SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Objective  To determine, in women with proteinuric pre-eclampsia, whether a discriminant value of proteinuria at the time of diagnosis predicts the presence or absence of subsequent adverse maternal and fetal outcomes.

Design  Retrospective cohort study.

Setting  One teaching hospital and two primary referral hospitals in Sydney, Australia.

Sample  Three hundred and twenty-one pregnant women with proteinuric pre-eclampsia, managed according to a uniform management protocol.

Methods  All women with the diagnosis of proteinuric pre-eclampsia in the years 1998–2001 were studied. After exclusion of women with pre-eclampsia superimposed on pre-existing hypertension, a twin pair, unavailable spot urine results, 353 women were analysed using logistic regression to determine separately the predictors of any adverse maternal or fetal outcomes at the time of delivery. Receiver operating characteristic (ROC) curves, sensitivity and specificity were then calculated from the data.

Main outcome measures  Adverse maternal outcomes: severe maternal hypertension (BP ≥ 170/110 mmHg), renal insufficiency, liver disease, cerebral irritation, haematological disturbances. Adverse fetal outcomes: small for gestational age, perinatal mortality.

Results  There were 108 (34%) adverse maternal outcomes and 60 (19%) adverse fetal outcomes including two stillbirths. In multivariate analysis, an adverse maternal outcome was significantly associated with higher spot urine protein/creatinine ratio at diagnosis (P < 0.0001) with an odds ratio (OR) of 1.003 per mg/mmol (95% confidence interval [CI] 1.002–1.004) and with older maternal age (P= 0.014) with OR 1.06 per year (95% CI 1.01–1.11). An increased risk of adverse fetal outcome was associated with higher spot urine protein/creatinine (P= 0.013; OR 1.44 per log [mg/mmol], 95% CI 1.08–1.92), gestation at diagnosis <34 weeks (P < 0.0001; OR 3.60, 95% CI 1.90–6.82) and early pregnancy systolic blood pressure ≤115 mmHg (P= 0.0002; OR 3.41, 95% CI 1.77–6.57). The area under the receiver operating characteristic (ROC) curve was 0.67 for adverse maternal outcomes and 0.72 for adverse fetal outcomes.

Conclusions  With increasing proteinuria, there is increased risk of adverse maternal and fetal outcomes. Although we did not identify a specific spot protein/creatinine ratio that could be used as a definitive screening value for adverse outcomes, it is possible to utilise data from this study to predict the likelihood of adverse maternal and fetal outcomes. A high spot urine protein/creatinine ratio in pre-eclamptic women of greater than 900 mg/mmol (∼9 g/day), or greater than 500 mg/mmol (∼5 g/day) in women over 35 years, is associated with a greatly increased likelihood of adverse maternal outcomes.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Pre-eclampsia is a major direct cause of both maternal and fetal morbidity and mortality worldwide.1 In Australia, it is estimated that between 4% and 10% of pregnant women will develop pre-eclampsia, causing four to five maternal deaths and up to 300 to 500 perinatal deaths per year.2

Pre-eclampsia is a multisystem disorder that affects the maternal kidneys, liver, brain, clotting systems and primarily, the placenta.3 Although more than 100 years have passed since pre-eclampsia was first described, knowledge of its aetiology and pathophysiology remains limited.4,5

It is known that women with proteinuric pre-eclampsia have a poorer outcome than those with gestational hypertension (non-proteinuric) alone.6 There is debate about whether proteinuria should remain a ‘sine qua non’ for the clinical diagnosis of pre-eclampsia but it nevertheless remains a hallmark of this disorder and is still a requirement for any research diagnosis of pre-eclampsia.7,8

Significant proteinuria is defined by the International Society for the Study of Hypertension in Pregnancy (ISSHP) as ≥300 mg/day of protein in a 24-hour urine collection or a spot urine protein/creatinine ratio ≥30 mg/mmol.8Not all women with proteinuric pre-eclampsia have adverse outcomes and, to the best of our knowledge, no study has determined whether a threshold value of protein excretion exists in pre-eclampsia, above which maternal and/or fetal complications are likely to be present, below which they are likely to be absent. Such knowledge would aid triage of pre-eclamptic women to require either inpatient or outpatient management and would therefore provide a considerable advance in the clinical management of these women.

