Dynamics and incidence patterns of maternal complications in early-onset hypertension of pregnancy

Authors


Dr W Ganzevoort, Department of Obstetrics and Gynaecology, Academic Medical Centre, Room H4-205, PO Box 22660; 1100 DD Amsterdam, the Netherlands. Email j.w.ganzevoort@amc.uva.nl

Abstract

Objective  To describe the variable disease expression and the patterns of development of major maternal morbidity and HELLP (haemolysis, elevated liver enzymes and low platelet count) syndrome in women with different subtypes of hypertensive disorders of pregnancy.

Design  Prospective cohort study.

Setting  Two university hospitals, tertiary care centres.

Population  Two hundred and sixteen women participating in a randomised trial of temporising management in early-onset hypertensive disease (PETRA trial). Women were between 24 and 34 completed weeks and had either HELLP syndrome, severe pre-eclampsia, eclampsia or hypertension and fetal growth restriction. Women were delivered in the event of fetal marked heart rate abnormalities, pulmonary oedema, therapy-resistant hypertension or recurrent HELLP syndrome.

Methods  Trial data were reanalysed to assess the time of onset of major maternal morbidity (e.g. pulmonary oedema, liver haematoma), HELLP syndrome and clinical disease. Associations between clinical parameters and prolongation of pregnancy were explored using logistic regression.

Main outcome measures  Diagnosis from admittance to discharge, major maternal morbidity and prolongation of pregnancy.

Results  The median time to delivery or fetal death was 8.2 (range 0.1–44) days. At study entry, 56 women (26%) had more than one diagnosis; this increased to 171 women (79%) by the time of discharge. The incidence of major maternal morbidity (total 26) was 4.2% at 2–4 days after inclusion and a mean of 1.7% (range 0–2%) thereafter per time frame of 3 days. The mean incidence of new or recurrent HELLP syndrome episodes was 5.5% (range 1.9–8.7%) per time frame of 3 days during the first 3 weeks after inclusion.

Conclusions  Pre-eclampsia is a dynamic disease, with extensive overlap of subtypes of the syndrome. Prolongation of pregnancy in early-onset hypertensive disorders results in the development of further HELLP syndrome episodes and reversible major maternal morbidity but may improve perinatal healthy survival.

Introduction

Hypertensive disorders of pregnancy (pre-eclampsia, eclampsia, HELLP [haemolysis, elevated liver enzymes and low platelet count] syndrome and fetal growth restriction) are associated with significant maternal and perinatal morbidities and mortalities.1–4 They are heterogeneous in their presentation and aetiology and might be characterised as a clinical syndrome rather than a single disease entity.5,6

In severe cases at early gestational ages, temporising management may be beneficial for the infant.7–9 Some maternal morbidity is inevitable with this strategy, although the risk of severe complications is low10–13 and these are generally reversible.14 Ideally, the choice of temporising management for each individual woman should be dependent on the balance of estimated risks and benefits for both the mother and infant. While adverse infant outcomes can, to a certain degree, be predicted by gestational age,14–16 the prediction of major maternal complications is difficult and even the severity of disease at admission is only of limited value.14

In a recent randomised trial (PETRA), we studied the effect of plasma volume expansion in women undergoing temporising management for early-onset pre-eclampsia.17,18 There were no significant differences in any of the primary outcomes, and so the treatment and control groups were combined for this post hoc analysis of variations in disease expression after study entry. We also analysed the patterns of development of major maternal morbidity and HELLP syndrome to assess if the short-term prediction of adverse events might be feasible.

Methods

The PETRA trial was a two-centre, randomised, open label trial that consecutively enrolled 216 women at a gestational age between 24 and 34 completed weeks with severe pre-eclampsia, HELLP syndrome or severe fetal growth restriction with associated pregnancy-induced hypertension or moderate pre-eclampsia.18 Inclusion criteria are specified in Table 1.19–21 The study was conducted at the Departments of Obstetrics of the Academic Medical Centre and the VU University Medical Centre in Amsterdam, The Netherlands, between 1 April 2000 and 31 May 2003. