Rothenbacher et al. BMC Medicine (2020) 18:300 https://doi.org/10.1186/s12916-020-01776-7 RESEARCH ARTICLE Open Access Contribution of cystatin C- and creatinine- based definitions of chronic kidney disease to cardiovascular risk assessment in 20 population-based and 3 disease cohorts: the BiomarCaRE project Dietrich Rothenbacher1,2*†, Martin Rehm1†, Licia Iacoviello3,4, Simona Costanzo3, Hugh Tunstall-Pedoe5, Jill J. F. Belch6, Stefan Söderberg7, Johan Hultdin8, Veikko Salomaa9, Pekka Jousilahti9, Allan Linneberg10, Susana Sans11, Teresa Padró12, Barbara Thorand13, Christa Meisinger14,15, Frank Kee16, Amy Jayne McKnight17, Tarja Palosaari9, Kari Kuulasmaa9, Christoph Waldeyer18, Tanja Zeller18,19, Stefan Blankenberg18,19, Wolfgang Koenig1,20,21 and on behalf of the BiomarCaRE consortium Abstract Background: Chronic kidney disease has emerged as a strong cardiovascular risk factor, and in many current guidelines, it is already considered as a coronary heart disease (CHD) equivalent. Routinely, creatinine has been used as the main marker of renal function, but recently, cystatin C emerged as a more promising marker. The aim of this study was to assess the comparative cardiovascular and mortality risk of chronic kidney disease (CKD) using cystatin C-based and creatinine-based equations of the estimated glomerular filtration rate (eGFR) in participants of population-based and disease cohorts. Methods: The present study has been conducted within the BiomarCaRE project, with harmonized data from 20 population-based cohorts (n = 76,954) from 6 European countries and 3 cardiovascular disease (CVD) cohorts (n = 4982) from Germany. Cox proportional hazards models were used to assess hazard ratios (HRs) for the various CKD definitions with adverse outcomes and mortality after adjustment for the Systematic COronary Risk Evaluation (SCORE) variables and study center. Main outcome measures were cardiovascular diseases, cardiovascular death, and all-cause mortality. (Continued on next page) * Correspondence: dietrich.rothenbacher@uni-ulm.de † Dietrich Rothenbacher and Martin Rehm contributed equally to this work. 1 Institute of Epidemiology and Medical Biometry, Ulm University, Helmholtzstr. 22, 89081 Ulm, Germany 2 Division of Clinical Epidemiology and Aging Research C070, German Cancer Research Center (DKFZ), Heidelberg, Germany Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Rothenbacher et al. BMC Medicine (2020) 18:300 Page 2 of 13 (Continued from previous page) Results: The overall prevalence of CKD stage 3–5 by creatinine- and cystatin C-based eGFR, respectively, was 3.3% and 7.4% in the population-based cohorts and 13.9% and 14.4% in the disease cohorts. CKD was an important independent risk factor for subsequent CVD events and mortality. For example, in the population- based cohorts, the HR for CVD mortality was 1.72 (95% CI 1.53 to 1.92) with creatinine-based CKD and it was 2.14 (95% CI 1.90 to 2.40) based on cystatin-based CKD compared to participants without CKD. In general, the HRs were higher for cystatin C-based CKD compared to creatinine-based CKD, for all three outcomes and risk increased clearly below the conventional threshold for CKD, also in older adults. Net reclassification indices were larger for a cystatin-C based CKD definition. Differences in HRs (between the two CKD measures) in the disease cohorts were less pronounced than in the population-based cohorts. Conclusion: CKD is an important risk factor for subsequent CVD events and total mortality. However, point estimates of creatinine- and cystatin C-based CKD differed considerably between low- and high-risk populations. Especially in low-risk settings, the use of cystatin C-based CKD may result in more accurate risk estimates and have better prognostic value. Keywords: Cohort study, Chronic kidney disease, Estimated glomerular filtration rate, Adverse outcome, Creatinine, Cystatin C Background also compared the strength of the associations and prog- Chronic kidney disease (CKD) represents a global public nostic values between population-based general and dis- health problem and affects a large proportion of the adult ease cohorts and in specific subgroups (e.g., older adults, population worldwide [1, 2]. CKD has a complicated rela- sex, participants with hypertension and diabetes). tionship with diabetes and hypertension and other associ- ated diseases, and it is an independent risk factor for Methods cardiovascular diseases (CVDs) as well as for all-cause mor- Study populations and study design tality [1]. Outcomes of CKD include not only progression The present study has been conducted within the MOR- to end-stage renal disease (ESRD) but also complications GAM/BiomarCaRE projects, described in detail previously such as hypertension, malnutrition, anemia, bone disease, [10, 11], with harmonized data from 20 population-based and a decreased quality of life [3, 4]. cohorts from 7 European countries and 3 CVD cohorts Also, subclinical CKD has been associated with a large from Germany. The harmonized data variables included burden of disease and mortality [5]. This finding is clin- baseline information on sex, age, smoking status, hyper- ically important because early detection and treatment tension (defined as systolic blood pressure > 140 mmHg or of CKD can prevent or delay the progression of CKD anti-hypertensive medication), and diabetes (defined as and its adverse health outcomes [6]. Meanwhile, it has self-report or antidiabetic medication). In addition, total been demonstrated that the addition of a cystatin C- cholesterol, C-reactive protein (CRP), N-terminal pro-B- based equation improves overall risk classification for type natriuretic peptide (Nt-proBNP), and troponin I death, cardiovascular disease, and end-stage renal disease were included, as well as study outcome information [7]. However, different equations, based on creatinine or (details below). Details of the studies are included in cystatin C measurements, for estimating CKD seem to Additional file 1, Table S1-S3, Box S1 [12–28] and have different performance characteristics in high-risk also can be found elsewhere [10, 11, 29]. and low-risk populations and subgroups such as older adults or patients with diabetes [8, 9]. An analysis of the Laboratory measurements clinical value in specific populations (e.g., for risk predic- In