Anda di halaman 1dari 6

| | |

Received: 3 December 2017    Revised: 23 March 2018    Accepted: 22 May 2018    First published online: 13 June 2018

DOI: 10.1002/ijgo.12536


Accuracy of 12-­hour urine collection in the diagnosis

of pre-­eclampsia

Rita M. Silva* | Sara R. Pereira | Susana Rego | Nuno Clode

Department of Obstetrics, Gynecology and

Reproductive Medicine, North Lisbon Hospital Abstract
Centre, Santa Maria University Hospital, Objective: To evaluate the accuracy of a 12-­hour urine collection to diagnose pre-­
Lisbon, Portugal
eclampsia and to determine whether time of collection influences the performance of
*Correspondence this test.
Rita M. Silva, Department of Obstetrics,
Gynecology and Reproductive Medicine, Methods: A prospective observational study was conducted in a tertiary obstetric
North Lisbon Hospital Centre, Santa Maria center in Portugal between November 1, 2015, and November 30, 2016. Pregnant
University Hospital, Lisbon, Portugal.
Email: women (≥20 weeks) admitted for observation with suspected pre-­eclampsia were eli-
gible for inclusion. Two consecutive 12-­hour urine samples were collected (07:00–
19:00 hours vs 19:00–07:00 hours). Protein levels were measured in each 12-­hour
sample, as well as in a pooled sample (07:00–07:00 hours). The diagnostic cut-­off val-
ues for pre-­eclampsia were 150 mg (12-­hour samples) and 300 mg (24-­hour sample).
Results: The study included 99 patients and diagnoses of pre-­eclampsia were con-
firmed among 42 (42%) patients. In all, 456 12-­hour urine samples were analyzed
(equivalent to 228 24-­hour samples). Qualitative analysis (pre-­eclampsia vs no pre-­
eclampsia) indicated substantial agreement between the 12-­ and 24-­hour samples
(Cohen κ 0.779). The sensitivity was 85.9% (95% confidence interval [CI] 81%–90%)
and the specificity was 91.7% (95% CI 88%–95%). No statistically significant differ-
ence was found between the two 12-­hour collections.
Conclusion: The 12-­hour test showed acceptable accuracy for detecting pre-­eclampsia,
regardless of the time of collection.

24-hour urine collection; Hypertensive disorders of pregnancy; Non-inferiority;
Pre-eclampsia; Proteinuria

1 |  INTRODUCTION equivalent by many obstetric societies.2,3 The protein-­to-­creatinine

ratio test is undoubtedly more practical and convenient to perform
Pre-­eclampsia and other hypertensive disorders of pregnancy account than the 24-­hour urine collection test but some controversy remains
for 10.3%–22.1% of all maternal deaths worldwide.1 The current diag- given the wide range of sensitivity (65%–96%), specificity (49%–
nostic criteria indicate that proteinuria should no longer be considered 100%), and clinical heterogeneity reported.4–6
an obligatory component of pre-­eclampsia ; however, quantification A systematic review and meta-­analysis by Stout et al.7 found that
of urinary protein levels is still routine practice. 12-­hour urine collection, with a 150 mg cut-­off level for proteinuria,
A 24-­hour urine collection, with a 300 mg cut-­off level for protein- performed well among women with hypertension during pregnancy.
uria, remains the gold standard to diagnose the presence of substan- However, the main limitations were the small sample sizes of the stud-
tial proteinuria. Nonetheless, alternative methods—including the urine ies included in the analysis, the heterogeneous methods used for urine
spot protein-­to-­creatinine ratio with a cut-­off of 0.3—are considered collection, and the lack of information on time of collection.

Int J Gynecol Obstet 2018; 142: 277–282 © 2018 International Federation of |  277  
Gynecology and Obstetrics
278       Silva ET AL.

