- Email: [email protected]
creatinine ratios and random testing with dipsticks in nephrotic children
Quantitation of proteinuria with urinary protein/creatinine ratios and random testing with dipsticks in nephrotic children Carolyn Abitbol, MD, Gaston Zilleruelo, MD, Michael Freundlich, MD, a n d Jose Strauss, MD From the Department of Pediatrics, Division of Nephrology, Universityof Miami School of Medicine, Miami, Florida We examined the relative feasibility of using random urinary dipstick testing and urinary protein/creatinine ratios in the quantitation of proteinuria. Sixty-four children with relapsing nephrotic syndrome, ranging in age from 14,~to 16 years, contributed 145 timed urine collections and 150 random specimens, which were analyzed by urinary protein dipstick, quantitation of protein and creatinine, or both. Total protein excretion was expressed as grams per surface area per day and the urinary protein/creatinine ratio as milligrams of protein per milligram of creatinine. Degrees of proteinuria were designated as physiologic (<0.1 g m / m2/day), intermediate (>0.1 and <1.0 gm/m2/day), or nephrotic (>1.0 g m / m2/day). The log regression analysis of the 24-hour urinary protein/creatinine ratio (y) and total protein excretion (x) was highly significant (r = 0.97; p <0.001). The upper and lower confidence limits of the urinary protein/creatinine ratio (1.0 and 0.1, respectively) closely approximated the designations for nephrotic and physiologic proteinuria, respectively. These values were therefore used to classify degrees of proteinuria by the urine protein/creatinine ratio. The validity of these tests was assessed by sensitivity, specificity, and predictive value, and compared with that of random testing by urinary dipstick. The urinary protein/ creatinine ratio offered g o o d reliability as a test for classifying degrees of proteinuria and accurately predicting nephrotic and physiologic proteinuria. The random dipstick testing was reliable only when results were distinctly positive and when sensitivity and specificity were low. The error in the total protein excretion value that was imposed by collection errors with high and low variations in the creatinine value (57% of samples) was largely corrected by normalization of the data by log transformation. When controlled for creatinine excretion, linear regression analysis allowed calculation of total protein excretion (TP) from the urinary protein/creatinine ratio (U P/Cr) by the following equation: TP (gm/m2/day) = 0.63 (U P/Cr) at all levels of proteinuria. The random urinary protein/creatinine ratio appeared to offer the most sensitive test for classification of proteinuria in children with nephrosis, with the advantages of ease and expediency not afforded by the 24-hour urine quantitation. (J PEDIATR1990;116:2437)
Submitted for publication April.11, 1988; accepted Aug. 23, 1989. Reprint requests: Carolyn Abitbol, MD, Department of Pediatrics (R- 131), Universityof Miami Schoolof Medicine,PO Box 016960, Miami, FL 33101. 9/20/16229
Estimation of proteinuria by urinary protein/creatinine ratios in random urine specimens seems feasible in both adults 1-5 and children.6,7 These observations have been made in small series of patients and usually not in those with proteinuria in the nephrotic range. The use of this mea-
243
242
"t al.
The Journal of Pediatrics February 1990
Table II. Sensitivity, specificity, and predictive value
,ssion with log-transformed data No.
9
~11ne collection 125 0.97 ~reatinine excretion Low 37 0 . 9 8 Normal 53 0.99 High 35 0.96 Random urine collection 20 0.97 x. Quantitativeproteinexcretion(gm/m2/day);y.
[
TP U P/Cr
Predictive value
Regression equation
Sensitivity Specificity Positive Negative (%) (%) (%) (%)
y = x + 0.2 y=x+0.5 y = x + 0.2 y = x - 0.01 y = x + 0.2 U P/Cr (mg/mg).
