Nephrol Dial Transplant (2006) 21: 77 83 doi:10.1093/ndt/gfi185 Advance Access publication 11 October 2005 Original Article The importance of standardization of creatinine in the implementation of guidelines and recommendations for CKD: implications for CKD management programmes Wim Van Biesen 1, Raymond Vanholder 1, Nic Veys 1, Francis Verbeke 1, Joris Delanghe 2, Dirk De Bacquer 3 and Norbert Lameire 1 1 Department of Internal Medicine, 2 Department of Clinical Biochemistry, University Hospital Ghent and 3 Department of Social Health, University Ghent, Belgium Abstract Background. In an attempt to reduce late referral and to improve the care of patients with chronic kidney disease (CKD), different organizations have issued guidelines on when to refer patients to the nephrologist. Most suggest referral of patients with a GFR below 60 ml/min/1.73 m 2, and demand referral if the GFR is below 30 ml/min/1.73 m 2. It is recommended to use the abbreviated MDRD equation to estimate GFR. This formula is, however, sensitive to the creatinine assay methodology. In addition, the impact of the implementation of such guidelines on the nephrology practice has never been evaluated. This study (i) identifies the true burden of CKD in a population and simulates the effects of a 100% implementation of the guidelines on the nephrology work load, and (ii) evaluates the validity of the estimated GFR using the abbreviated MDRD formula when routinely provided. Methods. Different laboratories (both hospital and private) in our region were asked to report on all the serum creatinine values performed during the first week of December 2004. If patients had more than one determination, only the lowest serum creatinine value was retained. Patients already known to a nephrology unit were not included. GFR was calculated using the abbreviated MDRD, using the serum creatinine as reported by these laboratories, or after correction to the MDRD-standard using different published equations. Results. 20 108 patients, with a mean age of 53.4± 16.2 years, 48% females, had at least one serum creatinine determination in the observation period. According to the K/DOQI CKD classification, 20.2, 1.6 and 0.8% of females and 13.3, 1.6 and 0.6% of Correspondence and offprint requests to: Wim Van Biesen, Renal Division, University Hospital Ghent, De Pintelaan 185, 9000 Ghent, Belgium. Email: wim.vanbiesen@ugent.be males were in stage 3, 4 and 5, respectively, when the abbreviated MDRD formula was used with the serum creatinine value as reported by the laboratories. Important differences in classifications were obtained when the different correction formulae for creatinine were applied. According to the current recommendations, this would lead to a mandatory referral of 1650 2400 CKD stage 4 patients per 100 000 inhabitants and a suggested referral of another 4100 15 360 CKD stage 3 patients per 100 000 inhabitants to a nephrology unit. Conclusion. Implementation of the current guidelines for referral of CKD patients to nephrologists would lead to an overload of the nephrology care capacities. Large differences in estimated GFRs with different corrections for serum creatinine are observed, resulting in important CKD classification differences. Standardization of serum creatinine assays is mandatory before guidelines, and especially the routine provision of the estimated GFR by the abbreviated MDRD formula, can be implemented in clinical practice. Keywords: chronic kidney disease; creatinine; end stage renal disease prevention; glomerular filtration; health economics Introduction End stage renal disease [ESRD, Chronic Kidney Disease (CKD) stage 5] is a severe condition, with a high morbidity and mortality, and a great burden both in terms of patient suffering and financial impact on the health care budget [1]. Evidence is available indicating that many patients suffer from secondary complications of chronic kidney disease (CKD) before ß The Author [2005]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