The primary aim of the present study was to test the hypothesis symptoms (e.g. dyspnea, epigastric pain, and cerebral or visual distur-
that a 12-­hour urine collection was non-­inferior to a 24-­hour urine col- bances). Physical exercise was limited as the participants spent the
lection for the diagnosis of pre-­eclampsia. The secondary goal was to majority of their time either sitting or lying on their beds or walking
determine whether time of collection (diurnal vs nocturnal) influenced within the ward. As per local protocols, women with suspected pre-­
the performance of the 12-­hour test. eclampsia underwent blood tests (i.e. complete blood cell count plus
measurement of liver enzymes and creatinine) and urinalysis (24-­
hour collection) three times per week, under the close supervision
2 |  MATERIALS AND METHODS of nursing staff.
In the present study, the 24-­hour urine collection was replaced by
The present prospective observational study was conducted in the two consecutive 12-­hour collections (Fig. 1). Collection vessels were
Department of Obstetrics, Gynecology and Reproductive Medicine, distributed to participants at 07:00 hours, following the first morning
North Lisbon Hospital Centre, Santa Maria University Hospital, Lisbon, void (which was not included in the collection). The first 12-­hour col-
Portugal, between November 1, 2015, and November 30, 2016. The lection took place from 07:00 hours until 19:00 hours (diurnal sample).
present sample comprised consecutive pregnant women (≥20 weeks) Women then received a new vessel for the second 12-­hour collection,
who were admitted to the obstetrics ward of Santa Maria University which took place from 19:00 hours until 07:00 hours (nocturnal sam-
Hospital for maternal–fetal surveillance owing to hypertension ple). This second sample included the first morning void.
(>140/90 mm Hg) and suspicion of pre-­eclampsia. The exclusion cri- The diurnal and nocturnal samples were first independently ana-
teria were known nephropathy and vaginal discharge or bleeding, as lyzed for the presence of substantial proteinuria (≥150 mg). These
these conditions could interfere with urinary protein excretion (UPE). two samples were then pooled to create a 24-­hour or “daily” sample
Urine cultures were requested whenever symptoms of urinary tract (07:00–07:00 hours) that was also analyzed for the presence of sub-
infections developed; women with positive test results were excluded stantial proteinuria (≥300 mg).
from the present analysis. The protocol was approved by the local eth- Each patient underwent a varying number of urine collections,
ics committee and informed consent was obtained from all participants. depending on the length of hospital stay; however, only the 24-­
The daily routine of the participants included at least three blood hour values were considered in the clinical-­decision making process.
pressure measurements and reports on the presence of relevant Demographic and clinical data were prospectively recorded.

Diurnal Nocturnal
urine urine
Urine collection, two consecutive 12-h periods
07:00– 19:00–
19:00 h 07:00h

Quantification of total proteinuria in each 12-h


Mixing and homogenization of both 12-h


urine Quantification of 24-h proteinuria in the total urine

F I G U R E   1   Schematic of study protocol.

Silva ET AL. |

The primary end point was a binary outcome of either having or

not having pre-­eclampsia. This outcome was based on the presence of
positive proteinuria using the cut-­off values outlined above for the 12-­
A total of 99 women were included in the present analysis. The mean
hour and 24-­hour samples. The secondary end point was testing the
number of urine collections during hospitalization was 2.3 (range
equivalence of diurnal and nocturnal proteinuria values.
1–11). In all, there were 456 12-­hour urine collections (equivalent to
As each 24-­hour measurement comprised two matching 12-­hour
228 24-­hour urine measurements).
measurements, the present analysis included three comparisons (Fig. 2).
The demographic and clinical characteristics of the participants
Comparison 1 considered all 12-­hour samples and all 24-­hour measure-
were summarized (Table 1). Most of the women were white, the
ments (2:1 sample ratio) as an indicator of the global performance of
median age was 34 years (range 18–50 years), and most patients
the 12-­hour test. Comparisons 2 and 3 considered the individual perfor-
mance of the diurnal and nocturnal samples, respectively. The qualita-
T A B L E   1   Characteristics of the participants (n=99).a,b
tive analysis assessed reliability and validity of the 12-­hour samples for
diagnosis of substantial proteinuria. The quantitative analysis assessed Characteristic Distribution
equivalence of the diurnal versus nocturnal values for proteinuria. Ethnicity
The data were analyzed using SPSS version 24 (IBM, Armonk, White 79 (79)
NY, USA). The Kolmogorov-­Smirnov test was used to assess normal
African 17 (17)
distribution of the variables. Systematic effects were assessed using
Other 3 (3)
the McNemar test (categorical) and the Wilcoxon signed-­rank test
Age, y 34 (18–50)
(numerical). Reliability of the 12-­hour method versus the 24-­hour
method was assessed using the Cohen κ test, where a κ coefficient of
Nulliparous 57 (57)
1 indicates perfect agreement and a κ coefficient of 0 indicates agree-
ment similar to chance. The validity of this concordance was evaluated Multiparous 41 (42)

