Skip to content

Risk prediction and prognosis

TL;DR — eGFR and albuminuria predict kidney failure, cardiovascular events and death jointly, but absolute risk depends on age, cause, trajectory and competing mortality (CKD Prognosis Consortium 2023, PMID 37787795). The four-variable Kidney Failure Risk Equation uses age, sex, eGFR and uACR; individual-participant validation across 31 cohorts and 721,357 people with CKD stages 3–5 in more than 30 countries gave a C-statistic of 0.90 (95% CI 0.89–0.92) at two years and 0.88 (0.86–0.90) at five, although it overestimated risk in several non-North American cohorts until a recalibration factor was added (Tangri 2016, PMID 26757465). A risk estimate is useful only when attached to a decision horizon such as referral, multidisciplinary preparation or replacement planning. Calibration must be checked locally, especially after equation changes and in older adults with high competing mortality (Grams 2023, PMID 36857500).

From stage to probability

G and A categories stratify population risk, but patients need an absolute probability over a defined period. KFRE translates common variables into two- and five-year kidney-failure risk (Tangri 2016, PMID 26757465).

Discrimination and calibration

A model can rank patients well while systematically over- or under-predicting absolute risk. Across 59 cohorts and 312,424 patients the KFRE stayed accurate and well calibrated using CKD-EPI 2021 eGFR, and was not improved by substituting one-year average ACR or eGFR, adding two-year prior eGFR slope, or adding cardiovascular comorbidities; but it underpredicted at eGFR 45–59 and overpredicted in older adults on the five-year horizon, which a spline-in-eGFR competing-risk model corrected in those subgroups without improving overall performance (Grams 2023, PMID 36857500).

Trajectory

Repeated eGFR and uACR add information, but short noisy slopes can mislead. Treatment-related acute dips should not be extrapolated as chronic decline (EMPA-KIDNEY Collaborative Group 2024, PMID 38061371).

Competing death

Older adults may die before kidney failure; standard survival models can overstate cumulative kidney-failure probability when death is treated as censoring.

Decision thresholds

Risk-based referral is more coherent than an eGFR threshold alone when albuminuria and trajectory differ. Thresholds should be tied to an action and capacity (KDIGO CKD Work Group 2024, PMID 38490803).

Communication

Natural frequencies and both kidney-failure and mortality horizons are preferable to labels such as high risk. Uncertainty and modifiable assumptions should be visible.

Prediction tools answer different questions

Tool / marker Horizon or outcome Strength Limitation
KDIGO G×A heat map Broad kidney/CV/mortality strata Simple, global framework Categories hide continuous risk (CKD Prognosis Consortium 2023, PMID 37787795)
Four-variable KFRE 2- and 5-year treated kidney failure C-statistic 0.90 (0.89–0.92) at 2 y across 31 cohorts, 721,357 people Overestimated risk outside North America without recalibration (Tangri 2016, PMID 26757465)
eGFR slope Rate of functional change Longitudinal and responsive Noise and acute drug dips
uACR change Glomerular-injury response Early and inexpensive Surrogate validity varies by disease and drug
Cause-specific models ADPKD, IgA or other disease progression Adds biological specificity Narrow transportability
Frailty/geriatric prognosis Death and functional decline Relevant to CKM/KRT choice Not integrated routinely with KFRE
Dynamic repeated-measure models Updated risk after new data Can incorporate trajectory Adding prior eGFR slope or averaged ACR did not improve KFRE performance in 59 cohorts (Grams 2023, PMID 36857500)

