Improved methods of assessing proteinuria in hypertensive pregnancy

Size: px

Start display at page:

Download "Improved methods of assessing proteinuria in hypertensive pregnancy"

Junior May
5 years ago
Views:

British Journal of Obstetrics and Gynaecology October 1997, Vol. 14, pp. 1159-1164 Improved methods of assessing proteinuria in hypertensive pregnancy *?Patrick J. Saudan MD, *?Mark A.

1 British Journal of Obstetrics and Gynaecology October 1997, Vol. 14, pp Improved methods of assessing proteinuria in hypertensive pregnancy *?Patrick J. Saudan MD, *?Mark A. Brown Associate Professor, $Tanya Farrell RN, $Loretta Shaw RN Departments of *Renal Medicine, fhfedicine, and $Obstetrics, St George Hospital, University of New South Wales, Kogarah, Australia Objective To determine whether use of an automated urinalysis device will improve the accuracy of detecting proteinuria, and whether spot urine protein to creatinine ratio will provide accurate quantitation of proteinuria in hypertensive pregnant women. Design Prospective studies assessing the accuracy of both detection and quantitation of proteinuria. Setting Antenatal ward and pregnancy day assessment unit of St George Hospital, a teaching hospital in Sydney, Australia. Population Hypertensive pregnant women admitted to hospital or day assessment unit for management of their hypertensive disorders. Methods I. Routine dipstick urinalysis and 2. Urinalysis by an automated device (Clinitek 1 Ames) on a midstream urine sample were compared with measurement of protein concentration on that sample (n = 13). In a third study, the prtein:creatinine ratio on a midstream (spot) urine sample was compared with protein excretion over the subsequent 24 hours (n = 1). Main outcome measures Relations between urine protein concentrations and 1. dipstick urinalysis and 2. automated urinalysis; 3. Positive and negative predictive values of spot protein : creatinine ratio for true proteinuria (2 3 mglday). Results Automated urinalysis improved the percentage of true positive urinalyses from 48% with visual urinalysis to 74% (P =.2). True negatives were 98% to 1% for both methods. Spot urine protein: creatinine ratio correlated well with subsequent 24-hour urine proteinuria (r =.93, P <.1). A prtein:creatinine ratio > 3 mg proteidmmol creatinine was the optimum discriminant value for true proteinuria, with sensitivity 93%, specificity 92%, positive predictive value 95% and negative predictive value 9%. Conclusions Use of an automated urinalysis device improved accurate detection of proteinuria, particularly reducing false positive tests. A random urine protein: creatinine ratio provides an accurate and rapid quantitation of proteinuria in hypertensive pregnant women. This should improve clinical care, especially when managing hypertensive pregnant women as outpatients. INTRODUCTION Detecting proteinuria is an integral part of the management of hypertensive pregnant women. The presence of proteinuria is required for the diagnosis of preeclampsia in some classifications'** and is associated with a higher rate of maternal and fetal complications among hypertensive pregnancies3s4. Moreover, quantitation of proteinuria is important in pre-eclampsia as the likelihood of complications increases with nephrotic range proteinuria5. Detecting proteinuria in pregnant women is usually done by routine visual dipstick urinalysis. However, recent studies have documented inaccuracies of this Correspondence: AProfessor M. A Brown, Department of Renal Medicine, St George Hospital, Kogarah, NSW 2217, Australia. RCOG 1997 British Journal of Obstetrics and Gynaecology method, giving high false positive6s7 and negative resultsb when compared with the gold standard of 24- hour urine measurement. In addition to the inaccuracies of detecting proteinuria, there are also practical difficulties quantifying proteinuria in hypertensive pregnant women. Problems are often encountered with 24-hour urine collection, such as inaccurate timing and incomplete collection due to ureteric dilatation. Moreover, it is often difficult waiting 24 or more hours to know whether proteinuria is truly present. We postulated that the use of an automated urindysis device might reduce observer variability and hence improve the false negative and positive results of dipstick urinalysis. Secondly, we proposed that a spot urine prtein:creatinine ratio could provide an accurate assessment of 24-hour urine proteinuria, 1 I59