by determination of sensitivity, specificity, positive predictive value Conception

(PPV), negative predictive value (NPV), and accuracy.8 The Pearson Spontaneous 83 (83)
coefficient and intraclass correlation coefficient were used to deter- IVF or ICSI 16 (16)
mine agreement among the diurnal and nocturnal samples (pairwise, No. of fetuses
quantitative) based on a single-­measure, absolute agreement, two-­ 1 87 (87)
way random-­effect model.9 An α value of 5% was considered to be sta- 2 12 (12)
tistically significant; all tests were two-­sided. The incidence of positive
proteinuria (24-­hour urinary protein level ≥300 mg) from a series of
Before pregnancy 27.4 (17.5–49.9)
100 collections among the population from the present study setting
During admission 32.7 (20.0–56.4)
(42%) was used to calculate the appropriate sample size. A sample size
Gestational age of the fetus, wk 35 (20–39)
of 372 measurements would allow testing of the non-­inferiority of the
Blood pressure during hospital stay, mm Hg
12-­hour method, with a 10% margin and 80% power at a 5% signifi-
cance level ( Multiple measurements were Normal (≤140/90 mm Hg) 14 (14)

obtained from the participants, which allowed a reduction in the total Raised (>140/90 mm Hg) 80 (85)

number of patients recruited.10 Diagnosis

Pre-­eclampsia 42 (42)
Gestational hypertension 54 (54)
24-h Chronic hypertension 3 (3)
24-­hour urine collection
Comparison ① No. of samples 228
Protein excretion, mg/24 hours 611.4 [48.6–6106.5]
12-­hour urine collection
12-h Diurnal 12-h Nocturnal No. of samples (diurnal + nocturnal) 456
Daily protein excretion, mg/24 hours 278.5 [16.9–3398.9]
Comparison ② Comparison ③ Nocturnal protein excretion, mg/24 hours 312.1 [18.6–2975.4]
“Diurnal and 24-h” “Nocturnal and 24-h”
Abbreviations: BMI, body mass index (calculated as weight in kilograms
divided by the square of height in meters); ICSI, intracytoplasmic sperm
24-h injection; IVF, in vitro fertilization.
Values are given as number (percentage), median (range), of mean [range].
F I G U R E   2   Summary of main comparisons made in the b
Percentages refer to valid data (missing data excluded). For each category,
present study. the missing data were inferior to 5%.
280       Silva ET AL.

were nulliparous. A diagnosis of pre-­eclampsia was made among

Accuracy (95% CI), %

42 (42%) women, after confirmation of positive proteinuria in the

89.3 (86.2–92.4)

86.4 (81.5–91.3)

92.1 (88.8–95.4)
24-­hour samples.
The McNemar test found no evidence of a systematic effect of
the data in the three comparisons (P>0.100; data not shown). The reli-
ability and validity of the 12-­hour test versus the 24-­hour test for a
diagnosis of pre-­eclampsia is shown in Table 2. Comparison 1 and 2
gave a substantial level of agreement (as assessed by the Cohen κ test)

90 (86.4–93.6)

86.9 (81.2–92.5)

93.2 (88.9–97.4)
NPV (95% CI), %
and comparison 3 an almost perfect agreement.
To validate these results, the sensitivity, specificity, PPV, NPV,
and accuracy for detection of pre-­eclampsia were assessed in each
of the three comparisons (Table 2). When compared with the 24-­hour
sample, the two 12-­hour samples showed 85.9% sensitivity, 91.7%

88.2 (83.6–92.9)

85.7 (78.5–92.9)

90.6 (84.8–96.5)
PPV (95% CI), %
specificity, and 89.3% accuracy. The diurnal sample slightly underesti-
mated the proportion of patient with proteinuria when compared with
the 24-­hour sample (39% vs 42%). The nocturnal sample detected the
same proportion of proteinuria as did the 24-­hour sample (42%). The
accuracy for this comparison was 92.1% (nine false positives and nine
false negatives).

