elevated urinary albumin levels have been found to be a predictor of cardiovascular disease in some studies. 5 9

Transcription

1 AJH 1998;11: Assessment of a New Dipstick Test in Screening for Microalbuminuria in Patients With Hypertension Linda M. Gerber, Karen Johnston, and Michael H. Alderman The prevalence of elevated urinary albumin levels is significantly greater in hypertensive than in normotensive subjects. To determine the sensitivity and specificity of a new dipstick test for microalbunimuria (the ), 171 hypertensive patients were studied at a unionsponsored hypertension treatment program. Sensitivity, specificity, predictive values, and correlation coefficients between urinary albumin concentration results obtained by the and by nephelometry were determined in three urine samples. Sensitivity values of the Micral- Test, compared with 24-h nephelometry, were 81%, 75%, and 92% in a 24-h, overnight, and random sample, respectively. Specificity values were 89%, 90%, and 63% in the three samples, respectively. Positive predictive value ranged from 41% to 67%, whereas negative predictive value ranged from 93% to 97%. Correlation coefficients between the logarithms of albumin concentrations obtained from the three different urine specimens using nephelometry fell between 0.71 and 0.78, whereas those obtained with the fell between 0.49 and 0.71, and across techniques, 0.29 to 0.53 (all P <.001). Results obtained with both nephelometry and the using overnight and random urine collections approximated those obtained with 24-h collection. These results, coupled with the ease and convenience of both specimen collection and the itself, support the use of the test as a valuable screening tool for microalbuminuria in patients with hypertension. Am J Hypertens 1998;11: American Journal of Hypertension, Ltd. KEY WORDS: Blood pressure, microalbuminuria, screening test, sensitivity, specificity. Elevated urinary albumin levels have been associated with increased mortality and have predicted clinical proteinuria in both insulin-dependent and non insulin-dependent diabetics. 1 4 In nondiabetic subjects with hypertension, elevated urinary albumin levels have been found to be a predictor of cardiovascular disease in some studies. 5 9 We have recently reported the prevalence of elevated urinary albumin levels to be significantly greater in hypertensive than in normotensive subjects. Furthermore, the level of urinary albumin concentration was positively related to blood pressure level among hypertensive subjects. 10 Because screening hypertensive subjects for microalbuminuria may become an important part of the management of hypertension, Received February 11, Accepted May 8, From the Department of Public Health, The New York Hospital Cornell University Medical College, New York, New York; and the Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine, Bronx, New York. This study was supported by Boehringer Mannheim Corporation, Mannheim, Germany. Address correspondence and reprint requests to Linda M. Gerber, PhD, Department of Public Health, Cornell University Medical College, 411 East 69th Street, New York, NY by the American Journal of Hypertension, Ltd /98/$19.00 Published by Elsevier Science, Inc. PII S (98)

2 1322 GERBER ET AL AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 we have assessed the sensitivity and specificity of a new dipstick test for microalbuminuria. Although found elsewhere to be an accurate, sensitive, and specific method in screening for microalbuminuria, the (Boehringer Mannheim, Indianapolis, IN) has not yet been used for screening among hypertensive subjects, nor has its sensitivity or specificity been tested in this population. Although many investigators view 24-h urine collections as providing the most accurate and consistent readings of urinary albumin concentrations, there is a concern for subject compliance. 5 As a result, random urine samples are often used due to their ease of collection, patient compliance, and convenience. 11 Finally, overnight urine collections provide another alternative in measuring urinary albumin excretion. 12 The sensitivity and specificity of the in relation to the 24-h immunochemistry by nephelometry method has not been determined in different timed urine samples (overnight, 24-h, and random urine specimens) in subjects with hypertension. This study was designed to address these questions. SUBJECTS AND METHODS Study subjects were 171 hypertensive patients who entered a New York City union-sponsored treatment program in 1994 with systolic blood pressure 140 or diastolic blood pressure 90 mm Hg. Patient evaluation and treatment plan have been described in detail elsewhere. 