the population-based cohorts of the MORGAM/Bio- tion) such as high-risk and low-risk CVD populations, marCaRE study, creatinine was measured with the older adults, or patients with diabetes would further help kinetic alkaline picrate Jaffe method with the isotope to assess the performance of the various estimated glom- dilution mass spectrometry (IDMS) traceable (NIST erular filtration rate (eGFR) estimation equations. SRM 967) Abbott Architect Assay CREATININE on the The aim of the study was to assess the prevalence of Architect c8000. Cystatin C was measured with the im- CKD using creatinine (Cr)- and cystatin C (cysC)-based munoassay cystatin C on an Abbott Diagnostics ARCH eGFR equations and their comparative risks for cardio- ITECT. All analyses were done at the BiomarCaRE vascular and mortality in participants of cohorts of the central laboratory at Mainz and after the move in the MORGAM/BiomarCaRE consortium representing the Medical University Center Hamburg-Eppendorf in general population and cohorts with manifest CVD. We Hamburg. The intra- and inter-assay coefficients of Rothenbacher et al. BMC Medicine (2020) 18:300 Page 3 of 13 variation (CVs) were measured using samples of coronary event included acute definite or possible myo- medium concentrations (creatinine: medium = 1.39–2.44 cardial infarction or coronary death, unstable angina mg/dL, high = 2.25–7.3 mg/dL; cystatin C: high = 2.95– pectoris, cardiac revascularization, or unclassifiable 4.77 mg/L). The intra-assay CVs for creatinine ranged death. Definition of CVD disease outcomes was based from 0.09 to 5.2% and for cystatin C from 0.78 to 4.0%, re- on data harmonized in the MORGAM project. For spectively. The inter-assay CV for creatinine ranged from population-based cohorts, participants with prevalent 2.3 to 8.1% and for cystatin C from 1.8 to 12.5% for the CVD at baseline were excluded from this outcome ana- measurements in the population-based cohorts. In the dis- lysis (note that these participants are still included in ease cohorts, creatinine was measured locally immediately the supplementary tables). For the disease cohorts, sub- by standardized routine methods in the respective labora- sequent CVD was defined as CVD as the main cause of tories of the participating centers. death and non-fatal stroke or myocardial infarction. (3) In addition, total cholesterol was measured locally by Total mortality as an endpoint was defined as death routine methods and subject to a central quality control in due to any cause during the follow-up time. More the general population cohorts (details under https://www. details of the event classification are provided elsewhere thl.fi/publications/monica/tchol/tcholqa.htm). C-reactive [11, 32] and in the MORGAM manual [29]. The follow- protein was measured on an Abbott Architect c8000 up started at the date of baseline examinations. system and the CRP Vario immunoassay (intra-assay CV Duration of follow-up in each cohort is described in 0.87–3.79%, inter-assay CV 2.57–4.71%, using samples of Additional file 1, Table S1. low concentration (2.8–4.2 mg/L)). N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels were measured on Statistical analysis an ELECSYS 2010 or a Cobas e411 using an electrochemi- The study populations were described with respect to luminescence immunoassay (ECLIA, Roche Diagnostics) baseline sociodemographic and medical characteristics. (intra-assay CV 1.48–7.04%, inter-assay CV 4.74–9.18% The prevalence of CKD was calculated and displayed using samples of low concentration (115.6–166.4 pg/mL)). across specific age categories. Cox proportional hazards models were used to assess hazard ratios (HRs) for the Assessment of chronic kidney disease various CKD definitions with adverse cardiovascular out- Kidney function was assessed by means of eGFR based comes and mortality after adjustment for the SCORE on the latest equations from the Chronic Kidney Disease variables (age, sex, smoking status, systolic blood pres- Epidemiology Collaboration (CKD-EPI), and included sure, total cholesterol) [33, 34] and also study cohort. creatinine and cystatin C [7]. The proportional hazards assumption was checked eGFRcrea according to the Chronic Kidney Disease graphically and based on the multivariable model. Be- Epidemiology Collaboration (CKD-EPIcrea) equation sides overall results, analyses were also done according [30]: eGFR = 141 × min (Cr/k, 1)a × max (Cr/k, to age (< 65, ≥ 65 years), sex, history of hypertension, −1.209 age 1) × (0.993) × (1.018 if female) × (1.159 if black) where and diabetes at baseline. In addition, the area under the k is 0.7 for females and 0.9 for males, a is − 0.329 for females curve (AUC) with 95% CI and the net reclassification and − 0.411 for males, min indicates the minimum of Cr/k improvement (NRI) [35] for events and non-events by or 1, and max indicates the maximum of Cr/k or 1. adding creatinine-based and cystatin C-based CKD to Cystatin C-based eGFR according to CKD-EPI collab- the ESC score variables’ adjusted model were calculated oration [31]: eGFR (CKD-EPIcysC) = 127.7 × (cysC)−1.17 × according to the risk strata of < 1%, 1 to < 5%, 5 to < 10, age−0.13 × (0.91 if female) × (1.06 if black). and ≥ 10% of estimated 10-year risk for the various CKD stage 3–5 was defined as eGFR of less than 60 events. Finally, we used a restricted cubic spline regres- mL/min/1.73 m2. In equations, Cr is given in mg/dL, sion with 3 degrees of freedom for multivariate analysis cysC in mg/L, age in years, weight in kg, and eGFR in within the context of the Cox proportional hazards mL/min/1.73 m2. Only participants with measurements models. SAS version 9.4 (SAS Institute Inc., Cary, NC) of creatinine and cystatin C were included in the final and R version 3.5.1 (R Foundation for Statistical Com- statistical analysis. puting) were used for all analyses. We used our own routines in SAS and R as well as the R package “rms” for Outcome definitions spline regression modeling. We ran most of the analyt- The following outcomes were included in the analysis: ical steps in both programs, underpinned by the 4-eye (1) cardiovascular mortality (fatal myocardial infarction, principle, to assure the quality of the results. fatal stroke, cardiac death, unclassified death) and (2) cardiovascular disease which was defined in the Results population-based