78 W. Van Biesen et al. Table 1. Participating laboratories, their methodology for serum creatinine determination and reference values Laboratory Methodology Reference values BIRNH-database Compensated Jaffé method using SMAC-Technicon 0.5 1.0 mg/dl MedicLab (ambulatory patients) Compensated Jaffé method using commercial agents (Gossin-Meyvis, 0.7 1.2 mg/dl Belgolab, Merck) on an Olympus analyser Hospital Maria Ziekenhuis Noord Limburg Compensated Jaffé method using commercial reagents (Roche, 0.5 1.3 mg/dl Mannheim, Germany) on a Cobas Integra 400 system Aalsters Stedelijk Ziekenhuis Rate-blanked compensated Jaffé method using commercial reagents 0.5 1.2 mg/dl (Roche, Mannheim) on a Cobas Integra 800 system University Hospital Ghent Rate-blanked compensated Jaffé method using commercial reagents (Roche, Mannheim, Germany) on a Modular P analyser (Hitachi, Tokyo, Japan) 0.5 1.2 mg/dl they enter a renal replacement programme. A timely correction of anaemia, maintenance of fluid balance, preservation of nutritional status, control of calcium/ phosphate homeostasis and a structured preparation of the patient for renal replacement therapy (RRT) have all proven to improve the outcome of CKD stage 5 patients. Moreover, for some causes of progressive CKD, like diabetic [2] or hypertensive [3] nephropathy, a correct and adequate treatment can delay the progression of CKD. Unfortunately, referral to the nephrologist is often delayed [4], making the implementation of all these preventive strategies cumbersome. Therefore, several scientific organizations have issued recommendations on screening and referral of patients with CKD. Although these recommendations have been conceived with the best of intentions, little is known on their effects on the treatment of ESRD on the micro-health level (effect on single patient outcome), on the macro-health level (cost benefit analysis for the health care system) or on the organizational level (impact on caregivers, i.e. nephrology units). In addition, these guidelines do not anticipate potential pitfalls encountered in the routine determination of GFR in large population groups, which mostly rely on estimations starting from serum creatinine values. Although the most commonly recommended formula, the abbreviated MDRD equation, has been validated in different patient groups, it also has its limitations [5]. Most importantly, the formula starts from serum creatinine values as determined by the Beckman Astra 8 methodology as used in the Cleveland clinic. The correction for the presence of non-creatinine chromogens is different in the different methodologies used to determine serum creatinine [6], necessitating correction formulae to convert routine serum creatinines to MDRD creatinines [5,7 9]. Although K-DOQI [10] advised already in 2000 to use only corrected creatinines in the abbreviated MDRD formula, the implementation of this advise is not widespread, and most laboratories, practitioners and nephrologists seem not to be aware of this important source of bias. This study (i) evaluates the burden of CKD; and (ii) evaluates the routine reporting of the abbreviated MDRD formula when no standardized serum creatinine assay is used. Starting from these prevalences, caveats in the practical interpretation of the guidelines and their impact are discussed, and potential solutions for the organization of CKD care are proposed. Patients and methods Five different sources of serum creatinine values were used. Four routine laboratories in Flanders were asked to provide us with a digital file containing all the serum creatinine as they were determined and reported in the first week of December 2004. All serum creatinine values requested by a nephrologist were discarded. As additional information, gender, age and a patient identification number were collected. If based on this patient identification number patients had more than one serum creatinine determination, only the lowest serum creatinine was maintained in the database. The methodology for serum creatinine determination and the normal range values for each of these laboratories are listed in Table 1. As a fifth source of creatinine values (n ¼ 10 617), we used the BIRNH (Belgian Interuniversity Registry for Nutritional Health) database [10]. This database was constructed using a random population sample based on voting lists. As we wanted to evaluate the real life impact of the current guidelines, we used the reported GFR and serum creatinine values as a basis. All the laboratories provided the requesting physician with an abbreviated MDRD-based [11] estimation of glomerular filtration (GFR): estimated GFR (in ml/min/1.73 m 2 ) ¼ 186 ms 1:154 crea age 0.203 0.742 if female 1.210 if Afro-American, as is recommended by EBPG, K-DOQI and KDIGO [1,10,12]. The MDRD formula is, however, based on serum creatinines measured with the Beckman-Astra technology. Use of other methodologies to determine serum creatinine can lead to important differences in obtained values, due