13,14 Demographic and clinical data (history and results of medical and physical examination) as well as routine laboratory tests were obtained for all participants. Blood pressure readings used in this study were taken by a nurse using a standard sphygmomanometer. Three blood pressure measurements were taken with the subject in a seated position and the average of the last two was used as the estimate of blood pressure. All blood pressures were measured before initiation of antihypertensive medication. Height and weight were determined with the subject in street clothes, and the body mass index was calculated as the weight in kilograms divided by the square of the height in meters. Informed consent was obtained from all participants at the time of recruitment, as approved by The New York Hospital Cornell Medical Center Committee on Human Rights in Research. Participants were instructed to collect consecutive overnight and 24-h urine samples and bring them to the clinic, at which time a random sample was obtained. Urinary albumin concentration was determined by nephelometry and by the Chemstrip Micral immunochemical dipstick () in each of the three samples. Aliquots of the three samples were refrigerated and sent to Metpath Laboratories (Teterboro, NJ). These samples were quantitated for urine TABLE 1. CHARACTERISTICS OF STUDY POPULATION Characteristics n Mean SD Range Age (years) Systolic BP (mm Hg) Diastolic BP (mm Hg) Body mass index (kg/m 2 ) Fasting blood glucose (mg/dl) History of diabetes (%) Male (%) Race (%) 170 White 24.7 Black 28.2 Hispanic 42.4 Other 4.7 albumin by rate nephelometry on the Array Protein System (Beckman Instruments, Inc., Brea, CA), with a detection limit of 2.3 g/ml and a coefficient of variation of 10%. The is an immunoassay test. The strip is immersed in the urine sample for 5 sec. A color reaction takes place and the resulting color is proportional to the albumin concentration in the urine. The color is read 5 min after dipping. The yields a range of semiquantitative results, which are determined by comparing the strip color with five color blocks on the vial label corresponding to concentrations approximating 0, 10, 20, 50, and 100 or greater. 15 According to the manufacturer s instructions, albumin concentrations 20 mg/l were classified as positive for microalbuminuria. Because urinary albumin concentrations were not normally distributed, these data were logarithmically transformed before correlation analyses. For the categorical results obtained by the, the logarithm of each category was estimated by taking the midpoint of the upper and lower boundaries of each category. To estimate the logarithm for the open- TABLE 2. COMPARISON OF URINARY ALBUMIN CONCENTRATIONS MEASURED BY NEPHELOMETRY AND MICRAL-TEST WITHIN THE SAME SPECIMEN USING A 24-H URINE SAMPLE Category (mg/l) Urine Albumin Concentration by Nephelometry (mg/l) >

3 AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 MICROALBUMINURIA SCREENING IN HYPERTENSIVES 1323 TABLE 3. COMPARISON OF URINARY ALBUMIN CONCENTRATIONS MEASURED BY NEPHELOMETRY AND MICRAL-TEST WITHIN THE SAME SPECIMEN USING AN OVERNIGHT URINE SAMPLE TABLE 4. COMPARISON OF URINARY ALBUMIN CONCENTRATIONS MEASURED BY NEPHELOMETRY AND MICRAL-TEST WITHIN THE SAME SPECIMEN USING A RANDOM URINE SAMPLE Category (mg/l) Urine Albumin Concentration by Nephelometry (mg/l) > 100 Category (mg/l) Urine Albumin Concentration by Nephelometry (mg/l) > ended upper category ( 100 mg/l), the logarithm of 200 was used. For the lowest category ( 1), a score of 0 was assigned (n 7 of 473 tests). Two-tailed probability levels for statistical significance are reported. RESULTS Table 1 describes the characteristics of the study population. Average blood pressures were within the mild range for hypertension. Although the mean fasting blood glucose was in the normal range, there was a good deal of variation in the group (standard deviation of 23.5 mg/dl), with six subjects having values 140 mg/dl. Additionally, 2.4% (n 3) of subjects had a history of diabetes, yielding a total of nine subjects with fasting glucose values 140 mg/dl or a history of diabetes. Tables 2 to 4 compare the results with the nephelometry results within the same specimen. Using a 24-h urine collection, 30 samples (of 37) with albumin concentrations 20 mg/l by nephelometry and 119 of the 134 having albumin concentrations 20 mg/l were correctly identified by the, giving a sensitivity of 81.1%, a specificity of 88.8%, a false positive rate of 8.8%, and a false negative rate of 4.1% ( coefficient , P.001) (Table 2). All the false negatives (n 7) fell in the 20 to 49.9 mg/l category (with a mean of 25.8 mg/l). Using an overnight specimen, the detected 21 of 25 samples with albumin concentrations 20 mg/l and correctly classified 125 of 145 having concentrations 20 mg/l, giving a sensitivity of 84.0%, a specificity of 86.2%, and false positive and false negative rates of 11.8% and 2.4%, respectively ( coefficient , P.001) (Table 3). Using a random specimen, 58 samples (of 66) with albumin concentrations 20 mg/l and 78 of the 104 having concentrations 20 mg/l were correctly classified by the, yielding a sensitivity and a specificity of 87.9% and 75.0%, respectively, a false positive rate of 15.3%, and a false negative rate of 4.7% ( coefficient , P.001) (Table 4). Figures 1 to 3 present these data graphically. Tables 5 to 7 compare each specimen (24-h, overnight, and random) against the albumin concentrations measured by nephelometry in the 24-h collection. Table 5 is therefore similar to Table 2 but further FIGURE 1. Scatterplot of Micral- Test results compared with urine albumin concentration in a 24-h sample.