Total protein Urinary protein/creatinine (ratio)
Dipstick assessments 24-Hour U P/Cr ratios Random U P/Cr ratios
I
surement as a rapid, quantitative assessment of proteinuria in children with nephrosis would be particularly advantageous in determining response to treatment regimens. We therefore attempted to assess the validity of applying the U P / C r ratio as an index for rapid quantitation of proteinuria in patients with relapsing nephrotic syndrome. METHODS Sixty-four children (45 male) with nephrotic syndrome who were followed at the University of Miami-Jackson Memorial Medical Center provided 145 timed, 24-hour urine collections and 150 random urine specimens. The children ranged in age from 18 months to 16 years (mean age 10.3 years). Focal segmental sclerosis was diagnosed by biopsy in 12 patients; minimal change disease was confirmed by biopsy in 29 or assumed by corticosteroid responsiveness8 in 23. All patients had normal creatinine clearance as estimated by serum creatinine concentration and height index. 9 Total protein determination was performed by the method of McIntosh 1~with the use of the standard Coomassie Blue Binding Technique (Bio-Rad Laboratories, Richmond, Calif.). The results were expressed as grams of protein per surface area per day, with degrees of proteinuria designated as physiologic (<0.1 gm/m2/day), intermediate (>0.1, and <1.0 gm/m2/day), or nephrotic (>1.0 gm/m2/day). 6"8 Creatinine was measured spectrophotometrically by modification of the Jaff6 method.ll Normal creatinine excretion was estimated to be 700 mg/m2/day. 12' 13 The upper and lower limits were set at 800 and 500 mg/mZ/day respectively. Timed 24-hour urine specimens were collected at home after instructions were provided in writing. One hundred forty-five single voided random specimens were obtained in the outpatient clinic within 12 hours of the beginning or end of a 24-hour collection period. No differentiation was made on the basis of the patient's supine or upright posture for any
70
68
89
60
90
84
85
97
95
93
93
100
of the random specimens, but the first morning and evening spot samples were avoided because they are k~or to correlate less well with 24-hour urine collections.1 One hundred random specimens were tested with the Multilabstix urine dipstick (Ames Co., Elkhart, Ind.) for semiquantitation of proteinuria. Fifty random specimens were assayed for protein and creatinine, 43 were correlated with the urine dipstick test results, and 20 with the 24-hour urine collections. Urinary protein/creatinine ratios were calculated from the relationship of protein (in milligrams) to creatinine (in milligrams). The sensitivity14 of the U P / C r ratio and urinary dipstick testing in assessing the degree of proteinuria was calculated as the percentage of 24-hour samples with proteinuria at either the physiologic or the nephrotic range that were properly designated by the method. Specificity14 was calculated as the percentage of 24-hour samples in each range of proteinuria that were properly designated by the test modality (U P / C r or dipstick). The positive predictive value 14 of the U P / C r ratio and of the dipstick test result was calculated as the percentage of 24-hour urine specimens that would have been accurately predicted with a U P / C r ratio >1.0 or a dipstick value >3 to 4+ to have nephrotic-range proteinuria (>1.0 g m / mZ/day). The negative predictive value 14 of the U P / C r ratio and the dipstick test result were calculated as the percentage of 24-hour samples that would have been accurately predicted to have physiologic-range proteinuria by a U P / C r ratio <0.1 or a dipstick value of zero to a trace amount. The data were analyzed by linear regression analysis before and after normalization by log transformation. 15 RESULTS
Daily collections. The linear regression of the U P / C r ratio from 125 twenty-four-hour urine collections to protein excretion was highly significant: r = 0.97, p <0.001, and linear regression equation y = x + 0.2, where y is loglo U P / C r ratio (in milligrams per milligram) and x is loglo TP
Volume 116 Number 2
Protein/creatinine ratios and dipsticks to assess proteinuria
/i
I 0 +z
245
/"
///'//" /I Y = X "1-0.2
o /~
,"
r = .97
o
iO*l
I/ /" o
9 /" /"
o o ,N", , ,,'~~" p" 9,'Q9 ~ /7~" o
// / o
mg/mg
o/"
//
~J
6 lo-I
9
o
9 9 //"
/
/
~ /"A 9 9/
,~'. 9/ 9
,/ 7 9
lo-2 /
;,,"
A, //
10-z
10 -I
0
10 *]
1 0 *z
Total Protein Excretion g/ mZl d
Figure. Logarithmic regression of U P/Cr ratio against total protein TP value, with 95% confidence limits shown. Samples with low (9 (<500 mg/m2/day), intermediate (O) (>500 to <800 mg/m2/day), and high (A) (>800 mg/m2/day) creatinine excretions are designated.
