The normal ageing process is characterized by a decrease in functional performance. "Functional normality" evolves throughout life and hence has no fixed value. The HUGE formula was developed to predict renal insufficiency.
Descrição original:
Título original
Haematocrit, Blood Urea and Gender (Huge) as a Screening Test for Cri
The normal ageing process is characterized by a decrease in functional performance. "Functional normality" evolves throughout life and hence has no fixed value. The HUGE formula was developed to predict renal insufficiency.
The normal ageing process is characterized by a decrease in functional performance. "Functional normality" evolves throughout life and hence has no fixed value. The HUGE formula was developed to predict renal insufficiency.
480 Introduction Traditionally, CRI has been identified by a decrease in GFR, calculated by clearance techniques (1). It is a widespread medical trend to predict CRI on the basis of calculations of the critical value of calculated or eGFR lower than 60 ml/min/1.73 m 2 . This procedure has at least two pitfalls. The first is that eGFR values do not always indicate the presence of CRI. The second is that the characteristics of the normal aging process are neglected. From the physiological point of view, the normal ageing process is characterized by a decrease in functional performance, including a decreased GFR, in comparison with healthy young individuals. Despite this fall, renal function retains its capability to maintain the equilibrium of the internal milieu of healthy elderly people (2). Thus, one of the most difficult tasks for physicians is to determine the presence of renal insufficiency in aged individuals, particularly those older than 70 years of age, since functional normality evolves throughout life and hence has no fixed value. GFR reaches a peak between 120-130 ml/min/ 1.73 m 2 at the age of 30, and thereafter decreases at a constant rate of 1 ml/min/year, although for persons aged 70-110 years, the decline averages 1.05 ml/min/year (3). Study design This was a predictive, transversal, non-randomized, multicentric and multinational study carried out with the collaboration of 11 nephrologists, 4 internal medicine physicians, 3 cardiologists, 3 geriatricians, 2 specialists in clinical biochemistry, 1 specialist in nuclear medicine and 3 family doctors from Argentina, Portugal and Spain. CRI diagnosis criteria Chronic renal insufficiency (CRI) and NCRI (patients without CRI) diagnosis were based on the decision of doctors after a holistic evaluation of the patient, including clinical findings, image, and biochemical variables, and not exclusively on a calculated or eGFR lower than 60 ml/min/1.73 m 2 . The assembly procedure, including data acquisition, data evaluation, the detection of variables with discriminative power, the development and validation of the HUGE formula, the achievement of our gold standard for the diagnosis of CRI or NCRI, the confirmed true diagnostic core database (CTDB), and the matching of the HUGE formula with the MDRD and CKD-EPI formulas in large database, is shown in the beside box diagram ( Figure 1). HAEMATOCRIT, BLOOD UREA AND GENDER (HUGE) AS A SCREENING TEST FOR CRI THE VALUE OF A FORMULA INCLUDING HAEMATOCRIT, BLOOD UREA AND GENDER (HUGE) AS A SCREENING TEST FOR CHRONIC RENAL INSUFFICIENCY J.A. ALVAREZ-GREGORI 1 , N. R. ROBLES 2 , C. MENA 3 , R. ARDANUY 4 , R. JAUREGUI 5 , J.F. MACIAS-NUNEZ 6 1. Honorific Professor of Geriatrics. Faculty of Medicine. University of Salamanca. Internal Resident. Casto Prieto Health Centre. Family and Community Medicine Unit. Salamanca. Spain; 2. Nephrologist. Renal. Unit. University Hospital. Infanta Cristina. Associated Director, Chair FIIPERVA of Vascular Risk Factors, Geriatrics. Badajoz. Spain; 3. Mena C. General Practitioner. County Hospital Don Benito, Badajoz. Spain; 4. Ardanuy R. Professor of Statistics Department Faculty of Sciences. University of Salamanca. Spain; 5. Family and Community Medicine Unit, Buenos Aires, Argentina; 6. Titular Professor of Medicine. Director Chair FIIPERVA of Vascular Risk Factors, Geriatrics. Chief