4 1324 GERBER ET AL AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 FIGURE 2. Scatterplot of Micral- Test results compared with urine albumin concentration in an overnight sample. provides a dichotomized table and a positive predictive value of 66.7%, negative predictive value of 94.4%, and coefficient of (P.001). Table 6 compares the result from an overnight specimen with the nephelometry-measured albumin concentration of the 24-h collection. The sensitivity and specificity are 75.0% and 89.6%, respectively, with positive predictive value of 65.9% and negative predictive value of 93.0%. The coefficient is (P.001). Table 7 compares the results obtained by the Micral- Test in a random specimen with the albumin concentration measured by nephelometry in the 24-h collection. The sensitivity and specificity are 91.9% and 62.7%, respectively, with a positive predictive value of 40.5%, negative predictive value of 96.6%, and coefficient of (P.001). The prevalence of microalbuminuria as defined by an albumin concentration of 20 mg/l in a 24-h collection measured by nephelometry is 21.6% (37/171). The subgroup of nine subjects with fasting blood glucose 140 mg/dl or a history of diabetes was examined separately. When urinary albumin was measured by nephelometry, two, one, and six subjects reached levels 20 mg/l in the 24-h, overnight, and random specimen, respectively. By the, values 20 mg/l were obtained for one, two, and seven subjects in the 24-h, overnight, and random specimen, respectively. As they were too small a group to analyze further, these subjects have been excluded from the following analyses. Correlation coefficients between the logarithms of albumin concentrations measured by the and nephelometry are shown in Table 8. All correlations are very highly significant, with the nephelometry-measured correlations ranging between 0.61 and The correlation between the 24-h and overnight albumin concentrations measured by the also fell within this range (0.71), whereas it was 0.49 FIGURE 3. Scatterplot of Micral- Test results compared with urine albumin concentration in a random sample.

5 AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 MICROALBUMINURIA SCREENING IN HYPERTENSIVES 1325 TABLE 5. RESULTS OF THE MICRAL-TEST: 24-H SPECIMEN 24-h Nephelometry Positive* Negative Total 24-h Positive* Negative Total Sensitivity 81.1% Specificity 88.8% Positive predictive value 66.7% Negative predictive value 94.4% coefficient (P.001) * Albumin concentrations 20 mg/l are considered positive. Albumin concentrations 20 mg/l are considered negative. TABLE 7. RESULTS OF THE MICRAL-TEST: RANDOM SPECIMEN 24-h Nephelometry Positive* Negative Total Random Positive* Negative Total Sensitivity 91.9% Specificity 62.7% Positive predictive value 40.5% Negative predictive value 96.6% coefficient (P.001) * Albumin concentrations 20 mg/l are considered positive. Albumin concentrations 20 mg/l are considered negative. and 0.53 when the 24-h and overnight results were compared with the random specimen. Correlation coefficients ranged from 0.29 to 0.53 when results obtained by the were compared with those obtained by nephelometry. DISCUSSION The compares well with the conventional laboratory measurement of albumin concentration. Within the same urine specimen, whether a timed 24-h, an overnight, or a random collection, sensitivity ranged from 81% to 88%, with a range of specificity from 75% to 89%. All urine-collection procedures had higher false positive than false negative rates, and as TABLE 6. RESULTS OF THE MICRAL-TEST: OVERNIGHT SPECIMEN 24-h Nephelometry Positive* Negative Total Overnight Positive* Negative Total Sensitivity 75.0% Specificity 89.6% Positive predictive value 65.9% Negative predictive value 93.0% coefficient (P.001) * Albumin concentrations 20 mg/l are considered positive. Albumin concentrations 20 mg/l are considered negative. such, would serve well as an initial screening test. Using all three types of collections, only one of 512 tests had a clearly elevated value ( 100 mg/l) that was not detected by the ; the other false negative results fell in the category approximating 20 mg/l. Although the overnight results more