The purpose of this study was to determine whether, in women with proteinuric pre-eclampsia, a discriminant value of spot urine protein/creatinine ratio exists at the time of antenatal diagnosis that predicts subsequent adverse maternal and fetal outcomes.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Data on all pregnant women with hypertension referred by their obstetricians or antenatal clinics to two renal physicians in the St George Hospital, Sydney have been entered into the HIP (Hypertension in Pregnancy) database since 1987. In general, all hypertensive pregnant women with proteinuria or maternal organs dysfunction (e.g. renal, liver disease, thrombocytopenia) are referred to this service. Demographic details, clinical and laboratory data at the time of referral and delivery and outcomes of mothers and babies have been collected prospectively on each woman.

Following confirmation that spot urine protein/creatinine ratio correlated well with 24-hour urine proteinuria in hypertensive pregnant women (r= 0.93, P < 0.001),9 spot urine testing was introduced in this unit in 1997. Using this database, we conducted a retrospective cohort study of women who delivered live births and stillbirths over a four-year period from 1998 to 2001.

This study was approved by the Southern section of the South East Health Human Research Ethics Committee.

The definitions used in this study are those of the ISSHP.8 Pre-eclampsia is defined clinically as the development of systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg after 20 weeks of gestation in women with no previous history of hypertension, cardiac or renal disease plus evidence of other organ system involvement, including one or more of: proteinuria, renal insufficiency, liver disease, neurological problems, haematological disturbances and fetal growth restriction (see below).

The ISSHP research definition of pre-eclampsia, which includes both hypertension and proteinuria (defined as ≥300 mg per 24 hours or a spot urine protein/creatinine ratio ≥30 mg/mmol), was used for this study.8

The spot urine protein/creatinine ratio used in all analyses was that measured at the initial diagnosis of proteinuric pre-eclampsia. Potential predictive or confounding variables included maternal age, parity, gestational age at time of diagnosis, ‘booking-in’, that is, early pregnancy, systolic and diastolic blood pressure and gestational diabetes.

Adverse maternal outcomes included any new episode after diagnosis of severe hypertension (≥170/110 mmHg), renal insufficiency (creatinine >90 μmol/L), liver disease (aspartate aminotransferase >40 U/L), cerebral irritation (hyperreflexia with clonus or repeated visual scotomata, requiring magnesium sulphate) and thrombocytopenia (platelets <150 × 109/L).

Adverse fetal outcomes included perinatal mortality and/or small for gestational age at the 10th centile using the Australian population growth centiles,10 corrected for sex.

All women with the diagnosis of proteinuric pre-eclampsia in the years 1998–2001 were studied. There were 390 babies delivered to women with this diagnosis, after excluding women with pre-eclampsia superimposed on pre-existing hypertension. Among these were 37 twins, of which one pair had small for gestational age and 12 pairs were discordant for small for gestational age at the 10th centile. One twin of each pair was randomly selected using Excel (7 out of 12 discordant small for gestational age were selected), leaving 353 women with one baby each. Of these, 15 did not have spot urine results available. Six women who had a postpartum diagnosis of proteinuric pre-eclampsia were excluded.

A computer-generated random check of 10% of the data found two incorrect diagnoses, both with booking-in blood pressure (BP) ≥ 140/90 mmHg, suggesting chronic hypertension. All booking-in blood pressures were then specifically checked, resulting in 11 women with pre-existing hypertension being excluded, and leaving a final sample of 321 women (Fig. 1). The primary analysis for fetal outcomes was repeated with twins excluded.

image

Figure 1. Pre-eclamptic women included in the study.

Download figure to PowerPoint

Data were analysed using SAS v8 program.11 Two logistic regression analyses were performed12 to determine separately the predictors of any adverse maternal or fetal outcomes at the time of delivery. For maternal outcomes only adverse outcomes that developed since diagnosis of proteinuric pre-eclampsia were counted. For example, women with renal insufficiency at diagnosis could not develop renal insufficiency, but could develop any of the other adverse maternal outcomes.

