The original sin of Cockcroft-Gault formula
E. Y. Park, Think‐You Kim · Nephrology Dialysis Transplantation · 2010
While reading the article by Etgen and colleagues which reported that ‘chronic kidney disease is associated with incident cognitive impairment in the elderly’ [1] based on the renal function evaluation using the Cockcroft–Gault (C–G) formula [2], we felt it was interesting to see that the decrease of renal function in the study subjects coincided with the increase of age. Generally in the same elderly patient group, C–G formula accurately estimates renal function in those with low creatinine clearance (Ccr) [3] while it calculates low results in those with normal Ccr [4]. In studies on young patients, it was reported that C–G formula highly evaluated renal function [4,5]. Because of all above findings, we thought it would be necessary to study how age and Ccr have any influence on C–G formula. We divided subjects with ages of 18 years old or older into three groups of Group I, II and III, according to Ccr levels. We then calculated C–G formula in each group, and drew scatter plots between age and Ccr, and between age and C–G formula (Figure 1A and B). In all three groups, C–G formula decreased as age increased with strong correlations between age and C–G formula. Scatter plot between (A) age and Ccr and (B) age and C–G formula (circles; Group I; n = 36, Ccr = 100–109 mL/min/1.73m2, age 23–78, Scr 0.63 ± 0.168 mg/dL, multiplication symbols; Group II; n = 46, Ccr = 60–69 mL/min/1.73m2, age 34–81, Scr 0.86 ± 0.27 mg/dL, squares; Group III; n = 42, Ccr = 10–19 mL/min/1.73m2, age 28–87, Scr 3.11 ± 1.15 mg/dL). (A) Correlation coefficients between age and Ccr for each Group I, II and III are r = −0.297 (P = 0.079), r = −0.059 (P = 0.711), r = −0.068 (P = 0.653) in the order and showed no statistical significance. (B) Regression equations and correlation coefficients between age and C–G formula for each Group I, II and III are as follows in the order: y = −1.385x + 194.2 (r = −0.555, P 1. While we were reading the article in which Cockcroft and Gault reported their formula, we were surprised to see the selected subjects for the study [2]. Ccr, shown next to creatinine excretion (CrE) in their summarized table, also was decreasing according to age. Renal function could be estimated just by age using the regression curve of Ccr, even without using the regression curve between age and CrE as they had done (Figure 2A and B). Regression curve between (A) mean age and mean CrE and (B) mean age and mean Ccr from the Cockcroft and Gault's study subjects (Table II. age, renal function and creatinine excretion in 249 patients from [2]). (A) CrE = −0.2 age + 28 (r = 0.988, P < 0.001). Cockcroft and Gault divided 249 subjects with ages between 18 and 92 into seven groups according to age, and found the above regression curve between CrE and age. By substituting the curve equation in place of CrE in the numerator of the equation for Ccr, they made the familiar C–G formula (millilitre per minute) = (140 − age in years) × (weight in kilogram) / 72 × Scr (milligram per decilitre). (B) Ccr = −1.4 age + 151 (r = 0.980, P < 0.001). Ccr as well as CrE showed strong correlation with age. In the study subjects for devising C–G formula, Ccr was already too low in old age subjects, and it was relatively higher in younger subjects. In conclusion, C–G formula was shown to be strongly interfered by age factor, and this interference was strongly influenced by Ccr. And the partial reason for such a result might lie with the non-random selection of study subjects. We are afraid that the effort by the authors to find the relationship between the renal function and the cognitive function might have ended as just an observation of the relationship between age and cognitive function because of the original sin of C–G formula. Conflict of interest statement. None declared.