Effect of the Refitted Race-Free eGFR Formula on the CKD Prevalence and Mortality in the Danish Population

Efforts to remove bias in health care services across races have prompted a refitting of the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula for eGFR (1). Although a simple removal of the race factor in the CKD-EPI formula led to an underestimation of GFR in Black people, refitting of the formula led to biases of similar magnitude but opposite direction between races (1). Among non-Black people, the refitted formula (eGFRcr[AS]) underestimated the GFR by a mean of 3.6 ml/min, which is considerably more than the old equation (0.1 ml/min) when the race term was excluded by assuming non-Black race (eGFRcr[ASR-NB]). The American Society of Nephrology has recommended immediate implementation of the refitted, race-free equation (2). In population studies, such implementation may affect the calculated prevalence of CKD, overall and in specific CKD stages, and the associated mortality. Therefore, we calculated and compared the prevalence and mortality of CKD using the eGFRcr(ASR-NB) and eGFRcr(AS) formulas in the predominantly non-Black Danish population using nationwide population-based laboratory data (3).

We used Danish population-based laboratory data from primary, secondary, and tertiary care (4), linked on the individual level with data on age, sex, and vital status from The Civil Registration System, as recently described (3).

We identified all adults alive in Denmark on January 1, 2019, included all of their outpatient plasma creatinine tests from 2016 to 2018, and calculated the eGFR using: (1) the original CKD-EPI formula assuming non-Black race, eGFRcr(ASR-NB), and (2) the refitted CKD-EPI formula, eGFRcr(AS) (1). From these tests, we identified all patients fulfilling the Kidney Disease Improving Global Outcomes eGFR criteria for CKD stage 3–5, that is, at least two reported eGFR <60 ml/min per 1.73 m² ≥90 days apart (5). Patients were categorized according to Kidney Disease Improving Global Outcomes CKD G-stages on the basis of the higher of the two lowest eGFRs ≥90 days apart. We tabulated sex, median age, and proportion with relevant comorbidities recorded up to 20 years before the index date for each cohort. We estimated the prevalence of CKD overall and by G-stages (%) as the number of patients fulfilling the criteria, per 100 adults alive in Denmark on January 1, 2019. Finally, using the Kaplan–Meier estimator, we estimated the 1-year mortality risk by following patients from January 1, 2019 until death, emigration, or December 31, 2019.

The study was reported to the Danish Data Protection Agency (record number 2015–57–0002) by Aarhus University (record number 2016–051–000001/812). According to Danish legislation, approval from an ethics committee or informed consent from patients is not required for registry-based studies.

Among 4,768,234 adults, we identified 261,625 and 198,367 patients with any CKD on January 1, 2019 using the eGFRcr(ASR-NB) and eGFRcr(AS) formulas, respectively (Table 1). The two CKD populations were comparable regarding age and sex; however, the refitted formula identified populations with slightly higher proportions of selected comorbidities. Moreover, the new refitted formula yielded a lower CKD prevalence (4.2%) than the original equation (5.5%). Although the prevalence of CKD stage 5 changed only minimally, the prevalence of stage 3a, 3b, and 4 decreased when compared with the original formula without correction for race. In addition, we observed a higher 1-year mortality in the cohort defined by the refitted formula compared with that on the basis of the original equation (8.9% versus 7.9%), mainly driven by an increase among those with CKD stages 3a and 3b. Absolute changes in prevalence and mortality were mainly observed in groups aged >50 years, and less so in younger adults (data not shown).

Table 1.

