Urianalisis | Statistical Hypothesis Testing | Meta Analysis - Scribd

May 9, 2018 | Author: Anonymous | Category: Documents
Report this link


Description

REVIEW

Accuracy of Urinalysis Dipstick Techniques in Predicting Significant Proteinuria in Pregnancy  Jason J. S. Waugh, MRCOG, T. Justin Clark, MD, MRCOG, T. G. Divakaran, MRCOG, Khalid S. Khan, MSc, MRCOG, and Mark D. Kilby, MD, MRCOG OBJECTIVE: To estimate the accuracy of point-of-care dip-

stick uri stick urinal nalysi ysiss in pred predict icting ing sig signifi nificant cant pro protei teinuri nuriaa in pregnancy. DATA SOURCES: Literature from 1970 to February 2002 was identified via 1) general bibliographic databases, that is, MEDLINE and EMBASE, 2) Cochrane Library and rele vant specialist register re gister of the Cochrane Collaboration, and 3) checking the reference lists of known primary and re view articles. METHODS OF STUDY SELECTION: Studies were selected if the accuracyy of dipsti accurac dipstick ck urinal urinalysis ysis techniques in predict predicting  ing  total protein excretion was estimated compared with a reference standard (laboratory estimation of protein excretion). The tests included visually read color-change dipsticks and autom automated ated dipsti dipstick ck urinal urinalysis. ysis. Study selecti selection, on, quality quali ty assessm assessment, ent, and data abstraction abstraction were perform performed ed independently and in duplicate. TABULATION, TABULATI ON, INTEGRA INTEGRATION, TION, AND RESULTS: Data from se-

lected stud lected studies ies were abs abstra tracted cted as 2  2 tab tables les com compari paring ng the test result with the reference standard. Test accuracy was expressed as likelihood ratios. Summary likelihood ratios  were generated as measures of diagnostic accuracy to determine posttest probabilities. The electronic search produced 1,543 citations. After independent review of published articles, a total of 34 articles was obtained for further  scrutiny, and 7 studies were considered eligible for inclusion in the review. The 6 studies evaluating visual dipstick urinalysis produced a pooled positive likelihood ratio of  3.48 (95% con confide fidence nce int interva ervall 1.6 1.66, 6, 7.27 7.27)) and a poo pooled led negative likelihood ratio of 0.6 (95% confidence interval 0.45, 0.8) for predicting 300 mg/24-hour proteinuria at the 1 or greater threshold. CONCLUSION: The accuracy of dipstick urinalysis with a 1  thresho thre shold ld in the pred predict iction ion of sig signifi nificant cant pro protei teinur nuria ia is poo poor  r  and therefore of limited usefulness to the clinician. Accuracy may be improved at higher thresholds (greater than 1 proteinuria), but available data are sparse and of poor  From the Department of Obstetrics and Gynecology, Leicester Warwick Medical  School, University of Leicester, Leicester, United Kingdom; Division of Reproduc-  tion and Child Health, Birmingham Women’s Hospital, Birmingham, United  Kingdom; and the Department of Fetal Medicine, Division of Reproduction and  Child Health, Birmingham Women’s Hospital, Birmingham, United Kingdom. VOL. 103, NO. 4, APRIL 2004 © 2004 by The American College of Obstetricians and Gynecologists. Published by Lippincott Williams & Wilkins.

methodolo method ologica gicall qua qualit lity. y. The Therefo refore, re, it is not poss possibl iblee to makee mea mak meanin ningful gful inf inferen erences ces abo about ut accu accurac racyy at hig higher  her  urine uri ne dip dipsti stick ck thr thresho esholds lds.. Ther Theree is an urge urgent nt need for  resea re searc rch h in th this is ar area ea of co comm mmon on ob obst stet etri ricc pr prac acti tice. ce. (Obstet Gyneco Gynecoll 2004;103: 2004;103:769 769–77. –77. © 2004 2004 by The AmerAmerican College of Obstetricians and Gynecologists.)

More than 20% of pregnancies will have at least 1 blood pressure recording of 140/90 mm Hg or greater after 20 weeks of gestation, and in half of these cases such a reading will lead to some form of intervention. 1 However, the incidence of proteinuric hypertension, or preeclampsia, is lower, between 2% and 4% of pregnancies.2,3 Pr Prot otei einur nuria ia is req requi uired red in th thee defi defini niti tion on of  2– 4 preeclampsia, althou although gh 2 recent reviews of the litera litera-ture found that its presence was a requirement in only 80% and 91% of reported articles, respectively. 5,6 However, its presence is an indepe independent ndent marker of maternal morbidity and perinatal morbidity and mortality. Dipsticks were used to confirm the presence of proteinuria (and hence preeclampsia) in 35% of studies and were the only measure of proteinuria in 21% of studies. 6  This method of detecting proteinuria has been demonstrated to vary in its positive and negative predictive value in the ide identi ntifica ficatio tion n of pree preecla clamps mpsia. ia.7–9  Therefore, the evaluation of point-of-care urine screening remains a subject of a continuing debate. 10 Furthermore, advances in technology have led to the introduction of automated semiquantit semiq uantitative ative dipsti dipstick ck urinal urinalysis. ysis.11 Sample sizes of  individual studies on this subject are small, leading to variab var iable le and imp imprec recise ise est estima imates tes of tes testt acc accurac uracy. y. A quantitativ quanti tativee system systematic atic review is necessa necessary ry to obtai obtain n moree pre mor precis cisee est estima imates tes and so est estima imate te the value of  point-of-c point -of-care are dipsti dipstick ck urinal urinalysis ysis in the predic prediction tion of significant proteinuria. SOURCES