excretion (in grams per square meter per day) (Figure; Table I). The lower and upper 95% confidence limits for physiologic and nephrotic proteinuria, 0.1 and 1.0 respectively, were designated as reference points to estimate degrees of proteinuria from the U P / C r ratio. A value for the U P / C r ratio <0.1 was considered the value corresponding to physiologic proteinuria; a U P / C r ratio >1.0 corresponded to nephrotic proteinuria. Random urine specimens. One hundred random specimens were assessed by dipstick testing and the results compared with the TP value of their respective 24-hour urine collections. The correlation was significant (r = 0.5; p <0.001). Forty-three random specimens with dipstick assessment were also assessed by the protein/creatinine index. The correlation coefficient between these two variables reached 0.6 (p <0.001) but was nevertheless lower than correlations drawn between either random or 24-hour U P / Cr ratio and TP value. Twenty random urine specimens were analyzed for U P / Cr ratio and correlated with the TP value of their respective 24-hour collections after log conversion (Table I). The correlation was highly significant (r < 0.95;p < 0.001). The
regression equation correlating U P / C r ratio of random specimens with the TP value was identical to that of the whole series (y = x + 0.2). Sensitivity, specificity, and predictive value. Sensitivity assessed the probability that the extreme ranges of proteinuria would be designated as either physiologic or nephrotic by the method specified (Table II). The urinary dipstick assessment was least sensitive; both random and 24-hour U P / C r ratios were highly sensitive. Specificity assessed the probability that any range of proteinuria would be appropriately designated by the test modality. Specificity was extremely low for the random dipstick assessment but high for the U P / C r ratio. Predictability, although not determined prospectively, attempted to assess the percentage of tests that would have accurately predicted nephrotic proteinuria (positive predictive value) or physiologic proteinuria (negative predictive value). The positive predictive value for all three methods was high. However, the high negative predictability of both the random and the 24-hour U P / C r ratio was in sharp contrast to the low negative predictability of the dipstick assessment.
246
Abitbol et al.
Variation in creatinine excretion. Since variation in creatinine excretion can markedly affect the U P/Cr ratio, the 125 timed urine collections were evaluated according to creatinine excretion rates. The mean creatinine excretion (___SD) for the entire series was 707 + 470 mg/m2/day. Normal creatinine excretion was taken as >500 <800 rag/ m2/day. When samples of low and high creatinine excretion were excluded from the analysis, the regression relationship for the remaining 53 samples was highly significant before and after log transformation of the data (r >0.98; p <0.001). As shown in the Figure, only abnormally low or high creatinine excretion accounted for dispersion of data outside the 95% confidence limits. Separate analysis by log regression of samples with low (<500 mg/m2/day) and high (>800 mg/m2/day) creatinine excretion resulted in regression lines that closely approximated the 95% confidence limits for the entire series. Hence, normalization of data by log transformation appeared to correct for variations in creatinine excretion.
DISCUSSION Measurement of proteinuria in an expedient and accurate manner in children with relapsing nephrotic syndrome is a difficult problem. Qualitative assessment by urinary dipsticks is commonly used 16 but has no proven reliability and is subject to error and poor standardization because of variations in urinary flow rates) The primary reference standard for quantitation of proteinuria remains the 24-hour urine collection, despite its inherent time delays and collection errors. In numerous articles, 1"7 authors have attempted to demonstrate the advantage of the protein/creatinine index over the timed urine collection. Ginsberg et al. 1 suggested that the random U P/Cr ratio may be more accurate in classifying degrees of proteinuria, with the qualification that a correction for urine creatinine levels be applied. According to our data, such a correction factor is not necessary if normalization by log transformation is performed. In selecting out those samples with inherent error (high and low creatinine excretions), we have shown a significant regression relationship between U P/Cr ratio and daily proteinuria whether the U P/Cr ratio is determined for a random or a 24-hour specimen. In our observations, the confidence limits for the U P/Cr ratio corresponding to physiologic (<0.1 gm/m2/day) and nephrotic (>1.0 gm/m2/day proteinuria closely approximated the identical numerals 0.1 and 1.0, respectively. The use of these limits in the rapid classification of degrees of proteinuria was assessed for sensitivity, specificity, and predictive value. The U P/Cr ratio demonstrated a high degree of reliability as a test in designating degrees of proteinuria,