studies as the first fatal or non-fatal A total of 20 population-based cohorts with 75,367 coronary heart disease event or cerebral infarction. The participants (median age 50 years, 50.9% men, 4.4% Rothenbacher et al. BMC Medicine (2020) 18:300 Page 4 of 13 diabetes) and with an average follow-up between 2.82 according to age. Notably, especially in the population- and 23.47 years were included in the study. In addition, based cohorts, the distribution in the older population three disease cohorts with 4982 patients with manifest (Fig. S1 panel C) is quite different between the two CVD (median age 63 years, 75.6% men, 18.7% diabetic) eGFR formulas. and an average follow-up time between 0.47 and 9.37 The overall prevalence of CKD stage 3–5 by CKD- years, respectively, were available within the MORGAM/ EPIcrea and CKD-EPIcys eGFR, respectively, was 3.3% and BiomarCaRE consortium for this analysis (for details, see 7.4% in the population-based cohorts and 13.9% and Table 1). Further details of the included cohorts, their 14.4% in the disease cohorts. In males and females, the main baseline characteristics, and details of renal func- prevalence of CKD showed a steep increase with age. In tion are displayed in Additional file 1, Tables S1-S3. In the population-based cohorts, prevalence was higher the population-based cohorts, the incidence of cardio- based on CKD-EPIcysC (Fig. 1), whereas patterns were vascular disease (non-fatal and fatal CVD events com- different based on CKD-EPI in the disease cohorts bined), cardiovascular mortality, and total mortality, (Fig. 2). respectively, was 8.2, 4.2, and 10.9 per 1000 person years, Figure 3 shows the association (HR and 95% CI) of whereas it was 21.2, 6.9, and 17.9 in the diseased co- both CKD-EPI equations with CVD mortality, fatal and horts, respectively (Table 1). non-fatal CVD events, and total mortality after adjust- The distribution of the eGFR as calculated with ment for the ESC score variables for the population- measurements of creatinine and cystatin C is shown in based cohorts in total, and stratified according to age Additional file 1, Fig. S1 for the population-based co- (cut point 65 years), sex, and hypertension and diabetes. horts in total (panel A) and also stratified according to The area under the curve (AUC) and net reclassification age (Fig. S1 panels B and C) and in Additional file 1, Fig. index (NRI) for events and non-events are also provided S2 for the disease cohorts, also in total and stratified (details in Additional file 1, Tables S4 for population- Table 1 Baseline characteristics of the study populations Population-based cohorts Disease cohorts Number of cohorts, n 20 3 Number of subjects, n 75,367 4982 Men, n (%) 38,350 (50.9%) 3766 (75.6%) Age at baseline, years Median (Q1, Q3) 50.0 (41.0, 59.0) 63.0 (54.0, 69.0) Proportion ≥ 65 years 13.1% 42.6% Daily smokers, n (%) 24,077 (31.9%) 892 (17.9%) Diabetes, n (%) 3286 (4.4%) 930 (18.7%) Hypertension, n (%) 30,811 (40.9%) 3514 (70.5%) Body mass index (kg/m2), mean (SD) 26.9 (4.6) 27.6 (4.1) Total cholesterol (mmol/L)* 5.7 (5.0, 6.5) 4.9 (4.1, 5.7) CRP (mg/L)* 1.3 (0.6, 2.9) 1.5 (0.4, 5.1) Nt-proBNP (pg/mL)* 45.9 (24.2, 86.3) 308.0 (116.0, 803.0) Troponin I (ng/L)* 2.3 (1.4, 3.7) 10.4 (4.6, 29.7) eGFR (mL/min/1.73m2)* CKD-EPIcrea 97.6 (85.9, 107.6) 82.7 (68.7, 94.6) CKD-EPIcysC 92.5 (76.4, 111.5) 91.1 (69.1, 116.2) CKD stage 3+, n (%) CKD-EPIcrea 2450 (3.3%) 691 (13.9%) CKD-EPIcysC 5562 (7.4%) 719 (14.4%) Endpoints (n, incidence rate per 1000 person years (95% CI)) Cardiovascular disease 6850, 8.2 (95% CI 8.0–8.4) 371, 21.2 (95% CI 19.2–23.5) Cardiovascular mortality 3796, 4.2 (95% CI 4.1–4.3) 132, 6.9 (95% CI 5.8–8.2) Total mortality 9840, 10.9 (95% CI 10.6–11.1) 343, 17.9 (95% CI 16.1–19.9) *Median (interquartile range, Q1, Q3) Rothenbacher et al. BMC Medicine (2020) 18:300 Page 5 of 13 Fig. 1 Prevalence of CKD based on different eGFR estimating equations in the population-based cohorts (histograms represent prevalence in % and bars 95% CIs) Fig. 2 Prevalence of CKD based on different eGFR estimating equations in the disease cohorts (histograms represent prevalence in % and bars 95% CIs) Rothenbacher et al. BMC Medicine (2020) 18:300 Page 6 of 13 Fig. 3 Association of CKD with various endpoints in population-based cohorts (squares represent HR and 95% CIs) Rothenbacher et al. BMC Medicine (2020) 18:300 Page 7 of 13 based cohorts). For example, the HR for CVD mortality well as high-risk populations. However, especially in the was 1.72 (95% CI 1.53 to 1.92) with CKD-EPIcrea and it (low-risk) population-based cohorts, it was evident that was 2.14 (95% CI 1.90 to 2.40) based on CKD-EPIcysC analysis of the creatinine- and cystatin C-based eGFRs after adjustment for covariates. In general, the HRs were and their consequences differed considerably. Cystatin higher for cystatin C-based eGFR compared to C-defined CKD showed a higher prevalence especially in creatinine-based eGFR for all three outcomes, especially older adults, and cystatin C-based eGFR was associated evident for total mortality. Mainly, the NRIne was always with higher CVD risk estimates. It seemed also to be as- larger for CKD-EPIcysC for all three outcomes. It was sociated with better risk classification, especially evident notably high for the age strata ≥ 65 years for cardiovas- for mortality in older adults. Notably, this difference was cular mortality and total mortality (4.2% and 6.5%, not obvious in the relatively older, high-risk disease respectively) and highest for the stratum diabetes for cohorts. Furthermore, we found evidence that the con- total mortality (NRIne 9.4%). ventional threshold of CKD (60 mL/min./1.73m2) seems Figure 4 shows the association of CKD with the valid also for participants aged 65 years and older, various outcomes in the diseased cohorts (details in although the increase in the estimated risk was less steep Additional file 1, Table S5 for disease cohorts). For ex- compared to the younger participants. Notably, higher ample, the HR for CVD mortality was 3.33 (95% CI 2.14 eGFR-creatinine (≥ 90 mL/min/1.73 m2) was associated to 5.19) with CKD-EPIcrea and 3.20 (95% CI 2.04 to 5.01) with higher hazard ratio of adverse events, while higher based on CKD-EPIcysC after adjustment for covariates. eGFR-cysC was not. Compared to a creatinine-based Especially