Epidemiologic studies in different Western countries have recently shown that prevalence of CKD, defined as GFR under 60 mL/min/1.73 m2, is about 10% in the global population [1, 16]. These data have been obtained with the MDRD study equation using a well calibrated serum creatinine [17, 18]. However, the use of this equation is not free from criticisms. We, and others, have demonstrated that this equation tends to strongly underestimate GFR in healthy populations and, more generally, in patients with normal or near normal creatinine values [2, 10, 11, 19]. Admitting this fact, the Levey's group has recently proposed a new equation which is thought to be especially better in the higher GFR range (over 60 ml/min/1.73 m2). In this view, the new CKD-EPI equations are different following the creatinine value (0.7 mg/dL for the women and 0.9 mg/dL for the men). This adaptation seems logical as relationship between GFR and creatinine is different in healthy as compared to CKD subjects. The better accuracy of the new equation is also certainly explained by the authors because the inclusion of healthy subjects in the equations development study (13% i.e. 694 healthy kidney donors). Logically, using the new CKD-EPI equations, the prevalence of stage 3 CKD in our population is significantly lower than if the MDRD study equation is used (7.98% versus 11.04%). Over 1992 patients screened, 61 (3.06%) were classified as having stage 3 CKD with the MDRD study equation compared to the CKD-EPI study equation. These data are not negligible from an epidemiological point of view.
Differences between MDRD and CKD-EPI equations are especially larger for the highest estimated GFR values. If we arbitrary fix a limit of 60 ml/min/1.73 m2 estimated by the MDRD study equations, differences between the two equations are significantly higher for the values over than below 60 ml/min/1.73 m2 (-2.3 ± 1.9 versus -2.7 ± 7.4 ml/min/1.73 m2, respectively). These differences are logical according to differences between equations. More astonishing is the difference observed in subgroup analysis according to gender. The difference between the two equations seems more impressive in women than in men. This is not explained by the GFR level or age as women have higher mean GFR and are younger (see below for the age effect). This difference could be explained by the lower cut-off chosen in women for the CKD-EPI equations. As relationship between creatinine and GFR is exponential, it could be logical that consequences on difference results between the two equations are more important in women. Nevertheless, more studies in the future are needed to explain such discrepancies between estimators regarding to gender because, obviously, one of the two equations is especially inaccurate in women.
In their presenting CKD-EPI article, Levey et al have also compared prevalence of CKD in the NHANES study. If we consider the same definition of stage 3 CKD, these authors have found that prevalence of CKD in the NHANES study was 9.88% with the CKD-EPI equation and 10.82% with the MDRD study equations (Appendix Table 6 in ). Difference in prevalence is more impressive in our own study. One explanation could be the higher mean age in our population. Indeed, one major difference between the two equations is the "age factor". In the MDRD study equation, a constant exponent is applied to age (age-0.203) whereas age is an exponent in the CKD-EPI equation (0.993age). Indeed, we find a significant correlation between age and difference between MDRD and CKD-EPI results (regression coefficient of 0.39 in multiple regression analysis). However, even if the MDRD study equation's performance in older population remains controversial [20, 21], the performance of the CKD-EPI equation in patients or subjects over 70 years old has not been studied (only 3% of patients between 70 and 75 years old in development data of the CKD-EPI equations study). Discrepancy between our results and NHANES study results could also be the result of potential anthropometrical differences between American and European populations.
The new CKD-EPI equation is certainly interesting and its performance will probably be better than the MDRD study equation in population free from renal disease. Yet, some limitations may be advanced. Firstly, the choice of the creatinine cut-off is logically different according to sex (0.9 mg/dL for men and 0.7 mg/dL for women). However, it would be also logical that equations vary with age as relationship between creatinine levels is also strongly influenced by age . Secondly, we underline, once again, the lack of data regarding older patients (more than 70 years old) that are clearly underrepresented in the CKD-EPI study. Thirdly, we have recently criticized the way the new "IDMS" traceable MDRD and more precisely the factor 175 has been obtained to make results IDMS traceable [9, 10]. This criticism is also valid for the CKD-EPI equation because serum creatinine measurements from the development data are coming from studies (MDRD and AASK for example) where creatinine had been measured with Jaffé methods. So, from our point of view, the factors used in the CKD-EPI equations (144 for women and 141 for men) are too low inducing a systematic overestimation of CKD prevalence. Lastly, the major criticism for the new CKD-EPI equations is its lack of advantage regarding its precision in estimating GFR. Indeed, in the Levey study, if for subjects with GFR over 60 ml/min/1.73 m2, the bias with measured GFR is improved when using the CKD-EPI equations as compared to the MDRD equation (median difference of 3.5 versus 10.6 mL/min/1.73 m2), however, the precision of the CKD-EPI equation in the same range of GFR doesn't appear better (and even seems slightly worse) than those of the MDRD (precision is reflected by interquartile range for differences: 25.7 versus 24.2 mL/min/1.73 m2, respectively) . So, if GFR estimation by CKD-EPI equation has an improved systematic bias and accuracy, this equation does not improve the precision of the estimation. This is disappointing but not astonishing as bias is, by nature, systematic although precision is random and especially linked to the precision of the creatinine measurement. The latter is not improved in the CKD-EPI equation in comparison with the MDRD equation (as already mentioned, creatinine has been measured with Jaffé methods) [8, 9].
There are some limitations to our study. First, the main limitation is linked to the fact that we have not measured GFR with a reference method. So, even if we have indirect arguments to affirm that MDRD study equations overestimate CKD prevalence in global populations, such an assertion can only be checked if a reference method GFR is used. Our data only underline potential strong discrepancies between results in epidemiological studies when either the MDRD or the CKD-EPI study equations are used. Epidemiological studies on renal function in the global population are still urgently waited but it represents heavy work. Second, our stage 3 CKD prevalence data are of interest only because they illustrate these discrepancies. As our population is clearly not representative of the Belgian population (because only volunteers are included), our stage 3 CKD prevalence results must not be considered for epidemiological considerations. For this reason, data regarding proteinuria or hypertension status that are lacking in this study are of relatively poor interest in our study. Third, we have no data on the ethnicity. As the ethnicity factor varies following the equations, this could be source of bias. However, in Belgium, Caucasians are, from far, the dominant ethnic group. Moreover, there is little doubt that differences observed in our study are not due to the ethnic factor (1.21 in the MDRD study equation and 1.16 in the CKD-EPI equations). Four, like in several epidemiological studies, our subjects have been tested only once, although definition of CKD sensu stricto implies that two or three testing have been undertaken over a three months period. Last, we have defined CKD as GFR less than 60 mL/min/1.73 m2. The definition of CKD is however subject of debate, notably in elderly population [2, 22]. Whatever, our data do not permit to bring to close this debate.