to the different strategies to compensate for the presence of noncreatinine chromogens [13 15]. To calibrate the reported values with those obtained in the original MDRD study, we used different calibration methods suggested in the literature to recalculate the serum creatinines and corresponding GFRs. Coresh et al. [16] suggested to decrease measured serum creatinine values with 0.23 mg/dl over the whole range of serum creatinine values. Froissart et al. [8] suggested that the serum creatinine obtained related to the MDRD serum creatinine by a linear equation: MDRD crea ¼ 1.151 (measured creatinine in mg%) 0.107. Hallan et al. [7] used another linear equation to correct their serum creatinines: MDRD crea ¼ 1.06 (measured creatinine
Implementation of guidelines in management programmes 79 Table 2. Serum creatinine values and GFR s as obtained after implementation of difficult correction strategies Reported Coresh Hallan Froissart Toffaletti (only BIRNH samples) Males Serum creatinine (mean±sd) (mg%) 1.13±0.28 0.93±0.28 1.01±0.31 1.19±0.33 1.26±0.18 MDRD (mean±sd) (ml/min/1.73 m 2 ) 76±16 98±32 90±30 73±18 66±10 Females Serum creatinine (mean±sd) (mg%) 0.93±0.25 0.73±0.26 0.79±0.28 0.97±0.29 1.09±0.17 MDRD (mean±sd) (ml/min/1.73 m 2 ) 72±18 133±56 121±50 94±27 77±14 in mg%) 0.16. These three correction formulae were used for the samples obtained in the four routine laboratories. For the serum creatinine values obtained from the BIRNH database, we were able to calibrate directly to the standard of the MDRD study, as a conversion formula from the SMAC-Technicon technology, as used in the BIRNH study, to the Beckman-Astra was reported by Toffaletti et al. [17]: MDRD crea in mg/l ¼ (0.89 measured creatinine in mg/l þ 2.4)/0.99. This allowed us to make a more precise estimation of the prevalence of the different CKD stages in the global population. A simulation of the number of ESRD patients over 5 years was made based on an assumed average decline of GFR with 4.4 ml/min/year, a value reported in the normal diet group of the MDRD study [18]. Results In total, 10 108 patients provided one serum creatinine determination. The mean age of this patient group was 53.4±16.3 years, and 48% were female. The data of the serum creatinine values as reported and after correction with the different formulae and the respective estimated GFRs are given in Table 2 and Figure 1. Patients were then classified in chronic kidney disease classes as proposed by the KDIGO work group [19] and K/DOQI [10] according to the GFRs obtained with the different calibration methods. Cross tabulation of these classifications demonstrated an important and statistically significant different classification with all four methods (Table 3). Overall, differences were most pronounced in stage CKD2 CKD3. In CKD stage 4 and 5, the classification based on the Coresh calibration was the most discordant when compared to the other three classifications. Reported serum creatinines were comparable in the laboratories analysing samples from ambulatory patients and the BIRNH population (Figure 1), whereas samples from the two large hospitals contained more elevated serum creatinines. This suggests a selection bias in the larger hospitals, where probably patients with more comorbid conditions are followed. When the reported estimated GFRs were compared with those of the BIRNH sample, the prevalence of stage CKD3 or higher was underestimated in males but overestimated in females (19.3 vs 27.5% and 21.3 vs 9.5% in males and females, respectively, if all samples were considered, P ¼ 0.001; and 13.1 vs 27.5% and 18.3 vs 9.5%, if only ambulatory patients were considered, P ¼ 0.001). This suggests the presence of a systematic methodological bias on top of the selection bias. When only patients with a GFR <60 ml were considered to have a decline in GFR of 4.4 ml/min/ year, over a 5 year period, 2.7, 1.8, 2.0 and 2.3% of the population would progress to RRT according to the projections based on the different methodologies for GFR estimation (reported, Coresh, Hallan and Froissart, respectively). However, mortality in this subgroup is high. Based on the prevalence of diabetes and CKD stage 3 or higher in the present database and on the mortality data reported by Foley et al. [20], we estimated the mortality risk to be 17.4 per 100 patient years at risk. Taking this into account, only 0.35, 0.23, 0.26 and 0.29%, respectively, would effectively end up on RRT, the others having died before they would reach