2 116 P. J. SAUDAN ET AL. thereby improving clinical management of hypertensive pregnant women. METHODS Pregnant women admitted to hospital or our pregnancy day assessment unit for management of their hypertensive disorders over a six-month interval were invited to participate in these studies. The studies were approved by the Southern Sydney Area Health Service Ethics Committee. Urinalysis studies Two studies were undertaken, one comparing routine dipstick urinalysis with laboratory urine protein concentration measurement, the second comparing urinalysis done by an automated device with laboratory urine protein concentration. In the first study, a routine visual dipstick urinalysis was done by one of our trained midwives on a carefully collected midstream urine sample, usually obtained in the morning. Multistix losg test strips (Bayer Diagnostics, Victoria, Australia) were used to grade urine protein concentration as nil, trace, 1+ (.3 &), 2+ (1 gl), or 3+ (13 &). For the second study, urinalysis was made on the same urine sample using an automated device (Clinitek 1 Ames). The printout result of this urinalysis was recorded and the same urine sample then sent to the biochemistry laboratory for formal protein concentration determination. Positive predictive values were calculated by considering urinalysis of 2 1+ (.3 g/l) to be positive and measured urine sample protein concentration of 2-3 g/l to be true positive. Negative predictive values were calculated by considering urinalysis of nil or trace to be negative and urine sample protein concentration <.3 gl to be true negative. Specificity and sensitivity were also determined on this basis. Differences in the frequency of positive and negative results were calculated by x2 testing using contingency tables with continuity correction. Urine protein : creatinine ratio study Our third study compared the spot urine protein : creatinine ratio with protein excretion over 24 hours. Following collection of a spot midstream urine sample usually, but not always, obtained in the morning, urine was then collected for 24 hours for measurement of protein and creatinine concentration and excretion. Urine protein was measured by a benzethoniwn chloride turbidometric method and urine creatinine by the Jaffe method, both using an Hitachi 911 autoanalyser - I 7 r I 5 3 z 2 1 n - Neghrace or 4+ Fig. 1. Accuracy of visual and automated urinalyses: by routine dipstick visual method (El) and by automated Urinalysis (M). The number of urine samples in which urinalysis did not reflect true negative (< 3 mg/l) or true positive Q 3 m a) proteinuria as measured by formal laboratory assay (i.e. a false result) is indicated by B. (Boehringer Manheim). The inter-assay coefficient of variation for this test in our laboratory is 1-6%, the intra-assay coefficient of variation being 3.4% at mean urine protein.21 g/l and.8% at -61 g/l. The urine protein: creatinine ratio was obtained by dividing the urine protein concentration (mgl) by the urine creatinine concentration (mmovl). This was calculated for both the spot and 24-hour urine samples. Protein : creatinine ratios obtained by both methods were compared by Bland-Altman plotsg. The latter method was used to compare urine protein:creatinine ratios obtained by each method firstly for all urine samples and then for only those urine samples which had 5 mglday proteinuria, a degree of proteinuria where it is most difficult to know whether true proteinuria is present from urinalysis alone6,. Sensitivity, specificity, positive and negative predictive values were determined for different protein : creatinine ratio values and a receiver-operator characteristic curve was established with proteinuria 1 3 mg/day as true positive and < 3 mg/day as true negative. RESULTS Urinalysis study Visual and automated dipstick urinalyses were performed on 13 samples and compared with the urine protein concentration of that sample. The distribution of proteinuria according to visual dipstick and automated urinalysis is shown in Fig. 1 and sensitivity, specificity, negative and positive predictive values are shown for each level of proteinuria in Table 1. True positive results were significantly more frequent with automated urinalysis ($ = 5.1, P =.24). With visual dipstick windysis, positive predictive values ranged from 24% at the RCOG 1997 Br J Obstet Gynaecol 14,