Clinician of Nephrology Department. University Hospital. University of Salamanca. Spain. Corresponding Author: Macas Nez JF. Titular Professor of Medicine. Director, Chair FIIPERVA of Vascular Risk Factors, Geriatrics. Chief Clinician of Nephrology. University Hospital. Faculty of Medicine. Avda. Alfonso X el Sabio s/n, Salamanca 37007, Spain. Phone: +34 923234540 ext. 1954, e-mail: jfmacias@usal.es Abstract: Introduction: Despite increasing use in clinical practice, an estimated glomerular filtration rate value (eGFR) of <60 ml/min/1.73 m 2 does not necessarily indicate the existence of chronic renal insufficiency (CRI) and this may lead to an over-estimate of CRI particularly in persons seventy years or older. Aim: To find a screening test able to differentiate CRI from the decrease in GFR normally associated with the renal ageing process. Methods: Medical information of 487 individuals of both sexes aged 16-102 was obtained from nephrologists, internal medicine physicians, cardiologists, geriatricians, family and nuclear medicine doctors from Argentina, Portugal and Spain. Data were assessed and statistically analysed using logistic regression techniques. From the discriminative variables it was derived the HUGE formula. Results: A formula including haematocrit , blood urea, and gender (HUGE), diagnoses CRI regardless of the variables of age, blood creatinine, creatinine clearance, or other eGFR. The HUGE formula is: L= 2.505458 (0.264418 x Hematocrit) + (0.118100 x Urea) [+ 1.383960 if male]. If L is a negative number the individual does not have CRI; if L is a positive number, CRI is present. Our data demonstrate that the HUGE formula is more reliable than MDRD and CKD-EPI, particularly in persons aged over 70. Conclusions: Our HUGE screening formula offers a straightforward, easily available and inexpensive method for differentiating between CRI and eGFR < 60 ml/min/1.73 m 2 that will prevent a considerable number of aged healthy persons, as much as 1.700.000 in Spain and 2.600.000 in UK, to be excluded from clinical assays or treatments contraindicated in CRI. Key words: Urea, chronic renal insufficiency, screening, haematocrit. Received June 30, 2010 Accepted for publication August 30, 2010 Figure 1 Expplanatory diagram showing the methodological pathway to achieve the HUGE formula (according to CONSORT statement) Data acquisition A database template was designed containing the variables of age, sex, body weight, height, abdominal girth, haemoglobin, haematocrit , plasma ferritin, vit B12, folic acid levels, blood urea, blood creatinine, 24-hour diuresis, urinary creatinine, proteinuria detectable with routine urinanalyses (> 300mg/24 h), blood uric acid, calcium, phosphate, PTH, and C- Cystatin. It was sent to our collaborators who were asked to include patients diagnosed with CRI (recorded as 1) and those with non chronic renal insufficiency (NCRI) (recorded as 0) seen and followed-up in their current out-patient and in-patient practice (first group of doctors). Evaluation of the predictive variables The received data were sent to a different group of collaborators (second group of doctors) asking them to classify persons as CRI or NCRI on the basis of analytical data without knowledge of the original diagnoses. Our gold standard for CRI or NCRI was the coincidence between the diagnoses of CRI or NCRI made for the first group of doctors with the prediction made for the second group of doctors in the 376 individuals who met all the criteria required for evaluation. Calculation of the HUGE formula The variables that had p-values > 0.10 (Student t test for independent data) lacked discriminative ability to detect CRI were: ferritin, folic acid, abdominal perimeter and height. The rest of the variables with a p value < 0.10, showing good ability to detect CRI, were: uric acid, vitamin B12, calcium, C- Cystatin, serum creatinine, urine creatinine, diuresis, age, weight, iron, haemoglobin, haematocrit, body mass index, transferrin, vitamin D3, phosphorus, pH (venous), HCO3- (venous), PTH and blood urea. A simple model of binary logistic regression was