Estimating the Prevalence of Low Glomerular Filtration Rate Requires Attention to the Creatinine Assay Calibration

To the Editor:

In the May issue of JASN, Clase et al. (¹) report a much higher estimate of the prevalence of low glomerular filtration rate (GFR) than recently published in the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (K/DOQI) Clinical Practice Guidelines, “Chronic Kidney Disease: Evaluation, Classification and Stratification” (²). The authors also conclude that “laboratory-generated clearance reports cannot be recommended without further study.” We disagree with this conclusion and the calculations upon which it is based.

Recently published data indicate that serum creatinine assays on the same samples were 0.23 mg/dl higher in the Third National Health and Nutrition Examination Survey (NHANES III) than in the Modification of Diet in Renal Disease (MDRD) study (³). The K/DOQI report calculates the prevalence of moderately (30 to 60 ml/min per 1.73m²) or severely (15 to 29 ml/min per 1.73m²) decreased GFR at 4.3% and 0.2%, respectively, among adults age 20 yr and older in the United States. This estimate is based on the NHANES III study after accounting for calibration differences between NHANES III and MDRD study laboratories. Interestingly, Clase et al. derive very similar estimates (4% and 0.17%) for nondiabetic black and white adult Americans under the hypothetical statement “if NHANES III–measured creatinine were systematically as much as 0.2 mg/dl (17.7 μmol/L) higher than MDRD-measured creatinine.” These estimates are markedly lower than the comparable estimates reported in their abstract and main results of 13% below 60 ml/min per 1.73m² and 0.26% below 30 ml/min per 1.73m². The latter estimates reflect methodologic bias, not the underlying GFR.

Calibration explains the “unexpected” nature of the authors’ findings; therefore, we do not think the data provide any evidence against the appropriate use of equations to estimate GFR. In fact, estimated GFR in NHANES III is generally similar to studies of inulin clearance in normal volunteers (see Figure 9 of K/DOQI report [²]). We agree with the authors that biased creatinine results similar to those from the NHANES III reference laboratory would be obtained by at least some clinical laboratories. However, we strongly disagree with their conclusion that it is premature for laboratories to report a calculated clearance estimate with each creatinine measurement. Serum creatinine is widely used to detect and follow the progression of kidney disease; therefore, the relevant question is whether reporting an estimated GFR would provide an improvement over current practice. We believe laboratory reporting of estimated GFR should be implemented for three main reasons: (1) estimated GFR provides a better estimate of GFR than serum creatinine; (2) estimated GFR is more readily interpretable than serum creatinine by both patients and physicians; and (3) calibration differences in serum creatinine measurements are more readily detected when estimated GFR is reported because GFR estimates account for differences in patients’ age and gender. It is our hope that as laboratories implement the calculation, they will also implement corrections for their own serum creatinine calibration. Initial progress in this regard is encouraging, with one major HMO moving in this direction and an NIH-organized workshop on standardization of the serum creatinine assay scheduled for this summer.

An estimated GFR cutoff of less than 60 ml/min per 1.73m² for diagnosing CKD in the absence of kidney damage was chosen by the K/DOQI Work Group, with interlaboratory variation in measurement of serum creatinine in mind. The Work Group recommended caution in interpreting GFR estimates of 60 to 90 ml/min per 1.73m² as low, especially in the elderly. The advantage of estimating GFR over reporting serum creatinine can be illustrated by the following example. A serum creatinine of 1.4 mg/dl corresponds to an estimated GFR of 40 ml/min per 1.73m² in a 70-yr-old white woman. This translates to moderate (Stage 3) chronic kidney disease and a higher prevalence of anemia and other co-morbidities. If serum creatinine measurements from a laboratory that overestimates serum creatinine by 0.23 mg/dl is used, the estimated GFR would be 49 ml/min per 1.73m². On the other hand, for a 40-yr-old African-American man, a serum creatinine of 1.4 mg/dl corresponds to an estimated GFR of 72 (89 in a lab calibrated high) ml/min per 1.73m². As expected, calibration takes on a greater importance at higher GFR and when individuals are close to a given GFR cutoff (³). However, in our opinion, physicians are much better equipped to make good decisions when provided with an estimated GFR in addition to the serum creatinine on laboratory printouts.

We would like to emphasize that the belief that 24-h urine is a better solution than estimating GFR from serum creatinine is not supported by most published data (²). For example, in the African-American Study of Kidney Disease and Hypertension (⁴), the median absolute value of percent difference from iothalamate GFR was 23% for 24-h creatinine clearance compared with 12% for the estimates from the MDRD study equation. This result was based on 1703 study screenees with a mean (SD) GFR of 57 (23) ml/min per 1.73m² (R² = 0.59 and 0.82, respectively). Errors in urine collection are frequent, and urinary creatinine clearance suffers similar sensitivity to assay calibration as equations because non-creatinine chromogens are present in serum but not urine. Reliance on a cancellation of serum non-creatinine chromogens with creatinine secretion has no scientific basis and may have led to the high interlaboratory variation in creatinine calibration. In our opinion, it is time to do away with routine 24-h urine collections to estimate creatinine clearance and to adopt estimation of GFR from standardized serum creatinine measurements. The recently published clinical practice guidelines (²) provide the rationale for standardized diagnosis and classification of chronic kidney disease, which considers early detection and treatment of complications of decreased GFR, including hypertension and other cardiovascular disease risk factors, as well as strategies to slow the progression of kidney disease. The evidence for specific interventions should be rigorously evaluated, but a number of recommendations already exist for this high-risk patient population. Efforts to improve the prediction and search for better markers are useful. The sooner this is done, the better care patients with kidney disease will receive.