 Two electronic bibliographic databases, MEDLINE and EMBASE, were searched from January 1970 to February 200 2002. 2. The MeSH (Medical (Medical sub subhead heading ings) s) ter terms ms

0029-7844/04/$30.00 doi:10.1097/01.AOG.0000118311.18958.63

769

“pregnancy,” “urine,” “proteinuria,” and “preeclampsia” were combined with the MeSH terms “urinalysis,” “dipstick,” and “near patient test” or “point-of-care test.” In addition, specialist computer databases—the Cochrane Library and relevant specialist registers of the Cochrane Collaboration—were consulted. Recent issues of the specialist journals, British Journal of Obstetrics and Gynaecology ,  American Journal of Obstetrics and Gynecology , and Obstetrics &  Gynecology , were searched manually, as were the reference lists of all known primary and review articles. An attempt to identify unpublished data was made through contact with individual experts or groups with an interest in the field, as well as contacting manufacturers of pointof-care urinalysis equipment. STUDY SELECTION

 The review focused on prospective observational studies or comparative cross-sectional studies in which the results of the diagnostic test of interest were compared with the results of a “reference standard.” The population of  interest was pregnant women. This included uncomplicated pregnancies, women with hypertension, and pregnancies complicated by preexisting renal disease. The diagnostic intervention was a point-of-care test for urine protein. The diagnostic reference standard was a laboratory assay for urine protein preferably from a 24-hour urine sample.  The studies identified were reviewed independently  by 2 of the authors (J.W. and T.G.D.). Identification of  potentially relevant studies was performed by scanning  the titles and abstracts obtained from the computer database searches or bibliography inspections. The full texts of these articles were retrieved, and final inclusion/ exclusion decisions were made on the basis of the information contained in these with reference to a checklist, the items of which were based on the selection criteria above. Disagreements about the inclusion/exclusion were initially resolved by consensus, and where this was not possible, it was resolved by using arbitration by a third reviewer (M.D.K.).  All articles meeting the eligibility criteria were rated for their methodological quality. We defined this as the confidence that the study design, conduct, and analysis minimized bias in estimation of diagnostic accuracy. 12–15 Based on the existing checklists the following features were targeted: Population: Prospective recruitment of the study population was considered adequate, whereas convenience sampling was deemed inadequate. In the absence of  any explicit information in the articles, these population details were considered reported unclearly. Population

770

Waugh et al

Accuracy of Dipstick Urinalysis

details were considered adequate if the population sampled contained pregnant women with proteinuria whether or not this variable was found in association with hypertension.  Test: The description of the test was considered adequate if  it was reported in enough detail to allow reproduction by other researches. Blinding was considered adequate if  the result of the diagnostic test under scrutiny was not known by the clinician performing the reference test (laboratory estimation of proteinuria). If the clinician performing the reference test was aware of the diagnostic test result or in the absence of such reporting, blinding  was considered inadequate or unclearly reported. Outcome: Proteinuria estimation from a 24-hour urine sample was considered adequate for use as a reference standard.The cutoff for significant proteinuria was taken as 300 mg/24 h. This is the level used in definitions of  preeclampsia. 2–6 Study quality was assessed by using a previously published hierarchy of evidence.16  A checklist was used to obtain and record the information.These assessments were performed independently and in duplicate (J.W. and  T.G.D.) and any disagreement resolved by consensus.  The results from each article were abstracted as a 2  2 table comparing results from the test (positive or negative) against those of the reference standard. This allowed the calculation of the true-positive rate (sensitivity), false-positive rate (1  specificity), likelihood ratios for both positive and negative tests, and pretest and posttest probability for each study.  An estimate of the pretest probability was obtained by calculating the prevalence of the outcome event in the individual studies. We then examined the implications of  the likelihood ratios generated for the different pretest probabilities by using Bayes’ theorem to generate posttest probabilities. The following equation was used to calculate posttest probability: posttest probability  likelihood ratio  pretest probability/[1  pretest probability  (1  likelihood ratio)]. Ranges of posttest proba bility were calculated by using lower and upper limits of  95% confidence intervals (CIs) of pretest probabilities and likelihood ratio. Summary likelihood ratios were generated as measures of diagnostic accuracy to determine posttest probabilities. The likelihood ratio represents the probability of a positive (or negative) test result in women with proteinuria ( 300 mg/24 h) to the probability of the same test result in those women without proteinuria. A correction factor of 0.5 was used when the data or a study included a zero value to allow calculation of the likelihood ratio and its CI. 15  The likelihood ratios indicate by how much a given test dipstick test finding raises