applied (4, 5) to variables with discriminative ability, resulting in the formula: Where L is a linear function of the discriminative variables: L= 2.505458 (0.264418 x Hematocrit) + (0.118100 x Urea) [+ 1.383960 if male]. Validation of the HUGE formula The Nuclear Medicine Unit of the University Hospital of Salamanca provided us with 111 patients in which the GFR had been obtained using the clearance of Diethylenetriami- nepentaacetic acid (DTPA) whose records were carefully revised by a nephrologist and a family doctor to confirm the already existing diagnosis of CRI or NCRI by clinical and laboratory means. HUGE was applied to those 111 subjects for validation purposes. Confirmed true diagnosis core database. (CTDB) CTDB was composed of 487 (376+111) individuals with the coincidence of CRI or NCRI, by the two groups of doctors already mentioned, which served as the basis for the calculations that allowed us to re-confirm the reliability of the HUGE formula and the comparison between HUGE and other formulae discussed here in. The cross tables technique was applied to compare the clinical diagnoses with the results obtained with the HUGE, MDRD and CKD-EPI formulas. Additional testing of the HUGE formula in large databases To determine whether the percentage differences of CRI and NCRI between the HUGE formula and MDRD was statistically different from that obtained in the 487 subjects of our confirmed CTDB, we tested larger databases with a total of 125.373 subjects provided by the University Hospital of Salamanca and our two partner centres, the Infanta Cristina University Hospital and the County Hospital of Don Benito (Extremadura, Spain) Statistics Variables with no discriminative power to detect subjects with CRI, (p>0.10) were not included in the model. A simple model of logistic binary regression was applied to variables with discriminative power (4, 5). The lowest limits of confidence for the well classified percentages were calculated by means of the exact formula that correlates the binomial distribution with the F of Snedecor (6, 7). The exact bilateral test of McNemar was applied on the cross tables to check whether there were statistically significant differences between CRI or NCRI classified in the validation of HUGE and the true presence of CRI or NCRI (8). Statistical evaluations of the data were carried out with SPSS program version 15.0.1 (Nov 2006) Copyright SPSS Inc. 1989-2009.Chicago, Illinois. Results Statistical tools were applied to the confirmed true diagnosis (CRI or NCRI) CTDB of 487 individuals (206 female and 281 male), aged 16-101 years. Biochemical variables from both groups (CRI and NCRI) are shown in Table 1. JNHA: GERIATRIC SCIENCE The Journal of Nutrition, Health & Aging Volume 15, Number 6, 2011 481 1 1 + e -L HAEMATOCRIT, BLOOD UREA AND GENDER (HUGE) AS A SCREENING TEST FOR CRI The Journal of Nutrition, Health & Aging Volume 15, Number 6, 2011 482 Table 1 Descriptive data of the variables in the study. Variables CRI CRI CRI NCRI NCRI NCRI t-Student N (cases) Mean St. Dev. N (cases) Mean St. Dev. AGE (years) 182 72.4011 12.04497 304 55.9079 16.26182 p<0.05 * Weight (Kgs) 98 68.8643 17.95198 261 84.3920 21.46188 p<0.05 * Height (cms) 97 163.1546 8.71820 261 163.4751 10.27003 p=0.785 Hb (g/dL) 169 12.3024 2.82276 207 14.5425 2.92542 p<0.05 * Hto (%) 180 36.3972 5.63808 305 43.3030 4.72725 p<0.05 * Fe(mg/dL) 85 66.1388 31.03267 131 85.8702 31.90139 p<0.05 * Folic acid (ng/ml) 45 10.4036 3.87976 124 9.5097 3.38640 p=0.147 B12 vit. (pg/ml) 44 412.0750 246.07483 123 497.2378 273.55242 p=0.071 Urea(mg/Dl) 179 109.5251 60.27049 304 40.5625 13.12152 p<0.05 * Serum Creat (mg/dL) 182 2.6991 1.96532 305 .8942 .19146 p<0.05 * Urine Creat (mg/dL) 141 58.0053 44.12296 181 84.0955 61.94116 p<0.05 * Calcium (mg/dL) 150 9.2307 .82144 156 9.4574 .48364 p<0.05 * Phosporus (mg/dL) 148 4.1890 2.77665 155 3.4861 .56861 p<0.05 * Uric Acid (mg/dL) 157 7.2420 2.15782 205 5.5748 1.47978 p<0.05 * PTH (pg/ml) 101 125.3603 135.87794 144 47.3243 18.16282 p<0.05 * Vit. D3 (ng/mL) 47 26.6862 20.35758 114 33.0798 13.04105 p<0.05 * Albumin (g/dL) 77 3.8431 .73161 