closely approximated those obtained by nephelometry using a 24-h collection ( coefficient ), the sensitivity of the using a random sample was much higher than that of the overnight sample when compared with the gold standard (91.9% v 75.0%). The probability that an individual is truly normoalbuminuric given a negative result with both the overnight and random collections is, however, extremely high (negative predictive values, 93% and 97%, respectively). This study is the first to transform results obtained from the semiquantitative into continuous quantitative results. This allowed both the timing of the urine collection and the technique in which urinary albumin was determined to be assessed simultaneously. Although recent studies of microalbuminuria have been conducted in patients with hypertension, 10,16 19 prior studies comparing results obtained from the with either radioimmunoassay or nephelometry have been conducted predominantly in diabetic samples 15,20 26 and in patients with kidney disease. 27 The results of the present study are similar to those of others, who found sensitivity and specificity values 80% to 90% when comparing results obtained from the with those obtained by laboratory methods within the same specimen. 15,20,22 24 Other investigators have found sensitivity values as great as

6 1326 GERBER ET AL AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 TABLE 8. CORRELATION COEFFICIENTS BETWEEN LOG URINARY ALBUMIN VARIABLES Variable Nephelometry 24-h Overnight Random 24-h Overnight Random : 24-h * * * * * : overnight * * * * : random * * * Nephelometry: 24-h * * Nephelometry: overnight * Nephelometry: random *P %, although these studies have been conducted on smaller numbers of diabetic subjects. 26 Tiu et al 25 compared the with radioimmunoassay using overnight urine samples in 75 diabetic patients and found the sensitivity and specificity of the to be 75% and 87.3%, respectively. The corresponding figures in the current study were 84% and 86.2%. Furthermore, when results of the overnight samples were compared with the 24-h collections measured by nephelometry in the current study, the sensitivity was 75% and specificity was 89.6%, almost identical to those found by Tiu et al. 25 To our knowledge, to date only two studies have used two different urine-collection procedures in comparing the with radioimmunoassay. 20,21 The study by Ward and Rao 20 compared urinary concentration measured by the with levels obtained with nephelometry using random and overnight samples. The reported correlations in that study between random and overnight results obtained by the two methods were higher for the and lower for nephelometry than those found in the present study; however, the difference in patient selection and sample size (diabetic men, years of age, N 29) may account for some differences observed. Furthermore, correlations obtained by Ward and Rao 20 were not based on logarithmically transformed albumin concentrations, as was the case in the present study. In another study comparing results obtained from the and nephelometry using a 24-h sample and a random one, Ward and colleagues 21 found no difference between mean results. The prevalence of microalbuminuria found in the present investigation ranged from 15% to 49%, depending on the type of urine collection and method of urinary albumin measurement. Using the gold standard of nephelometry in the 24-h collection, the prevalence of microalbuminuria was 21.6% when defined as a urinary albumin concentration 30 mg/l. This corresponds to an earlier study conducted in a different sample drawn from the same union-sponsored hypertension treatment program. Where the prevalence of urinary albumin levels of 15 mg/24 h in the earlier study was found to be 31.1% as determined by radioimmunoassay in a 24-h urine collection, 10 the equivalent prevalence in the current study based on times the number of hypertensive subjects is 29.1%. These results suggest that the offers a simple and easy method to screen for microalbuminuria in patients with hypertension. Although the false positive rates are not trivial, at 12% and 15% with an overnight and random specimen, respectively, they