Urinary protein and albumin excretion corrected by creatinine and specific gravity

Clinica Chimica Acta 294 (2000) 139–155www.elsevier.com/ locate /clinchim

Urinary protein and albumin excretion corrected bycreatinine and specific gravity

a b c ,*David J. Newman , Michael J. Pugia , John A. Lott ,b bJane F. Wallace , Andrew M. Hiar

aSW Thames Institute for Renal Research, St. Helier Hospital, Wrythe Lane, Carshalton,Surrey SM5 1AA, UK

bDiagnostics Business Group, Bayer Corporation, 1884 Miles Avenue, Elkhart, IN 46514, USAcDepartment of Pathology, The Ohio State University Medical Center, Starling Loving M-029, Columbus,

OH 43210, USA

Received 17 September 1999; received in revised form 3 December 1999; accepted 17 December 1999

Abstract

Timed urine collections are difficult to use in clinical practice owing to inaccurate collectionsmaking calculations of the 24-h albumin or protein excretion questionable. One of our goals wasto assess the ‘correction’ of urinary albumin and (or) protein excretion by dividing these by either

1the creatinine concentration or the term, (specific gravity 2 1) 3 100 . The 24-h creatinineexcretion can be estimated based on the patients’ gender, age and weight. We studied the influenceof physiological extremes of hydration and exercise, and protein and creatinine excretion inpatients with or suspected kidney disorders. Specimens were collected from healthy volunteersevery 4 h during one 24-h period. We assayed the collections individually to give us an assessmentof the variability of the analytes with time, and then reassayed them after combining them to givea 24-h urine. For all volunteers, the mean intra-individual CVs based on the 4-h collectionsexpressed in mg/24 h were 80.0% for albumin and 96.5% for total protein (P . 0.2). The CVswere reduced by dividing the albumin or protein concentration by the creatinine concentration orby the term, (SG-1) 3 100. This gave a CV for mg albumin/g creatinine of 52% (P , 0.1 vs.albumin mg/g creatinine); mg protein /g creatinine of 39% (P , 0.05 vs. mg protein /g creatinine);

Abbreviations: AER, albumin excretion; PER, protein excretion; ACR, albumin creatinine ratio; PCR,protein /creatinine ratio; ASG, albumin/(SG-1)3100 ratio; PSG, protein /(SG-1)3100 ratio; TP, true positive;FP, false positive; TN, true negative; FN, false negative; Alb, albumin; Cre, creatinine; SGU, (1-SG)3100

*Corresponding author. Tel.: 11-614-293-5383; fax: 11-614-293-5984.E-mail address: lott.1@osu.edu (J.A. Lott)

1Note that the term, (SG-1)3100 increases with the increasing concentration of all dissolved solids. Because(1-SG)3100 is dimensionless, we called (SG-1)3100 ‘SG units’, i.e. SGU.

0009-8981/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0009-8981( 00 )00181-9

140 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

mg albumin/ [(SG-1) 3 100] of 49% (P , 0.1 vs. albumin) / [(SG-1) 3 100]; and mg protein / [(SG-1) 3 100] of 37% (P , 0.05 vs. mg protein) / [(SG-1) 3 100]. For the 68 subjects in the study, thestrongest correlation was between the creatinine concentrations and the 24-h urine volume:r 5 0.786, P , 0.001. The correlation of (SG-1) 3 100 vs. the 24-h urine volume was: r 5 0.606,P , 0.001; for (SG-1) 3 100 and the creatinine concentration, the correlation was: r 5 0.666,P , 0.001. Compared to the volunteers, the albumin and protein excretion in mg/24 h were morevariable in the patients. The same was true if the albumin or protein concentrations were dividedby the creatinine concentration or by (SG-1) 3 100. Protein and albumin concentrations werelower in dilute urines. Dividing the albumin or protein concentrations by the creatinineconcentration reduced the number of false negative protein and albumin results. Dividing thealbumin or protein values in mg/24 h by (SG-1) 3 100 eliminated fewer false negatives. Albuminconcentrations increased significantly after vigorous exercise. The increase was almost eliminatedwhen the albumin result was divided by the creatinine concentration suggesting that a decreasedurine flow and not increased glomerular permeability causes an increase of post-exercisealbuminuria. The same was true for proteinuria. A dipstick test plus an optical strip reader that canmeasure urine protein, albumin, and creatinine and calculate the appropriate ratios provides abetter screening test for albuminuria or proteinuria than one measuring only albumin or protein. 2000 Elsevier Science B.V. All rights reserved.