155 4.5013 .37329 p<0.05 * Cystatin C 50 1.6554 .55881 154 .8403 .19900 p<0.05 * Venous Ph 65 7.3143 .08522 42 7.3824 .05127 p<0.05 * HCO3 (mmol/L) 113 22.8345 4.60589 68 25.0588 3.39865 p<0.05 * * p < 0.05 There is a significant difference between variables of both groups; (Students t test for the comparison of means of independent data; CI 95%) Table 2 Comparison between CRI predicted with HUGE, MDRD and CKD-EPI versus real observed CRI in the population of the study (under and over 70 years old) Under 70 years Predicted CRI HUGE Predicted CRI MDRD Predicted CRICKD-EPI N=285 NCRI CRI NCRI CRI NCRI CRI OBSERVED NCRI 223 7 223 8 226 5 CRI 10 42 2 52 2 52 Chi-Square of Pearson p<0.001 p<0.001 p<0.001 McNemar P=0,629 P=0,109 P=0,453 Phi 0.796 0.892 0.922 Over 70 years Predicted CRI HUGE Predicted CRI MDRD Predicted CRI CKD-EPI N=201 NCRI CRI NCRI CRI NCRI CRI OBSERVED NCRI 68 5 60 13 55 18 CRI 9 116 1 127 0 128 Chi-Square of Pearson p<0.001 p<0.001 p<0.001 McNemar P=0,424 P=0,002 P=0,00 Phi 0.851 0.852 0.813 NCRI (Non Chronic Renal Insufficiency); CRI (Chronic Renal Insufficiency) Table 3 Cross Tables in all subjects, younger and older 70. HUGE vs. MDRD Whole population MDRD Under 70 MDRD Over 70 MDRD N=125326 N=84957 N=40369 NCRI CRI NCRI CRI NCRI CRI HUGE NCRI 90430 7263 HUGE NCRI 72305 2349 HUGE NCRI 18125 4914 CRI 5825 21808 CRI 2936 7367 CRI 2889 14441 483 The Journal of Nutrition, Health & Aging Volume 15, Number 6, 2011 JNHA: GERIATRIC SCIENCE We found that the association of haematocrit, urea and gender (HUGE) according to the following formula: L= 2.505458 (0.264418 x Hematocrit) + (0.118100 x Urea) [+ 1.383960 if male] showed the highest ability to discriminate CRI from NCRI individuals: If L is lower than 0, it means that the individual does not have CRI. If L is higher than 0, it means that the individual suffers from CRI. In the population below 70 years of age (Table 2), no significant differences were found by the McNemar exact test in the cross tables for HUGE, MDRD and CKD-EPI. In persons aged over 70 years MDRD showed significant differences with the diagnosis of CRI (p=0.002, MacNemar test), wrongly classifying 13 subjects as CRI instead of NCRI. In the case of the CKD-EPI, this mistaken discrimination occurred in 18 individuals out of 201 (9%) (P= 0.001, McNemar test), whereas HUGE only misclassified 5 subjects, with no statistical differences with the true diagnosis of CRI from our CTDB (p= 0,424. MacNemar tes)t (Table 2). Comparison between HUGE and MDRD formulae in large data base are given in Table 3. Out of the 125,373 individuals included, 40,369 of them were older than 70 years. In this group, the use of MDRD formula diagnosed 19,355 as suffering from CRI, whereas the use of HUGE only diagnosed 17,330 as having CRI, a difference of 10.46 %. Data concerning sensitivity, specificity, positive predictive value and negative predictive value of persons aged > 70 years of the HUGE, MDRD and CKD-EPI formulae with the true diagnosis of CRI or NCRI in our CTDB are given in Table 4. Table 4 Comparison of the accuracy between formulae in people over 70 years old Over 70 years old Sensitivity Specificity PPV NPV HUGE 92.8 93.15 95.87 88.31 MDRD 99.22 82.19 90.71 98.36 CKD-EPI 100 75.34 87.67 100 Discussion We believe this study establishes a straightforward, readily available and inexpensive tool for differentiating CRI from NCRI based on eGFR < 60 ml/min/1.73 m 2 . The L value calculated by the HUGE formula is particularly useful in persons over 70 years of age. It should be noted that MDRD and CKD-EPI formulae show differences in the prediction of CRI when compared with the true observed diagnosis in 6.96% (MDRD) and 8.95 % (CKD- EPI) of subjects older than 70 years (Table 2). In the large database of 125.373 subjects, the differences in the prediction for CRI between the HUGE formula and the MDRD in the population older than 70 years rose to 10.46 % (see Table 3). Taking into account the above percentages and the demographic data of Spain and UK, we have estimated that in Spain with a population of 45.000.000 inhabitants, 20% of whom are older than 