Variation of urinary protein to creatinine ratio during the day in women with suspected pre-eclampsia

Authors

  • K Verdonk,

    1. Department of Obstetrics and Gynaecology, Division of Obstetrics and Prenatal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    2. Department of Internal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • IC Niemeijer,

    1. Department of Obstetrics and Gynaecology, Division of Obstetrics and Prenatal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • WCJ Hop,

    1. Department of Biostatistics, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • YB de Rijke,

    1. Department of Clinical Chemistry, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • EAP Steegers,

    1. Department of Obstetrics and Gynaecology, Division of Obstetrics and Prenatal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • AH van den Meiracker,

    1. Department of Internal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    Search for more papers by this author
  • W Visser

    Corresponding author
    1. Department of Obstetrics and Gynaecology, Division of Obstetrics and Prenatal Medicine, Erasmus Medical Centre, Rotterdam, the Netherlands
    • Correspondence: W Visser, Department of Obstetrics and Gynaecology, Division of Obstetrics and Prenatal Medicine, Erasmus Medical Centre, Dr Molewaterplein 60, 3015 GJ Rotterdam, the Netherlands. Email Willy.visser@erasmusmc.nl

    Search for more papers by this author

Abstract

Objective

To investigate the stability throughout the day of the protein to creatinine ratio (PCR) in spot urine, to demonstrate whether the PCR is a valid alternative for 24-hour protein investigation in pregnant women.

Design

Prospective study.

Setting

Tertiary referral university centre.

Population

Women suspected of having pre-eclampsia, admitted to the Erasmus Medical Centre.

Methods

Twenty-four-hour urine collections and simultaneously three single voided 5-ml aliquots were obtained at 8 a.m., 12 a.m. (noon) and 5 p.m. A PCR was measured in each specimen and compared with the 24-hour protein excretion.

Main outcome measures

The 24-hour proteinuria and PCR measured in spontaneous voids.

Results

The PCRs correlated strongly with each other and with the 24-hour protein excretion but did show variation throughout the day (mean coefficient of variation 36%; 95% confidence interval 31–40%). The coefficient of variation was unrelated to the degree of 24-hour proteinuria. Receiver operating characteristics curves to discriminate between values below and greater than or equal to the threshold of 0.3 g protein per 24-hour had an area under the curve of respectively 0.94 (8 a.m.), 0.96 (noon) and 0.97 (5 p.m.). Sensitivities at 8 a.m., noon and 5 p.m. were respectively 89%, 96% and 94%; specificities were 75%, 78% and 78% with the proposed PCR cut-off of 30 mg/mmol (0.26 g/g) (National Institute for Health and Care Excellence guidelines).There is no evidence of a difference between the three measurement times regarding the sensitivities and specificities.

Conclusion

The PCR determined in spot urine varies throughout the day but is a valid alternative for 24-hour urine collections in pregnant women. It is especially useful to rapidly identify clinically relevant proteinuria.

Introduction

Pre-eclampsia is diagnosed when after 20 weeks of gestation repeated blood pressure measurements equal or exceed 140/90 mm Hg in combination with the de novo proteinuria of ≥0.3 g/24-hour.[1, 2] The reference standard for the latter measurement is the quantity of protein in a 24-hour urine collection, a very cumbersome procedure, prone to errors related to inaccurate and/or incomplete collections, problems that become exaggerated during pregnancy.[3-5] Indeed, imprecise outpatient urine collections have been reported to approach 50% in some studies; as a consequence, patients are therefore often admitted to the hospital to quantify the degree of proteinuria.[6-9]