for mortality, HRs were more often higher for definition of CKD, the cystatin C-based definition of CKD-EPIcrea. For both CKD variables in disease cohorts, CKD seemed to have a higher specificity, a measure espe- NRIne was highest in the strata female sex, but the pat- cially important in populations with low prevalence of terns were less clear than those in the population-based CKD and therefore resulted in many fewer false positives. cohorts. Details of reclassification in both types of cohorts are Prevalence of CKD and differences of eGFRs shown in Additional file 1, Table S6. In the population- Based on a recent review, the prevalence of CKD stage based cohorts, a large proportion (47.4%) of the CKD-crea- 3–5 varied considerably in studies from 19 general defined cases were re-graded by the CKD-cysC-defined populations from 13 European countries and was be- CKD to no CKD. The pattern, in general, was similar to the tween 1.0 and 5.9% in the adult population. Many fac- diseased cohorts, although the much lower numbers of the tors including measurement issues certainly contribute disease cohorts have to be considered. to this variation besides differences in comorbidity Figure 5 shows adjusted HR for creatinine-based eGFR [36]. These numbers are in line with our estimates in (left side) and for cystatin C-based eGFR (right side) for the population-based cohorts. Estimates of cystatin C total mortality in the population-based cohorts overall CKD were in general slightly higher, but striking dif- (panel A) and after stratification for age (panel B age up ferences occurred especially in the older adults. In the to 64 years, panel C age 65+ years). Although the in- population-based cohorts, cystatin C-based CKD crease of the HR with decreasing eGFR is steeper with prevalence was much higher in older adults, whereas the cystatin C-based eGFR, this difference is mainly in the diseased cohorts CKD-EPIcrea resulted in a higher driven by the trend in the population up to 65 years. CKD prevalence in participants aged 65 years or older, Furthermore, a clear increase of the HRs is evident especially in females. below 60 mL/min/1.73m2, even in the age strata 65+ Cystatin C is more sensitive than creatinine especially years, pointing to the validity of the current threshold in the detection of early kidney dysfunction among for CKD stage 3+. Additional file 1, Fig. S7 shows the re- various patient groups such as diabetes, in sarcopenia, sults for mortality in the disease cohorts showing similar and also in the older adults. Potential confounders for patterns for creatinine- and cystatin C-based eGFR. creatinine-based eGFR are muscle mass, dehydration, Additional file 1, Fig. S8 shows the AUC for cystatin C- dietary factors, and other comorbid diseases. As sug- and creatinine-based eGFR with consistently higher gested in other studies that have included older adults, AUC values for the cystatin C-based eGFR definition in the creatinine-based definition of CKD in high-risk both cohorts in all strata. groups may result in a higher prevalence estimate be- cause of the association of creatinine levels with many Discussion comorbid conditions such as muscle mass, frailty, or This large cohort study conducted within the framework diabetes [37, 38]. Cystatin C-based eGFR is less influ- of the MORGAM/BiomarCaRE consortium clearly dem- enced by age or ethnicity, but other factors such as onstrated that CKD is an important risk factor for subse- obesity, inflammation, and smoking as well as intake of quent CVD events and total mortality, both in low- as glucocorticoids may affect serum values. Rothenbacher et al. BMC Medicine (2020) 18:300 Page 8 of 13 Fig. 4 Association of CKD with various endpoints in disease cohorts (squares represent HR and 95% CIs) Rothenbacher et al. BMC Medicine (2020) 18:300 Page 9 of 13 Fig. 5 Spline regression representing creatinine (left side)- and cystatin C (right side)-based eGFR associated with hazard ratio (HR) for mortality after adjustment for SCORE variables and study cohort for all and after stratification for age in the population-based cohorts (for details, see the “Methods” section) Differences in risk estimation for various outcomes plots that higher creatinine-based eGFR again showed Our results show that in middle-aged low-risk popula- an adverse association with the outcomes has been tions, cystatin C-based CKD classification had a stronger shown previously [7, 42]. It probably reflects the associa- association with risk of CVD morbidity and especially tions with comorbid conditions such as lower muscle mortality, as seen in other studies [7, 39, 40]. CKD mass, frailty, or diabetes and implies that creatinine- should be considered as an equally relevant risk factor based eGFR, especially in elderly, has severe limitations, for mortality and end-stage renal disease for patients particularly when it exceeds 100 mL/min/1.73m2 [43]. with, as well as those without, hypertension. Notably, Thus, it should be investigated further as it was also very associations were even stronger in patients without obvious in subjects aged 64 or younger in our analysis. hypertension [41]. For patients with diabetes, the relative Furthermore, especially in younger and low-risk risks were much the same as for non-diabetic persons, populations, the cystatin C-based eGFR may deliver pointing to the importance of CKD for adverse out- more valid results [7] compared to creatinine-based comes [34]. As suggested in other studies, in patients eGFR, an observation in line with our data. In high- with diabetes, the use of a cystatin C-based CKD defin- risk populations or older adults, this may be different ition offered better clinical utility for risk prediction than [44]. However, evidence from studies including older creatinine-based equations [8]. In addition, a systematic adults is so far inconsistent. A study including 1639 review including 23 studies came to the conclusion that British men aged 71 to 92 years [45], a study includ- cystatin C-based eGFR represents measured GFR well in ing 1741 participants with CKD and a mean age of patients with diabetes [9]. The observation in our spline 73 years [44], and a study conducted in older adult Rothenbacher et al. BMC Medicine (2020) 18:300 Page 10 of 13 women (mean age 75.2 years) [46] showed no benefit Furthermore, we