The Journal of Pediatrics February 1990
whereas the random dipstick assessment tended to result in more misclassification of proteinuria. The predictive value of dipstick testing showed a high degree of reliability when results were definitively positive; however, when the value range was other than nephrotic, there was little correlation with degree of proteinuria. This finding confirms the limited usefulness of the dipstick in the evaluation and management of relapsing nephrosis. Ideally, any result from a timed urine collection with either high or low creatinine content should be discarded. With it, of course, goes the anticipated laboratory information, the expense, and the wasted effort of the patient or family. In our series, 58% of the samples had to be discounted for this reason. When one weighs the potential errors and the tedium involved in timed collections against the accuracy, simplicity, and expediency of the U P / C r ratio, it seems reasonable to suggest abandonment of timed specimen collection. In fact, it is likely that the U P / C r ratio may be a more accurate reflection of magnitude of proteinuria, because rates of creatinine excretion are poorly standardized in children. Calculation of actual proteinuria from the U P/Cr ratio by use of the regression equation seems valid from several standpoints. The log regression equation (y = x + 0.2) appeared three times in the analysis of our data on both timed and random specimens (Table I). Further substantiation for use of this equation comes from Housel',6' 7 who reported the same regression relationship derived from the specimens of 20 subjects (15 children) after comparing ratios of protein and albumin to creatinine. Our series relates only to children with relapsing nephrotic syndrome and normal glomerular filtration rates. Schwab et al. 5 demonstrated an excellent correlation of U P/Cr ratios with TP values in a large number of patients with a broad spectrum of renal diseases and varying levels of renal function. Although it seems reasonable to assume that these relationships can be extrapolated to other diseases and circumstances, care must be taken to validate individual cases. Orthostatic proteinuria could be most deceptive. Random urine specimens obtained while patients are in the supine or upright posture should be carefully compared. Ideally, the supine U P/Cr ratio would be <0. I, whereas the upright might exceed 1.0. In assessments of severely malnourished patients and small infants, erroneously low creatinine excretion rates with normal urine flow rates are probable. Although log transformation of the U P/Cr ratio in such patients might correct for creatinine excretion, the validity of its application in these subjects should be specifically investigated. Our observations confirmed the advantage of the U P/Cr ratio over the urinary dipstick assessment in both random and 24-hour collections for assessment of proteinuria in
Volume 116 Number 2
Protein/creatinine ratios and dipsticks to assess proteinuria
children with nephrosis. Physiologic proteinuria is designated by a U P / C r ratio <0.1, and nephrotic proteinuria is defined by a U P / C r ratio > 1.0. Quantitation of proteinuria from the linear regression is provided by the following simple equation: T P ( g m / m 2 / d a y ) -- 0.63 (U P / C r ) .
REFERENCES 1. Ginsberg JM, Chang BS, Matares RA, Garella S. Use of single voided urine samples to estimate quantitative proteinuria. N Engl J Med 1983;309:1543-6. 2. Sessoms S, Mehta K, Kovarsky J. Quantitation of proteinuria in systemic lupus erythematosus by use of a random spot urine collection. Arthritis Rheum 1983;26:918-20. 3. Shaw AB, Risdon P, Lewis-Jackson JD. Protein: creatinine and Albustix in assessment of proteinuria. Br Med J 1983; 287:929-32. 4. Nathan DM, Rosenblum C, Protasowicki VD. Single-voided urine samples can be used to estimate quantitative microalbuminuria. Diabetes Care 1987;10:414-8. 5. Schwab S.I, Christensen RL, Dougherty K, Klahr S. Quantitation of proteinuria by the use of the protein-to-creatinine ratios in single urine samples. Arch Intern Med 1987;147:943-4. 6. Houser M. Assessment of proteinuria using random urine samples. J PEDIATR 1984;104:845-8. 7. Houser M. Characterizationofproteinuria using random urine samples. Int J Pediatr Nephrol 1986;7:197-202.