Keywords: Albuminuria; Effects of exercise; Microalbuminuria; Proteinuria; Renal failure;Specific gravity; Test variation

1. Introduction

Urine dipstick testing is widely used for the detection of early proteinuria andpossible early renal damage [1]. There are a variety of analytical methodsavailable for the measurement of both urine total protein and albumin con-centrations [2–4]. They differ in sensitivity and specificity for the proteins inurine, and their relative analytical and clinical performances have been evaluatedin this context. The analytical measurement of protein may be simple, but theestimation of the amount excreted in 24 h is fraught with error. The volume ofurine excreted can be highly variable depending mainly on the individual’s fluidintake and physical activity. In a dilute urine, the total protein excretion may beunderestimated. If the urine is concentrated, as frequently occurs after strenuousphysical activity, an increased protein concentration could be misinterpreted. Toavoid this problem, accurately timed urine specimens have been proposed,expressing protein excretion in units of mg/min. The difficulty is in collecting anaccurate 24-h specimen [5–7]. To reduce the uncertainty of the timing, acollection is made of the early morning first voiding [1]. But a means ofcorrecting for urine concentration and (or) volume is needed when the collectionaccuracy is in doubt. Two techniques used to compensate for variation in theexcretion volumes are to divide the albumin or protein concentrations by thecreatinine concentration or by the term, (specific gravity 2 1) 3 100 [8].

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 141

Urinary creatinine excretion in normal persons depends on gender, age, andmuscle mass [9,10]. The 24-h creatinine excretion in a patient with stableglomerular filtration is fairly constant [11], and is the basis of the Cockroft–Gault calculation for creatinine excretion as is given here [12]. The ratios,milligram of albumin or milligram of protein per gram of creatinine, are readilydetermined. If we calculate the creatinine excretion per 24 h, then the urinaryalbumin and (or) protein loss are easily calculated. An estimate of the 24-halbumin (or protein) excretion is given by:

(g creatinine /24 h) 3 (mg albumin/g creatinine) 5 mg albumin/24 h.

The first term is measured or calculated from the Cockroft–Gault equation,and the second term is obtained from the assay of albumin (or protein) andcreatinine. The error in this approach is much smaller than the typical errorsmade in ‘24-h urine’ collections.

Various pathologies influence the urinary creatinine excretion. It is oftendecreased in advanced renal disease, acromegaly, strenuous exercise, hyper-thyroidism and muscle damage potentially causing an underestimation of proteinexcretion. Creatinine excretion is commonly increased in diabetes mellitus andhypothyroidism causing overestimation of protein excretion [10]. A few studiescompared creatinine and specific gravity as correction factors when an incom-plete collection occurs. The term ‘correcting for SG’ indicates dividing thealbumin or protein concentration by (SG-1)3100 [10].

The urinary excretion of creatinine has been used for many years to get anestimate of the 24-h excretion volume and to ‘correct’ the quantitativemeasurement of other analytes such as albumin, proteins, cortisol, and catechol-amines to a 24-h basis [13,14]. Until recently, no point of care or dipstick testswere available to make the correction [15,16]. A combined measurement

dipstick is now available on the CLINITEK 50 instrument (Bayer Corp.,Tarrytown, NY) that uses novel analytical techniques for both albumin andcreatinine. These methods are more accurate when read by a reflectometer ratherthan visually [15–17]. We used these dipsticks to assess the utility of dividingthe results by creatinine or (SG-1)3100 for adjusting results on random urines.

One of our goals was to measure intra- and inter-individual variation ofurinary albumin and protein concentrations in healthy individuals and in patientswith impaired renal function. Comparisons were made between excretion inmg/24 h and the ratios of albumin or protein to creatinine or to (SG-1)3100.The correlation of urinary volumes to creatinine or (SG-1)3100 was calculatedfrom the 24-h collection data of the healthy volunteers and patients. We alsotested other healthy individuals following strenuous exercise with physicalcontact. The latter commonly leads to dehydration and overestimation ofalbumin or protein concentrations. Finally we determined the utility of a new

142 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

urinalysis strip for albumin, protein, and creatinine by comparison to quantita-tive methods.

2. Materials and methods

2.1. Healthy volunteers

A total of 118 specimens were collected during one 24-h period at 4 hintervals from 20 healthy volunteers. Some of the volunteers were awakened ifneeded to provide a urine specimen during their sleeping time. One subjectprovided only four specimens. We also recorded the gender, age and weight ofthe volunteers; these are required for the Cockcroft–Gault equation [12]. Weperformed quantitative assays for albumin, protein, creatinine, and measured thevolumes for all individual collections in a 24-h period; these were thencombined to simulate a 24-h urine collection. SG was determined by refrac-tometry. An additional 19 specimens were collected from 16 of the samevolunteers after 12 h of fluid deprivation. Three volunteers gave us specimensafter 12 and 14 h of fluid deprivation. The volunteers then drank 1 l of water,and the first voiding was collected 1 h later from 12 volunteers and assayed asabove.