The qualitative proteinuria dipstick test is not a reliable alternative. Its readings are affected by whether the urine is concentrated or diluted, as well as by its pH, resulting in an unacceptable number of false-positive and false-negative test results.[10] In recent years, caregivers have turned to measuring the protein to creatinine ratio (PCR) in single voided urine and this has become the preferred and recommended approach for quantifying proteinuria in non-pregnant subjects.[3, 11, 12] In this respect, the recently published National Institute for Health and Care Excellence guidelines recommend using a spot PCR ≥ 30 mg/mmol as an alternative to ≥0.3 g/day in 24-hour collections to diagnose clinically relevant proteinuria.[13] However, use of the PCR on a single voided urine as an alternative to the 24-hour measurement assumes that the urinary excretion rates of protein and creatinine throughout the day are relatively constant. Creatinine excretion shows very little variation, but we could not find verification for protein excretion. We therefore explored the constancy of spot PCR at different time points during the day and their relationship with simultaneously obtained 24-hour urinary protein excretion. Specifically we assessed the diagnostic accuracy of the PCR in women hospitalised to determine whether or not they had pre-eclampsia.

Methods

Women suspected of having developed pre-eclampsia, admitted to the Department of Obstetrics and Gynaecology's inpatient service at the Erasmus Medical Centre in Rotterdam were recruited for a protocol, the study was completed in 2 years. All volunteers gave consent following detailed explanation of the study goal using a protocol approved by the hospital's ethics review committee. Suspected pre-eclampsia was defined as de novo hypertension, with blood pressure ≥140/90 mmHg after 20 weeks of gestation and a urine protein dipstick reading ≥1+. Pregnant women with chronic hypertension who developed new-onset proteinuria after mid-gestation were also asked to participate. Women with urinary tract infection, pre-existing proteinuria or having a delivery before the 24-hour urinary collection was completed, were excluded.

During the urine collection period the women were on bed rest with bathroom privileges. The collection began at midnight with 5-ml separate aliquots saved for PCR testing from requested spontaneous voids at approximately 8 a.m., 12 a.m. (noon) and 5 p.m. Nurses monitored the completeness of the 24-hour collection, and when errors occurred, the procedure was stopped and restarted at midnight the next day.

Creatinine was measured by an enzymatic assay (CREA plus; Roche Diagnostics, Germany; inter-assay coefficient of variation < 3%) and protein was measured by a colorimetric assay (Roche Diagnostics; inter-assay coefficient of variation < 1.6%). Dipsticks (Albustix; Siemens Healthcare Diagnostics Inc., Tarrytown, NY, USA) were visually analysed. A single plus represented 0.3 g/l. Statistical analysis was performed with SPSS Statistics 18 (SPSS Inc., Chicago, IL, USA). Proteinuria, defined as ≥ 0.3 g in the 24-hour collection was used as the reference standard to compare with the respective PCRs. A cut-off value of the PCR of 30 mg/mmol (0.26 g/g)[1, 13] was used to calculate the sensitivity and specificity of PCR at the three time points.

PCRs at 8 a.m., noon and 5 p.m. were compared using the Wilcoxon signed-rank test and for each subject the coefficient of variation for the three time points of the PCR was computed. For the whole group, Spearman's bivariate analysis was used to calculate correlation coefficients between the PCRs at the three time points and between the PCRs and the 24-hour urinary protein excretion. For the three PCRs, receiver operating characteristics (ROC) curves were constructed and areas under the curve (AUC) were calculated. The Cochran Q-test was used to test if there were differences in sensitivity and specificity at the different time points.

Optimal cut-off levels were determined with the Youden's index by finding the maximum value of the sum of sensitivity and specificity (thereby taking equal weights of the sensitivity and specificity). Intra-class correlation coefficients were calculated to evaluate the agreement of the three PCRs at the different time points. Values were log-transformed in all analyses to obtain normal distributions, except for calculating the intra-individual coefficient of variation of the PCR. A value of ≤ 0.05 (two-sided) was considered to indicate a significant difference.

Results

One hundred and twelve women were recruited, six were excluded because of incomplete data, and one because of pre-existing proteinuria. None had evidence of a urinary tract infection. Of the 105 evaluable women, five had a creatinine excretion of < 5.5 mmol (0.6 g) per 24-hour, suggesting incomplete 24-hour urine collection.[14] In one of these patients PCR was positive. However, excluding these patients had no effect on the main outcomes of the study so these patients were not excluded and their data are shown in the Results.