247
8. International Study for Kidney Disease in Children. The nephrotic syndrome in children: prediction of histopathology from clinical and laboratory characteristics at the time of diagnosis. Kidney Int 1987;13:159-65. 9. Schwartz G J, Haycock GB, Edelmann CM Jr, et al. A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 1976;58:259-63. 10. Mclntosh JC. Application of dye-binding method to the determination of protein in urine and cerebrospinal fluid. Clin Chem 1977;23:1939-40. 11. Slot C. Plasma creatinine determination: a new and specific Jaff6 reaction method. Scand J Clin Lab Invest 1965;17: 381-7. 12. Heymfield Sb, Arteaga C, McManus C, Smith J, Moffitt HS. Measurement of muscle mass in humans: validity of the 24-hour urinary creatinine method. Am J Clin Nutr 1983; 37:478-94. 13. Chantler C, Barratt TM. Laboratory evaluation. In Holliday, Barratt, Vernier, eds. Pediatric Nephrology. Baltimore: Williams & Wilkins, 1987:293. 14. Sochett E, Daneman D. Screening tests to detect microalbuminuria in children with diabetes. J PEDIATR1988;112:744-8. 15. Snedecor G, Cochran. Statistical methods. 6th ed. Ames, Iowa: Iowa State University Press, 1972:59. 16. McCracken BH. Estimation of quantitative proteinuria [Letter]. N Engl J Med 1984;310:1464.
Submitted for publication April.11, 1988; accepted Aug. 23, 1989. Reprint requests: Carolyn Abitbol, MD, Department of Pediatrics (R- 131), Universityof Miami Schoolof Medicine,PO Box 016960, Miami, FL 33101. 9/20/16229
Estimation of proteinuria by urinary protein/creatinine ratios in random urine specimens seems feasible in both adults 1-5 and children.6,7 These observations have been made in small series of patients and usually not in those with proteinuria in the nephrotic range. The use of this mea-
243
242
"t al.
The Journal of Pediatrics February 1990
Table II. Sensitivity, specificity, and predictive value
,ssion with log-transformed data No.
9
~11ne collection 125 0.97 ~reatinine excretion Low 37 0 . 9 8 Normal 53 0.99 High 35 0.96 Random urine collection 20 0.97 x. Quantitativeproteinexcretion(gm/m2/day);y.
[
TP U P/Cr
Predictive value
Regression equation
Sensitivity Specificity Positive Negative (%) (%) (%) (%)
y = x + 0.2 y=x+0.5 y = x + 0.2 y = x - 0.01 y = x + 0.2 U P/Cr (mg/mg).
Total protein Urinary protein/creatinine (ratio)
Dipstick assessments 24-Hour U P/Cr ratios Random U P/Cr ratios
I
surement as a rapid, quantitative assessment of proteinuria in children with nephrosis would be particularly advantageous in determining response to treatment regimens. We therefore attempted to assess the validity of applying the U P / C r ratio as an index for rapid quantitation of proteinuria in patients with relapsing nephrotic syndrome. METHODS Sixty-four children (45 male) with nephrotic syndrome who were followed at the University of Miami-Jackson Memorial Medical Center provided 145 timed, 24-hour urine collections and 150 random urine specimens. The children ranged in age from 18 months to 16 years (mean age 10.3 years). Focal segmental sclerosis was diagnosed by biopsy in 12 patients; minimal change disease was confirmed by biopsy in 29 or assumed by corticosteroid responsiveness8 in 23. All patients had normal creatinine clearance as estimated by serum creatinine concentration and height index. 9 Total protein determination was performed by the method of McIntosh 1~with the use of the standard Coomassie Blue Binding Technique (Bio-Rad Laboratories, Richmond, Calif.). The results were expressed as grams of protein per surface area per day, with degrees of proteinuria designated as physiologic (<0.1 gm/m2/day), intermediate (>0.1, and <1.0 gm/m2/day), or nephrotic (>1.0 gm/m2/day). 6"8 Creatinine was measured spectrophotometrically by modification of the Jaff6 method.ll Normal creatinine excretion was estimated to be 700 mg/m2/day. 12' 13 The upper and lower limits were set at 800 and 500 mg/mZ/day respectively. Timed 24-hour urine specimens were collected at home after instructions were provided in writing. One hundred forty-five single voided random specimens were obtained in the outpatient clinic within 12 hours of the beginning or end of a 24-hour collection period. No differentiation was made on the basis of the patient's supine or upright posture for any
70
68
89
60
90
84
85
97
95
93
93
100