could not include measurements of pro- from using cystatin C-based eGFR equations when tein or albumin in urine as they were not available in a compared to creatinine-based ones. However, the standardized manner in all included studies. Proteinuria large study of Shlipak and colleagues, including 11 or albuminuria is indeed an import prognostic factor and cohorts of the general population (with a mean age of used as an indicator for renal damage especially in CKD 60 years) and 5 cohorts with CKD (mean age 55 stages 1 and 2 and explains much of the risk increase years), found that cystatin C-based eGFR could better within these stages. However, we dichotomized CKD and categorize risk than creatinine-based eGFR, especially summarized all stages 3 and above versus the rest, accord- around the threshold of 60 mL/min/1.73m2 [7], in line ing to the clinically used threshold. with recommendations of KDIGO [5]. However, the younger age in that study has to be considered. Conclusions Implications CKD is an important risk factor for subsequent CVD All methods to estimate GFR are associated with system- events and total mortality. However, point estimates of atic and random error; however, on a positive note, they creatinine-based definition and cystatin C-based CKD have less time-to-time variability than measured GFR differed considerably between low- and high-risk popu- [47]. Furthermore, it is important to consider the under- lations. Especially in low-risk settings, the use of cystatin lying risk profile of the population on the impact of the C-based eGFR may result in more accurate risk esti- relative risk estimates. Especially in low-risk settings, mates and have better prognostic value compared to cystatin C-based eGFR seems more accurate and delivers creatinine-based CKD definition. Therefore, the clinical fewer false positives. The lower risk estimates associated utility of both equations in different risk populations and with CKD in the diseased cohorts probably reflect the risk groups has to be considered and should be evalu- higher prevalence of comorbid conditions and the higher ated further. baseline risk within this population. As suggested by some authors, both CKD-EPI equations may have simi- Supplementary information lar accuracy but show bias in opposite directions and the Supplementary information accompanies this paper at https://doi.org/10. combination of both may deliver the best results in older 1186/s12916-020-01776-7. adults [48, 49]. However, currently, measurement of cystatin C is more expensive than creatinine. Therefore, Additional file 1: The Additional file 1 contains additional descriptive and analysis results and information about the included studies. Table further research should focus on the cost-effectiveness S1. List of included population-based and disease-cohorts. Table S2. of cystatin C-based measurements of CKD, especially in Main baseline characteristics of population-based and disease-cohorts in- low-risk settings and older populations with the aim of cluded. Table S3. Renal function at baseline of population-based and disease-cohorts included. Figure S1. Distribution of the estimated glom- finding ways to reduce the costs of measurement within erular filtration rate (eGFR) based on creatinine and cystatin C in the routine medical care. population based cohorts. Figure S2. Distribution of the estimated glomerular filtration rate (eGFR) based on creatinine and cystatin C in the disease cohorts. Table S4. Association of CKD with various endpoints Strength and limitations (HR and 95% CI) including area-under the curve (AUC with 95% CI) and We included a large number of general population- net-reclassification index (NRI with 95% CI) in population-based cohorts. based studies which had been assembled within the Table S5. Association of CKD with various endpoints (HR and 95% CI) in- cluding area-under the curve (AUC with 95% CI) and net-reclassification MORGAM/BiomarCaRE consortium, used harmonized index (NRI with 95% CI) in disease cohorts. Table S6. Agreement of re- data, and could rely on centralized measurements for classification by the CKD-EPIcysC based definition compared to CKD- creatinine and cystatin C. However, unfortunately, cysta- EPIcrea based definition in the population-based cohorts (Panel A) and in the disease cohorts (Panel B). Figure S7. Spline regression representing tin C measurements had not been standardized to the creatinine (left side) and cystatin C (right side) based eGFR associated global WHO reference material, which however should with hazard ratio’s (HR) for mortality after adjustment for SCORE-variables not affect internal validity of results, but external compar- and study center for all and after stratification for age in the disease co- horts. Box S1. Further Description of Study Cohorts. ability. We also had cohorts of patients with prevalent CVD available, although the numbers of included partici- pants were much lower compared to the population-based Abbreviations studies and they only came from Germany. Unfortunately, AUC: Area under the ROC curve; BMI: Body mass index; CHD: Coronary heart the laboratory measures of the disease cohorts had been disease; CI: Confidence interval; Cr: Creatinine; cysC: Cystatin C; CKD: Chronic kidney disease; CKD-EPI: Chronic Kidney Disease Epidemiology Collaboration; measured at each study center. Furthermore, we relied on CV: Coefficient of variation; CVD: Cardiovascular disease; eGFR: Estimated one single measurement to define CKD, and therefore, glomerular filtration rate; ESRD: End-stage renal disease; hs-cTnT: High- some of the participants would not have been classified as sensitivity cardiac troponin T; hs-cTnI: High-sensitivity cardiac troponin I; hs- CRP: High-sensitivity C-reactive protein; HR: Hazard ratio; NT-proBNP: N- having CKD if measured twice and if a time period of 6 terminal pro-B-type natriuretic peptide; NRI: Net reclassification improvement; months had been applied to define the chronicity of CKD. ROC: Receiver-operating characteristic Rothenbacher et al. BMC Medicine (2020) 18:300 Page 11 of 13 Acknowledgements Medicine Unit, Institute of Cardiovascular Disease, University of Dundee, Not applicable Dundee, UK. 7Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden. 8Department of Medical Biosciences, Clinical Authors’ contributions Chemistry, Umeå University, Umeå, Sweden. 