ESPN 54th Annual Meeting

ESPN 2022


 
Imprecision in GFR estimates - impact on CKD staging
JANUSZ FEBER 1 IVAN BLASUTIG 2 ROBERT L. MYETTE 3 ROBERT GOW 1

1- CHILDRENS HOSPITAL OF EASTERN ONTARIO, UNIVERSITY OF OTTAWA
2- CHEO, UNIVERSITY OF OTTAWA, EASTERN ONTARIO REGIONAL LABORATORY ASSOCIATION
3- CHEO, UNIVERSITY OF OTTAWA, KIDNEY RESEARCH CENTER, THE OTTAWA HOSPITAL RESEARCH CENTER
 
Introduction:

Schwartz GFR (SchwGFR) is estimated from height (cm) and serum creatinine (SCr) and is used for staging of chronic kidney disease (CKD). The SchwGFR formula may introduce additional propagation error in GFR estimation, in addition to intrinsic variability/error in measuring SCr and cm. 

The aim of the study was to analyze the propagation error (PE) of SchwGFR and its impact on CKD classification. 

Material and methods:

All available SCr results obtained from June 2021 to December 2021 were retrieved from the lab (924 samples from 648 patients). SchwGFR (point estimates, ml/min/1.73 m2) were calculated using the formula published by Schwartz et al (2009). Lab specific standard deviations (SD) of SCr assays (mean SD = 5 umol/l) and variability of height measurements (mean SD = 0.5 cm) were used to calculate PE in SchwGFR (composite error/SD propagated from SCr and height measurements), which resulted in interval-based GFR (SchwGFRi) with individual 95% confidence intervals. The agreement/disagreement between SchwGFRi and traditional point estimated GFR (SchwGFRp) was then analyzed (Cohen’s kappa, test of proportions) to inform classification of CKD stages. Samples were classified as CKD if the lower limit of the SchwGFRi was below the CKD classification threshold of 90 ml/min/1.73 m2.

Results:

Mean SchwGFR PE (%) [89% prediction intervals] were: 1.49 [1.04-1.95], 1.83 [1.37-2.31], 2.17 [1.67-2.64] and 2.51 [2.01-2.95] at GFR levels of 30, 60, 90 and 120 ml/min/1.73 m2, respectively. SchwGFRp identified CKD (GFR < 90) in 411/924 samples, whereas SchwGFRi detected CKD in an additional 49/924 (5.3%) samples (p<0.02), mostly CKD2.

Conclusions:

The propagation error of SchwGFR estimates was relatively small (up to 3%) but increased with increasing GFR and allowed for calculation of individual GFR “trusted” intervals.  This led to a significantly higher number (additional 5%) of GFR samples classified as CKD by interval estimates compared to traditional point estimates.