OBSTETRICS & GYNECOLOGY

or lowers the probability of having proteinuria. This is important in clinical decision making because the estimated probability of disease (or not having disease) is a prime factor determining whether to withhold treatment, undertake further diagnostic testing, or treat without further testing. For a positive result, a likelihood ratio of  more than 1 increases the probability that proteinuria will be present. The greater the likelihood ratio, the larger the increase in probability of the adverse event and the more clinically useful the test result. For a negative test result, a likelihood ratio of less than 1 decreases the probability that proteinuria is present: The smaller the likelihood ratio, the larger the decrease and the more clinically useful the test result. It is generally considered that a likelihood ratio of more than 10 for a positive test or less than 0.1 for a negative test results in conclusive changes in prior probability and is thus use-

ful in informing clinical decision making. 17 Thus, the generation of likelihood ratios and posttest probabilities represents a more relevant method of establishing the utility of a test. Pooling of likelihood ratios was performed where possible by weighting the log likelihood ratio from each study in inverse proportion to its variance.  An exploration for publication bias (preferential reporting of studies with positive or statistically significant results) was performed by producing a funnel plot, which is a scatter plot of individual study accuracy against corresponding precision (inverse of variance).  The log of diagnostic odds ratio was used as the accuracy measure because it accommodates likelihood ratios for  both positive and negative test results. When no publication bias is present, the plots will be shaped like a funnel because studies of smaller size are expected to have increased variation in the estimates of accuracy.

Figure 1. Study selection process.  Taylor AA, Davison JM. Albumin excretion in normal pregnancy [letter]. Am J Obstet Gynecol 1997;177:1159 – 60. † Waugh J, Kilby M, Bell S, Seed P, Shennan A, Halligan A. Bedside urine albumin/creatinine ratio testing in hypertensive pregnancy [abstract]. Hypertens Pregnancy 2002;21. Waugh. Accuracy of Dipstick Urinalysis. Obstet Gynecol 2004.

VOL. 103, NO. 4, APRIL 2004

Waugh et al

Accuracy of Dipstick Urinalysis

771

Table 1. Diagnostic Accuracy of Dipstick Urinalysis in Detecting Significant Proteinuria: Studies Selected for Systematic Review Population Study (year published)

Study design

Visual dipstick proteinuria



Patient selection

Test Pregnancy status (n )

Description of Reference technique standard

Consecutive Hypertensive (230)

Adequate

Unknown mix (690)  Adequate hypertensive/ uncomplicated 8 Meyer et al (1994) Retrospective Unreported Hypertensive (300) Adequate

No

Paruk et al 26 (1997) Prospective

Unreported Hypertensive (150)

Adequate

 Waugh et al44 (2001)  Waugh et al46 (2002)

Prospective

Consecutive Hypertensive (197)

Adequate

Prospective

Consecutive Hypertensive (171)

Adequate

Saudan et al11 (1997)  Waugh et al44 (2001)



Reference standard cutoff

Blinding of results Follow-up

1 visual dipstick vs total protein excretion (300 mg/24 h)

Brown and Buddle 9 Prospective (1995) Higby et al 39 (1995) Prospective

Visual dipstick proteinuria

Outcome

24-h urine protein 24-h urine protein

300 mg/24h Yes

 90%

300 mg/24h Yes

 90%

24-h urine protein 24-h urine protein 24-h urine protein 24-h urine protein

300 mg/24h No

 90%

300 mg/24 h Unreported

 90%

300 mg/24h Yes

 90%

300 mg/24h Yes

 90%

24-h urine protein 24-h urine protein

300 mg/L

Yes

 90%

300 mg/L

Yes

 90%

1, 2 , 3  visual dipstick vs protein concentration (300 mg/L)

Prospective

Consecutive Hypertensive (103)

Adequate

Prospective

Consecutive Hypertensive (197)

Adequate

 Automated dipstick proteinuria  1, 2, 3 automated dipstick vs protein concentration (300 mg/L) and/or total protein excretion (300 mg/24 h) Saudan et al11 (1997)  Waugh et al*

Prospective

Consecutive Hypertensive (103)

Adequate

Prospective

Consecutive Hypertensive (171)

Adequate

24-h urine protein 24-h urine protein

Yes

 90%

300 mg/24h Yes

 90%

300 mg/L

* Waugh et al. Hypertens Pregnancy 2002;21.