Predicting who gets CKD, not only who progresses

KFRE begins after the diagnosis. The CKD Prognosis Consortium built the complementary tool from 34 multinational cohorts and 5,222,711 individuals in 28 countries, modelling incident eGFR below 60 separately by diabetes status because predictor availability differs. Among 4,441,084 participants without diabetes (mean age 54, 38% women) there were 660,856 incident cases (14.9%) over mean 4.2 years; among 781,627 with diabetes (mean age 62, 13% women), 313,646 cases (40%) over mean 3.9 years. The 5-year equations use age, sex, race/ethnicity, eGFR, cardiovascular history, ever-smoking, hypertension, BMI and albuminuria, adding diabetes medications, HbA1c and their interaction in the diabetes model; median C-statistics were 0.845 (IQR 0.789–0.890) without diabetes and 0.801 (0.750–0.819) with diabetes, tested in 9 external cohorts (n = 2,253,540) (Nelson 2019, PMID 31703124). These cohort incidences describe the development datasets and should not be read as population-wide five-year risks.

External validation: discrimination travels, calibration does not

The 0.90 C-statistic quoted above is a pooled figure; single-country validations are consistently lower and calibration is the recurring failure.

Validation cohort Population Discrimination Calibration finding
31 multinational cohorts, 721,357 patients (Tangri 2016, PMID 26757465) CKD G3–G5, >30 countries C 0.90 (0.89–0.92) at 2 y Overestimated outside North America until recalibration added
59 cohorts, 312,424 patients (Grams 2023, PMID 36857500) CKD, CKD-EPI 2021 eGFR Maintained Underpredicted at eGFR 45–59; overpredicted in older adults at 5 y
Urban China, 4,587 (4-variable) / 1,414 (8-variable) (Pan 2024, PMID 38689834) Community CKD, eGFR <60 C 0.750 (0.615–0.885) at 2 y; 0.766 (0.625–0.907) at 5 y Acceptable; decision-curve net benefit exceeded an eGFR <30 referral rule
Peru, 7,519 (G3a–4) and 2,798 (G3b–4) (Bravo-Zúñiga 2024, PMID 38184316) Primary care, competing risk of death modelled Time-dependent AUC and C >0.8 Underestimated 2-year risk and overestimated 5-year risk; authors advise recalibration before clinical use
CARE FOR HOMe, 403 patients (Lennartz 2016, PMID 26787778) Tertiary referral G2–G4, mean 4.4 y C 0.91 (0.83–0.99) Adding renal resistive index improved model fit (p<0.001) but not the C-statistic, sensitivity or specificity, and failed to replicate in an independent 162-patient cohort

Two lessons follow. First, the widely quoted C-statistic of 0.90 comes from pooled multinational data with wide case-mix; expect something nearer 0.75–0.85 in a single local population, which is still usefully better than an eGFR threshold. Second, the Peruvian result — underestimating short-term and overestimating long-term risk in the same cohort — is the pattern that most damages clinical use, because the two horizons drive different decisions (urgent preparation versus counselling about lifetime risk).

The Lennartz result is worth stating as a negative finding: routine Duplex ultrasound parameters did not add usable prognostic information to KFRE and the apparent improvement did not replicate (Lennartz 2016, PMID 26787778). It belongs alongside the Grams finding that averaged ACR, prior eGFR slope and cardiovascular comorbidity also failed to improve KFRE (Grams 2023, PMID 36857500): the four-variable model has proved unusually resistant to enrichment.

Machine learning: better discrimination, unproven decisions

The Klinrisk model, trained on routinely collected laboratory data, was externally validated in the pooled CANVAS/CREDENCE trial populations for the outcome of ≥40% eGFR decline or kidney failure, achieving AUC 0.81 (95% CI 0.78–0.83) at 1 year and 0.88 (0.86–0.89) at 3 years, with Brier scores 0.020 (0.018–0.022) and 0.056 (0.052–0.059) respectively, and outperforming the KDIGO heat map at every interval (p < 0.01) (Tangri 2024, PMID 38807510). Beating the heat map is a low bar — the heat map is a categorical population tool, not an individual predictor — and validation inside trial populations does not establish performance in unselected care, where laboratory data are missing non-randomly. No trial has yet randomized care to a machine-learning risk output versus KFRE.