2.2. Hospitalized patients

A total of 68, 24-h urine specimens were collected from patients withconfirmed or likely kidney disorders. Among these, 34 had chronic renal failurewith and without nephrotic syndrome, previous kidney transplant, or diabeticnephropathy. Six had hypertension without diabetes or known kidney disorderbut spilled an abnormal amount of protein. Eighteen had diabetes mellitus withrenal insufficiency, and 10 had cancer and a serum creatinine .20 mg/ l thatsuggested renal insufficiency.

2.3. Football players

The effect of exercise on the urinary concentrations of creatinine and albuminwas determined on specimens from 61 members of the varsity football team atthe University of St. Francis (Joliet, IL), 36 participated in a 2-h football game(players), and 25 sat on the bench during the entire game (non-players). Spoturine specimens were collected just prior to the game and within 1 h after thegame from all 61 subjects. The specimens were analyzed immediately foralbumin and creatinine both quantitatively and with the new dipsticks on aBayer CLINITEK 50 analyzer.

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 143

2.4. Quantitative laboratory methods

For albumin, we used the Dade–Behring immunonephelometric method intheir Analyzer II instrument (Dade Behring, Miami, FL). For creatinine, we useda kinetic Jaffe method in the Cobas Mira analyzer (Roche Diagnostic Systems,Sommerville, NJ) with the reagents and procedure from Roche. For proteindeterminations, the Sigma (St. Louis, MO) Microprotein pyrogallol red methodwas performed in the Mira analyzer according to the manufacturer’s instructions.Specific gravity was determined using a TS meter (Cambridge Instruments Inc.,Buffalo, NY). Corrections for specific gravity were calculated by dividing themg/ l albumin or protein by the term, (SG-1)3100. Specimens were refrigeratedat 48C and tested the next day or frozen on the day of collection and stored forup to 14 days, thawed overnight at 48C, and assayed. The above tests wereperformed in duplicate on each specimen. Pre- and post-game specimens fromall 61 football team members were assayed only once for the above tests.

2.5. Point-of-care methods

1The Bayer DCA-2000 Analyzer was used for the quantitative determi-nation of albumin and creatinine in pre- and post-exercise urine specimens. Thispoint-of-care analyzer permitted on-site testing with results available in 5 min.Duplicate results were also obtained for all specimens using a new urinalysisdipstick from Bayer that has test pads for albumin, protein and creatinine. Thestrips were read in a CLINITEK 50 reflectometer [16]. This instrument providessemiquantitative results for albumin and gives readings of 10, 30, 80, 150 and300 mg/ l; the creatinine pads give values of 100, 500, 1000, 2000 and 3000mg/ l; and the protein pads give readings of 0 or negative, 150 or trace, 300,1000 and 3000 mg/ l. The Clinitek 50 dipstick reader gives only one of thevalues stated above for each test and no in-between results.

2.6. Statistical analyses

Nonparametric comparisons were performed using the Mann–Whitney U-testcurve-fitting algorithm (Microsoft Excel) to obtain the equations and coefficient

2of determination (R ) for the line of best fit for urine creatinine vs. urine volumeand (SG-1)3100 vs. urine volume. The Cockroft–Gault equations for estimat-ing the 24-h creatinine excretion are [16]:

• For men, creatinine5[(1402age in years)(weight in kg)] /5000.• For women, creatinine5[(1402age in years)(weight in kg)]30.85/5000.

For example, application of the equation to a 60-year-old man weighing 70 kggives a 24-h creatinine excretion of [(140260)3(70)] /500051.12 g.

144 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

3. Results

3.1. Intra-individual variation in healthy adults

Fig. 1 shows the intra-individual variations in albumin, protein, creatinine andSG in the 4-h collections as determined in 20 healthy volunteers during a 24-hperiod. The concentrations were calculated based on quantitative results andexpressed as mg/24 h, mg/g creatinine, or mg/[(SG-1)3100]. Statisticalcomparisons are in the legend for Fig. 1.

3.2. Inter-individual variation of all subjects

Mean excretions including ranges and inter-individual variations for 24-hcollections in our five groups are shown in Table 1. The creatinine excretion was

Fig. 1. Mean intra-individual variation in urine parameters in 20 healthy volunteers during a 24-hperiod. Albumin excretion vs. protein excretion, P,0.2; albumin excretion vs. albumin/creatinineratio, P,0.1; protein excretion vs. protein /creatinine ratio, P,0.1; albumin excretion vs.albumin/SGU ratio, P,0.05; protein excretion vs. protein /SGU ratio, P,0.05.

D.J.

New

man

etal.

