design.
Supplemental Figure 2. Flow diagram.
Supplemental Figure 3. Forest plots for sensitivity analyses.
References
- 1.Khan S, Amedia CA, Jr. Economic burden of chronic kidney disease. J Eval Clin Pract. 2008;14(3):422–434. doi: 10.1111/j.1365-2753.2007.00883.x. [DOI] [PubMed] [Google Scholar]
- 2.Australian Institute of Health and Welfare. Projections of the Prevalence of Treated end-stage Kidney Disease in Australia 2012-2020. AIHW; 2014. [Google Scholar]
- 3.Sarnak MJ Levey AS Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American heart association councils on kidney in cardiovascular disease, high blood pressure research, clinical cardiology, and epidemiology and prevention. Circulation. 2003;108(17):2154–2169. doi: 10.1161/01.CIR.0000095676.90936.80. [DOI] [PubMed] [Google Scholar]
- 4.Hanafusa N Abe M Joki N, et al. Annual dialysis data report 2019, JSDT renal data registry. Ren Replace Ther. 2023;9(1):47. doi: 10.1186/s41100-023-00478-z [DOI] [Google Scholar]
- 5.Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004;351(13):1296–1305. doi: 10.1056/NEJMoa041031 [DOI] [PubMed] [Google Scholar]
- 6.Weiner DE Tighiouart H Amin MG, et al. Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: a pooled analysis of community-based studies. J Am Soc Nephrol. 2004;15(5):1307–1315. doi: 10.1097/01.asn.0000123691.46138.e2 [DOI] [PubMed] [Google Scholar]
- 7.Oshima M Jun M Ohkuma T, et al. The relationship between eGFR slope and subsequent risk of vascular outcomes and all-cause mortality in type 2 diabetes: the ADVANCE-ON study. Diabetologia. 2019;62(11):1988–1997. doi: 10.1007/s00125-019-4948-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Inker LA Heerspink HJL Tighiouart H, et al. GFR slope as a surrogate end point for kidney disease progression in clinical trials: a meta-analysis of treatment effects of randomized controlled trials. J Am Soc Nephrol. 2019;30(9):1735–1745. doi: 10.1681/ASN.2019010007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Neuen BL, Weldegiorgis M, Herrington WG, Ohkuma T, Smith M, Woodward M. Changes in GFR and albuminuria in routine clinical practice and the risk of kidney disease progression. Am J Kidney Dis. 2021;78(3):350–360.e1. doi: 10.1053/j.ajkd.2021.02.335 [DOI] [PubMed] [Google Scholar]
- 10.Kidney Disease Improving Global Outcomes KDIGO CKD Work Group. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int. 2024;105(4S):S117–S314. doi: 10.1016/j.kint.2023.10.018 [DOI] [PubMed] [Google Scholar]
- 11.Levey AS, Coresh J. Chronic kidney disease. Lancet. 2012;379(9811):165–180. doi: 10.1016/S0140-6736(11)60178-5 [DOI] [PubMed] [Google Scholar]
- 12.Jun M Harris K Heerspink HJL, et al. Variability in estimated glomerular filtration rate and the risk of major clinical outcomes in diabetes: post hoc analysis from the ADVANCE trial. Diabetes Obes Metab. 2021;23(6):1420–1425. doi: 10.1111/dom.14351 [DOI] [PubMed] [Google Scholar]
- 13.Al-Aly Z, Balasubramanian S, McDonald JR, Scherrer JF, O’Hare AM. Greater variability in kidney function is associated with an increased risk of death. Kidney Int. 2012;82(11):1208–1214. doi: 10.1038/ki.2012.276 [DOI] [PubMed] [Google Scholar]
- 14.Malhotra R Katz R Jotwani V, et al. Estimated GFR variability and risk of cardiovascular events and mortality in SPRINT (systolic blood pressure intervention trial). Am J Kidney Dis. 2021;78(1):48–56. doi: 10.1053/j.ajkd.2020.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fravel MA Ernst ME Webb KL, et al. GFR variability, survival, and cardiovascular events in older adults. Kidney Med. 2023;5(2):100583. doi: 10.1016/j.xkme.2022.100583 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jun M Wick J Neuen BL, et al. The prevalence of CKD in Australian primary care: analysis of a national general practice dataset. Kidney Int Rep. 2024;9(2):312–322. doi: 10.1016/j.ekir.2023.11.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Busingye D Gianacas C Pollack A, et al. Data resource profile: medicineinsight, an Australian national primary health care database. Int J Epidemiol. 2019;48(6):1741–1741h. doi: 10.1093/ije/dyz147 [DOI] [PubMed] [Google Scholar]
- 18.Australian Commission on Safety and Quality in Health Care. Medicineinsight Australia: Australian Commission on Safety and Quality in Health Care; 2024. https://www.safetyandquality.gov.au/our-work/indicators-measurement-and-reporting/medicineinsight. [Google Scholar]
- 19.Levey AS Stevens LA Schmid CH, et al.; CKD-EPI Chronic Kidney Disease Epidemiology Collaboration. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–612. doi: 10.7326/0003-4819-150-9-200905050-00006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.SAS Institute Inc. SAS/STAT® 15.2 User’s Guide: The MI Procedure. SAS Institute Inc; 2020. [Google Scholar]
- 21.Tseng CL Lafrance JP Lu SE, et al. Variability in estimated glomerular filtration rate values is a risk factor in chronic kidney disease progression among patients with diabetes. BMC Nephrol. 2015;16:34. doi: 10.1186/s12882-015-0025-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nishiwaki H Missikpode C Ricardo AC, et al.; CRIC Study Investigators. Time-updated estimated GFR variability is associated with mortality, cardiovascular disease, and end-stage kidney disease in patients with CKD: findings from the CRIC study. Am J Kidney Dis. 2025;85(6):695–703.e1. doi: 10.1053/j.ajkd.2025.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Suzuki A Obi Y Hayashi T, et al. Visit-to-visit variability in estimated glomerular filtration rate predicts hospitalization and death due to cardiovascular events. Clin Exp Nephrol. 2019;23(5):661–668. doi: 10.1007/s10157-019-01695-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Perkins RM, Kirchner HL, Hartle JE, Bucaloiu ID. Estimated glomerular filtration rate variability and risk of end-stage renal disease among patients with stage 3 chronic kidney disease. Clin Nephrol. 2013;80(4):256–262. doi: 10.5414/CN107905 [DOI] [PubMed] [Google Scholar]
- 25.Denic A, Glassock RJ, Rule AD. Structural and functional changes with the aging kidney. Adv Chronic Kidney Dis. 2016;23(1):19–28. doi: 10.1053/j.ackd.2015.08.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. The need to obtain informed patient consent was waived.
Data belong to a third party, and authors are not authorized to share the data. Identity of Third Party: The current study is based on data from MedicineInsight, a national general practice data collection developed by NPS MedicineWise and managed by the Australian Commission on Safety and Quality in Health Care from 2023. Data from MedicineInsight used in the current study are not publicly available, but researchers may express an interest in MedicineInsight data to the Australian Commission on Safety and Quality in Health Care.