Any variable found to be associated with the outcome on univariate analysis with P < 0.25 was included in the initial multivariate logistic regression model for that outcome. Potential collinearity problems were examined by assessing the stability of parameter estimates and standard errors in this model. The linearity assumption for continuous variables was checked in both models.12 For fetal outcomes none of the predictors satisfied this assumption, so spot urine values were transformed using natural logarithms, gestation at diagnosis was dichotomised at 34 weeks and booking-in systolic blood pressure (SBP) at 115 mmHg. Each variable was assessed for effect modification with spot urine protein/creatinine. Variables that were not confounders of the effect of spot urine protein/creatinine and were not statistically significant at P < 0.05 were then progressively eliminated, starting with the least significant, prior to checking the fit of the final model.

Receiver operating characteristic (ROC) curves, sensitivity and specificity were calculated from the data. Positive likelihood ratio was calculated as sensitivity/(1 − specificity).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Characteristics of the mothers are summarised in Table 1. Of the 321 women, 108 (34%) had one or more adverse maternal outcomes by delivery (excluding maternal organ system involvement at the time of diagnosis) and 60 (19%) had adverse fetal outcomes including two perinatal deaths, one to a 23 year old woman with spot urine of 138 mg/mmol at 27 weeks of gestation, the other to a 30 year old with spot urine of 637 mg/mmol at 35 weeks of gestation. There were no maternal deaths.

Table 1.  Characteristics of 321 women with proteinuric pre-eclampsia. Data are expressed as mean [SD], median [range] or n (percentage).
Age (years)30 [5]
‘Booking’ systolic BP (mmHg)115 [11]
‘Booking’ diastolic BP (mmHg)70 [8]
Spot urine protein/creatinine (mg/mmol)78 [30–2075]
Nulliparity233 (73)
Gestational diabetes23 (7)

In this study, 73/321 (23%) were <34 weeks of gestation at diagnosis and 33/321 (10%) were <34 weeks of gestation at delivery. The types of maternal complications at the time of diagnosis and outcomes at delivery are described in Table 2.

Table 2.  Outcomes of 321 pregnancies with proteinuric pre-eclampsia. Data are presented as n (%).
 At diagnosisAt delivery
  • *

    Lack of SGA at time of diagnosis due to lack of scan data.

Severe hypertension89 (28)
Neurological problems19 (6)
Renal insufficiency15 (5)28 (9)
Liver dysfunction19 (6)39 (12)
Thrombocytopenia33 (10)53 (17)
Small for gestational age*59 (18)
Perinatal mortality2 (0.6)

In our study, the median spot protein/creatinine ratio was 78 mg/mmol: 190 women (59%) presented with spot protein/creatinine ratio of 30 to 100 mg/mmol, 100 (31%) with 101 to 500 mg/mmol, 22 (7%) with 501 to 900 mg/mmol of whom two were over 35 years and 9 (3%) above 900 mg/mmol.

Age and spot urine protein/creatinine were the only variables that were significant (P < 0.05) in both univariate and multivariate analyses of adverse maternal outcomes (Table 3).Figure 2 shows that the probability of an adverse maternal outcome increases with increasing age and spot urine protein/creatinine. For example, the probability reaches 0.8 when maternal age is 45 years and spot urine is 600 mg/mmol. For the same probability at 20 years of age, the spot urine would need to be 1120 mg/mmol. As maternal age decreases, an increasing value of spot urine protein/creatinine is required to give the same predicted probability of adverse maternal outcome. The area under the ROC curve for spot urine and age is 0.67 (Fig. 3).

Table 3.  Significant predictors in the logistic regression models for adverse maternal and fetal outcomes. OR odds ratio adjusted for other predictors shown, using logistic regression.
PredictorsOR95% CIP
  1. OR = odds ratio adjusted for other predictors shown, using logistic regression.