Clinical characteristics of the study population are presented in Table 1. The 24-hour proteinuria varied from 0.05 to 17.04 g with 73 women (70%) having a 24-hour proteinuria exceeding 0.3 g. Median (interquartile range) PCR was 60.0 mg/mmol (23.3–178.5 mg/mmol) at 8 a.m., 79.2 mg/mmol (28.2–233.9 mg/mmol) at noon and 73.3 mg/mmol (24.6–747.1 mg/mmol) at 5 p.m. The PCR of 8 a.m. was lower than the PCR at noon (60 mg/mmol versus 79.2 mg/mmol; = 0.002) but did not differ from the PCR at 5 p.m. (60 mg/mmol versus 73.3 mg/mmol; = 0.167), the PCR at noon was higher than the PCR at 5 p.m. (79.2 mg/mmol versus 73.3 mg/mmol; = 0.044). The average coefficient of variation of the three PCRs was 36% (95% confidence interval [95% CI] 31–40%). The intra-individual coefficient of variation did not correlate with the amount of proteinuria (Spearmans rho = −0.08, = 0.41). The intra-class correlation coefficients were very good with 0.91 (95% CI 0.86–0.94) for 8 a.m. versus noon, 0.90 (95% CI 0.86–0.93) for noon versus 5 p.m. and 0.90 (95% CI 0.85–0.93) for 8 a.m. versus 5.00 p.m.

Table 1. Characteristics of the study population (total and subdivided according to values < 0.3 g/24 hours and ≥ 0.3 g/24 hours)
Proteinuria<0.3 g/24 hours≥0.3 g/24 hoursTotal
  1. Values shown are median (25th centile – 75th centile), or numbers (percentage) of patients.

Number3273105
Nullipara19 (59%)40 (55%)59 (56%)
Gestational age (weeks)33.1 (30–36)31 (28–35)31 (29–35)
Age (years)30 (27–35)31 (29–34)31 (28–34)
Serum creatinine (μmol/l)57 (48–61)60 (52–67)58 (50–65)
Serum uric acid (mmol/l)0.36 (0.28–0.45)0.36 (0.29–0.40)0.36 (0.29–0.40)
Proteinuria (g/day)0.18 (0.13–0.21)1.27 (0.6–3.01)0.61 (0.24–1.70)

The distribution of the PCR values stratified for the positive and negative outcomes of the 24-hour urinary protein collection is shown in Figure 1. The PCR in one or more of the samples was a false negative in ten women (7.3%) at a PCR cut-off of 30 mg/mmol. The 24-hour proteinuria in these ten women was relatively low, varying from 0.3 to 0.6 g with an average value of 0.4 g. The ROC curves of the three PCRs to discriminate between 24-hour proteinuria of <0.3 versus ≥0.3 g/24 hours are shown in Figure 2. The three PCRs discriminated well, with AUCs varying from 0.94 to 0.97. Optimal cut-off values were 50.4 mg/mmol at 8 a.m., 42.4 mg/mmol at noon and 35.4 mg/mmol at 5 p.m. as determined by the Youden's index. Sensitivities and specificities for different cut-off points of PCR to diagnose proteinuria are given in Table 2. There were no significant differences between the three measurement times regarding the specificities (Cochran Q-test: = 0.89). The same applied to the sensitivities (= 0.12).

Table 2. Sensitivities and specificities, for different cut points of PCR to diagnose proteinuria (≥0.3 g/24 hours)
 PCR 8 a.m.PCR 12 a.m.PCR 5 p.m.
  1. a

    PCR cut-off 30 mg/mmol.

  2. b

    Classifying according to the optimal cut-off for the protein creatine ratio (PCR cut-off 8 a.m. 50.4 mg/mmol, PCR cut-off 12 a.m. 42.4 mg/mmol, PCR cut-off 5 p.m. 35.4 mg/mmol).

  3. c

    PCR at 8 a.m. is lower than PCR at noon P = 0.002.