3 PROTEINURIA IN HYPERTENSIVE PREGNANCY 1161 Table 1. Sensitivity, specificity and predictive values ( h) of visual and automated urinalyses (n = 13). True positives are the percentage of urine samples with corresponding urine protein concentration 2 3 mg/l. True negatives are the percentage of urine samples with corresponding urine protein concentration < 3 m a. More urinalyses by the automated method were true positives than by ward urinalysis (P =.24). PPV = positive predictive value; NPV = negative predictive value; WU =ward urinalysis; Auto = automated. Urinalysisresult Sensitivity Specificity PPV NPV WU Auto WU Auto WU Auto WU Auto Negative or trace 1 98 I+ (.3 g/l) (1 g/l) (13&) I+ level to 93% at the 3+/4+ level. With automated urinalysis, these values ranged from 47% at the 1+ level to 1% at the 3+ level. Overall sensitivity was 1% and 97%, respectively, for visual dipstick and automated analysis while specificity was 55% and 85%, respectively. Urine protein : creatinine ratio study Twenty-four hour urine collections with a spot (random) urine sample before starting collection were provided by 1 women. Spot urine protein: creatinine ratio correlated well with 24-hour urine protein excretion (Y =.93, P <.1). Fourteen percent of our patients had proteinuria 2 3 g/day and all of these had a prtein:creatinine ratio > 38 mg proteidmmol creatinine. The limits of agreement between the two prtein:creatinine ratios for the 62 samples with proteinuria < 5 mg were -2 to 18 mg protedmmol creatinine (Fig. 2a). When calculated for all samples, the limits of agreement were -284 to 272 mg protein/ mmol creatinine (Fig. 2b). On the basis of the ROC (Fig. 3), the most discriminant spot urine protein : creatinine ratio value for detecting true proteinuria (2 3 mg/day) was 3 mg protedmmol creatinine, based on defining the protein: creatinine ratio value which gave the best combination of specificity, sensitivity, positive and negative predictive values. This is evident in Table 2 which shows the sensitivity, specificity, positive and negative predictive values for protein : creatinine ratio values over the range 2 to 45 mg proteidmmol creatinine. The 3 mg proteidmmol creatinine protein: creatinine ratio value yielded a positive predictive value of 95% and a negative predictive value of 9%, while maintaining sensitivity and specificity above 9%. None of the false negatives according to this protein : creatinine value had proteinuria greater than 4 mg/day. Table 2. Predictive values of spot urine prtein:creatinine values for true proteinuria 2 3 mg/day (n = 1). Cutoff value is given in mg proteinlmmol creatinine. Key as for Table 1. Cutoff value Sensitivity (%) Specificity ( h) PPV (%) NPV ( h) I DISCUSSION Accurate and rapid detection and quantitation of proteinuria are essential for the management of hypertensive disorders in pregnancy. In our study, use of an automated analyser significantly increased the accuracy of proteinuria detection in hypertensive pregnant women and quantitation was made rapidly and reliably using the protein : creatinine ratio in a spot urine sample. Dipstick urinalysis, which is usually done as the screening test for proteinuria, has been found to give high false po~itive~.~ and false negative rates8, casting doubt on the reliability of dipstick urinalysis for detecting true proteinuria in clinical and research practice. None of these previously published studies assessing urinalysis in pregnant women used an automated urinalyser for detecting proteinuria. Our study confirmed the high false positive rate found in previous studies and showed that automated urinalysis improved the predictive power of urinalysis at the lower levels of I+ (.3 gl) and 2+ (1 a). In an obstetric clinic accuracy of dipstick urinalysis relies on midwives and doctors interpretation, careful collection of midstream samples to avoid contamination with vaginal secretions, and urine concentration. The automated device eliminated inter- and intra-observer variabilities and midwives always attempted to obtain a carefully collected midstream sample. Even so, there were persistently high rates of false positives encountered at the 1+ level with the automated analyser, though only to half the extent of the visual urinalysis. This implies that the chemical reaction in the dipstick, set to reflect low levels of proteinuria (around.3 gl), is oversensitive. This is advantageous in managing pre-eclampsia as it avoids missing cases of proteinuria but it will be incorrect at least half the time. (Table 1) When this automated analyser was first evaluated (in nonpregnant women) it gave only a 2% to 3% incidence of false positive results but this was testing contrived urine samples containing known amounts of albumin O. The automated analyser also compared favourably with visual readings when it was used for determination of proteinuria from nephrotic patients. RCOG 1997 Br J Obstet Gynaecol 14,