/C

linicaC

himica

Acta

294(2000)

139–155

145

Table 1Inter-individual variation in urinary parameters for creatinine, SG, volume, albumin, and protein

a bUrinary parameters Healthy Kidney damage Hypertension Diabetes Cancer All(units) (n520) (n534) (n56) (n518) (n510) (n588)

c d e dCreatinine (mg/ l) Avg. (Range) 1083 (412–2562) 545 (163–1496) 1008 (311–1926) 600 (165–1927) 508 (189–1357)c%CV 55 48 51 66 65 67

e dCreatinine (mg/24 h) Avg. (Range) 1638 (1114–3697) 1162 (235–2667) 1248 (230–1908) 1636 (45–11735) 1112 (783–1686)%CV 26 43 47 155 25 90

Specific gravity Avg. (Range) 1.015 (1.007–1.031) 1.010 (1.003–1.021) 1.015 (1.008–1.023) 1.015 (1.005–1.032) 1.012 (1.006–1.023)%CV 64 41 54 65 57 59

Urine volume (l) Avg. (Range) 1.90 (0.44–3.61) 2.33 (0.45–4.17) 1.33 (0.62–2.45) 2.36 (0.90–4.26) 2.88 (0.88–4.67)%CV 46 38 51 55 46 48

d f fAlbumin Avg. (Range) 6.1 (2.2–10.1) 783 (4.2–4289) 125 (3.9–525) 697 (6.3–4780) 410 (9.1–2163)hAER (mg/24 h) %CV 34 142 153 185 148 195

d f fAlbumin Avg. (Range) 3.9 (1.3–8.0) 660 (3.4–2875) 407 (3.3–2283) 1033 (1.3–4736) 379 (5.4–1965)hACR (mg/g) %CV 42 120 215 155 144 183

d eAlbumin Avg. (Range) 2.6 (1.0–5.2) 376 (2.2–1895) 163 (2.7–888) 352 (0.8–2007) 187 (1.8–1046)hASG (mg/SGU) %CV 45 123 208 172 159 178

g e f dProtein Avg. (Range) 79 (38–179) 1743 (50–8169) 358 (83–779) 1357 (57–7961) 1282 (198–3119)hPER (mg/24 h) %CV 38 104 83 153 70 143

d f dProtein Avg. (Range) 50 (26–108) 1895 (55–16527) 739 (72–3388) 1850 (26–8025) 1302 (117–3116)hPCR (mg/g) %CV 40 150 170 141 76 173

d f dProtein Avg. (Range) 33 (19–63) 889 (29–4107) 325 (53–1317) 638 (16–3400) 492 (40–1509)hPSG (mg/SGU) %CV 32 102 146 154 84 144

a Patients with cancer and serum creatinine concentrations of $20 mg/ l.b The percentage CV of 24-h specimens from all 88 subjects, i.e. the healthy volunteers and the patients.c The average and percentage CV observed with 24-h specimens for a group of healthy volunteers and for different groups of hospitalized patients with known or

likely kidney disorders.d P,0.05 vs. Healthy.e P,0.1 vs. Healthy.f P,0.2 vs. Healthy.g P,0.01 vs. Healthy.h Abbreviations: AER, albumin excretion; PER, protein excretion; ACR, albumin/creatinine ratio; PCR, protein /creatinine ratio; ASG, albumin/(SG-1)3100 ratio;

PSG, protein /(SG-1)3100 ratio.

146 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

significantly lower in patients with kidney diseases, diabetes, and cancer.Creatinine concentrations varied inversely with the urine volume, as expected(Fig. 2a). The fit of the points to the curvilinear regression line was good, and

Fig. 2. (a) Relationship of 24-h urine volume and urine creatinine concentration in healthy2volunteers and patients. R 50.618, r50.786, P,0.001, n588. (b) Relationship of 24-h urine

2volume with urine (SG-1)3100 in healthy volunteers and patients. R 50.367, r50.606, P,

0.001, n588.

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 147

the number of outliers can be judged from Fig. 2a. The term, (SG-1)3100 was aless efficient predictor of urine volume (Fig. 2b).

In order to test the relationship between the urine volume and creatinine,specimens with creatinine concentrations of ,250 mg/ l were segregated, andthe albumin, protein, specific gravity and volume were determined. Creatinineconcentrations of ,250 mg/ l were observed in only 10 patients with kidney

disorders (Table 2). Five of the 10 had albumins of ,30 mg/ l, and four of thesehad normal albumins when the albumin was divided by the creatinine. Nine ofthe 10 negatives had a normal protein, i.e. ,150 mg/ l, and the remainingpatient had a normal protein when divided by creatinine. Creatinine con-centrations of $2500 mg/ l were observed in two of the 20 volunteers; none ofthe 68 patients had such high values. Summaries of the data from the volunteers(‘Healthy’) and all the patients are given in Table 1.