are offset by the extremely low false negative rates (2% and 5% for overnight and random specimens, respectively). The determination of urinary albumin can be easily accomplished by the in both the clinical and epidemiologic setting, providing important prognostic data for patients with hypertension. REFERENCES 1. Mogensen CE: Microalbuminuria predicts clinical proteinuria and early mortality in maturity-onset diabetes. N Engl J Med 1984;310: Mogensen CE, Poulsen PL: Epidemiology of microalbuminuria in diabetes and in the background population. Curr Opin Nephrol Hypertens 1994;3: Jarrett RJ, Viberti GC, Argyropoulos A, et al: Microalbuminuria predicts mortality in non insulin-dependent diabetes. Diabet Med 1984;1: Viberti G: Etiology and prognostic significance of albuminuria in diabetes. Diabetes Care 1988;11: Metcalf PA, Scragg RKR: Epidemiology of microalbuminuria in the general population. J Diabetes Complications 1994;8: Bulpitt CJ, Beevers DG, Butler A, et al: The survival of treated hypertensive patients and their causes of death: a report from the DHHS Hypertensive Care Computing Project (DHCCP). J Hypertens 1986;4: Gioconi S, Levanti C, Fommei E, et al: Microalbuminuria and casual and ambulatory blood pressure monitoring in normotensives and in patients with borderline and mild essential hypertension. Am J Hypertens 1989; 2: Haffner SM, Stern MP, Gruber MKK, et al: Microalbu-

7 AJH NOVEMBER 1998 VOL. 11, NO. 11, PART 1 MICROALBUMINURIA SCREENING IN HYPERTENSIVES 1327 minuria. Potential marker for increased cardiovascular risk factors in non-diabetic subjects? Arteriosclerosis 1990;10: Yudkin JS, Forrest RD, Jackson CA: Microalbuminuria as predictor of vascular disease in non-diabetic subjects. Lancet 1988;2: Gerber LM, Shmukler, C, Alderman MH: Differences in urinary albumin excretion rate between normotensive and hypertensive, white and nonwhite subjects. Arch Intern Med 1992;152: Marshall SM: Screening for microalbuminuria: which measurement? Diabet Med 1991;8: Marshall SM, MacLeod JM: Urinary albumin excretion. Timed urine collections advisable. Br Med J 1993;306: Alderman MH, Ooi WL, Madhavan S, Cohen H: Treatment-induced blood pressure reduction and the risk of myocardial infarction. JAMA 1989;262: Madhavan S, Ooi WL, Cohen H, Alderman MH: Relation of pulse pressure and blood pressure reduction to the incidence of myocardial infarction. Hypertension 1994;23: Spooren PFMJ, Lekkerkerker JFF, Vermes I. Micral- Test : a qualitative dipstick test for micro-albuminuria. Diabetes Res Clin Pract 1992;18: Agewall S, Wikstrand J, Ljungman S, et al on behalf of the Risk Factor Intervention Study Group: Does microalbuminuria predict cardiovascular events in nondiabetic men with treated hypertension? Am J Hypertens 1995;8: Summerson JH, Bell RA, Konen JC: Racial differences in the prevalence of microalbuminuria in hypertension. Am J Kidney Dis 1995;26: Olinic N, Vida-Simiti L, Cristea A, Muresan A. Microalbuminuria in hypertensive patients. Rom J Intern Med 1994;32: Bigazzi R, Bianchi S, Campese VM, Baldari G: Prevalence of microalbuminuria in a large population of patients with mild to moderate essential hypertension. Nephron 1992;61: Ward KM, Rao MB: Clinical utility of screening for microalbumin with Micral (abst). Clin Chem 1993;39(6): Ward KM, Bossetti BM, Cataland S, et al: Screening for microalbuminuria with Micral : 24-hour vs. random urine samples. Diabetes 1995;4(suppl 1):121A. 22. Hasslacher C: Clinical significance of microalbuminuria and evaluation of the. Clin Biochem 1993;26: Hermans MP, Selvais P, van Ypersele de Strihou M, Ketelslegers J-M: Detection of low-range diabetic microalbuminuria: revisited. Diabet Med 1994;11: Phillipou G:. A new semiquantitative test for urinary albumin. Diabetes Care 1993;16: Tiu SC, Lee SS, Cheng MW: Comparison of six commercial techniques in the measurement of microalbuminuria in diabetic patients. Diabetes Care 1993;16: Marshall SM, Shearing PA, Alberti KGMM: strips evaluated for screening for albuminuria. Clin Chem 1992;38: Minetti EE, Cozzi MG, Granata S, Guidi E: Accuracy of the urinary albumin titrator stick in kidney-disease patients. Nephrol Dial Transplant 1997;12: