Estimated glomerular filtration rate (eGFR) becomes less reliable in patients with advanced chronic kidney disease (CKD).Methods.
Using the Swedish CKD Registry (2005-11), linked to the national inpatient, dialysis and death registers, we compared the performance of plasma-iohexol measured GFR (mGFR) and urinary clearance measures versus eGFR to predict death in adults with CKD stages 4/5. Performance was assessed using survival and prognostic models.Results.
Of the 2705 patients, 1517 had mGFR performed, with the remainder providing 24-h urine clearances. Median eGFR (CKD-EPIcreatinine) was 20 mL/min/1.73 m2 [interquartile range (IQR) 14-26], mGFR 18 mL/min/1.73 m2 (IQR 13-23) and creatinine clearance 23 mL/min (IQR 15-31). Median follow-up was 45 months (IQR 26-59), registering 968 deaths (36%). In fully adjusted Cox models, a rise in mGFR of 1 mL/min/1.73 m2 was associated with a 5.3% fall in all-cause mortality compared with a 1.7% corresponding fall for eGFR [adjusted hazard ratio (aHR) 0.947 (95% CI, 0.930-0.964) versus aHR 0.983 (95% CI, 0.970-0.996)]. mGFR was also statistically superior in prognostic models (discrimination using logistic regression and integrated discrimination improvement). Urinary clearance measures showed a stronger aetiological relationship with death than eGFR, but were not statistically superior in the prognostic models.Conclusions.
The performance of mGFR was superior to eGFR, in both aetiological and prognostic models, in predicting mortality in adults with CKD stage 4/5, demonstrating the importance of GFR per se versus non-GFR determinants of outcome. However, the relatively modest enhancement suggests that eGFR may be sufficient to use in everyday clinical practice while mGFR adds important prognostic information for those where eGFR is believed to be biased.