3.3. Agreement of dipstick results with quantitative assays

The agreement of the semiquantitative dipstick results with the quantitativelaboratory results for albumin, creatinine, and protein was tested with specimensfrom the volunteers and patients. All assays for albumin, protein and creatininewere performed in duplicate. The albumin and protein data are shown in Tables3 and 4. All but one of the dipstick tests on the healthy volunteers was classifiedas normal for albumin. The data for the 68 patients from Tables 3 and 4 alongwith sensitivity and specificity calculations are summarized in Table 5. For bothalbumin and protein concentrations expressed in the three ways, the sensitivitieswere not statistically different for any of the comparisons (P.0.05). Two of thespecificities were statistically different (Table 5).

3.4. Effect of reduced fluid intake on creatinine and SG in healthy volunteers

The creatinine and SG for 16 healthy volunteers were determined after 12 h offluid depravation that increased the creatinine concentrations by an average of1587 mg/ l. With the quantitative method, creatinines in the subjects were all.750 mg/ l with 47%.1500 mg/ l, and 25%.2000 mg/ l. One subject had aurine creatinine concentration of 3980 mg/ l, our highest value. By dipstick, 63%were $2000 mg/ l. SG was also increased to $1.020 in 16 out of 19 specimens.

All specimens collected within 1 h following hydration with 1 l of water hadcreatinine values of ,920 mg/ l with an average of 493 mg/ l. By the dipstick,all were #500 mg/ l. SG was not significantly reduced (P.0.05) and was only,1.010 in two out of 12 cases. The 24-h urines had an average quantitative

148D

.J.N

ewm

anet

al./

Clinica

Chim

icaA

cta294

(2000)139

–155

Table 2Albumin and protein concentrations for 10 patients with ,250 mg creatinine / l in a 24-h collection

Condition Urinary albumin concentration Urinary protein concentration

Creatinine Albumin AER ACR ASG Protein PER PCR PSG

(mg/ l) (mg/ l) (mg/24 h) (mg/g) (mg/SGU) (mg/ l) (mg/24 h) (mg/g) (mg/SGU)

a aKidney damage 163 Neg. 35 148 34 Neg. 572 2436 567a aKidney damage 164 Neg. 54 79 43 Neg. 666 973 532a a a a a a a aDiabetes 165 Neg. Neg. Neg. Neg. Neg. Neg. Neg. Neg.a a a aKidney damage 175 Neg. 46 86 Neg. Neg. 614 1146 Neg.

aCancer 189 69 291 365 98 Neg. 1542 1931 520a aCancer 207 32 148 153 45 Neg. 933 964 Neg.aDiabetes 213 600 1052 2817 400 Neg. 2048 5484 779aKidney damage 232 344 1166 1484 687 Neg. 1906 2425 1123

Kidney damage 249 218 434 877 218 411 8169 16527 4107a a aCancer 249 Neg. 60 71 Neg. Neg. 2661 3116 432

a Reference (‘normal’) ranges: ‘Neg.’ (negative) used here are defined here as: ,30 mg of albumin/ l or ,30 mg albumin/g creatinine or ,300 mg of protein / l, or,300 mg protein /g creatinine. The correction for (SG-1)3100 is expressed here as ‘mg/SGU’; i.e. the albumin or protein values were divided by the term,(SG-1)3100. Abbreviations: AER, albumin excretion; PER, protein excretion; ACR, albumin/creatinine ratio; PCR, protein /creatinine ratio; ASG, albumin/ [(SG-1)3100] ratio; PSG, protein / [(SG-1)3100] ratio; Neg., negative.

D.J.

New

man

etal.

/C

linicaC

himica

Acta

294(2000)

139–155

149

Table 3aAlbumin concentrations in specimens from volunteers and patients. Albumin results by quantitative laboratory methods and dipsticks

Condition Number Albumin No. quant. No. strip No. quant. No. strip No. quant. No. strip

results results results results results results

Units ,30 ,30 30–149 $150 30–149 ,30 30–149 $150 $150 ,30 30–149 $150

bHealthy 20 mg/ l 19 19 0 0 1 0 1 0 0 0 0 0

20 mg/24 h 20 20 0 0 0 0 0 0 0 0 0 0c20 mg/g 20 19 1 0 0 0 0 0 0 0 0 0

aPatients 68 mg/ l 25 18 5 2 15 2 8 5 28 0 5 23

68 mg/24 h 19 15 3 1 23 4 15 4 26 1 4 21c68 mg/g 20 18 2 0 20 3 11 6 28 0 4 24

a Patients with various disorders as given in Table 1.b For example, of the 20 subjects in the Healthy group, 19 had a quantitative albumin of ,30 mg/ l, and negative dipstick values. One Healthy subject had a

quantitative and dipstick albumin between 30 and 149 mg/ l. All tests were performed in duplicate and the results averaged. Duplicate dipsticks agreed in all cases.c Strip results in mg/g are based on ratio of the concentrations of albumin to that of creatinine.