70, potentially 1.700.000 wrong diagnosis of CRI would be made based on the analyses performed. For UK the number would reach 2.600.000, where general practitioners are asked to carry a register with patients eGFR <60 ml ml/min/ 1.73 m 2 (9). The similarity in the value of overdiagnosed percentage of CRI in subjects included in our CTDB and that observed in larger data bases may be interpreted as a good index of the reliability of HUGE. In an attempt to find out the causes that may explain the differences observed among HUGE and MDRD formulae we believe that there are two possible sources of errors. The first is that as tacitly expressed in the original paper describing MDRD formula (10): This prediction equation (the MDRD formula) was not tested in persons without renal disease ., persons < 18 years, elderly persons (persons > 70 years of age) The second is the widespread tendency to confuse screening tests with diagnostic tests. For a screening test to be accurate, its sensitivity, specificity and predictability values, either positive or negative, must be measured. Otherwise, the use of eGFR as a screening method may be a double-edged sword, sometimes wielded clumsily by the well intended (11). The sensitivity for the detection of early chronic kidney disease has been enhanced at the expense of lower specificity; this is why some healthy subjects are wrongly classified as CRI by eGFR testing (12). In the present study, persons aged over 70 years, included in our CTDB the specificity was 85% for MDRD and 75% for CKD-EPI, while for HUGE it was 93.15% (Table 4). In light of the foregoing considerations, MDRD formula estimations are undoubtedly valuable for managing patients with proven chronic kidney disease; for example in staging and monitoring progress. However, the use of eGFR to test patients without previously recognized CKD is not only controversial, but maybe ill-advised (13). Establishing an incorrect diagnosis of CRI by estimating GFR lower than 60 ml/min/1.73 m 2 provided by routine laboratory analyses will have, at least three undesirable effects. First, over diagnosis of CRI may provoke emotional anguish. Second, the unnecessary referral to nephrology outpatient clinics will increase expenditure of time and money. Finally, the wrong diagnosis of CRI may deny individuals appropriate treatment for other diseases (e.g. oncological, haematological or others) and also prevent subjects, particularly the elderly, from being included in clinical trials because the tendency to equalize CRI to GFR less than 60 ml/min/1.73 (14-16). Furthermore blood urea correlated better than blood creatinine with GFR calculated as iohexol clearance (17). Similarly, the reciprocal of blood urea has been previously identified as an independent predictor for GFR (10, 18). Proteinuria is worldwide accepted as a marker of renal injury, and in our study proteinuria was absent in the NCRI group, supporting the reliability of HUGE for CRI screening (19, 20). Conclusions The HUGE formula with data obtained from a general population, offers a straightforward readily available and inexpensive method based on haematocrit plasma urea and gender to differentiate more accurate than MDRD formulae to differentiate CRI from eGFR < 60 ml/min/1.73 m 2 , particularly useful in persons aged over 70 years of age and overcoming the disadvantages derived from the use of serum creatinine to calculate GFR. Conflict of interest: The authors declare no conflict of interest. Limitations: Our database excluded patients suffering from known liver, heart or lung insufficiency, acute renal failure, bleeding conditions, anaemic syndromes different from CRI, undernourished, pregnant, severely ill patients, patients admitted to intensive care units, and patients on programs of end-of-life geriatric treatment or receiving palliative oncologic care. Research Group: Nephrologist: Garcia N., Juncos L. and Musso C. G. (from Cordoba and Buenos Aires, Argentina), Garrido J. (Viseu, Portugal), Fraile P., Lerma J.L. and Rodriguez Commes J.L. (Salamanca, Spain), Grande J. (Zamora, Spain), Robles N.R. (Badajoz, Spain), Rodriguez