4 1162 P. J. SAUDAN ET AL. = E,E - E" t I I I I I I I*r QAA,. IL ' " " -5-5 t -1 1 I I I I I Average PIC ratio (rng/rnrnol) Fig. 2. Limits of agreement between protein:creatinine ratios (PIC) on 24-h urine collection and spot urine samples: (a) samples with < 5 mg protewday; (b) all samples. Another study compared the accuracy of urinalyses obtained by a semi-automated urinalysis device with quantitative manual urinalysis performed by trained technicians12. Their visual readings were as precise as those achieved by the reflectance photometer. However, this result is unlikely to be reproducible in the stressfil working conditions of an antenatal ward, where levels of training vary and there is greater scope for observer RCOG 1997 Br J Obstet Gynaecol 14,

5 PROTEINURIA IN HYPERTENSIVE PREGNANCY 1163 variabilitity. The importance of midwives training and their understanding of the importance of the urinalysis probably reflects the final accuracy of routine ward urinalyses. As evidence for this, the same midwives involved in this study took part in our previous study assessing the accuracy of visual dipstick urinalysis7. Using the same teststrip, the sensitivity of visual dipstick urinalysis increased from 86% in the prior study to 1% and specificity from 39% to 55%. In our third study we found that a random urine protein: creatinine ratio was an accurate indicator of the subsequent 24-hour urine proteinuria. This has been found previously in nonpregnant healthy volunteers and in patients with renal impairment13, kidney transplant patientsi4 and in type I diabetic patients with nephropathyi5. In three previous studies of pregnant women use of a random urine protein : creatinine ratio gave conflicting results. A longitudinal study of pregnant women with renal disease showed excellent correlation of random daytime urine protein: creatinine ratio with 24-hour urine proteinuria throughout pregnancy, but this study did not involve hypertensive patients16. Another study enrolling pre-eclamptic patients also noted a good correlation between the random urine protein: creatinine ratio and 24-hour urine proteinuria. This study, however, included a limited number of proteinuric patients (n = 35) and the spot urine was only collected in the morning, not reflecting usual clinical practice where urinalysis may be required at other times during the day or night. Lindow and DaveyI8 concluded that the protein: creatinine ratio did not reliably predict the amount of proteinuria due to variation of protein and creatinine excretion during daytime. In that study, using catheterised pregnant women, the protein : creatinine ratio had a coefficient of variation of 27% among eight hourly collections, but the number of patients involved was small (n = 22) and all had significant proteinuria (2 2+ on dipstick analysis). Our study involved only hypertensive pregnant women and the excellent correlation coefficient between the spot urine protein: creatinine ratio and subsequent 24-hour protein excretion was similar to that of Quadri et al. I6 and Jaschevatsky et al. 17. More importantly, the Bland-Altman plots showed good agreement between the two methods of assessing proteinuria, especially for patients with < 5 mg proteinuridday. When all samples were included,in the analysis, the limits of agreement increased to *227 mg proteidmmol creatinine, demonstrating a reduced accuracy of the prtein:creatinine ratio in the presence of heavy proteinuria. Nevertheless, all our patients with heavy proteinuria (2 3 g daily, 1 times the upper limit of normal) had a protein : creatinine ratio more than 38 mg proteidmmol creatinine (12 times the upper limit of normal). In the majority of women, who had > c) u) cn lower levels of proteinuria, the agreement between methods was good and the few women who had false negative results using cutoff protein: creatinine value of 3 mg proteidmmol creatinine all had < 4 mg proteinuridday (i.e. none had very high protein excretion). This is reassuring in that the few false negatives had only minimally increased proteinuria. Including our data, there is now evidence in three out of four studies in pregnant women that the spot urine prtein:creatinine ratio is an accurate indicator of 24- hour proteinuria, particularly in hypertensive pregnant women. CONCLUSION 2 I I I I Our results show that use of an automated urinalysis device increases the positive predictive value of urinalysis from 24% to 47% at the 1+ (.3 g/l) concentration and from 53% to 83% at the 2+ (1 gl) concentration, without significantly altering the false negative rate. Further, random urine protein: creatinine ratio of > 3 mg proteidmmol creatinine has an excellent predictive value for proteinuria above 3 mg/day. Its rapid determination in the same sample used for dipstick analysis means that clinicians should now be able to screen hypertensive pregnant women for proteinuria more accurately than before and then estimate the quantity of proteinuria within a few hours. These practical changes should improve clinical care, particularly when managing pregnant women with mild hypertension as outpatients. RCOG 1997 Br J Obstet Gynaecol 14,