91.7 (88.3–95.0)

90.2 (85.0–95.2)

93.2 (88.9–97.5)
Although nocturnal proteinuria appeared to be more accurate than

(95% CI), %
diurnal proteinuria for detecting pre-­eclampsia, this difference was not
statistically significant (P=0.151; related-­samples Wilcoxon signed-­
rank test, data not normally distributed). The equivalence between
both measurements was also tested the using Pearson and intraclass
85.9 (81.0–90.9)

81.3 (73.4–89.1)

90.6 (84.8–96.5)
coefficient correlation (Fig. 3). Both measures showed good (but not
excellent) correlation (<0.9).
(95% CI), %

Some discrepancy was observed between the 12-­hour and 24-­

T A B L E   2   Reliability and validity of 12-­hour urine collection for the diagnosis of pre-­eclampsia.a

hour absolute values. Theoretically, the difference between the 24-­

Abbreviations: CI, confidence interval; NPV, negative predictive value; PPV, positive predictive value.
hour measure and the sum of two 12-­hour measures should be close
P valued

to zero. However, the mean difference was 5.2 mg/dL (95% confidence



interval –58 to 101), which was attributed to both measurement vari-
ability and measurement error. Therefore, all of the statistical tests

Values are given as mean ± SD or number (percentage), unless indicated otherwise.

Cohen κ

were repeated after excluding the automatically generated outlier val-




ues (n=10). This approach did not affect the results reported above.

Defined as a protein level of ≥150 mg/12 hours or ≥300 mg/24 hours.


187 (41)
96 (42)

91 (39)
96 (42)

96 (42)
96 (42)

The three comparisons are outlined schematically in Figure 2.

The present study found that a 12-­hour urine collection provided

acceptable accuracy for the detection of proteinuria when compared
Proteinuria, mg

278.5 ± 437.9
611.3 ± 918.6

278.4 ± 437.8
611.3 ± 918.6

312.1 ± 490.0
611.3 ± 918.6

with the current gold standard (24-­hour urine collection). These find-
ings suggested that the 12-­hour test could offer a reasonable alterna-
tive to the 24-­hour test in the diagnosis of pre-­eclampsia.
A 12-­hour sample (07:00–19:00 hours).

A 12-­hour sample (19:00–07:00 hours).

A 24-­hour sample (07:00–07:00 hours).

Pre-­eclampsia is generally characterized by raised blood pressure

in association with increased UPE as a marker of proteinuria. However,
All 12-­hour samples (n=456)

in the absence of proteinuria, a diagnosis of pre-­eclampsia can still be

24-­hour samples (n=228)

made on the basis of new-­onset hypertension with signs of end-­organ

Nocturnal (n=288)g

lesion, including thrombocytopenia, renal insufficiency, impaired liver

Diurnal (n=288)e

function, pulmonary edema, and cerebral or visual symptoms.2,11

Daily (n=288)f

Daily (n=288)f
Comparison 1

Comparison 2

Comparison 3

Elevated levels of UPE are detected among most women with pre-­
eclampsia owing to increased vascular permeability and altered tubu-
χ test.

lar charge selectivity of the glomerular filtration barrier.12,13 Adverse

d 2

obstetric outcomes were thought to be related to elevated UPE;



Silva ET AL. |

Nocturnal proteinuria 19:00–07:00h (mg/12h)

the day. Regarding blood pressure, the available records only allowed
4000 the identification of women with either normal values or raised values
during their whole hospital stay. A detailed record of blood pressure
values could aid understanding of the importance of hypertension for
UPE variance between the diurnal and nocturnal periods.
2000 r (Pearson): 0.852
(P<0.001) An optimum method to both quickly and accurately diagnose
1000 ICC: 0.845
95%CI 0.803–0.879
pre-­eclampsia is still lacking; however, some promising work is being