Table 2. Accuracy of Urine Dipstick at 1 Threshold in Predicting Significant Proteinuria in Pregnant Hypertensive Women Using a Reference Standard Cutoff of 300 mg/24 h Level of urine dipstick and study

1 (visual) Brown and Buddle9 Higby et al39 Meyer et al8 Paruk et al26*  Waugh et al44  Waugh et al46  All studies (6) (Pooled)  1 (automated)  Waugh et al†

Prevalence  300 mg/24 h (pretest probability) (%)

Dipstick proteinuria-positive test (sensitivity)

Dipstick proteinuria-negative test (specificity)

53 7 81.0 5 70 45.0 39 (37, 42)

60/70 (0.86) 24/51 (0.47) 162/243 (0.67) 59/101 (0.58) 31/138 (0.22) 39/77 (0.51) 375/680 (0.55) (0.37, 0.72)

62/160 (0.39) 633/639 (0.99) 42/57 (0.74) 24/49 (0.49) 58/59 (0.98) 73/94 (0.78) 892/1,058 (0.84) (0.57, 0.95)

63/77 (0.82)

76/94 (0.81)



45.0

 An estimate of the pretest probability was obtained by calculating the prevalence of the outcome event in the population studied. The following  equation was used for calculating posttest probability: posttest probability  likelihood ratio  pretest probability/ 1–pretest probability  (1–likelihood ratio) . Ranges of posttest probability were calculated by using lower and upper limits of 95% confidence intervals of pretest probabilities and likelihood ratios. * Data estimated from sensitivity and specificity given in Paruk study. †  Waugh et al. Hypertens Pregnancy 2002;21. 772

Waugh et al

Accuracy of Dipstick Urinalysis

OBSTETRICS & GYNECOLOGY

 The adjusted rank correlation method was used to test the correlation between estimated diagnostic odds ratio and their inverse variance.  We also plotted the estimates of true positive rates against false positive rates from the included studies to develop a summary receiver operating characteristic curve that characterizes the performance of the test as found in the multiple studies.18  A receiver operating  characteristic area greater than 0.5 suggests some degree of test accuracy, with higher accuracy suggested by a receiver operating characteristic area closer to 1.0 (representing perfect test accuracy). Meta-analysis was not possible for all studies because of variation in study methods and differences in diagnostic tests cutoff levels. RESULTS

 The combined MEDLINE and EMBASE search produced 1,543 citations and, after initial screening, there were 34 studies both reviewers thought were potentially relevant. After independent review of these articles, 7 studies were eligible for inclusion in the review (Figure 1). There were 1,841 pregnant women’s urine samples that were reported in these 7 primary studies.  The methodological details and quality criteria of the selected studies are shown in Table 1. The population studied was deemed adequate in 6 of 7 studies, and patient selection was reported and felt to be accurate in 5 studies. The description of the urinalysis technique was adequate in all studies; however, masking of test results was only adequately reported in 5 studies. Follow-up was more than 90% for all studies selected. As such, 3

studies (Waugh et al. Hypertens Pregnancy 2002;21) 9,44 were classified as level 1 for quality, and 1 study was level 2.39 One study was level 3,26 1 level 4,8 and 1 level 5.11 Six studies (Waugh et al. Hypertens Pregnancy 2002;21)8,9,26,39,44 reported the use of a 24-hour urine protein quantification as the reference standard whereas one11 reported protein concentration.  All studies used the same dipsticks (Multistix; Bayer Corporation, Elkhart, IN) for visual testing except  Waugh et al,44 who used a Boehringer dipstick (Boehringer-Mannheim, Mannheim, Germany). However,  because the thresholds on the protein pads on these 2 dipsticks are the same (1 is equivalent to 30 mg/dL), the pooling of data from both types of dipstick was considered to be appropriate. Some of the studies reported on the use of more than 1 urinalysis technique for the estimation of proteinuria (Table 1). Brown and Buddle9 studied the effect of timing of the random urine void in relation to the 24-hour urine sample. For the purpose of the systematic review, the results from the random void that preceded the 24-hour urine sample are included in the pooled data for consistency with other included studies, which all tested random urine samples that preceded the 24-hour urine collection. Higby et al 39 studied the Multistix 10SG and a dipstick for microalbumin. Although it was possible to extract data for the Multistix 10SG, it was not possible to compare the microalbumin dipstick to the 24-hour urine collection  because a 2  2 table could not be constructed. Waugh et al44 compared dipstick accuracy when different reference standard assays were used and also reported on the

Posttest probability (%) Likelihood ratio positive (95% confidence interval)

Likelihood ratio negative (95% confidence interval)

Test positive (95% confidence interval)

Test negative (95% confidence interval)

0.54 (0.45, 0.66) 50.12 (21.47, 117.02) 2.53 (1.63, 3.95) 1.14 (0.83, 1.58) 13.25 (1.85, 94.84) 2.27 (1.47, 3.51) 3.48 (1.66, 7.27)

5.49 (2.97, 10.16) 0.53 (0.40, 0.69) 0.45 (0.36, 0.57) 0.85 (0.59, 1.23) 0.79 (0.72, 0.87) 0.64 (0.49, 0.82) 0.6 (0.45, 0.8)

38 80 92 70 97 65 72 (53, 86)

86 4 66 64 65 34 30 (23, 40)