RRT. In patient numbers, however, this still represents a burden of new RRT patients ranging between 230 and 350 per 100 000 inhabitants over a 5 year period, or an annual take on rate between 46 and 70 per 100 000 inhabitants. For comparison, the actual annual take on rate in Flanders between 2000 2005 varied between 12 and 14 per 100 000 inhabitants. A treatment that would reduce GFR deterioration by 50% and decrease cardiovascular mortality by 10% would result in evolution to RRT in 0.54, 0.38, 0.47 and 0.55% or an annual take-on rate between 76 and 110 per 100 000 inhabitants. Discussion This study thus yields two important conclusions: (i) the prevalence of CKD is higher than expected and (ii) routine reporting of estimated GFRs based on the abbreviated MDRD formula can only be advocated if all laboratories use a uniform standardized methodology to determine serum creatinine. These findings have important implications when a strategy to tackle the burden of chronic kidney disease is to be developed. It is clear that for screening and prevention programmes, nephrologists will have to collaborate with the primary care physicians, by educating them on strategies to preserve residual renal function. For the RRT programmes, strategies like integrated care
80 W. Van Biesen et al. Fig. 1. (A) Serum creatinine levels as reported by the different laboratories. (B) Estimated GFRs based on the reported creatinine levels using the abbreviated MDRD equation. 1, BIRNH population (random population sample); 2, MedicLab population (private laboratory, ambulatory patients); 3, Ziekenhuis Noord Limburg population (local peripheral hospital based laboratory, mix of hospitalized and ambulatory patients); 4, Aalsters Stedelijk Ziekenhuis Laboratory (Large regional hospital based laboratory, mix of hospitalized and ambulatory patients); 5, University Hospital Ghent laboratory (ambulatory patients); 6, University Hospital Ghent laboratory (hospitalized patients).
Implementation of guidelines in management programmes 81 Table 3. Prevalence of difficult stages of CKD as defined by K-DOQI Reported Coresh Hallan Froissart Toffaletti (only BIRNH samples) Males CKD1 (% of total) 42.2 69.9 61.5 37.9 1.4 CKD2 (% of total) 38.4 17.2 22.7 40.6 71.1 CKD3 (% of total) 14.5 8.7 11.0 15.8 27.4 CKD4 (% of total) 3.4 3.0 3.3 4.0 0.1 CKD5 (% of total) 1.4 1.2 1.5 1.7 0.0 Females CKD1 (% of total) 38.7 84.5 81.7 69.1 16.4 CKD2 (% of total) 40.0 7.8 9.6 19.8 74.1 CKD3 (% of total) 16.6 5.0 6.0 8.0 9.4 CKD4 (% of total) 3.0 2.0 1.9 2.1 0.1 CKD5 (% of total) 1.7 0.6 0.8 1.1 0 Chi square statistic for classification: P<0.0001 for all 5 CKD classes between the 5 different methods. approach [21] will have to be developed to obtain an economically and physically sustainable growth of the RRT population. Reporting of estimated GFRs based on the abbreviated MDRD equation, resulting in complete referral to and treatment by nephrologists of all patients according to current recommendations [10,13,19], would result in an overload of the nephrological capacity. This overload would be on two levels. First of all, an impressive number of patients in stage CKD3 or 4 would need consultation by a nephrologist. Second, the number of patients on RRT would dramatically increase. The projected numbers, however, differ quite impressively, especially in the moderate CKD stage 3 level, according to the methodology used for the determination of serum creatinine, as there is no standardization as yet for the correction of noncreatinine chromogens [15]. The projected number of patients in need of RRT is less dependent on the correction method used, because the abbreviated MRDR formula is more robust in the higher serum creatinine range [22], and because these patient numbers depend essentially on the percentage of patients who will survive until CKD stage 5 [23]. Surprisingly, the prevalence of CKD stage 3 or higher was highest in the BIRNH-database, a random population sample where an exact correction of the serum creatinine to match the MDRD methodology was performed. Implementing the guidelines supposes a valid test which discriminates those who need further investigation from those who certainly don t. This test should be cheap, sensitive and consistent. In addition, a nonexpensive first confirmation with a higher specificity to substantially decrease the number of false positives without creating false negatives should be available. This is especially warranted when the golden standard confirmation test is