  2. CI = confidence interval.

Adverse maternal outcomes
Spot urine protein/creatinine (mg/mmol)1.0031.002–1.004<0.0001
Age (years)1.061.01–1.110.014
 
Adverse fetal outcomes
Spot urine protein/creatinine (log mg/mmol)1.441.08–1.920.013
Booking SBP ≤115 mmHg3.411.77–6.570.0002
Gestation at diagnosis <34 weeks3.601.90–6.82<0.0001
image

Figure 2. Relationship between probability of adverse maternal outcome and spot urine and maternal age.

Download figure to PowerPoint

image

Figure 3. ROC curves for adverse maternal outcomes for age, spot urine and both together.

Download figure to PowerPoint

At all ages, higher spot urine protein/creatinine value increased the probability of an adverse maternal outcome. When the spot urine protein/creatinine ratio was 900 mg/mmol or more for all ages, the positive likelihood ratio (LR+) was 7 or greater. That is, the odds of an adverse maternal outcome are seven times greater in women with a spot urine >900 mg/mmol. In women over 35 years, a protein/creatinine ratio greater than 500 mg/mmol was associated with a LR+ of 7 or more.

In multivariate analysis, spot urine protein/creatinine, booking-in systolic blood pressure and gestation at diagnosis were all significant (Table 3). The probability of an adverse fetal outcome is increased with increasing spot urine. The risk is higher when the gestation at consultation is <34 weeks, or when maternal booking-in systolic blood pressure ≤ 115 mmHg. The primary analysis for predicting adverse fetal outcomes was repeated after excluding twins and the results were not altered.

The area under the ROC curve for adverse fetal outcomes is 0.72. When the spot urine protein/creatinine ratio is 700 mg/mmol, the LR+ is about 4 for all women with gestation at consultation <34 weeks or booking-in systolic blood pressure ≤ 115 mmHg.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Prior studies have shown that proteinuric hypertension has worse maternal and fetal outcomes than non-proteinuric hypertensive pregnancy.6,13–16 This study shows that in women with proteinuric pre-eclampsia, the probability of an adverse maternal outcome increases steadily with both increasing maternal age and increasing spot urine protein/creatinine. The threshold value of spot urine therefore decreases as maternal age increases. Overall, although the spot urine protein/creatinine ratio at diagnosis was found to be statistically significant (P < 0.0001) for predicting adverse maternal outcome, the OR was low (1.003 per mg/mmol) and unlikely to be of clinical significance. This differed at high levels of spot urine. When the spot urine protein/creatinine ratio was 900 mg/mmol or more for all ages, there was an associated clinically important increase in risk of developing adverse maternal outcomes (LR+ of 7 or more). When maternal age was over 35 years, the same increase in risk occurred at a protein/creatinine ratio greater than 500 mg/mmol.

The probability of an adverse fetal outcome was also increased with increasing spot urine protein/creatinine level but maternal age had no influence on this outcome. The risk of an adverse fetal outcome was significantly greater when the gestation at initial presentation was <34 weeks, or when maternal booking-in systolic blood pressure was at or below 115 mmHg. In the presence of these two strong predictors of adverse fetal outcome, spot urine was a statistically significant predictor but added little to the discriminatory power of the model. In this sample, a screening test of women with gestation at consultation <34 weeks or maternal booking-in systolic blood pressure ≤ 115 mmHg yielded a sensitivity of 49/60 = 82% and a specificity of 39%. When the spot urine protein/creatinine ratio is 700 mg/mmol or above, 4 out of 16 women had adverse fetal outcomes. All these four women had gestation at consultation <34 weeks or maternal booking-in systolic blood pressure ≤ 115 mmHg and would therefore be identified by the screening test. Spot urine did not provide incremental value to the test in this sample.

Although the research did not identify a specific spot protein/creatinine ratio that could be used as a definitive screening test for adverse outcomes, the data suggest that finding a high spot protein/creatinine ratio does have some clinical predictive utility for adverse maternal outcomes.

A screening procedure could identify all women with a spot urine ratio above 900 mg/mmol and those over 35 years with a spot urine above 500 mg/mmol as at greatly increased risk of an adverse maternal outcome, as only 10% of our patients have spot protein/creatinine ratio >500 mg/mmol. In this sample, such a test would have a low sensitivity of 11/108 = 10% but a specificity of 100%.