0.22 (0.14, 0.36)

77.7

15.6

4.27 (2.78, 6.56)

VOL. 103, NO. 4, APRIL 2004

Waugh et al

Accuracy of Dipstick Urinalysis

773

Table 3. Accuracy of Urine Dipstick at Various Thresholds in Predicting Significant Proteinuria in Pregnant Hypertensive Women Using a Reference Standard Cutoff of 300 mg/L Level of urine dipstick (no. of Prevalence  300 mg/24 h Dipstick proteinuria Dipstick proteinuria evaluations) and study (year published) (pretest probability) (%) negative test (sensitivity) positive test (specificity)

1 (visual) Saudan et al11 (1997)*  Waugh et al44 (2001) Both studies (2) (Pooled)  2 (visual) Saudan et al11 (1997)*  3 (visual) Saudan et al11 (1997)*  1 (automated) Saudan et al11 (1997)*  2 (automated) Saudan et al11 (1997)*  3 (automated) Saudan et al11 (1997)* 

65.0 54.3 33.7† (31.5, 36.0)

67/67 (1.0) 31/107 (0.29) 98/174

14/36 (0.38) 1/90 (0.01) 23/126

65.0

67/67 (1.0)

5/36 (0.14)

65.0

67/67 (1.0)

1/36 (0.02)

65.0

60/67 (0.9)

5/36 (0.14)

65.0

56/67 (0.83)

1/36 (0.02)

65.0%

62/67 (0.93)

0/36 (1.0)

 An estimate of the pretest probability was obtained by calculating the prevalence of the outcome event in the population studied. The following  equation was used for calculating posttest probability: posttest probability  likelihood ratio  pretest probability/ 1–pretest probability  (1–likelihood ratio) . Ranges of posttest probability were calculated by using lower and upper limits of 95% confidence intervals of pretest probabilities and likelihood ratios. * Data estimated from sensitivity and specificity given in Saudan studies. † Prevalence derived from Table 1 for comparison (pooled urine dipstick studies  1 protein with 300 mg/24 h reference standard.

difference in accuracy when dipsticks are compared with either total protein excretion in 24 hours or protein concentration.  Two studies11,44 reported on the predictive values of  urine dipstick testing with varying thresholds for urinalysis.9,11  The reference standard protein threshold used to define significant proteinuria (300 mg/24 hours) was the same in 6 of the 7 included studies, thereby allowing data to  be pooled (Waugh et al. Hypertens Pregnancy 2002;21) (Table 2).8,9,26,39,44 In the remaining study, protein concentration was used as a reference standard, 300 mg/L. One study44 presented data with both total 24-hour protein excretion and protein concentration. In the evaluation of  visual dipstick urinalysis, the most commonly used threshold to predict significant proteinuria (300 mg/24 h) was 1 or greater. These 6 studies produced a pooled positive likelihood ratio of 3.48 (95% CI 1.66, 7.27) and a negative likelihood ratio of 0.6 (95% CI 0.45, 0.8) (Table 3). The receiving operator characteristic area was 0.7. Heterogeneity of test performance was confirmed across all studies (P   .001). Univariate subgroup analyses stratified for items of study quality did not provide an explanation for the observed variation in diagnostic performance. Restricting meta-analysis to the 2 studies of  highest methodological quality (Waugh et al. Hypertens Pregnancy 2002;21)9 generated pooled likelihood ratios of 2.26 (95% CI 1.01, 5.05) and 0.01 (95% CI 0.38, 0.97) for positive and negative tests, respectively. Using these

774

Waugh et al

Accuracy of Dipstick Urinalysis

accuracy estimates changed the pretest probability of  significant proteinuria from 39% (95% CI 37%, 42%) to 59% (95% CI 37%, 79%) for a positive test and 28% (95% CI 18%, 41%) for a negative test result.  Two studies looked at the effect of automation (Waugh et al. Hypertens Pregnancy 2002;21). 11 However, these 2 studies used different reference standards so that data could not be pooled. The study of highest quality (Waugh et al. Hypertens Pregnancy 2002;21) evaluated both automated and visual techniques in a  blinded comparison. Estimates of accuracy were better for both positive and negative test results for automated compared with visual urinalysis (positive likelihood ratio of 4.27 [95% CI 2.78, 6.56]) and a negative likelihood ratio of 0.22 [95% CI 0.14, 0.36] for automated urinalysis compared with a positive likelihood ratio of 2.27 [95% CI 1.47, 3.51] and a negative likelihood ratio of 0.64 [95% CI 0.49, 0.82] for visual urinalysis).  A funnel plot (not shown) did not show asymmetry, and statistical tests (rank correlation) to explore for publication and related biases, found that funnel plot asymmetry was not statistically significant ( P   .4). Publication bias is therefore unlikely to be a problem. CONCLUSION

 The use of dipsticks to screen urine for protein is an integral part of current antenatal care strategies. Our

OBSTETRICS & GYNECOLOGY

Posttest probability (%) Likelihood ratio positive (95% confidence interval)