expensive because of the cost of a single test on itself, or because the high number of false positives to be tested. For CKD, most guidelines advocate as a screening test the use of serum creatinine and the abbreviated MDRD formula-derived GFR. The abbreviated MDRD formula has been validated in different populations [14], and is considered to be very precise [22]. However, the accuracy of the test is questionable, as the test is extremely sensitive to the fact that the equation has been developed using the Beckman CX3 technology, resulting in very precise serum creatinine values with a nearly complete correction for non-creatinine chromogens. Most laboratories use at present other technologies, with less complete correction for the presence of non-creatinine chromogens, resulting in higher reported serum creatinine values [15,24]. Using these uncorrected serum creatinines in the MDRD equation results in spuriously low GFR rates, especially in the near normal serum creatinine range, as the impact of non-creatinine chromogens on the measurement is highest in this range. Bostom et al. [22] found that the accuracy of Cockroft Gault-derived GFR was superior to the MDRD-derived GFR if serum creatinine values were not calibrated to the Beckman CX3 technology. These calibration differences can be very substantial [5,7,8]. As demonstrated by this article, every single laboratory should elaborate its own specific calibration formula, even if they use a commercial creatinine assay, as even small changes in methodology will result in deviations of the final result. Thus, even if every laboratory in the world would correct by a general formulae its serum creatinine assay to the Beckman CX3 methodology to make the abbreviated MDRD formula result in valid GFRs, it appears that this correction does not result in a valid, reproducible estimate of GFR. To avoid such misinterpretations, KDIGO recommends to provide only a MDRD-based GFR to nonnephrologists if the estimated GFR is below 60 ml/min [19], but even in the 30 60 ml range, substantial errors can occur. In their guidelines, K/DOQI stresses the importance of the creatinine correction [10], but according to our data, this recommendation seems not to work in routine clinical practice. If no standardized methodology to measure serum creatinine is adopted, the routine use of serum creatinine-based equations to determine GFR will remain unreliable. Although the abbreviated MDRD equation is more robust in the low GFR ranges, there is still some
82 W. Van Biesen et al. deviation according to the used correction method. This underscores the danger to take decisions on starting of RRT only based on the estimated clearances. This also implies that studies on the impact of time of start of RRT based on estimated clearances [25] are probably biased. For progression of chronic kidney disease, prevention can slow down the rate of decline of renal function [3,18,26,28]. Most interventions relate to life style changes, like diet and smoking restriction, or result in prescription of medications in asymptomatic patients, like for hypertension, proteinuria, lipid disturbances or insulin resistance. Therefore, motivating patients to implement these preventive measures takes time and commitment. In view of the patient numbers deduced from this survey, it is impossible that all these patients can exclusively be followed by a nephrologist. Therefore, the nephrology community should establish educational sessions for the general practitioners, not only on how to screen for CKD, but also which preventive measures to take. Organizations like KDIGO, and national and international societies have a responsibility here to provide the nephrologist with educational tools to instruct primary care physicians. With regard to secondary prevention, it is of note that the impact of decreased GFR on mortality already starts at very modest degrees of renal impairment [29,30]. It is unclear how a perfect creatinine measurement and its derived perfect GFR would relate to increased mortality, as these new values will be different from the ones we are used to. Nevertheless, the cardiovascular impact of early renal impairment will, of course, not be different. Slowing down decline of residual renal function does not necessarily result in a decrease in the number of patients on RRT [23]. Indeed, preventive strategies like e.g. ACE-i do not only result in a slower decline of RRF, but also in an improved cardiovascular outcome. The impact of the latter