Prevalence of CKD in Denmark at January 1, 2019, overall and by stage, when using two different formulas to calculate eGFR on the basis of plasma creatinine

Covariate	eGFRcr(ASR-NB) (original)					eGFRcr(AS) (refitted)
	Cohort					Cohort
	Any CKD (stages 3–5)	CKD Stage 3a	CKD Stage 3b	CKD Stage 4	CKD Stage 5	Any CKD (stages 3–5)	CKD Stage 3a	CKD Stage 3b	CKD Stage 4	CKD Stage 5
Individuals, n	261,625	168,089	67,771	20,511	5254	198,367	126,679	51,513	15,502	4673
Prevalence, %	5.49	3.53	1.42	0.43	0.11	4.16	2.66	1.08	0.33	0.10
1-year mortality, %	7.9	5.7	10.3	16.1	17.7	8.9	6.7	11.2	16.5	17.6
Male sex, %	44	43	44	46	60	44	43	43	47	61
Median age, yrs	78	77	80	81	70	78	78	81	80	69
Age category
18–49	1.8	1.4	1.5	2.7	14.2	2.0	1.4	1.8	3.2	15.6
50–69	18.0	20.2	12.6	13.8	33.4	16.8	18.0	12.7	15.4	34.9
70–89	70.8	71.4	72.2	67.4	48.0	70.8	72.1	71.1	66.5	46.0
90+	9.4	6.9	13.7	16.1	4.4	10.4	8.5	14.4	14.9	3.5
Comorbidity <20 yrs before index, %
Diabetes	24.3	20.1	29.3	38.8	38.4	26.5	22.3	31.7	39.7	37.7
Liver disease	2.0	1.8	2.2	2.5	3.6	2.1	1.9	2.3	2.6	3.7
Chronic pulmonary disease	14.3	12.9	16.3	19.1	15.7	15.1	13.8	17.2	18.8	15.7
Connective tissue disease	7.8	7.3	8.4	8.8	9.7	8.0	7.6	8.4	9.0	9.7
Heart failure	1.0	0.6	1.3	2.6	4.9	1.2	0.7	1.6	2.8	5.1
Thromboembolic disease	31.6	27.9	36.8	41.9	43.3	33.6	30.2	38.3	42.2	43.5
Cancer	20.9	20.1	22.5	22.7	20.3	21.4	20.8	22.6	22.7	20.1

eGFRcr(ASR-NB), eGFR creatinine (age+sex+race-non-Black); eGFRcr(AS), eGFR creatinine (age+sex).

In a large nationwide Danish population, implementation of the new race-free eGFRcr(AS) formula reduced estimated CKD prevalence by almost 25% and led to the identification of patients with a higher 1-year mortality compared with the eGFRcr(ASR-NB) formula. In addition to the lower eGFR, the greater comorbidity in the populations using the refitted formula may in part explain the higher mortality. This is expected in a predominantly non-Black population in which the eGFRcr(AS) formula results in a lower eGFR when compared with the eGFRcr(ASR-NB) formula, which has frequently been used in such populations when information on race is not available (3). It will have implications for epidemiologic studies on CKD and the associated outcomes and, thus, for public health measures and initiatives that are often directed by disease prevalence. On the individual level, implementation of the new formula may affect diagnostic workup, evaluation, and treatment of CKD, especially in those with an eGFR close to CKD cutoffs. Thus, the eGFR criteria determining the evaluation and management of CKD should be re-evaluated if the refitted CKD-EPI formula is implemented (5). In conclusion, in a predominantly non-Black population, the introduction of the refitted formula had a considerable effect on the resulting population identified with CKD.

Disclosures

S. Vestergaard, U. Heide-Jørgensen, and C. Christiansen report employment at The Department of Clinical Epidemiology, and H. Birn at The Department of Biomedicine and the Department of Renal Medicine, which are involved in studies with funding from various companies as research grants to (and administered by) Aarhus University or Aarhus University Hospital. H. Birn reports consultancy agreements with AstraZeneca, Galapagos, and Vifor Pharma; reports receiving research funding from GlaxoSmithKline and Vifor Pharma; reports receiving honoraria from Alexion, AstraZeneca, Merck Sharp & Dohme Corp., Novartis Healthcare, and Novo Nordisk; reports having advisory or leadership roles with the Augustinus Foundation, Danish Health Authority, Helen and Ejnar Bjørnow Foundation, and Kidney and Dialysis; and reports speakers bureau for BestPractice Nordic (a medical information service).

Funding

This work was supported by Det Frie Forskningsråd grant 0134-00407B.