150 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

Table 4Protein concentrations in specimens from volunteers and patients. Protein results by quantitative

alaboratory method and dipsticks

Condition Number Protein No. quant. No. strip No. quant. No. stripresults results results results

Units ,300 ,300 $300 $300 ,300 $300

Healthy 20 mg/ l 20 20 0 0 0 020 mg/24 20 20 0 0 0 0

b20 mg/g 20 20 0 0 0 0

Patients 68 mg/ l 29 22 7 39 0 39with 68 mg/24 h 21 17 4 47 2 45

bconditions 68 mg/g 23 21 2 45 3 42

a Number of dipstick results for 20 healthy subjects and 68 patients with various disorders. All tests wereperformed in duplicate and the results averaged. Duplicate dipsticks results agreed in all cases.

b Strip results in milligrams per gram are based on ratio of the concentrations of protein to creatinine.

creatinine of 716 mg/ l, with all values between 250 and 2500 mg/ l; thedipsticks gave 500–2000 mg/ l. The 24-h concentration agreed with the averageof all the individual specimens as expected. The individual specimens wereportions of the combined 24-h specimens. The average 24-h creatinine excretionagreed with the Cockcroft–Gault value, 615% for all specimens suggesting thatthese were accurately collected 24-h specimens.

The albumin and protein results were not as affected by the specimen volumewhen the ratio to creatinine was used. Individual albumin results ranged from0.2 to 83 mg/ l. All albumin values above the reference range of .30 mg/ lreturned to normal after correction by creatinine. The albumin/creatinine ratioswere 1.0–28.5 mg/g for all specimens. We consider those with ratios #20 mg/g

Table 5Albumin and protein concentrations in specimens from patients. Sensitivity and specificity for

aalbumin and protein results and ratios in 68 patients

Test Units No. TP No. FP No. TN No. FN % Sensitivity % Specificity

Albumin mg/ l 41 7 18 2 95.3 72.0mg/24 h 44 4 15 5 89.8 78.5

bmg/g creatinine 45 2 18 3 93.8 90.0

Protein mg/ l 39 7 22 0 100.0 75.9mg/24 h 45 4 17 2 95.7 81.0

cmg/g creatinine 42 2 21 3 93.3 91.3

a TP, true positive; FP, false positive; TN, true negative; FN, false negative.b For the specificity data, P,0.05 for albumin in mg/ l vs. mg albumin/g creatinine.c P,0.05 for protein in mg/ l vs. mg protein /g creatinine.

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 151

as borderline normal. The expected albumin excretion of 2.0–10.4 mg/24 hmatched the observed 24-h albumin excretions of 2.2–10.1 mg/24 h. Individualprotein results ranged from 0 to 364 mg/ l. The average protein /creatinine ratiowas similar for all specimens, i.e. 46.2–51.7 mg/g. The expected proteinexcretion was 40–133 mg/24 h based on the Cockcroft–Gault calculatedcreatinine concentration. This agreed reasonably well with the observed 24-hprotein excretion of 30–176 mg/24 h.

3.5. The effect of exercise on albumin and creatinine excretion in footballplayers

The mean creatinine concentrations of urine specimens from football playerswere greatly increased by exercise from pre-game mean (S.D.) values of 1590mg/ l (680) to 3400 mg/ l (870) post-game, P,0.001. For the non-players, the

Fig. 3. Albumin concentrations, creatinine concentrations and ratios of the albumin/creatinineconcentrations by dipstick for 25 non-players and 36 players before and after a football game. Thealbumin/creatinine ratio was not significantly different (P,0.2) between the pre- or post-gamevalues for the players or the non-players.

152 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

mean creatinine was 1780 mg/ l (600) pre-game and 1900 mg/ l (1020),post-game (P.0.5). In the players, the albumin increased from 18.2 to 295.0mg/ l giving an average gain of 110.6 mg/ l from pre-game to post-game(P,0.001). The non-players showed a change in the albumin ranging from23.5 to 66.3 mg/ l (average gain of 12.8 mg/ l, P,0.001). The albumin/creatinine ratio reduced the effect of exercise. The players had a mean of 6.7mg/g (12.2) before and a mean of 37.9 mg/g (27.2) post-game or a changeranging from 215.5 to 109.0 mg/g (average gain of 31.2 mg/g, P,0.005).Non-players had a mean of 4.1 mg/g (1.0) before and a mean of 14.4 mg/g(18.6) post-game or a change ranging from 21.0 to 83.0 mg/g (average gain of8.8 mg/g, P.0.2). The dipsticks also showed increased creatinine and albuminconcentrations post exercise in the players but not in the non-players (Fig. 3).All post-game albumin results of players were $30 mg/ l, and most had .3000mg/ l creatinine concentrations. The albumin/creatinine ratio was not sig-nificantly different (P.0.2) between the pre- or post-game values for eitherplayers or non-players.