of the random specimens, but the first morning and evening spot samples were avoided because they are k~or to correlate less well with 24-hour urine collections.1 One hundred random specimens were tested with the Multilabstix urine dipstick (Ames Co., Elkhart, Ind.) for semiquantitation of proteinuria. Fifty random specimens were assayed for protein and creatinine, 43 were correlated with the urine dipstick test results, and 20 with the 24-hour urine collections. Urinary protein/creatinine ratios were calculated from the relationship of protein (in milligrams) to creatinine (in milligrams). The sensitivity14 of the U P / C r ratio and urinary dipstick testing in assessing the degree of proteinuria was calculated as the percentage of 24-hour samples with proteinuria at either the physiologic or the nephrotic range that were properly designated by the method. Specificity14 was calculated as the percentage of 24-hour samples in each range of proteinuria that were properly designated by the test modality (U P / C r or dipstick). The positive predictive value 14 of the U P / C r ratio and of the dipstick test result was calculated as the percentage of 24-hour urine specimens that would have been accurately predicted with a U P / C r ratio >1.0 or a dipstick value >3 to 4+ to have nephrotic-range proteinuria (>1.0 g m / mZ/day). The negative predictive value 14 of the U P / C r ratio and the dipstick test result were calculated as the percentage of 24-hour samples that would have been accurately predicted to have physiologic-range proteinuria by a U P / C r ratio <0.1 or a dipstick value of zero to a trace amount. The data were analyzed by linear regression analysis before and after normalization by log transformation. 15 RESULTS
Daily collections. The linear regression of the U P / C r ratio from 125 twenty-four-hour urine collections to protein excretion was highly significant: r = 0.97, p <0.001, and linear regression equation y = x + 0.2, where y is loglo U P / C r ratio (in milligrams per milligram) and x is loglo TP
Volume 116 Number 2
Protein/creatinine ratios and dipsticks to assess proteinuria
/i
I 0 +z
245
/"
///'//" /I Y = X "1-0.2
o /~
,"
r = .97
o
iO*l
I/ /" o
9 /" /"
o o ,N", , ,,'~~" p" 9,'Q9 ~ /7~" o
// / o
mg/mg
o/"
//
~J
6 lo-I
9
o
9 9 //"
/
/
~ /"A 9 9/
,~'. 9/ 9
,/ 7 9
lo-2 /
;,,"
A, //
10-z
10 -I
0
10 *]
1 0 *z
Total Protein Excretion g/ mZl d
Figure. Logarithmic regression of U P/Cr ratio against total protein TP value, with 95% confidence limits shown. Samples with low (9 (<500 mg/m2/day), intermediate (O) (>500 to <800 mg/m2/day), and high (A) (>800 mg/m2/day) creatinine excretions are designated.
excretion (in grams per square meter per day) (Figure; Table I). The lower and upper 95% confidence limits for physiologic and nephrotic proteinuria, 0.1 and 1.0 respectively, were designated as reference points to estimate degrees of proteinuria from the U P / C r ratio. A value for the U P / C r ratio <0.1 was considered the value corresponding to physiologic proteinuria; a U P / C r ratio >1.0 corresponded to nephrotic proteinuria. Random urine specimens. One hundred random specimens were assessed by dipstick testing and the results compared with the TP value of their respective 24-hour urine collections. The correlation was significant (r = 0.5; p <0.001). Forty-three random specimens with dipstick assessment were also assessed by the protein/creatinine index. The correlation coefficient between these two variables reached 0.6 (p <0.001) but was nevertheless lower than correlations drawn between either random or 24-hour U P / Cr ratio and TP value. Twenty random urine specimens were analyzed for U P / Cr ratio and correlated with the TP value of their respective 24-hour collections after log conversion (Table I). The correlation was highly significant (r < 0.95;p < 0.001). The
regression equation correlating U P / C r ratio of random specimens with the TP value was identical to that of the whole series (y = x + 0.2). Sensitivity, specificity, and predictive value. Sensitivity assessed the probability that the extreme ranges of proteinuria would be designated as either physiologic or nephrotic by the method specified (Table II). The urinary dipstick assessment was least sensitive; both random and 24-hour U P / C r ratios were highly sensitive. Specificity assessed the probability that any range of proteinuria would be appropriately designated by the test modality. Specificity was extremely low for the random dipstick assessment but high for the U P / C r ratio. Predictability, although not determined prospectively, attempted to assess the percentage of tests that would have accurately predicted nephrotic proteinuria (positive predictive value) or physiologic proteinuria (negative predictive value). The positive predictive value for all three methods was high. However, the high negative predictability of both the random and the 24-hour U P / C r ratio was in sharp contrast to the low negative predictability of the dipstick assessment.