The cardiovascular half of the prognosis

CKD prognosis is not mainly about kidney failure, and the risk tools have moved to reflect that. The AHA PREVENT equations were derived from 3,281,919 individuals across 25 datasets and externally validated in 3,330,085 from 21 more, are sex-specific and race-free, are adjusted for the competing risk of non-CVD death, and include eGFR as a core predictor with uACR and HbA1c as optional additions; median external C-statistics for total CVD were 0.794 (IQI 0.763–0.809) in women and 0.757 (0.727–0.778) in men, with calibration slopes 1.03 and 0.94 (Khan 2024, PMID 37947085). Multinational testing across 44 observational cohorts and 18 randomized trials — 293,737 PREVENT events among 6,422,714 individuals and 258,086 SCORE2 events among 5,437,384 — found similar discrimination and generally good calibration for both algorithms across North America, Europe and Asia/other regions, with published scaling factors for 1–9-year horizons (Neuen 2026, PMID 42086979).

Putting eGFR (and optionally uACR) inside a general cardiovascular risk equation is the practical resolution of the "which risk do I quote?" problem: for most people with CKD, PREVENT-type cardiovascular risk is the larger number and KFRE the smaller one, and quoting either alone misrepresents the prognosis.

A modifiable exposure with a dose–response

Among 5,459,014 adults in 39 general-population cohorts from 40 countries followed a mean 8 years, 246,607 (5.6%) had GFR decline (defined as ≥40% eGFR decline, KRT initiation or eGFR <10) and 782,329 (14.7%) died. Adjusted for age, sex, race and smoking, hazard ratios for GFR decline at BMI 30, 35 and 40 versus BMI 25 were 1.18 (95% CI 1.09–1.27), 1.69 (1.51–1.89) and 2.02 (1.80–2.27), consistent across eGFR subgroups; adjusting further for comorbidities attenuated these to 1.03 (0.95–1.11), 1.28 (1.14–1.44) and 1.46 (1.28–1.67) (Chang 2019, PMID 30630856). The attenuation shows that diabetes, hypertension and other comorbidities account for part of the association, but it does not quantify causal mediation. Mortality showed a J-shaped relation with a minimum at BMI 25 in the general population, and the GFR-decline associations were weaker in the high-cardiovascular-risk and established-CKD cohorts.

Frailty: the prognostic axis that KFRE does not contain

Frailty is common in CKD and predicts the outcomes that compete with kidney failure. Meta-analysis of 18 cohort studies with 22,788 participants found a median reported prevalence of frailty of 41.8% (range 2.8–81.5%) and of pre-frailty of 43.9% (19.1–62.7%). Pooled hazard ratios for mortality were 1.68 (95% CI 1.46–1.94; p < 0.001) for pre-frailty and 1.48 (1.21–1.81; p < 0.001) for frailty; for hospitalisation, 1.56 (1.37–1.76; p < 0.001) for pre-frailty and 1.21 (0.79–1.85; p = 0.38) for frailty; and for falls, 1.83 (1.40–2.37; p < 0.001) for frailty and 1.19 (0.44–3.22; p = 0.73) for pre-frailty (Mei 2021, PMID 33218914).

Two things are odd in these pooled numbers and should be stated rather than smoothed. The prevalence range spans 2.8% to 81.5%, which reflects incompatible frailty instruments rather than genuine population variation. And pre-frailty carried a higher pooled mortality hazard than frailty, which is biologically implausible as a dose–response and most likely reflects differential adjustment and competing risk across the included studies.

Even taking the direction rather than the magnitude, the implication for this page is concrete: a KFRE estimate and a frailty assessment answer different questions, and the frailty axis is the one that determines whether a projected kidney-failure risk will be reached before death. No published model integrates the two.