9Finnish Institute for Health and All authors were involved in the design of the study. DR, MR, and WK did Welfare, Helsinki, Finland. 10Center for Clinical Research and Prevention, the statistical analysis and wrote the first draft. All authors contributed to the Bispebjerg and Frederiksberg Hospital, The Capital Region, Copenhagen, further drafts. The corresponding author attests that all listed authors meet Denmark. 11Catalan Department of Health, 08005 Barcelona, Spain. 12 the authorship criteria. DR is the guarantor. All authors read and approved Cardiovascular ICCC-Program, Research Institute Hospital de la Santa Creu i the final manuscript. Sant Pau, IIB-Sant Pau, Barcelona, Spain. 13Institute of Epidemiology, Helmholtz Zentrum München, German Research Center for Environmental Funding Health, Neuherberg, Germany. 14Independent Research Group Clinical This work was supported by Seventh Framework Programme collaborative Epidemiology, Helmholtz Zentrum München, German Research Center for project [grant agreement no. HEALTH-F2-2011-278913]. The MORGAM Environmental Health, Neuherberg, Germany. 15 project has been funded through several grants from the EU, the latest ones Ludwig-Maximilians-Universität München, Chair of Epidemiology at being from the Seventh Framework Programme (FP7/2007-2013) under UNIKA-T Augsburg, Augsburg, Germany. 16Queen’s University of Belfast, UK grant agreement no. HEALTH-F2-2011-278913, BiomarCaRE, FP7/2007-2013 Clinical Research Collaboration Centre of Excellence for Public Health, Belfast, [HEALTH-F4-2007-201413, ENGAGE and HEALTH-F3-2010-242244, CHANCES], UK. 17Centre for Public Health, School of Medicine, Dentistry and Biomedical and the Medical Research Council London [G0601463, No 80983: Serum Sciences, Queen’s University of Belfast, Belfast, UK. 18Clinic for General and Biomarkers in the MORGAM Populations]. This has supported central Interventional Cardiology, University Heart Center Hamburg, Hamburg, coordination, workshops, and part of the activities of the MORGAM Data Germany. 19German Center for Cardiovascular Research (DZHK e.V.), partner Centre, at THL in Helsinki, Finland, and the MORGAM Biomarker Laboratory site Hamburg, Lübeck, Kiel, Hamburg, Germany. 20Deutsches Herzzentrum at Johannes Gutenberg University in Mainz, Germany. MORGAM Participating München, Technische Universität München, Munich, Germany. 21DZHK Centres are funded by regional and national governments, research councils, (German Centre for Cardiovascular Research), partner site Munich Heart charities, and other local sources. Dr. Zeller was supported by the German Alliance, Munich, Germany. Center of Cardiovascular Research (DZHK e.V.) under the grant number 81Z1710101. Stefan Blankenberg reports grants from Abbott Diagnostics, Received: 11 May 2020 Accepted: 3 September 2020 during the conduct of the study. Open access funding provided by Projekt DEAL. Availability of data and materials References Due to ethical restrictions, the data cannot be made publicly available but 1. Chronic Kidney Disease Prognosis Consortium, Matsushita K, van der Velde may be available upon request. The request should be directed to Prof. M, Astor BC, Woodward M, Levey AS, de Jong PE, Coresh J, Gansevoort RT. Rothenbacher (dietrich.rothenbacher@uni-ulm.de). Association of estimated glomerular filtration rate and albuminuria with all- cause and cardiovascular mortality in general population cohorts: a Ethics approval and consent to participate collaborative meta-analysis. Lancet Lond Engl. 2010;375:2073–81. All studies were carried out according to the Declaration of Helsinki. Each of 2. Zhang Q-L, Rothenbacher D. Prevalence of chronic kidney disease in the 23 included contributing studies had previously obtained ethics approval population-based studies: systematic review. BMC Public Health. 2008;8:117. from their respective institutional review boards, and all participants 3. Astor BC, Matsushita K, Gansevoort RT, Velde M Van der, Woodward M, provided informed consent. Details of each study can also be found in Levey AS, Jong PE de, Coresh J, Chronic Kidney Disease Prognosis Additional file 1 Box S1: Further Description of Study Cohorts. Consortium. Lower estimated glomerular filtration rate and higher albuminuria are associated with mortality and end-stage renal disease. A Consent for publication collaborative meta-analysis of kidney disease population cohorts. Kidney Int Not applicable 2011;79:1331–1340. 4. Nitsch D, Mann AG, Bulpitt C, Roderick PJ, Fletcher A. Impairment of kidney Competing interests function and reduced quality-of-life in older people: a cross-sectional study. All authors have completed the ICMJE uniform disclosure form at www. Age Ageing. 2011;40:381–7. icmje.org/coi_disclosure.pdf and declare the following: no support from any 5. Levey AS, de Jong PE, Coresh J, El Nahas M, Astor BC, Matsushita K, organization for the submitted work, no financial relationships with any Gansevoort RT, Kasiske BL, Eckardt K-U. The definition, classification, and organizations that might have an interest in the submitted work in the prognosis of chronic kidney disease: a KDIGO Controversies Conference previous 3 years, and no other relationships or activities that could appear to report. Kidney Int. 2011;80:17–28. have influenced the submitted work. 6. Chen TK, Knicely DH, Grams ME. Chronic kidney disease diagnosis and Outside the submitted work, S.B. reports personal fees from Lumira management: a review. JAMA. 2019;322:1294–304. Diagnostics, ThermoFisher, Siemens, Abbott Diagnostics, and Roche 7. Shlipak MG, Matsushita K, Ärnlöv J, Inker LA, Katz R, Polkinghorne KR, Diagnostics and grants from ThermoFisher, Siemens, and Abbott Diagnostics. Rothenbacher D, Sarnak MJ, Astor BC, Coresh J, Levey AS, Gansevoort RT, S.S. reports personal fees from Actelion Ltd.; V.S. reports personal fees from Prognosis Consortium CKD. Cystatin C versus creatinine in determining risk Novo Nordisk and Sanofi and grants from Bayer AG. W.K. reports personal based on kidney function. N Engl J Med. 2013;369:932–43. fees from AstraZeneca, Novartis, Pfizer, The Medicines Company, DalCor, 8. Schöttker B, Herder C, Müller H, Brenner H, Rothenbacher D. Clinical utility Kowa, Amgen, Corvidia, Berlin-Chemie, Sanofi, and Daichii-Sankyo, in addition of creatinine- and cystatin C-based definition of renal function for risk to grants and non-financial support from Beckmann, Singulex, Abbott, and prediction of primary cardiovascular events in patients with diabetes. Roche Diagnostics. D.R., M.R., J.J.F.B., J.H., P.J., A.L., S.Sa., T.Pad., B.T., C.M., F.K., Diabetes Care. 2012;35:879–86. A.J.M., H.T.P., L.I., S.C., C.W., and T.Pal. declare no conflict of interest. 