246
Abitbol et al.
Variation in creatinine excretion. Since variation in creatinine excretion can markedly affect the U P/Cr ratio, the 125 timed urine collections were evaluated according to creatinine excretion rates. The mean creatinine excretion (___SD) for the entire series was 707 + 470 mg/m2/day. Normal creatinine excretion was taken as >500 <800 rag/ m2/day. When samples of low and high creatinine excretion were excluded from the analysis, the regression relationship for the remaining 53 samples was highly significant before and after log transformation of the data (r >0.98; p <0.001). As shown in the Figure, only abnormally low or high creatinine excretion accounted for dispersion of data outside the 95% confidence limits. Separate analysis by log regression of samples with low (<500 mg/m2/day) and high (>800 mg/m2/day) creatinine excretion resulted in regression lines that closely approximated the 95% confidence limits for the entire series. Hence, normalization of data by log transformation appeared to correct for variations in creatinine excretion.
DISCUSSION Measurement of proteinuria in an expedient and accurate manner in children with relapsing nephrotic syndrome is a difficult problem. Qualitative assessment by urinary dipsticks is commonly used 16 but has no proven reliability and is subject to error and poor standardization because of variations in urinary flow rates) The primary reference standard for quantitation of proteinuria remains the 24-hour urine collection, despite its inherent time delays and collection errors. In numerous articles, 1"7 authors have attempted to demonstrate the advantage of the protein/creatinine index over the timed urine collection. Ginsberg et al. 1 suggested that the random U P/Cr ratio may be more accurate in classifying degrees of proteinuria, with the qualification that a correction for urine creatinine levels be applied. According to our data, such a correction factor is not necessary if normalization by log transformation is performed. In selecting out those samples with inherent error (high and low creatinine excretions), we have shown a significant regression relationship between U P/Cr ratio and daily proteinuria whether the U P/Cr ratio is determined for a random or a 24-hour specimen. In our observations, the confidence limits for the U P/Cr ratio corresponding to physiologic (<0.1 gm/m2/day) and nephrotic (>1.0 gm/m2/day proteinuria closely approximated the identical numerals 0.1 and 1.0, respectively. The use of these limits in the rapid classification of degrees of proteinuria was assessed for sensitivity, specificity, and predictive value. The U P/Cr ratio demonstrated a high degree of reliability as a test in designating degrees of proteinuria,
The Journal of Pediatrics February 1990
whereas the random dipstick assessment tended to result in more misclassification of proteinuria. The predictive value of dipstick testing showed a high degree of reliability when results were definitively positive; however, when the value range was other than nephrotic, there was little correlation with degree of proteinuria. This finding confirms the limited usefulness of the dipstick in the evaluation and management of relapsing nephrosis. Ideally, any result from a timed urine collection with either high or low creatinine content should be discarded. With it, of course, goes the anticipated laboratory information, the expense, and the wasted effort of the patient or family. In our series, 58% of the samples had to be discounted for this reason. When one weighs the potential errors and the tedium involved in timed collections against the accuracy, simplicity, and expediency of the U P / C r ratio, it seems reasonable to suggest abandonment of timed specimen collection. In fact, it is likely that the U P / C r ratio may be a more accurate reflection of magnitude of proteinuria, because rates of creatinine excretion are poorly standardized in children. Calculation of actual proteinuria from the U P/Cr ratio by use of the regression equation seems valid from several standpoints. The log regression equation (y = x + 0.2) appeared three times in the analysis of our