Assessment of renal function in rheumatoid arthritis: validity of a new prediction method

Medical treatment of patients with rheumatoid arthritis (RA) requires assessment of renal function. Because determination of endogenous creatinine clearance from a 24-hour urine collection is an unreliable and time-consuming procedure, several formulae that predict creatinine clearance from clinical and serum parameters have been developed. However, because of muscular atrophy, these formulae show lower correlations with measured creatinine clearance in patients with RA than in the healthy population. Recently, a new formula has been derived from the large Modification of Diet in Renal Disease (MDRD) study and has been shown reliably to predict renal function in individuals with renal dysfunction. To investigate the validity of this method in the RA population, estimates of creatinine clearance were derived using the most commonly used Cockcroft-Gault formula and the new MDRD method in control subjects and patients with RA. Age, height, serum albumin, blood urea nitrogen, and creatinine clearance were similar in both groups, but patients with RA had a lower body weight as well as serum and urinary creatinine concentrations. In control subjects, both methods showed comparable correlations with measured creatinine clearance (r = 0.82 and 0.83, respectively). In patients with RA, the Cockcroft-Gault formula revealed a lower correlation (r = 0.69) with a moderate bias (mean error = -10.7) and prediction accuracy (mean squared error = 342). For the MDRD method, r was still lower at 0.41, the mean error was -18.9, and the mean squared error was 479. We conclude that in patients with RA, the Cockroft-Gault formula is preferable to predict creatinine clearance before use of drugs such as methotrexate or nonsteroidal anti-inflammatory drugs.