A retrospective study by Schiff et al.,17 demonstrated that no differences in maternal or fetal outcomes were found between pregnancies with marked increases in proteinuria and those with modest or no increases. A recent study which assessed this issue differently demonstrated that the magnitude of proteinuria was not associated with increased morbidity for mother or neonate in pre-eclampsia.18 Newman et al.18 sought to determine whether delayed delivery of pre-eclampsia with massive proteinuria increased maternal or neonatal morbidity, a different research question to our study. They retrospectively reviewed parturients admitted with pre-eclampsia at ≤37 weeks of gestation, divided into three groups of proteinuria (mild, severe or massive) and compared the outcomes. In our study, we analysed spot urine as a continuous variable to give a more sensitive analysis.

It is possible that analyses using other measures of proteinuria may have yielded different results. Detecting proteinuria by routine dipstick analysis has been shown to be inaccurate, giving high false positive19,20 and negative results21 when compared with the reference standard of 24-hour urine measurement. However, 24-hour urine collections are cumbersome, subject to collection error, require good patient compliance and result in greater than 24-hour delay in diagnosis. Several prospective studies have now shown that a spot urine protein/creatinine ratio correlates well with 24-hour urine proteinuria in pregnancy,9,22–25 with two studies giving a value of r= 0.93, P < 0.001.9,22 Although one recent study did not confirm a correlation,26 this test has now been accepted by both the Australasian Society for the Study of Hypertension in Pregnancy (ASSHP) and ISSHP as a reliable measure of proteinuria in pre-eclampsia.7,8 Thus, it is both practical and valid to test whether spot urine protein/creatinine (as a quick and accurate surrogate for the 24-hour urinary protein) at the time of diagnosis of pre-eclampsia helps predict maternal and fetal outcomes.

The limitations of our study must be acknowledged. First, the quality of the data is dependent on diagnostic consistency of practitioners. This is unlikely to be a significant problem in this study as our unit uses one protocol-driven, uniform, standardised management and only two physicians in the unit code all diagnoses. A second limitation is that information abstraction from the HIP database, searching for missing values, and analyses were done by one person (the first author) not blinded to the hypothesis or outcome. Third, selection bias may also arise as the criteria for entry to the HIP database included referral for physician care. It is likely that cases not referred, who have milder disease, would reduce the complication rate but it is general practice in our unit that all proteinuric cases are referred for joint obstetric and physician care. Finally, as 90% of women were ≥34 weeks at delivery, adverse fetal outcomes seen in those <34 weeks may be under-estimated, giving rise to type II errors, delivery at this gestation being known to be associated with higher perinatal mortality and morbidity resulting from prematurity.13,27,28

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

In summary, it is clear that in women with pre-eclampsia there is an association between the degree of proteinuria at the time of diagnosis and subsequent adverse maternal or fetal outcomes.

With increasing proteinuria, there is increased risk of adverse maternal and fetal outcomes. This risk is augmented with increased maternal age. Although the research did not identify a specific spot protein/creatinine ratio to use as a cut-point for a screening test, it is possible to use data from this study to predict the likelihood of adverse maternal and fetal outcomes. For all ages, if the spot urine protein/creatinine ratio is greater than 900 (∼9 g/day) or maternal age over 35 years and protein/creatinine ratio greater than 500 (∼5 g/day), there is a greatly increased likelihood of developing an adverse maternal outcome. However, in practice, the majority of women (over 90%) present with lower proteinuria than this. Therefore, all cases of proteinuric pre-eclampsia should continue to be treated with equal care until new factors are found which allow more reliable prediction of maternal and fetal outcome in pre-eclamptic women.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

The authors would like to thank the obstetricians at St George Hospital for their willing co-operation, Caroline Homer for help with HIP data, Rod Isaacs and Tom Park for assistance with Access & SAS, and family and friends for their continual support.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSIONS
  8. Acknowledgements
  9. References

Accepted 15 June 2004