4. Discussion

In clinical practice, it is difficult to verify that a timed collection of urine iscomplete [17,18]. The inter-individual creatinine excretion is variable and isaffected by age, gender and muscle mass. The many causes of renal insuf-ficiency and failure are also factors. In the patients, using the creatinineexcretion gives an estimate of whether the urine is representative of a complete24-h urine collection [9,17,18]. Intra-individual creatinine excretions are reason-ably steady at a constant hydration. The uncertainty in estimating the complete-ness of a urine collection is about 615% [12]. Nevertheless, this is still superiorto an albumin or protein assay on a random specimen. Most nephrologists insistthat creatinine be determined on all random urines, and the validity of thisapproach is well documented [19].

In our study of a group of healthy individuals, the 24-h creatinine excretionshowed greater variability than the urine volume that in turn showed greatervariability than the specific gravity (Fig. 1). The creatinine excretion was alsosignificantly different between the different patients groups and showed a widervariation than did the urine volume (Table 1). Calculating the ratios of analyteto creatinine is an imperfect tool, nevertheless, we were able to demonstrate thecompleteness of our collections in the volunteers by demonstrating agreementbetween the predicted and measured creatinine excretions.

Urine specific gravity was significantly less variable than either creatinineexcretion or urine volume (Table 1) Plots of creatinine vs. the urine volume

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 153

showed less scatter (Fig. 2a) than did SG vs. the urine volume (Fig. 2b). Urinealbumin excretion reflects primarily glomerular permeability, whereas totalprotein excretion reflects a combination of permeability, tubular leakage, tubularsecretion, and normal protein in the urine shed by the kidneys, e.g. theTamm–Horsfall protein. For the volunteers, the CV of the total protein excretion(96.5%) was not more variable than that of albumin (80.0%) (P.0.2). Howeyreported a similar intra-individual variation of 103% for albumin in randomurines [18]. The variability of protein and albumin excretion in different patientpopulations is shown in Table 2. The table shows that albumin is more variablethan total protein in the patients but not in the volunteers. The manner in whichprotein or albumin excretion was expressed, i.e. mg/24 h, mg/g, etc. made noconsistent difference within patient groups.

Exercise is known to increase the urine concentrations of albumin, protein,and creatinine. Exercise causes water loss, and the individuals typically producea more concentrated urine that is evident from the creatinine concentration[20–24]. We observed this phenomenon in the football players that did vs. thosethat did not play. Our quantitative analyses confirmed previous reports that thereis a degree of exercise-induced proteinuria. The proteinuria would be grosslyoverestimated if corrections for the decreased urinary volume were not made.The semi-quantitative dipstick test confirmed the influence of exercise. The ratioof the protein or albumin to the creatinine concentration eliminated false positivealbumin and protein values owing to dehydration.

In conclusion, we found for the 88 subjects in the study, the strongestcorrelation was between the creatinine concentrations and the 24-h urinevolume: r50.786, P,0.001. The correlation of (SG-1)3100 vs. the 24-h urinevolume was: r50.606, P,0.001, and the correlation for (SG-1)3100 vs. thecreatinine concentration was: r50.666, P,0.001. Intra- and inter-individualvariation in protein and albumin excretion were reduced by dividing by thecreatinine concentration, or by (SG-1)3100. Dividing by creatinine alsocorrected for exercise-induced albuminuria. The dipstick results had a lowincidence of false negatives. Dipsticks with protein, albumin and creatinine padsgave good agreement with the quantitative methods. The ratio of albumin orprotein to creatinine improved the specificity as compared to the value inmilligrams per liter. The sensitivity increased slightly, but the change was notsignificant.

Acknowledgements

We thank Dr Donald R. Parker of Bayer Corporation, Elkhart, Indiana, for hisenthusiastic support during the study.

154 D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155

References

[1] Murakami M. Screening for proteinuria and hematuria in school children: methods andresults. Acta Paediatr Jpn 1990;32:682–9.

[2] Parsons M, Newman DJ, Pugia MJ, Newall RG, Price CP. Performance of a reagent stripdevice for quantitation of the urine albumin: creatinine ratio in a point-of-care setting. ClinNephrol 1999;51:228–32.