might outweigh the former, resulting in a paradoxical increase instead of a decrease of patients in need of RRT. Based on our survey, the take-on-rate for RRT in Flanders can increase 5-fold within the next five years after a complete implementation of the current recommendations. It is neither economically nor physically sustainable if all these patients would be treated by hospital-based haemodialysis. Strategies to increase the use of low cost dialysis modalities, like peritoneal dialysis [21], home haemodialysis and transplantation should be encouraged. Also for this purpose, the nephrology community needs to establish management protocols. At our unit, we have established a Chronic Kidney Disease clinic, where patients are informed about kidney disease, and modalities of renal replacement, allowing the patient to make an informed decision on his/her preferred modality. It has been demonstrated that such an approach results in an important number of patients opting for home-based treatments [31,32]. The present survey has several limitations, and does not intend to be a thorough epidemiological evaluation. The samples from the different laboratories are probably not completely representative of the general population, as we have only samples from patients sent for a laboratory investigation. The BIRNH sample, in contrast, is a random population sample based on voting lists, and is thus representative for the overall population, and in this sample, with an appropriate correction of the serum creatinine. Surprisingly, by this approach, the projected numbers of patients with CKD stage 3 or more are even higher. Our results are also concordant with other reported series [9,29]. The calculation of the decline of residual renal function and mortality during the progression to ESRD has been based on values reported by large series like the MDRD study, and are not real measurements. Nevertheless, the projections give an idea of what might happen in the near future, and are especially valid with regard to the observed differences obtained with the different correction methods, as these are independent of the limitations of this survey. In conclusion, the current guidelines as issued are valuable, but their practical implementation is hampered by the lack of a standardized serum creatinine assay. A new simple and accurate global standard for the determination of kidney function is clearly needed. Although the projected numbers differ widely, it is clear that an epidemic of CKD and ESRD is to be expected, with enormous medical and financial consequences. Collaboration of the nephrology community with primary care organizations will be mandatory. The number of patients on RRT will also grow impressively, and this despite the measures taken to delay progression of CKD in the individual patient, mostly because of a better cardiovascular survival of these patients. Structured programmes should be established to increase the share of low cost renal replacement therapies. It is clear that in order to improve the outcome of chronic disease, management skills, organizational strength and political vision will be as important as scientific eminence. Acknowledgements. We thank the clinical biochemistry laboratories from the University Hospital Ghent (J. D.), from the Aalsters Stedelijk Ziekenhuis (Dr De Bie and Dr Coppens), from the Maria Ziekenhuis Noord Limburg (Dr E. Van Den Driessche) and from Mediclab Aalst (Dr De Clerck F.) for their invaluable cooperation and support. Conflict of interest statement. None declared. References 1. Eknoyan G, Lameire N, Barsoum R et al. The burden of kidney disease: improving global outcomes. Kidney Int. 2004; 66: 1310 1314 2. Hunsicker LG. Emerging trends for prevention and treatment of diabetic nephropathy: blockade of the RAAS and BP control. J Manag Care Pharm 2004; 10: S12 S17 3. Bakris GL, Williams M, Dworkin L et al. Preserving renal function in adults with hypertension and diabetes: a consensus approach. National Kidney Foundation Hypertension and