Likelihood ratio negative (95% confidence interval)

Test positive (95% confidence interval)

Test positive (95% confidence interval)

2.57 (1.71, 3.87) 26.07 (3.63, 187.26) 2.53 (1.86, 3.44)

0.01 (0.00, 0.19) 0.72 (0.64, 0.81) 0.55 (0.48, 0.64)

82.7 96.9 56.3 (46.1, 65.9)

0.0 46.1 21.9 (18.1, 26.5)

7.20 (3.19, 16.24)

0.01 (0.00, 0.14)

93.1

0.0

36.00 (5.21, 248.66)

0.01 (0.00, 0.12)

98.5

0.0

6.45 (2.85, 14.60)

0.12 (0.06, 0.25)

92.3

18.4

30.09 (4.34, 208.45)

0.17 (0.10, 0.29)

98.2

23.9

68.01 (4.33, 1068.16)

0.07 (0.03, 0.17)

100.0

12.23

review of the literature has demonstrated that the accuracy of dipstick urinalysis with a 1  threshold in the prediction of significant proteinuria is poor. Neither a positive nor negative dipstick result (1  threshold) substantially raises or lowers the probability of having clinically significant proteinuria, thereby limiting its usefulness in informing clinical decision making. Accuracy may be improved at higher thresholds (greater than 1  proteinuria), but available data are sparse and of poor methodological quality. Therefore, it is not possible to make meaningful inferences about accuracy at higher urine dipstick thresholds.  The validity of our conclusion depends on the strength of methods used in our review. We strictly complied with the criteria for a systematic review of  diagnostic tests.13 We used a prospective protocol, posed a clear research question, and performed a comprehensive literature search. The inclusion and exclusion criteria were set a priori. Information on study characteristics and methodology was extracted in duplicate to minimize errors. Test accuracy was measured by using clinically meaningful measures and possible sources of heterogeneity explored. Heterogeneity relates to the presence of  differences in results between individual studies. Homogeneity of results from studyto study is one of the criteria for meta-analysis, but presence of inconsistency itself  does not always invalidate a meta-analysis. In this situation, it is important to consider possible reasons for heterogeneity and then try to explain it. Exploration for sources of heterogeneity was performed by taking into account differences in methodological quality and study characteristics, by using univariable analytic techniques.

VOL. 103, NO. 4, APRIL 2004

However, this approach did not explain the observed variation. However, these exploratory analyses are restricted by the small number of available studies. Only 3 of the 7 primary studies that were included satisfied criteria for high methodological quality. Restricting  pooled data to 3 higher quality studies did not materially alter the results, although it did reduce the estimates of  accuracy slightly. Our inferences that accuracy of dipstick testing at the 1  level for significant proteinuria is poor is unchanged whether we base inferences on the overall pooled results or restrict inferences to the 3  best-quality studies. Thus, in view of the lack of satisfactory explanations for heterogeneity between studies we  believe it to be reasonable to base inferences on the overall pooled results. Only one of the studies included in this systematic review studied an unselected obstetric population 39 with a low prevalence of proteinuria (7%). The remaining  studies restricted the obstetric population of interest to hypertensive women, in whom the prevalence of proteinuria varied, but was generally higher (pooled prevalence 39%). However, the ability of urine dipstick to predict clinically significant proteinuria was poor regardless of disease prevalence. Dipstick manufacturers claim that technology has improved in the past 20 years, and the introduction of  automated dipstick readers has been encouraged but without supportive evidence as demonstrated in this review. Several authors have previously discussed the limited usefulness of urine dipsticks as diagnostic tests for proteinuria (Waugh et al. Hypertens Pregnancy 2002;21),5–11 in addition to practical drawbacks such as

Waugh et al

Accuracy of Dipstick Urinalysis

775

the effect of variation in urine concentration on the accuracy of dipstick urinalysis. 44 Consequently, some authorities have recommended the mandatory use of  24-hour urine protein estimations for women with hypertension in pregnancy.10 However, 24-hour urine collections are often incomplete and as such may have hidden errors that clinicians will be unaware of. Saudan et al11 and others have suggested that use of protein-tocreatinine ratios as a method of quantifying proteinuria rather than cumbersome 24-hour urinary protein measurement. Protein-to-creatinine ratio determination is at present a laboratory-based investigation and so its potential use as a screening test for proteinuria in pregnancy may be limited compared with point-of-care tests such as urinary dipsticks. However, it is now also possible to test for creatinine at point of care with a dipstick pad and this remains to be assessed in combination with protein testing to see if it can improve the dipstick performance with a point of care protein/creatinine ratio. Preeclampsia or proteinuric hypertension is a common complication of pregnancy 2,3 with the potential for serious complications for both mother and baby. Optimal management depends on accurate and timely diagnosis. Our review has shown that significant proteinuria, with point-of-care urine dipstick analysis, cannot be accurately detected or excluded at the 1 threshold and is not recommended for diagnosing preeclampsia. Further research is necessary to determine the prediction of  proteinuria using higher dipstick thresholds.