  4. d

    PCR at 5 p.m. is lower than PCR at noon P = 0.044.

PCR, median (25th–75th centile)60 (26–173)c79 (29–229)73 (25–193)d
True positives/false positivesa65/870/768/7
True negatives/false negativesa24/825/325/4
True positives/false positivesb59/169/467/3
True negatives/false negativesb31/1428/429/5
Area under the ROC curve (95% CI)94 (90–98)96 (92–99)97 (94–100)
Sensitivity,% (95% CI)a89 (80–95)96 (88–99)94 (86–98)
Specificity,% (95% CI)a75 (57–89)78 (60–91)78 (60–91)
Sensitivity,% (95% CI)b81 (72–90)95 (89–100)93 (90–96)
Specificity,% (95% CI)b97 (93–100)88 (76–99)91 (81–100)
Figure 1.

Scatterplot of individual protein–creatinine ratios obtained at 8 a.m., 12 a.m. and 5 p.m. stratified for proteinuria <0.3 (inverse triangles) or ≥0.3 g per 24-hour (triangles). Note the logarithmic scale on the vertical axis.

Figure 2.

Receiver operating characteristics (ROC) curve to diagnose proteinuria (≥0.3 g/24-hour) according to PCR 8 a.m., PCR 12 a.m. and PCR 5 p.m. The areas under the curve of PCR at 8 a.m., 12 a.m. and 5 p.m. are 0.94 (95% CI 0.90–0.98), 0.96 (95% CI 0.92–1.00) and 0.97 (95% CI 0.94–1.00), respectively.

Discussion

Main findings

Studying inpatients suspected of having pre-eclampsia we found that the PCRs during the day are well correlated but are subject to variation. Nonetheless, ROC analyses indicate that PCRs at different daytime points are quite accurate to confirm or reject the presence of the threshold for diagnosing 24-hour proteinuria, as reflected by the AUCs of >0.94 or more. In 7.3% of women with a 24-hour proteinuria of ≥0.3 g one or two of the PCRs was below the generally accepted cut-off value of 30 mg/mmol and hence was false negative. In these women proteinuria, albeit above the cut-off value of ≥0.3 g, was relatively low with an average value of 0.4 g per day.

Strengths and limitations

To our knowledge no other studies have looked at the variation of the spot PCR during the day in the manner described here. The difference from other studies is that we not only compared three spot PCRs against the 24-hour urine collection, but also looked at the variation of the three spot PCRs throughout the day and measured the within-subject variation in PCR.

Although the nurses closely monitored the urinary collection, the total creatinine excretion in the 24-hour collection was relatively low in several subjects suggesting under-collection of urine. Another limitation concerning the generalisability of this study is that subjects were hospitalised for 24 hours and had bed rest. However, even in a situation where there is little variance in the activity pattern, there was still a considerable variability in the PCR during the day. One could emphasise that the activity pattern of non-hospitalised patients is more diverse, which could also affect the variation of the PCR.[4] Moreover, 70% of our patients had proteinuria. The performance of the test will be worse in a cohort of patients with a relatively low prevalence of proteinuria. This study might be underpowered for detecting differences in the sensitivity and specificity of the PCR during the day, because the primary objective of this study was to determine the possible variation of the PCR during the day in a spot urine. Further studies are needed in women with minimal proteinuria and in ambulant patients.

Interpretation

Multiple studies have been performed to investigate the accuracy of the PCR in pregnant women. In general these studies compared the spot PCR against 24-hour urine collection. In a systematic review and meta-analysis Cote et al. [8] concluded that the accuracy of the PCR to rule-out proteinuria is reasonable using a cut-off value of 30 mg/mmol. With this cut-off point, sensitivity and specificity were respectively 84% and 76%.[8] In another systematic review it has been estimated that the optimum threshold for PCR has to be between 35 and 40 mg/mmol (0.30 and 0.35 mg/g) for diagnosing clinically relevant proteinuria, but that sensitivity (83.6%) and specificity (76.3%) of the PCR are relatively poor.[15] In the latter review the authors performed a subgroup analysis for hypertensive pregnant women with a positive dipstick test for proteinuria. In this subgroup, sensitivity and specificity improved with respective values of 86% and 95% when using a threshold of the PCR above 23 mg/mmol.[15]