Pregnancy as an unmodelled kidney exposure

Standard CKD prognostic models contain no term for pregnancy, yet its effect on trajectory is quantified. In 178 pregnancies among 159 women with CKD stages 3–5 across six UK tertiary renal centres (2003–2017, including 43 transplant recipients), the live birth rate was 98% but 56% of babies were born preterm. Chronic hypertension was the strongest predictor of delivery before 34 weeks: 32% (31/96) in women with confirmed chronic hypertension versus 0% (0/25) in normotensive women, and the risk doubled from 20% (95% CI 9–36) to 40% (26–56) when the gestational fall in serum creatinine was less than 10% of pre-pregnancy values. A pre-pregnancy or early-pregnancy urine protein:creatinine ratio above 100 mg/mmol raised the odds of birthweight below the 10th centile (OR 2.57, 95% CI 1.20–5.53).

The trajectory cost was measurable: eGFR fell 4.5 mL/min/1.73 m² between pre-pregnancy and post-partum against a pre-pregnancy annual decline of 1.8 mL/min/1.73 m²/year — equivalent to 1.7, 2.1 and 4.9 years of disease progression in CKD stages 3a, 3b and 4–5 respectively, and larger in women with chronic hypertension and in those whose serum creatinine fell by less than 10% during gestation (Wiles 2021, PMID 33313680).

The reverse direction is also prognostic. Among 27,800 adults with deliveries in one US health system (1996–2019, mean age 28), 2,977 (10.7%) had at least one preeclamptic pregnancy; after propensity matching, preeclampsia was associated with subsequent chronic hypertension (HR 1.77, 95% CI 1.45–2.16), eGFR below 60 (HR 3.23, 1.64–6.36), albuminuria above 300 mg/g (HR 3.60, 2.38–5.44) and recurrent preeclampsia (HR 24.76, 12.47–48.36). Post-partum follow-up testing was poor: in the first six months, 31% versus 14% had a serum creatinine measured, and only 26% of both groups had urine protein tested (Srialluri 2023, PMID 37516302).

A more-than-threefold hazard for reduced eGFR combined with a 26% follow-up testing rate identifies preeclampsia as a potentially actionable follow-up gap; the observational association does not establish which surveillance strategy improves outcomes.

Decision and interpretation matrix

Dimension Question Guardrail
Diagnostic axis Cause + G category + A category Avoid treating eGFR as the diagnosis
Time axis Chronicity and trajectory Separate acute change from persistent disease
Risk axis Kidney failure + cardiovascular events + death Show competing events
Treatment axis Eligibility, absolute benefit, harm, burden Do not rank drugs by relative effect alone
Measurement axis Assay, equation, repeatability State what was actually measured
Equity axis Testing, referral, access, affordability Audit downstream care, not labels only
Patient axis Symptoms, function, life participation Include outcomes patients prioritize
Evidence axis RCT, cohort, model, guideline Do not collapse designs

Evidence ledger

This ledger makes the page’s evidentiary mix inspectable. It does not imply that every source answers every question.