Wilcoxon: P=0.151
developed using angiogenesis-­related biomarkers. Consistent evi-
0 1000 2000 3000 4000 dence indicates that the levels of placental growth factor decrease,
Diurnal proteinuria 07:00–19:00h (mg/12h) whereas the levels of soluble fms-­like tyrosine kinase 1 (also known
as sFLT-­1) increase, among pre-­eclamptic women before the clinical
F I G U R E   3   Agreement between diurnal and nocturnal 12-­hour
values; data were not normally distributed. Abbreviation: ICC, onset of this condition.17 Prospective data have already suggested a
intraclass correlation coefficient. potential role for the ratio of these two growth factors as a screening
method for pre-­eclampsia that offers high NPV (99.3%–99.9%) and a
however, studies have failed to demonstrate a strong causality.14,15 78.6% detection rate of the disease within 1 week.17 These biomarkers
Given that the absence of proteinuria could delay prompt intervention, provide rapid results (approximately 30 minutes) and seem to be cost-­
substantial proteinuria is no longer considered to be either an indica- saving18 in identifying patients who would benefit from hospitaliza-
tor of disease severity or an obligatory component for the diagnosis of tion. However, recommendations on their use have yet to be approved.
pre-­eclampsia in the presence of another sign of end-­organ dysfunc- In conclusion, the present study demonstrated that a 12-­hour
tion.2 Nevertheless, proteinuria undoubtedly remains an important urine collection could offer an alternative to the gold-­standard method
clinical feature of pre-­eclampsia and is fundamental to the diagnosis of 24-­hour urine collection when a fast result is required or when it
for many patients. Quantification of proteinuria in a 24-­hour urine is the only financially viable solution. Further studies—including those
collection is the current gold-­standard test because it is representa- designed to assess cost-­effectiveness—are now required to investi-
tive of circadian variability in UPE. Collection of urine samples in an gate this possibility.
inpatient setting is probably more reliable than collection in an outpa-
tient setting. Indeed, many women with suspected pre-­eclampsia are
hospitalized for optimum clinical management. By contrast, the cost of
collection in an inpatient setting will be higher than that incurred in an RMS contributed to the conception and design of the study, the
outpatient setting. acquisition, analysis, and interpretation of data, literature review, and
The present study design followed the suggestions of Stout writing and revising the manuscript. SRP and SR contributed to the
et al.7 by obtaining a large sample size, ensuring that the collection of acquisition of data and to revising the manuscript. NC contributed to
urine was homogeneous, and stratifying the samples by diurnal ver- the conception and design of the study, the acquisition, analysis, and
sus nocturnal. An effort was made to implement only minor changes interpretation of data, and revising the manuscript.
in the established routines of the obstetric ward so as to minimize
procedure errors.
In the qualitative analysis, the 12-­hour test detected a similar pro-
portion of patients with proteinuria as the 24-­hour test, with a dis- The authors have no conflicts of interest.
crepancy of less than 10% (as was proposed to test with the sample
size). However, global reliability was not as satisfactory as expected,
with Cohen k values of less than 0.8.
Nocturnal proteinuria seemed to perform better than diurnal pro- 1. Say L, Chou D, Gemmill A, Tunçalp Ö, Moller AB, Alkema L. Global
causes of maternal death: A WHO systematic analysis. Lancet Global
teinuria for the diagnosis of pre-­eclampsia, with high sensitivity, speci-
Health. 2014;2:323–333.
ficity, PPV, NPV, and accuracy. Nevertheless, the quantitative analysis 2. American College of Obstetricians and Gynecologists. Task Force
(comparing absolute values of proteinuria) found no statistically sig- – ACOG Guidelines: Hypertension in pregnancy. Published 2012.
nificant difference between the two collection times. Therefore, it
was not possible to make firm conclusions about the performance Work%20Group%20Reports/public/HypertensioninPregnancy.pdf.
Accessed November 20, 2017.
of nocturnal collection versus diurnal collection. Levels of UPE are
3. National Institute for Health and Care Excellence. NICE Clinical
known to be raised by physical exercise, high blood pressure, urinary Guideline no 107. Hypertension in pregnancy: diagnosis and man-
tract infection, massive obesity, cardiac failure, reduced water intake, agement. Published 2011.
hematuria, uterine bleeding, vaginal discharge, and some severe Accessed November 20, 2017.
4. Henderson JT, Thompson JH, Burda BU, Cantor A, Beil T, Whitlock
stressful situations.16 In the present study, several of these confound-
E. Screening for Preeclampsia: A Systematic Evidence Review for the U.S.
ers were controlled by the exclusion criteria but also by the fact that Preventive Services Task Force, Vol. 148. Rockville (MD): Agency for
the participants undertook only limited physical exercise, even during Healthcare Research and Quality (US); 2017:18–26.
282       Silva ET AL.