9. Cheuiche AV, Queiroz M, Azeredo-da-Silva ALF, Silveiro SP. Performance of cystatin C-based equations for estimation of glomerular filtration rate in Author details diabetes patients: a prisma-compliant systematic review and meta-analysis. 1 Sci Rep. 2019;9:1418. Institute of Epidemiology and Medical Biometry, Ulm University, Helmholtzstr. 22, 89081 Ulm, Germany. 2Division of Clinical Epidemiology 10. Zeller T, Hughes M, Tuovinen T, Schillert A, Conrads-Frank A, den Ruijter H, and Aging Research C070, German Cancer Research Center (DKFZ), Schnabel RB, Kee F, Salomaa V, Siebert U, Thorand B, Ziegler A, Breek H, Heidelberg, Germany. 3Department of Epidemiology and Prevention, IRCCS Pasterkamp G, Kuulasmaa K, Koenig W, Blankenberg S. BiomarCaRE: Neuromed, Pozzilli, Italy. 4Research Center in Epidemiology and Preventive rationale and design of the European BiomarCaRE project including 300,000 Medicine (EPIMED), Department of Medicine and Surgery, University of participants from 13 European countries. Eur J Epidemiol. 2014;29:777–90. Insubria, Varese, Italy. 5Cardiovascular Epidemiology Unit, Institute of 11. Evans A, Salomaa V, Kulathinal S, Asplund K, Cambien F, Ferrario M, Perola Cardiovascular Research, University of Dundee, Dundee, UK. 6Vascular M, Peltonen L, Shields D, Tunstall-Pedoe H, Kuulasmaa K, MORGAM Project. Rothenbacher et al. BMC Medicine (2020) 18:300 Page 12 of 13 MORGAM (an international pooling of cardiovascular cohorts). Int J 30. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI, Kusek Epidemiol. 2005;34:21–7. JW, Eggers P, Van Lente F, Greene T, Coresh J, CKD-EPI (Chronic Kidney 12. Löwel H, Lewis M, Hörmann A, Keil U. Case finding, data quality aspects and Disease Epidemiology Collaboration). A new equation to estimate comparability of myocardial infarction registers: results of a south German glomerular filtration rate. Ann Intern Med. 2009;150:604–12. register study. J Clin Epidemiol. 1991;44:249–60. 31. Inker LA, Eckfeldt J, Levey AS, Leiendecker-Foster C, Rynders G, Manzi J, 13. Meisinger C, Strassburger K, Heier M, Thorand B, Baumeister SE, Giani G, Waheed S, Coresh J. Expressing the CKD-EPI (Chronic Kidney Disease Rathmann W. Prevalence of undiagnosed diabetes and impaired glucose Epidemiology Collaboration) cystatin C equations for estimating GFR with regulation in 35-59-year-old individuals in Southern Germany: the KORA F4 standardized serum cystatin C values. Am J Kidney Dis Off J Natl Kidney Study. Diabet Med J Br Diabet Assoc. 2010;27:360–2. Found. 2011;58:682–4. 14. Rodes A, Sans S, Balaña LL, Paluzie G, Aguilera R, Balaguer-Vintro I. 32. Waldeyer C, Makarova N, Zeller T, Schnabel RB, Brunner FJ, Jørgensen T, Recruitment methods and differences in early, late and non-respondents in Linneberg A, Niiranen T, Salomaa V, Jousilahti P, Yarnell J, Ferrario MM, the first MONICA-Catalonia population survey. Rev Epidemiol Sante Veronesi G, Brambilla P, Signorini SG, Iacoviello L, Costanzo S, Giampaoli S, Publique. 1990;38:447–53. Palmieri L, Meisinger C, Thorand B, Kee F, Koenig W, Ojeda F, Kontto J, 15. Sans S, Paluzie G, Balañá L, Puig T, Balaguer-Vintró I. Trends in prevalence, Landmesser U, Kuulasmaa K, Blankenberg S. Lipoprotein(a) and the risk of awareness, treatment and control of arterial hypertension between 1986 cardiovascular disease in the European population: results from the and 1996: the MONICA-Catalonia study. Med Clin (Barc). 2001;117:246–53. BiomarCaRE consortium. Eur Heart J. 2017;38:2490–8. 16. Sans S, Puigdefábregas A, Paluzie G, Monterde D, Balaguer-Vintró I. 33. Conroy RM, Pyörälä K, Fitzgerald AP, Sans S, Menotti A, De Backer G, De Increasing trends of acute myocardial infarction in Spain: the MONICA- Bacquer D, Ducimetière P, Jousilahti P, Keil U, Njølstad I, Oganov RG, Catalonia Study. Eur Heart J. 2005;26:505–15. Thomsen T, Tunstall-Pedoe H, Tverdal A, Wedel H, Whincup P, Wilhelmsen L, 17. Santimone I, Di Castelnuovo A, De Curtis A, Spinelli M, Cugino D, Gianfagna Graham IM, SCORE project group. Estimation of ten-year risk of fatal F, Zito F, Donati MB, Cerletti C, de Gaetano G, Iacoviello L, Project cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003;24: Investigators MOLI-SANI. White blood cell count, sex and age are major 987–1003. determinants of heterogeneity of platelet indices in an adult general 34. Fox CS, Matsushita K, Woodward M, Bilo HJG, Chalmers J, Heerspink population: results from the MOLI-SANI project. Haematologica. 2011;96: HJL, Lee BJ, Perkins RM, Rossing P, Sairenchi T, Tonelli M, Vassalotti JA, 1180–8. Yamagishi K, Coresh J, Jong PE de, Wen C-P, Nelson RG, Chronic 18. Eriksson M, Holmgren L, Janlert U, Jansson J-H, Lundblad D, Stegmayr B, Kidney Disease Prognosis Consortium. Associations of kidney disease Söderberg S, Eliasson M. Large improvements in major cardiovascular risk measures with mortality and end-stage renal disease in individuals factors in the population of northern Sweden: the MONICA study 1986- with and without diabetes: a meta-analysis. Lancet Lond Engl 2012; 2009. J Intern Med. 2011;269:219–31. 380:1662–1673. 19. Köster M, Asplund K, Johansson Å, Stegmayr B. Refinement of Swedish 35. Pencina MJ, D’Agostino RB, Pencina KM, Janssens ACJW, Greenland P. administrative registers to monitor stroke events on the national level. Interpreting incremental value of markers added to risk prediction models. Neuroepidemiology. 2013;40:240–6. Am J Epidemiol. 2012;176:473–81. 20. Rolandsson O, Norberg M, Nyström L, Söderberg S, Svensson M, Lindahl B, 36. Brück K, Stel VS, Gambaro G, Hallan S, Völzke H, Ärnlöv J, Kastarinen M, Weinehall L. How to diagnose and classify diabetes in primary health care: Guessous I, Vinhas J, Stengel B, Brenner H, Chudek J, Romundstad S, lessons learned from the Diabetes Register in Northern Sweden Tomson C, Gonzalez AO, Bello AK, Ferrieres J, Palmieri L, Browne G, (DiabNorth). Scand J Prim Health Care. 2012;30:81–7. Capuano V, Van Biesen W, Zoccali C, Gansevoort R, Navis G, Rothenbacher 21. Yarnell JW. The PRIME study: classical risk factors do not explain the D, Ferraro PM, Nitsch D, Wanner C, Jager KJ, European CKD Burden severalfold differences in risk of coronary heart disease between France and Consortium. CKD prevalence varies across the European general population. Northern Ireland. Prospective epidemiological study of myocardial J Am Soc Nephrol JASN. 2016;27:2135–47. infarction. QJM Mon J Assoc Physicians. 