PMID Record used Role and boundary
37787795 Estimated GFR, Albuminuria, and Adverse Outcomes: individual-participant data meta-analysis. (CKD Prognosis Consortium 2023, PMID 37787795) Synthesis; heterogeneity and included-study definitions constrain transport.
26757465 Multinational assessment of equations predicting kidney failure. (Tangri 2016, PMID 26757465) Synthesis; heterogeneity and included-study definitions constrain transport.
36857500 Kidney Failure Risk Equation evaluation with novel inputs in 59 cohorts. (Grams 2023, PMID 36857500) Observational or conceptual evidence; association is not treatment effect.
38061371 Effects of empagliflozin on progression of chronic kidney disease: a prespecified secondary analysis from the EMPA-KIDNEY trial. (EMPA-KIDNEY Collaborative Group 2024, PMID 38061371) Intervention study; eligibility, comparator, endpoint and follow-up bound inference.
38490803 KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. (KDIGO CKD Work Group 2024, PMID 38490803) Guideline or commentary; recommendation evidence depends on its review.
38519239 Executive summary of the KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease: known knowns and known unknowns. (Levin 2024, PMID 38519239) Guideline or commentary; recommendation evidence depends on its review.
32061315 Global, regional, and national burden of chronic kidney disease, 1990-2017. (GBD CKD Collaboration 2020, PMID 32061315) Modelled projection; the estimate follows from the model inputs and assumptions, not from observed randomized follow-up.
22038337 A population-based approach for the definition of chronic kidney disease: CKD Prognosis Consortium. (Cirillo 2012, PMID 22038337) Synthesis; heterogeneity and included-study definitions constrain transport.
23243116 Cohort profile: the chronic kidney disease prognosis consortium. (Matsushita 2013, PMID 23243116) Observational or conceptual evidence; association is not treatment effect.
30348535 Relationship of Estimated GFR and Albuminuria to Concurrent Laboratory Abnormalities. (Inker 2019, PMID 30348535) Synthesis; heterogeneity and included-study definitions constrain transport.
34554658 New Creatinine- and Cystatin C-Based Equations to Estimate GFR without Race. (Inker 2021, PMID 34554658) Observational or conceptual evidence; association is not treatment effect.
34563581 A Unifying Approach for GFR Estimation: Recommendations of the NKF-ASN Task Force on Reassessing the Inclusion of Race in Diagnosing Kidney Disease. (Delgado 2022, PMID 34563581) Guideline or commentary; recommendation evidence depends on its review.
26028594 eGFR and albuminuria for prediction of cardiovascular outcomes: individual-participant meta-analysis. (Matsushita 2015, PMID 26028594) Synthesis; heterogeneity and included-study definitions constrain transport.
32970396 Dapagliflozin in Patients with Chronic Kidney Disease. (Heerspink 2020, PMID 32970396) Intervention study; eligibility, comparator, endpoint and follow-up bound inference.
36331190 Empagliflozin in Patients with Chronic Kidney Disease. (EMPA-KIDNEY Collaborative Group 2023, PMID 36331190) Intervention study; eligibility, comparator, endpoint and follow-up bound inference.
30990260 Canagliflozin and Renal Outcomes in Type 2 Diabetes and Nephropathy. (Perkovic 2019, PMID 30990260) Intervention study; eligibility, comparator, endpoint and follow-up bound inference.
41205219 Chronic Kidney Disease Prevalence and Awareness Among US Adults. (Gong 2026, PMID 41205219) Observational or conceptual evidence; association is not treatment effect.
38213490 Cost-effectiveness of screening for chronic kidney disease: evidence and gaps. (van Mil 2024, PMID 38213490) Guideline or commentary; recommendation evidence depends on its review.
39137037 Screening for chronic kidney disease: change of perspective and novel developments. (van Mil 2024, PMID 39137037) Observational or conceptual evidence; association is not treatment effect.
38186904 Cost-effectiveness of screening for CKD in the general adult population: systematic review. (Yeo 2024, PMID 38186904) Synthesis; heterogeneity and included-study definitions constrain transport.
40227684 Balancing Efficiency and Equity in Population-Wide CKD Screening. (Cusick 2025, PMID 40227684) Modelled projection; the estimate follows from the model inputs and assumptions, not from observed randomized follow-up.
37403003 Chronic kidney disease of unknown aetiology: a global review. (Rao 2023, PMID 37403003) Observational or conceptual evidence; association is not treatment effect.
33116757 Mesoamerican Nephropathy: What We Know so Far. (Sanchez Polo 2020, PMID 33116757) Observational or conceptual evidence; association is not treatment effect.
18161745 Cellular and molecular mechanisms of fibrosis. (Wynn 2008, PMID 18161745) Observational or conceptual evidence; association is not treatment effect.
7246778 Hyperfiltration in remnant nephrons: a potentially adverse response to renal ablation. (Hostetter 1981, PMID 7246778) Observational or conceptual evidence; association is not treatment effect.
24522492 Relative risks of CKD for mortality and end-stage renal disease across races are similar. (Wen 2014, PMID 24522492) Observational or conceptual evidence; association is not treatment effect.