J.C. (Las Palmas de Gran Canaria, Spain). Internal Medicine doctors: Calabria F., Funtowic G., Scibona P., Waissman G. (Buenos Aires, Argentina), and Bueno C., Romero Requena J. (Badajoz, Spain); Cardiologist: Ingaramo R. (Trelew, Argentina), Puppi L.M., (Buenos Aires, Argentina) and Martin Luengo C. (Salamanca, Spain); Geriatricians: Jaramillo E. (Cceres, Spain), Rodriguez Maas L. and Pablos Hernandez C (Salamanca, Spain); Clinical Biochemist: Zaro Bastanzuri M.J. (Badajoz, Spain) and Garca Garca C. (Salamanca, Spain). Family and Community Doctor: Jauregui R. (Buenos Aires, Argentina) and from Nuclear Medicine, Ruano R. (Salamanca, Spain) References 1. Smith, H.W., Disease of the kidney and urinary tract, in The Kidney: Structure and Function in Health and Disease H.W. Smith, Editor. 1951 Oxford University Press: New York. p. 836-886. 2. Macias Nunez, J.F., Ribera Casado, J.M.,De la Fuente del Rey, M et al, Biology of the ageing process and its clinical consequences, in The aging kidney in health and disease, J.F. Macias Nunez, Cameron, J.S., Oreopoulos, D.G., Editor. 2008, Springer: New York. p. 55-91. 3. Davies, D.F. and N.W. Shock, Age changes in glomerular filtration rate, effective renal plasma flow, and tubular excretory capacity in adult males. J Clin Invest, 1950. 29(5): p. 496-507. 4. Hosmer, D.W. and S. Lemeshow, Applied logistic regression. 2nd ed. Wiley series in probability and statistics. Texts and references section. 2000, New York: Wiley. xii, 373 p. 5. Kleinbaum, D.G., L.L. Kupper, and H. Morgenstern, Epidemiologic research : principles and quantitative methods. 1982, Belmont, Calif.: Lifetime Learning Publications. xix, 529 p. 6. Johnson, N.L. and S. Kotz, Discrete distributions. Their Distributions in statistics. 1969, Boston,: Houghton Mifflin. xvi, 328 p. 7. Johnson, N.L. and S. Kotz, Continuous univariate distributions. Their Distributions in statistics. 1970, New York,: Hougton Mifflin. 2 v. 8. McNemar, Q., Psychological statistics. 1949, New York,: Wiley. vii, 364 p. 9. DH Renal NSF Team: The National Service Framework for Renal Services Part Two: Chronic Kidney Disease, Acute Renal Failure and End of Life Care T.N.S.F.f.R. Services, Editor. 2005, Department of Health: London. 10. Levey, A.S., 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(6): p. 461-70. 11. Grimes, D.A. and K.F. Schulz, Uses and abuses of screening tests. Lancet, 2002. 359(9309): p. 881-4. 12. Poggio, E.D. and A.D. Rule, A critical evaluation of chronic kidney disease--should isolated reduced estimated glomerular filtration rate be considered a 'disease'? Nephrol Dial Transplant, 2009. 24(3): p. 698-700. 13. Giles, P.D., P.B. Rylance, and D.C. Crothers, New results from the Modification of Diet in Renal Disease study: the importance of clinical outcomes in test strategies for early chronic kidney disease. QJM, 2008. 101(2): p. 155-8. 14. Stevens, L.A. and A.S. Levey, Impact of reporting estimated glomerular filtration rate: it's not just about us. Kidney Int, 2009. 76(3): p. 245-7. 15. Jain, A.K., et al., When laboratories report estimated glomerular filtration rates in addition to serum creatinines, nephrology consults increase. Kidney Int, 2009. 76(3): p. 318-23. 16. Himmelfarb, J., Chronic kidney disease and the public health: gaps in evidence from interventional trials. JAMA, 2007. 297(23): p. 2630-3. 17. Fehrman-Ekholm, I. and L. Skeppholm, Renal function in the elderly (>70 years old) measured by means of iohexol clearance, serum creatinine, serum urea and estimated clearance. Scand J Urol Nephrol, 2004. 38(1): p. 73-7. 18. Swedko, P.J., et al., Serum creatinine is an inadequate screening test for renal failure in elderly patients. Arch Intern Med, 2003. 163(3): p. 356-60. 19. Hallan, S.I., et al., Combining GFR and albuminuria to classify CKD improves prediction of ESRD. J Am Soc Nephrol, 2009. 20(5): p. 1069-77. 20. Ikizler, T.A., CKD classification: time to move beyond KDOQI. J Am Soc Nephrol, 2009. 20(5): p. 929-30. HAEMATOCRIT, BLOOD UREA AND GENDER (HUGE) AS A SCREENING TEST FOR CRI The Journal of Nutrition, Health & Aging Volume 15, Number 6, 2011 484