data on both timed and random specimens (Table I). Further substantiation for use of this equation comes from Housel',6' 7 who reported the same regression relationship derived from the specimens of 20 subjects (15 children) after comparing ratios of protein and albumin to creatinine. Our series relates only to children with relapsing nephrotic syndrome and normal glomerular filtration rates. Schwab et al. 5 demonstrated an excellent correlation of U P/Cr ratios with TP values in a large number of patients with a broad spectrum of renal diseases and varying levels of renal function. Although it seems reasonable to assume that these relationships can be extrapolated to other diseases and circumstances, care must be taken to validate individual cases. Orthostatic proteinuria could be most deceptive. Random urine specimens obtained while patients are in the supine or upright posture should be carefully compared. Ideally, the supine U P/Cr ratio would be <0. I, whereas the upright might exceed 1.0. In assessments of severely malnourished patients and small infants, erroneously low creatinine excretion rates with normal urine flow rates are probable. Although log transformation of the U P/Cr ratio in such patients might correct for creatinine excretion, the validity of its application in these subjects should be specifically investigated. Our observations confirmed the advantage of the U P/Cr ratio over the urinary dipstick assessment in both random and 24-hour collections for assessment of proteinuria in
Volume 116 Number 2
Protein/creatinine ratios and dipsticks to assess proteinuria
children with nephrosis. Physiologic proteinuria is designated by a U P / C r ratio <0.1, and nephrotic proteinuria is defined by a U P / C r ratio > 1.0. Quantitation of proteinuria from the linear regression is provided by the following simple equation: T P ( g m / m 2 / d a y ) -- 0.63 (U P / C r ) .
REFERENCES 1. Ginsberg JM, Chang BS, Matares RA, Garella S. Use of single voided urine samples to estimate quantitative proteinuria. N Engl J Med 1983;309:1543-6. 2. Sessoms S, Mehta K, Kovarsky J. Quantitation of proteinuria in systemic lupus erythematosus by use of a random spot urine collection. Arthritis Rheum 1983;26:918-20. 3. Shaw AB, Risdon P, Lewis-Jackson JD. Protein: creatinine and Albustix in assessment of proteinuria. Br Med J 1983; 287:929-32. 4. Nathan DM, Rosenblum C, Protasowicki VD. Single-voided urine samples can be used to estimate quantitative microalbuminuria. Diabetes Care 1987;10:414-8. 5. Schwab S.I, Christensen RL, Dougherty K, Klahr S. Quantitation of proteinuria by the use of the protein-to-creatinine ratios in single urine samples. Arch Intern Med 1987;147:943-4. 6. Houser M. Assessment of proteinuria using random urine samples. J PEDIATR 1984;104:845-8. 7. Houser M. Characterizationofproteinuria using random urine samples. Int J Pediatr Nephrol 1986;7:197-202.
247
8. International Study for Kidney Disease in Children. The nephrotic syndrome in children: prediction of histopathology from clinical and laboratory characteristics at the time of diagnosis. Kidney Int 1987;13:159-65. 9. Schwartz G J, Haycock GB, Edelmann CM Jr, et al. A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 1976;58:259-63. 10. Mclntosh JC. Application of dye-binding method to the determination of protein in urine and cerebrospinal fluid. Clin Chem 1977;23:1939-40. 11. Slot C. Plasma creatinine determination: a new and specific Jaff6 reaction method. Scand J Clin Lab Invest 1965;17: 381-7. 12. Heymfield Sb, Arteaga C, McManus C, Smith J, Moffitt HS. Measurement of muscle mass in humans: validity of the 24-hour urinary creatinine method. Am J Clin Nutr 1983; 37:478-94. 13. Chantler C, Barratt TM. Laboratory evaluation. In Holliday, Barratt, Vernier, eds. Pediatric Nephrology. Baltimore: Williams & Wilkins, 1987:293. 14. Sochett E, Daneman D. Screening tests to detect microalbuminuria in children with diabetes. J PEDIATR1988;112:744-8. 15. Snedecor G, Cochran. Statistical methods. 6th ed. Ames, Iowa: Iowa State University Press, 1972:59. 16. McCracken BH. Estimation of quantitative proteinuria [Letter]. N Engl J Med 1984;310:1464.