1998;91:667–76. 37. Rothenbacher D, Klenk J, Denkinger M, Karakas M, Nikolaus T, Peter R, 22. Kendrick S, Clarke J. The Scottish record linkage system. Health Bull (Edinb). Koenig W, ActiFE Study Group. Prevalence and determinants of chronic 1993;51:72–9. kidney disease in community-dwelling elderly by various estimating 23. Tunstall-Pedoe H, Woodward M, Tavendale R, A’Brook R, McCluskey MK. equations. BMC Public Health. 2012;12:343. Comparison of the prediction by 27 different factors of coronary heart 38. Canales MT, Blackwell T, Ishani A, Taylor BC, Hart A, Barrett-Connor E, Lewis disease and death in men and women of the Scottish Heart Health Study: C, Beyth RJ, Stone K, Ensrud KE, Outcomes of Sleep Disorders In Older Men cohort study. BMJ. 1997;315:722–9. (Mros Sleep) Study. Estimated GFR and mortality in older men: are all eGFR 24. Blankenberg S, Rupprecht HJ, Bickel C, Torzewski M, Hafner G, Tiret L, Smieja formulae equal. Am J Nephrol. 2016;43:325–33. M, Cambien F, Meyer J, Lackner KJ, Investigators AG. Glutathione peroxidase 39. Svensson-Färbom P, Ohlson Andersson M, Almgren P, Hedblad B, Engström 1 activity and cardiovascular events in patients with coronary artery disease. G, Persson M, Christensson A, Melander O. Cystatin C identifies N Engl J Med. 2003;349:1605–13. cardiovascular risk better than creatinine-based estimates of glomerular 25. Rothenbacher D, Koenig W, Brenner H. Comparison of N-terminal pro-B- filtration in middle-aged individuals without a history of cardiovascular natriuretic peptide, C-reactive protein, and creatinine clearance for disease. J Intern Med. 2014;275:506–21. prognosis in patients with known coronary heart disease. Arch Intern Med. 40. Lees JS, Welsh CE, Celis-Morales CA, Mackay D, Lewsey J, Gray SR, Lyall DM, 2006;166:2455–60. Cleland JG, Gill JMR, Jhund PS, Pell J, Sattar N, Welsh P, Mark PB. Glomerular 26. Breitling LP, Salzmann K, Rothenbacher D, Burwinkel B, Brenner H. Smoking, filtration rate by differing measures, albuminuria and prediction of F2RL3 methylation, and prognosis in stable coronary heart disease. Eur cardiovascular disease, mortality and end-stage kidney disease. Nat Med. Heart J. 2012;33:2841–8. 2019;25:1753–60. 27. Keller T, Zeller T, Ojeda F, Tzikas S, Lillpopp L, Sinning C, Wild P, Genth-Zotz 41. Mahmoodi BK, Matsushita K, Woodward M, Blankestijn PJ, Cirillo M, Ohkubo S, Warnholtz A, Giannitsis E, Möckel M, Bickel C, Peetz D, Lackner K, Baldus S, T, Rossing P, Sarnak MJ, Stengel B, Yamagishi K, Yamashita K, Zhang L, Münzel T, Blankenberg S. Serial changes in highly sensitive troponin I assay Coresh J, Jong PE de, Astor BC, Chronic Kidney Disease Prognosis and early diagnosis of myocardial infarction. JAMA. 2011;306:2684–93. Consortium. Associations of kidney disease measures with mortality and 28. Keller T, Zeller T, Peetz D, Tzikas S, Roth A, Czyz E, Bickel C, Baldus S, end-stage renal disease in individuals with and without hypertension: a Warnholtz A, Fröhlich M, Sinning CR, Eleftheriadis MS, Wild PS, Schnabel RB, meta-analysis. Lancet Lond Engl 2012;380:1649–1661. Lubos E, Jachmann N, Genth-Zotz S, Post F, Nicaud V, Tiret L, Lackner KJ, 42. Shastri S, Katz R, Rifkin DE, Fried LF, Odden MC, Peralta CA, Chonchol M, Münzel TF, Blankenberg S. Sensitive troponin I assay in early diagnosis of Siscovick D, Shlipak MG, Newman AB, Sarnak MJ. Kidney function and acute myocardial infarction. N Engl J Med. 2009;361:868–77. mortality in octogenarians: Cardiovascular Health Study All Stars. J Am 29. Kulathinal S, Niemelä M, Niiranen T, Saarela O, Palosaari T, Tapanainen H, Geriatr Soc. 2012;60:1201–7. Kuulasmaa K, Participating Centres for the MProject contributors from. 43. Canney M, Sexton DJ, O’Connell MDL, Kenny RA, Little MA, O’Seaghdha CM. Description of MORGAM Cohorts. MORGAM Project e-publications. https:// Kidney function estimated from cystatin C, but not creatinine, is related to www.thl.fi/publications/morgam/index.html. Accessed 14 Sep 2020. objective tests of physical performance in community-dwelling older adults. J Gerontol A Biol Sci Med Sci. 2017;72:1554–60. Rothenbacher et al. BMC Medicine (2020) 18:300 Page 13 of 13 44. Shardlow A, McIntyre NJ, Fraser SDS, Roderick P, Raftery J, Fluck RJ, McIntyre CW, Taal MW. The clinical utility and cost impact of cystatin C measurement in the diagnosis and management of chronic kidney disease: a primary care cohort study. PLoS Med. 2017;14:e1002400. 45. Zonoozi S, Ramsay SE, Papacosta O, Lennon LT, Ellins EA, Halcox JPJ, Whincup P, Wannamethee SG. Chronic kidney disease, cardiovascular risk markers and total mortality in older men: cystatin C versus creatinine. J Epidemiol Community Health.2019; 73:645–51. 46. Lim WH, Lewis JR, Wong G, Turner RM, Lim EM, Thompson PL, Prince RL. Comparison of estimated glomerular filtration rate by the chronic kidney disease epidemiology collaboration (CKD-EPI) equations with and without Cystatin C for predicting clinical outcomes in elderly women. PLoS One. 2014;9:e106734. 47. Rowe C, Sitch AJ, Barratt J, Brettell EA, Cockwell P, Dalton RN, Deeks JJ, Eaglestone G, Pellatt-Higgins T, Kalra PA, Khunti K, Loud FC, Morris FS, Ottridge RS, Stevens PE, Sharpe CC, Sutton AJ, Taal MW, Lamb EJ, eGFR-C Study Group. Biological variation of measured and estimated glomerular filtration rate in patients with chronic kidney disease. Kidney Int. 2019;96:429–35. 48. Fan L, Levey AS, Gudnason V, Eiriksdottir G, Andresdottir MB, Gudmundsdottir H, Indridason OS, Palsson R, Mitchell G, Inker LA. Comparing GFR estimating equations using cystatin C and creatinine in elderly individuals. J Am Soc Nephrol JASN. 2015;26:1982–9. 49. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, Kusek JW, Manzi J, Van Lente F, Zhang YL, Coresh J, Levey AS, Investigators CKD- EPI. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med. 2012;367:20–9. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.