It is only an estimate. A significant error is possible. eGFR is most likely to be inaccurate in people at extremes of body type, for example malnourished, amputees, etc. See Race, below. It is not valid in pregnant women or in children (see Age, below).
Confidence intervals: 90% confidence intervals are quite wide, e.g. 90% of patients will have a measured GFR within 30% of their estimated GFR. 98% have measured values within 50% of the estimated value. For an individual patient values will be much more consistent than this, just as creatinine values are – e.g. a 20% fall in eGFR is certain to reflect an important change.
Race: Some racial groups may not fit the MDRD equation well. It was originally validated for US white and black patients. For Afro Caribbean black patients, eGFR was 21% higher for any given creatinine in the MDRD study. So if race was not included in the estimate you have, it should be increased by approx. 20% for a black patient. In the UK white population the equation seems to work quite well. It may not perform so well in all racial groups.
Not so good near normal: The MDRD equation tends to underestimate normal or near-normal function, so slightly low values should not be over-interpreted. Furthermore, laboratory differences in creatinine estimations may make significant differences. Routine reporting of eGFR values >90 is not recommended and many labs are now reporting all values over 60 as >60. Note however that a significant (e.g. 20%) rise in creatinine while eGFR is >60 may still be important as it will usually reflect a real change in GFR.
Creatinine level must be stable: eGFR calculations assume that the level of creatinine in the blood is stable over days or longer. They are not valid if it is changing.
Age: The MDRD equation is not valid for under-18s. Use the Counahan-Barrat method for children – e.g. the paediatric calculator from Steve Fadem at nephron.com
Which MDRD equation? in the UK, most laboratories calculate eGFR on all samples sent for creatinine measurement. Since 2006 the equation they use should take into account local variations in accuracy of creatinine assays, so eGFR values obtained in this way should be a little more accurate than those generated by any of the online calculators (including ours). Technically, this is the “175” equation used with a correction factor when necessary, rather than the original “186” equation.