REFERENCES

1. National High Blood Pressure Education Program Working Group Report on High Blood Pressure in Pregnancy [review]. Am J Obstet Gynecol 1990;163:1691–712 2. Redman CW, Jefferies M. Revised definition of preeclampsia. Lancet 1988;1:809–12. 3. Perry IJ, Beevers DG. The definition of pre-eclampsia. Br J Obstet Gynaecol 1994;101:587–91. 4. Brown MA, Lindheimer MD, de Swiet M, Van Assche A, Moutquin JM. The classification and diagnosis of the hypertensive disorders of pregnancy: statement from the International Society for the Study of Hypertension in Pregnancy (ISSHP) [review]. Hypertens Pregnancy 2001; 20:IX–XIV. 5. Chappell L, Poulton L, Halligan A, Shennan AH. Lack of consistency in research papers over the definition of  pre-eclampsia [review]. Br J Obstet Gynaecol 1999;106: 983–5. 6. Harlow FH, Brown MA. The diversity of diagnoses of  preeclampsia [review]. Hypertens Pregnancy 2001;20: 57–67.

776

Waugh et al

Accuracy of Dipstick Urinalysis

7. Kuo VS, Koumantakis G, Gallery ED. Proteinuria and its assessment in normal and hypertensive pregnancy. Am J Obstet Gynecol 1992;167:723– 8. 8. Meyer NL, Mercer BM, Friedman SA, Sibai BM. Urinary dipstick protein: a poor predictor of absent or severe proteinuria. Am J Obstet Gynecol 1994;170:137–41. 9. Brown MA, Buddle ML. Inadequacy of dipstick proteinuria in hypertensive pregnancy. Aust N Z J Obstet Gynaecol 1995;35:366 –9. 10. Halligan AW, Bell SC, Taylor DJ. Dipstick proteinuria: caveat emptor. Br J Obstet Gynaecol 1999;106:1113–5. 11. Saudan PJ, Brown MA, Farrell T, Shaw L. Improved methods of assessing protienuria in hypertensive pregnancy. Br J Obstet Gynaecol 1997;104:1159–64. 12. Dunn G, Everitt B. Clinical biostatistics: an introduction to evidence based medicine. London, England: Edward  Arnold; 1995. 13. Khan K, ter Riet G, Popay J, Nixon J, Kleijnen J. Study quality assessment. In: Khan KS, ter Riet G, Glanville J, Snowden AJ, Kleijnen J, editors. Undertaking systematic reviews of research on effectiveness. CRD’s guidance for carrying out or commissioning reviews. 2nd ed. CRD Report No. 4. York (UK): NHS Centre for Reviews and Dissemination (CRD), University of York; 2001. 14. Khan KS, Daya S, Jahad A. The importance of quality of  primary studies in producing unbiased systematic reviews.  Arch Intern Med 1996;156:661– 6. 15. Sankey SS, Weissfeld LA, Fine MJ, Kapoor W. An assessment of the use of continuity correction for sparse data in meta-analysis. Commun Stat 1996;25:1031–56. 16. Clark TJ, Mann CH, Shah N, SongF, Khan KS, Gupta JK.  Accuracy of outpatient endometrial biopsy in the diagnosis of endometrial cancer [review]. BJOG 2002;109:313–21. 17. Jaeschke R, Guyatt G, Sackett DL. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? The Evidence-Based Medicine Working Group. JAMA 1994; 271:389–91. 18. Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data analytical approaches and some additional considerations. Stat Med 1993;12:1293–316. 19. Young R, Buchanan RJ, Kinch RAH. Use of the protein/ creatinine ratio of a single voided urine specimen in the evaluation of suspected pregnancy induced hypertension [Erratum in J Fam Pract 1997;44:21]. J Fam Pract 1996;42: 385–9. 20. Robert M, Sepandj F, Liston RM, Dooley KC. Random protein-creatinine ratio for the quantitation of proteinuria in pregnancy. Obstet Gynecol 1997;90:893–5. 21. Konstantin-Hansen KF, Hesseldahl H, Pedersen SM. Microalbuminuria as a predictor of pre-eclampsia. Acta Obstet Gynecol Scand 1992;71:343– 6.