What can and cannot be concluded

  • Risk associations do not by themselves establish that changing the marker changes risk.
  • A relative effect must be paired with baseline risk, follow-up and the exact endpoint.
  • Albuminuria, acute eGFR change, chronic eGFR slope and kidney failure are not interchangeable.
  • Subgroup consistency is not evidence that every subgroup had adequate power.
  • Guideline recommendations combine evidence with values, feasibility, cost and service capacity.
  • Older adults require competing-mortality and treatment-burden framing.
  • Dialysis and transplantation comparisons are vulnerable to eligibility and immortal-time bias.
  • Modelled lifetime benefit is not a randomized observed benefit.
  • A biochemical response without a patient-important outcome remains a surrogate result.
  • This page is research synthesis, not individualized medical advice.

Research-design checklist

  • Define CKD cause, G category, A category and chronicity at baseline.
  • Report the creatinine or cystatin C equation and laboratory calibration.
  • Prespecify acute and chronic eGFR slopes when haemodynamic effects are expected.
  • Keep sustained GFR decline, kidney failure and replacement therapy separable.
  • Report absolute event risks, follow-up and confidence intervals with relative effects.
  • Treat death as a competing event where it can preclude kidney failure.
  • Measure hyperkalaemia, acute kidney injury and treatment discontinuation consistently.
  • Include symptoms, function, life participation and treatment burden.
  • Describe background RAS, SGLT2, MRA and GLP-1 therapy explicitly.
  • Prespecify albuminuria and cause strata without over-reading underpowered interactions.
  • Record screening, prescribing, persistence and monitoring as separate implementation steps.
  • Report representation, access and affordability variables needed for equity analysis.

Open questions

  • Should the KFRE be routinely reported with a competing-mortality companion estimate? A spline-plus-competing-risk model improved calibration in older adults and at eGFR 45–59 but not overall, leaving the trade-off unresolved (Grams 2023, PMID 36857500).
  • Which risk threshold, tied to which action, actually improves outcomes? Risk-based referral is more coherent than an eGFR threshold, but no trial has randomised the threshold itself (KDIGO CKD Work Group 2024, PMID 38490803).
  • Does trajectory information add to a single eGFR–uACR pair once acute drug effects are excluded? Adding two-year prior slope did not improve the KFRE (Grams 2023, PMID 36857500).

  • How should absolute risk be communicated so that patients and clinicians reach the same decision? Natural frequencies are recommended by convention, not by comparative trial evidence in CKD.

  • Why has the four-variable KFRE resisted every attempted enrichment — averaged ACR, prior eGFR slope, cardiovascular comorbidity (Grams 2023, PMID 36857500), renal resistive index (Lennartz 2016, PMID 26787778) — and does that indicate a ceiling on predictable variance or a failure of the candidate predictors?

  • Should KFRE be recalibrated per country before use? Peruvian data show simultaneous 2-year underestimation and 5-year overestimation in the same cohort (Bravo-Zúñiga 2024, PMID 38184316), and Chinese community C-statistics are ~0.75 against a pooled 0.90 (Pan 2024, PMID 38689834) (Tangri 2016, PMID 26757465).
  • Does acting on a machine-learning risk score improve outcomes relative to KFRE? Klinrisk discriminates better than the KDIGO heat map (Tangri 2024, PMID 38807510) but no trial has randomized care to either output.

  • How much of the adiposity–GFR-decline association is independent of diabetes and hypertension? Comorbidity adjustment roughly halves the hazard ratios but does not eliminate them (Chang 2019, PMID 30630856).