5. Sanchez-Ramos L, Gillen G, Zamora J, Stenyakina A, Kaunitz AM. The 12. Moran P, Baylis P, Lindheimer M. Glomerular ultrafiltration in normal
protein-­to-­creatinine ratio for the prediction of significant proteinuria and preeclamptic pregnancy. J Am Soc Nephrol. 2003;14:648–652.
in patients at risk for preeclampsia: A meta-­analysis. Ann Clin Lab Sci. 13. Côté AM, Mallapur A, Katageri G, et al. Measurement of proteinuria.
2013;43:211–220. In: Magee LA, vonDadelszen P, Stones W, Mathai M, eds. The FIGO
6. Cote AM, Brown MA, Lam E, et  al. Diagnostic accuracy of urinary Textbook of Pregnancy Hypertension. London: The Global Library of
spot protein:creatinine ratio for proteinuria in hypertensive pregnant Women’s Medicine; 2016:19–32.
women: Systematic review. BMJ. 2008;336:1003–1006. 14. Newman MG, Robichaux AG, Stedman CM, et al. Perinatal outcomes
7. Stout MJ, Conner SN, Colditz GA, Macones GA, Tuuli MG. The utility of in preeclampsia that is complicated by massive proteinuria. Am J
12-­hour urine collection for the diagnosis of preeclampsia: A system- Obstet Gynecol. 2003;188:264–268.
atic review and meta-­analysis. Obstet Gynecol. 2015;126:731–736. 15. Schiff E, Friedman SA, Kao L, Sibai BM. The importance of urinary pro-
8. Watson PF, Petrie A. Method agreement analysis: A review of correct tein excretion during conservative management of severe preeclamp-
methodology. Theriogenology. 2010;73:1167–1179. sia. Am J Obstet Gynecol. 1996;175:1313–1316.
9. Koo TK, Li MY. A guideline of selecting and reporting intraclass 16. Mogensen CE, Vestbo E, Poulsen PL, et  al. Microalbuminuria and
correlation coefficients for reliability research. J Chiropr Med. potential confounders. Diabetes Care. 1995;18:572–581.
2016;15:155–163. 17. Dragan I, Georgiou T, Prodan N, Akolekar R, Nicolaides KH.
10. Mowinckel P, Hagen K, Heiberg T, Kvien T. Repeated measures in Screening for pre-­eclampsia using sFlt-­1/PlGF ratio cut-­off of 38 at
rheumatoid arthritis reduced the required sample size in a two-­armed 30–37 weeks’ gestation. Ultrasound Obstet Gynecol. 2017;49:73–77.
clinical trial. J Clin Epidemiol. 2008;61:940–944. 18. Vatish M, Strunz-McKendry T, Hund M, Allegranza D, Wolf C, Smare
11. Sibai B, Dekker G, Kupferminc M. Pre-­eclampsia. Lancet. 2005;365: C. sFlt-­1/PlGF ratio test for pre-­eclampsia: An economic assessment
785–799. for the UK. Ultrasound Obstet Gynecol. 2016;48:765–771.