OBSTETRICS & GYNECOLOGY

22. Gribble RK, Fee SC, Berg RL. The value of routine dipstick screening for protein at each pre-natal visit. Am J Obstet Gynecol 1995;173:214–7. 23. North RA, Taylor RS, Schellenberg JC. Evaluation of a definition of pre-eclampsia. Br J Obstet Gynaecol 1999; 106:767–73. 24. Bell SC, Halligan AW, Martin A, Ashmore J, Shennan  AH, Lambert PC, et al. The role of observer error in dipstick proteinuria. Br J Obstet Gynaecol 1999;106: 1177–80. 25. Gaucherand P, Salle B, Sergeant P, Guibaud S, Rudigoz RC. Micro-albuminuria analysis and pregnancy: an approach to detect placental insufficiency? Eur J Obstet Gynecol Reprod Biol 1996;119:70:49 –52. 26. Paruk F, Moodley J, Daya PKS, Meineke K. Screening for proteinuria in hypertensive disorders of pregnancy. J Obstet Gynaecol 1997;17:528–30. 27. Lindow SW, Davey DA. The variability of urinary protein and creatinine excretion in patients with gestational proteinuric hypertension. Br J Obstet Gynaecol 1992;99: 869–72. 28. Brown MA, Wang MX, Buddle ML, Carlton MA, Cario GM, Zammit VC, et al. Albumin excretory rate in normal and hypertensive pregnancy. Clin Sci 1994;86:251–5. 29. Uttendorfsky OT, Veersema D, Mooij PN, Stolte LA. Protein/creatinine ratio in the assessment of proteinuria during pregnancy. Eur J Obs Gynecol Reprod Biol 1988; 27:221–6. 30. Holm J, Hemmingsen L, Irgens-Moller L. Diagnostic value of hyperuricemia and microalbuminuria in preeclampsia. Med Sci Res 1988;16:123–4. 31. Zuijderhoudt FM, Kluitenberg WE, Satink JG, Huikeshoven FJ. Measuring low-grade albuminuria in pregnancy. Br J Obstet Gynaecol 1989;96:619 –20. 32. Nakamura T, Ito M, Yoshimura T, Mabe K, Okamura H. Usefulness of the urinary microalbumin/creatinine ratio in predicting pregnancy-induced hypertension. Int J Gynecol Obstet 1992;37:99–103. 33. Thong KJ, Howie AF, Smith AF, Greer IA, Johnstone FD. Microalbuminuria in random daytime specimens in pregnancy-induced hypertension. J Obstet Gynaecol 1991;11: 324–7. 34. Irgens-Moller L, Hemmingsen L, Holm J. Diagnostic value of microalbuminuria in pre-eclampsia. Clin Chem Acta 1986;157:295–8. 35. Rodriguez MH, Masaki DI, Mestman J, Kumar D, Rude

VOL. 103, NO. 4, APRIL 2004

R. Calcium/creatinine ratio and microalbuminuria in the prediction of preeclampsia. Am J Obstet Gynecol 1988; 159:1452–5. 36. Jaschevatzky OE, Rosenberg RP, Shalit A, Zonder HB, Grunstein S. Protein/creatinine ratio in random urine specimens for quantitation of proteinuria in preeclampsia. Obstet Gynecol 1990;75:604–6. 37. Ramos JG, Martins-Costa SH, Mathias MM, Guerin YL, Barros EG. Urinary protein creatinine ratio in hypertensive pregnant women. Hypertens Pregnancy 1999;18: 209–18. 38. Das V, Bhargava T, Das SK, Pandey S. Microalbuminuria: a predictor of pregnancy induced hypertension. Br J Obstet Gynaecol 1996;103:928–30. 39. Higby K, Suiter CR, Siler-Khodr T. A comparison  between two screening methods for detection of micr oproteinuria. Am J Obstet Gynecol 1995;173:1111–4. 40. Higby K, Suiter CR, Phelps JY, Siler-Khodr T, Langer O. Normal values of urinary albumin and total protein excretion during pregnancy. Am J Obstet Gynecol 1994;171: 984–9. 41. Baker PN, Hackett GA. The use of urinary albumincreatinine ratios and calcium-creatinine ratios as screening  tests for pregnancy induced hypertension. Obstet Gynecol 1994;83:745–9. 42. Quadri KH, Bernardini J, Greenberg A, Laifer S, Syed A, Holley JL. Assessment of renal function during pregnancy using random urine protein to creatinine ratio and Cockcroft-Gault formula. Am J Kidney Dis 1994;24:416–20. 43. Bell SC, Armstrong CA, Shennan AH, Boyce T, Halligan  AW. Reliable urinalysis in the management of hypertensive pregnancies. Eur J Obstet Gynaecol Reprod Biol 2000;93:181–3. 44. Waugh J, Bell SC, Kilby M, Lambert P, Shennan A, Halligan A. Effect of concentration and biochemical assay on the accuracy of urine dipsticks in hypertensive pregnancies. Hypertens Pregnancy 2001;20:205–17. 45. Goh JT, Krause H. A prospective observational study on the accuracy or patient self-testing of urine at an antenatal clinic. Aust N Z J Obstet Gynaecol 2002;42:1:67–8  Address reprint requests to: Mark Kilby, Professor in Maternal and Fetal Medicine, Division of Reproductive and Child Health, Birmingham Women’s Hospital, Edgbaston, Birmingham B15 2TG, United Kingdom; e-mail: [email protected]

Received August 12, 2003. Received in revised form December 19, 2003. Accepted January 8, 2004.

Waugh et al

Accuracy of Dipstick Urinalysis

777



Comments

Copyright © 2018 ADOC Inc.