In our study there was, in fact, some variation in the PCR throughout the day, which appeared unrelated to the degree of proteinuria. Apart from the underlying renal lesion, a variety of factors such as posture, exercise, emotional factors and blood pressure may influence protein excretion, which is in accordance with the finding that the 8 a.m. PCR, after an overnight rest, is lower than the PCR at noon. Furthermore, this resulted in a somewhat higher optimal cut-off value of 50.4 mg/mmol for diagnosing clinically relevant proteinuria. In our population the prevalence of proteinuria was large, with 70% of the women having a proteinuria of at least 0.3 g per 24 hours, and in all women the urinary dipstick test for proteinuria was positive. Given the variation in PCR throughout the day, it should be emphasised that in a cohort of patients with a relatively low prevalence of proteinuria the performance of a PCR test result ≥30 mg/mmol will be worse.[15]

Twenty-four hour urine collection is time-consuming and often inaccurate. Incomplete outpatient collections have been reported to approach 50% in some studies.[3] Even in this study, where 24-hour urine collections were performed while patients were hospitalised and supervised by the nurses, collections were probably incomplete in almost 5% of the participants, as reflected by the low 24-hour urinary creatinine excretion. Collection errors may be due not only to over- and under-collection, but also to gestation-related changes in the urinary tract, including dilatation or transient obstruction of the ureters when gravidas assume certain postures and the enlarged uterus presses on the point where the iliac arteries cross the ureters.[4, 5] It has been proposed therefore, that the 24-hour or any other form of timed urine collection should be replaced by the determination of the PCR in a single voided urine, as has become the standard approach in non-pregnant subjects.[12]

Two studies have investigated the variation of the PCR throughout the day in 6- to 8-hour collections.[14, 16] Gonsales Valério et al. [14] investigated 75 hypertensive pregnant women after a gestational age of 20 weeks, recording the PCR at arrival as well as during four subsequent 6-hour periods. The mean at each period was fairly constant, but, unfortunately, no information about the within-subject variation in PCR was reported. In concordance with our study PCRs at the different time-points and 24-hour protein excretion were strongly correlated with correlation coefficients of 0.8 or higher.[14] Contrary data emerged from the study by Lindow and Davey, who measured PCR values in three 8-hour samples in 22 pre-eclamptic women with indwelling catheters, comparing these values to the protein excreted in 24 hours. The PCR appeared to vary considerably and the correlation with 24-hour collections was poor.[16] Few details were given but if these patients were ill enough to require catheterisation, than a changing glomerular filtration rate might have invalidated the results.

Conclusion

PCR determined in single voided urine can be used as an alternative for 24-hour urine collections in pregnant women with pre-eclampsia or suspected of having this condition. However, the PCR measured in spontaneous voids varies throughout the day, which appears to be independent of the amount of proteinuria. While this knowledge should be taken into account when it is used as a substitute of a 24-hour proteinuria measurement in women suspected of having preeclampsia, the ratio appears an acceptable alternative, especially when time is a factor to distinguish gestational hypertension from pre-eclampsia. Since all our patients were hospitalised further studies are necessary to determine whether our results also apply to the outpatient setting.

Disclosure of interests

The authors declare that they have no conflict of interest.

Contribution to authorship

WV and YR conceived and designed the study; WV, performed the experiments; YR; contributed reagents, materials and analysis tools; WV, WH, IN and KV analysed the data; WV KV and IN drafted the manuscript. KV, IN, WH, YR, ES, AM and WV participated in the writing of the final version of the manuscript.

Details of ethics approval

This study was approved by the Erasmus MC Local Research Ethics Committee. The approval registration number is: MEC-2012-475.

Funding

No financial support was received for this study.

Acknowledgements

None.

Ancillary