  • Why does pre-frailty carry a higher pooled mortality hazard (1.68, 1.46–1.94) than frailty (1.48, 1.21–1.81) in CKD (Mei 2021, PMID 33218914)? The inversion suggests differential adjustment or competing risk rather than a dose–response, and no harmonised frailty instrument exists across the 2.8–81.5% prevalence range.

  • Should a pregnancy be modelled as a discrete increment of CKD progression? Its measured cost is equivalent to 1.7, 2.1 and 4.9 years of pre-pregnancy decline at stages 3a, 3b and 4–5 (Wiles 2021, PMID 33313680), and no prognostic model includes it.
  • Why is post-partum kidney follow-up performed in only about a quarter of people after preeclampsia when the subsequent hazard for eGFR below 60 is 3.23 (1.64–6.36) (Srialluri 2023, PMID 37516302)?

References

  1. CKD Prognosis Consortium et al. Estimated GFR, Albuminuria, and Adverse Outcomes: individual-participant data meta-analysis. JAMA. 2023;330(13):1266-1277. PMID 37787795
  2. Tangri et al. Multinational assessment of equations predicting kidney failure. JAMA. 2016;315(2):164-174. PMID 26757465
  3. Grams et al. Kidney Failure Risk Equation evaluation with novel inputs in 59 cohorts. J Am Soc Nephrol. 2023;34(3):482-494. PMID 36857500
  4. EMPA-KIDNEY Collaborative Group et al. Effects of empagliflozin on progression of chronic kidney disease: a prespecified secondary analysis from the EMPA-KIDNEY trial. Lancet Diabetes Endocrinol. 2024;12(1):39-50. PMID 38061371
  5. KDIGO CKD Work Group et al. KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int. 2024;105(4S):S117-S314. PMID 38490803
  6. Levin et al. Executive summary of the KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease: known knowns and known unknowns. Kidney Int. 2024;105(4):684-701. PMID 38519239
  7. GBD CKD Collaboration et al. Global, regional, and national burden of chronic kidney disease, 1990-2017. Lancet. 2020;395(10225):709-733. PMID 32061315
  8. Cirillo et al. A population-based approach for the definition of chronic kidney disease: CKD Prognosis Consortium. J Nephrol. 2012;25(1):7-12. PMID 22038337
  9. Matsushita et al. Cohort profile: the chronic kidney disease prognosis consortium. Int J Epidemiol. 2013;42(6):1660-1668. PMID 23243116
  10. Inker et al. Relationship of Estimated GFR and Albuminuria to Concurrent Laboratory Abnormalities. Am J Kidney Dis. 2019;73(2):206-217. PMID 30348535
  11. Inker et al. New Creatinine- and Cystatin C-Based Equations to Estimate GFR without Race. N Engl J Med. 2021;385(19):1737-1749. PMID 34554658
  12. Delgado et al. A Unifying Approach for GFR Estimation: Recommendations of the NKF-ASN Task Force on Reassessing the Inclusion of Race in Diagnosing Kidney Disease. Am J Kidney Dis. 2022;79(2):268-288.e1. PMID 34563581
  13. Matsushita et al. eGFR and albuminuria for prediction of cardiovascular outcomes: individual-participant meta-analysis. Lancet Diabetes Endocrinol. 2015;3(7):514-525. PMID 26028594
  14. Heerspink et al. Dapagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2020;383(15):1436-1446. PMID 32970396
  15. EMPA-KIDNEY Collaborative Group et al. Empagliflozin in Patients with Chronic Kidney Disease. N Engl J Med. 2023;388(2):117-127. PMID 36331190
  16. Perkovic et al. Canagliflozin and Renal Outcomes in Type 2 Diabetes and Nephropathy. N Engl J Med. 2019;380(24):2295-2306. PMID 30990260