[3] Parsons MP, Newman DJ, Newall RG, Price CP. Validation of a point-of-care assay for theurinary albumin:creatinine ratio. Clin Chem 1999;45:414–7.

[4] Garg S, Hiar C, Pennington L, Osberg I, Hamilton R. A new reliable and rapid methodfor estimation of urinary albumin concentration. Diabetologica [Abstract] 1997;40(Suppl1):2096.

[5] Coresh J, Toto RD, Kirk KA, Whelton PK, Massry S, Jones C et al. Creatinine clearance as ameasure of GFR in screens for the African American Study of Kidney Diseases andhypertension pilot study. Am J Kidney Dis 1998;32:32–42.

[6] Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method toestimate glomerular filtration rate from serum creatinine: a new prediction equation.Modification of Diet in Renal Disease study group. Ann Intern Med 1999;130:461–70.

[7] Ruggenenti P, Gaspari F, Perna A, Remuzzi G. Cross-sectional longitudinal study of spotmorning urine protein:creatinine ratio, 24-h urine excretion rate, glomerular filtration rate,and end-stage renal failure in chronic renal disease in patients without diabetes. Br Med J1998;316:504–9.

[8] Moore Jr. RR, Hirate-Dulas CA, Kasiske BL. Use of urine specific gravity to improvescreening for albuminuria. Kidney Int 1997;52:240–3.

[9] Mason HJ, Williams NR, Morgan MG, Stevenson AJ, Armitage S. Influence of biologicaland analytical variation on urine measurements for monitoring exposure to cadmium. OccupEnviron Med 1998;55:132–7.

[10] Lenter C. Body fluids, urine, nitrogenous substances. In: Lenter C, editor, Geigy ScientificTables, West Caldwell NJ: Ciba-Geigy Corporation, 1981, p. 63.

[11] Vestergaard P, Leverett R. Constancy of urinary creatinine excretion. J Lab Clin Med1958;51:211–8.

[12] Cockcroft DW, Gault MH. Prediction of the creatinine clearance from serum creatinine.Nephron 1976;16:311–41.

[13] Corcuff JB, Tabarin A, Rashedi M, Duclos M, Roger P, Ducassou D. Overnight urinary freecortisol determination: a screening test for the diagnosis of Cushing’s syndrome. ClinEndocrinol (Oxford) 1998;48:503–8.

[14] Bjorgaas M, Sagen E, Johnson H, Vik T, Sager G. Urinary excretion of catecholamines inhospitalized and non-hospitalized healthy children and adolescents. Scand J Clin Lab Invest1998;58:339–46.

[15] Pugia MJ, Lott JA, Kajimi J, Takaaki S, Sasaki M, Kuromoto K et al. Screening schoolchildren for albuminuria, proteinuria and occult blood with dipsticks. Clin Chem Lab Med1999;37:149–57.

[16] Pugia MJ, Lott JA, Luke KE, Shihabi ZK, Wians Jr. FH, Phillips L. Comparison ofinstrument-read dipsticks for albumin and creatinine in urine with visual results andquantitative methods. J Clin Lab Anal 1998;12:280–4.

[17] Gowans EM, Fraser CG. Biological variation of serum and urine creatinine and creatinineclearance: ramifications for interpretation of results and patient care. Ann Clin Biochem1988;25:259–63.

D.J. Newman et al. / Clinica Chimica Acta 294 (2000) 139 –155 155

[18] Howey JE, Browning MC, Fraser CG. Biologic variation of urinary albumin: consequencesfor analysis, specimen collection, interpretation of results, and screening programs. Am JKidney Dis 1989;13:35–7.

[19] Ginsberg JM, Chang BS, Matarese RA, Garella S. Use of single voided urine samples toestimate quantitative proteinuria. New Engl J Med 1983;309:1543–6.

[20] Poortmans JR, Engels MF, Sellier M, Leclercq R. Urine protein excretion and swimmingevents. Med Sci Sports Exercise 1991;7:831–5.

[21] Poortmans JR, Vancalck B. Renal glomerular and tubular impairment during strenuousexercise in young women. Eur J Clin Invest 1978;8:175–8.

[22] Robertshaw M, Cheung CK, Fairly I, Swaminathan R. Protein excretion after prolongedexercise. Ann Clin Biochem 1993;30:34–7.

[23] Yaguchi H, Ishigooka M, Hayami S, Kobayashi T, Nakada T, Mitobe K. The effect oftriathlon on urinary excretion of enzymes and proteins. Int Urol Nephrol 1998;30:107–12.

[24] Armstrong LE, Maresh CM, Castellani JW, Mergeron MF, Kenefick RW, LaGasse KE et al.Urinary indices of hydration status. Int J Sport Nutr 1994;4:265–79.