Implementation of guidelines in management programmes 83 Diabetes Executive Committees Working Group. Am J Kidney Dis 2000; 36: 646 661 4. Lameire N, Van Biesen W. The pattern of referral of patients with end-stage renal disease to the nephrologist a European survey. Nephrol Dial Transplant 1999; 14 [Suppl 6]: 16 23 5. Coresh J, Eknoyan G, Levey AS. Estimating the prevalence of low glomerular filtration rate requires attention to the creatinine assay calibration. J Am Soc Nephrol 2002; 13: 2811 2812 6. Schwartz GJ, Brion LP, Spitzer A. The use of plasma creatinine concentration for estimating glomerular filtration rate in infants, children, and adolescents. Pediatr Clin North Am 1987; 34: 571 590 7. Hallan S, Asberg A, Lindberg M, Johnsen H. Validation of the Modification of Diet in Renal Disease formula for estimating GFR with special emphasis on calibration of the serum creatinine assay. Am J Kidney Dis 2004; 44: 84 93 8. Froissart M, Rossert J, Jacquot C, Paillard M, Houillier P. Predictive performance of the modification of diet in renal disease and Cockcroft-Gault equations for estimating renal function. J Am Soc Nephrol 2005; 16: 763 773 9. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004; 351: 1296 1305 10. Kornitzer M, Dramaix M. The Belgian Interuniversity Research on Nutrition and Health (B.I.R.N.H.): general introduction. For the B.I.R.N.H. Study Group. Acta Cardiol 1989; 44: 89 99 11. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002; 39: S1 266 12. Section I. Measurement of renal function, when to refer and when to start dialysis. Nephrol Dial Transplant 2002; 17 [Suppl 7]: 7 15 13. Levey AS, Bosch JP, Lewis JB et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: 461 470 14. Stevens LA, Levey AS. Measurement of kidney function. Med Clin North Am 2005; 89: 457 473 15. Hanser AM, Hym B, Michotey O et al. Comparison of methods for the determination of blood creatinine. Ann Biol Clin (Paris) 2001; 59: 737 742 16. Coresh J, Astor BC, McQuillan G et al. Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis 2002; 39: 920 929 17. Toffaletti J, Blosser N, Hall T, Smith S, Tompkins D. An automated dry-slide enzymatic method evaluated for measurement of creatinine in serum. Clin Chem 1983; 29: 684 687 18. Levey AS, Greene T, Beck GJ et al. Dietary protein restriction and the progression of chronic renal disease: what have all of the results of the MDRD study shown? Modification of Diet in Renal Disease Study group. J Am Soc Nephrol 1999; 10: 2426 2439 19. Levey AS, Eckardt KU, Tsukamoto Y et al. Definition and classification of chronic kidney disease: a position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int 2005; 67: 2089 2100 20. Foley RN, Murray AM, Li S et al. Chronic kidney disease and the risk for cardiovascular disease, renal replacement, and death in the United States Medicare population, 1998 to 1999. J Am Soc Nephrol 2005; 16: 489 495 21. Van Biesen W, Vanholder R, Lameire N. The role of peritoneal dialysis as the first-line renal replacement modality. Perit Dial Int 2000; 20: 375 383 22. Bostom AG, Kronenberg F, Ritz E. Predictive performance of renal function equations for patients with chronic kidney disease and normal serum creatinine levels. J Am Soc Nephrol 2002; 13: 2140 2144 23. Hsu CY, Vittinghoff E, Lin F, Shlipak MG. The incidence of end-stage renal disease is increasing faster than the prevalence of chronic renal insufficiency. Ann Intern Med 2004; 141: 95 101 24. Wuyts B, Bernard D, Van den NN et al. Reevaluation of formulas for predicting creatinine clearance in adults and children, using compensated creatinine methods. Clin Chem 2003; 49: 1011 1014 25. Korevaar JC, Jansen MA, Dekker FW et al. When to initiate dialysis: effect of proposed US guidelines on survival. Lancet 2001; 358: 1046 1050 26. Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. UK Prospective Diabetes Study Group. BMJ 1998; 317: 703 713 27. Effects of ramipril on cardiovascular and microvascular outcomes in people with diabetes mellitus: results of the HOPE study and MICRO-HOPE substudy. Heart Outcomes Prevention Evaluation Study Investigators. Lancet 2000; 355: 253 259 28. Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: the HOPE randomized trial. Ann Intern Med 2001; 134: 629 636 29. Shlipak MG, Sarnak MJ, Katz R et al. Cystatin C and the risk of death and cardiovascular events among elderly persons. N Engl J Med 2005; 352: 2049 2060 30. Vanholder R, Massy Z, Argiles A et al. Chronic kidney disease as cause of cardiovascular morbidity and mortality. Nephrol Dial Transplant 2005; 20: 1048 1056 31. Schreiber M, Ilamathi E, Wolfson M et al. Preliminary findings from the National Pre-ESRD Education Initiative. Nephrol News Issues 2000; 14: 44 46 32. Goovaerts T, Jadoul M, Goffin E. Influence of a Pre-Dialysis Education Programme (PDEP) on the mode of renal replacement therapy. Nephrol Dial Transplant 2005; 20(9): 1842 1847 Received for publication: 15.7.05 Accepted in revised form: 6.9.05