6 1164 P. J. SAUDAN ET AL. Acknowledgements The authors would like to thank the midwives for their enthusiastic involvement in these studies, the obstetricians of St George Hospital for allowing us to study their patients and Mrs J. Hendley for her secretarial assistance. The study was supported by grants from the Division of Medicine and Southpath Pathology services, St George Hospital. Dr P. Saudan is a recipient of the fonds de perfectionnement from the University Hospital, Geneva, Switzerland. References Davey DA, MacGillivray I. The classification and definition of the hypertensive disorders of pregnancy. Am J Obstef Gynecol1988; 158: National High Blood Pressure Education Program Working Group. Report on high blood pressure in pregnancy. Am J Obstef Gynecol 199,163: Ferrazani S, Caruso A, De Carolis S, Martino W, Mancuso S. Proteinuria and outcome of 444 pregnancies complicated by hypertension. Am JObstef Gynecoll99; 162: Brown MA, Buddle ML. The importance of nonproteinuric hypertension in pregnancy. Hypertens Pregnancy 1995; 14: Chua S, Redman CWG. Prognosis for pre-eclampsia complicated by 5 g or more of proteinuria in 24 hours. Eur J Obsfet Gynaecol 1988; 29: Kuo VS, Koumantakis G, Gallery EDM. Proteinuria and its assessment in normal and hypertensive pregnancy. Am J Obsfet Gynecol 1992; 17: Brown MA, Buddle ML. Inadequacy of dipstick proteinuria in hypertensive pregnancy. Aust NZJObstet Gynecol1995; 35: 4: Meyer NL, Mercer BM, Friedman SA, Sibai BM. Urinary dipstick protein: a poor predictor of absent or severe proteinuria. Am J Obstet Gynecoll994; 17: Bland JM, Altman DG. Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet 1995; 346: Peele JD, Gadsden RH, Crews R. Evaluation of Ames Chi-Tek. Clin Chem 1977; 23: Peele JD, Gadsden RH, Crews R. Semi-automated vs visual reading of urinalysis dipsticks. Clin Chem 1977; 23: Elin RJ, Hosseini JM, Kestner I, Nishi HH. Comparison of automated and manual methods for urinalysis. Am J Clin Path 1986; Ginsberg JM, Chang BS, Matarese RA, Garella S. Use of single voided urine samples to estimate quantitative proteinuria. N Engl J Med 1983; 39: Steinhauslin F, Wauters JP. Quantitation of proteinuria in kidney transplant patients: accuracy of the urinary proteidcreatinine ratio. Clin Nephmll995; 43: Brodby RA, Rohde RD, Zeev S, Pohl MA, Bain RP, Lewis EJ. The urine protein to creatinine ratio as a predictor of 24-hour urine protein excretion in type 1 diabetic patients with nephropathy. Am J Kid Dis 1995; 26: Quadri KHM, Bemardini J, Greenberg A, Laifer S, Syed A, Holley JL. Assessment of renal hnction during pregnancy using a random urine protein to creatinine ratio and Cockcroft-Gault formula. Am J KidDis 1994; 24: Jaschevatzky OE, Rosenberg P, Shalit A, Zonder HB, Grunstein S. Proteidcreatinine ratio in random urine specimens for quantitation of proteinuria in pre-eclampsia. Obsfet GynecoZl99; 75: Lindow SW, Davey DA. The variability of urinary protein and creatinine excretion in patients with gestational hypertension. Br J Obstet Gynaecoll992; 99: Received 21 October 1996 Returnedfor revision 18 March 1997 Accepted 25 April 1997 Q RCOG 1997 Br J Obstet Gynaecol 14,

The Relationship between Random Urinary Protein-to-Creatinine Ratio and 24-hours Urine Protein in Diagnosis of Proteinuria in Mild Preeclampsia

Received: 8/22/2004 Accepted: 4/20/2005 Original Article The Relationship between Random Urinary Protein-to-Creatinine Ratio and 24-hours Urine Protein in Diagnosis of Proteinuria in Mild Preeclampsia