  17. Gong et al. Chronic Kidney Disease Prevalence and Awareness Among US Adults. JAMA Cardiol. 2026;11(1):77-81. PMID 41205219
  18. van Mil et al. Cost-effectiveness of screening for chronic kidney disease: evidence and gaps. Clin Kidney J. 2024;17(1):sfad254. PMID 38213490
  19. van Mil et al. Screening for chronic kidney disease: change of perspective and novel developments. Curr Opin Nephrol Hypertens. 2024;33(6):583-592. PMID 39137037
  20. Yeo et al. Cost-effectiveness of screening for CKD in the general adult population: systematic review. Clin Kidney J. 2024;17(1):sfad137. PMID 38186904
  21. Cusick et al. Balancing Efficiency and Equity in Population-Wide CKD Screening. JAMA Netw Open. 2025;8(4):e254740. PMID 40227684
  22. Rao et al. Chronic kidney disease of unknown aetiology: a global review. Trop Med Int Health. 2023;28(8):588-600. PMID 37403003
  23. Sanchez Polo et al. Mesoamerican Nephropathy: What We Know so Far. Int J Nephrol Renovasc Dis. 2020;13:261-272. PMID 33116757
  24. Wynn et al. Cellular and molecular mechanisms of fibrosis. J Pathol. 2008;214(2):199-210. PMID 18161745
  25. Hostetter et al. Hyperfiltration in remnant nephrons: a potentially adverse response to renal ablation. Am J Physiol. 1981;241(1):F85-F93. PMID 7246778
  26. Wen et al. Relative risks of CKD for mortality and end-stage renal disease across races are similar. Kidney Int. 2014;86(4):819-827. PMID 24522492
  27. Nelson RG, et al. Development of Risk Prediction Equations for Incident Chronic Kidney Disease. JAMA. 2019;322(21):2104-2114. PMID 31703124
  28. Pan L, et al. External Validation of the Kidney Failure Risk Equation Among Urban Community-Based Chinese Patients With CKD. Kidney Med. 2024;6(5):100817. PMID 38689834
  29. Bravo-Zuniga J, et al. Multicentre external validation of the prognostic model kidney failure risk equation in patients with CKD stages 3 and 4 in Peru: a retrospective cohort study. BMJ Open. 2024;14(1):e076217. PMID 38184316
  30. Lennartz CS, et al. External Validation of the Kidney Failure Risk Equation and Re-Calibration with Addition of Ultrasound Parameters. Clin J Am Soc Nephrol. 2016;11(4):609-615. PMID 26787778
  31. Tangri N, et al. Machine learning for prediction of chronic kidney disease progression: Validation of the Klinrisk model in the CANVAS Program and CREDENCE trial. Diabetes Obes Metab. 2024;26(8):3371-3380. PMID 38807510
  32. Khan SS, et al. Development and Validation of the American Heart Association's PREVENT Equations. Circulation. 2024;149(6):430-449. PMID 37947085
  33. Neuen BL, et al. Multinational validation of the PREVENT and SCORE2 cardiovascular risk equations across 6.4 million individuals. Nat Med. 2026;32(7):2629-2638. PMID 42086979
  34. Chang AR, et al. Adiposity and risk of decline in glomerular filtration rate: meta-analysis of individual participant data in a global consortium. BMJ. 2019;364:k5301. PMID 30630856
  35. Mei F, et al. Frailty as a Predictor of Negative Health Outcomes in Chronic Kidney Disease: A Systematic Review and Meta-Analysis. J Am Med Dir Assoc. 2021;22(3):535-543.e7. PMID 33218914
  36. Wiles K, et al. The impact of chronic kidney disease Stages 3-5 on pregnancy outcomes. Nephrol Dial Transplant. 2021;36(11):2008-2017. PMID 33313680
  37. Srialluri N, et al. Preeclampsia and Long-term Kidney Outcomes: An Observational Cohort Study. Am J Kidney Dis. 2023;82(6):698-705. PMID 37516302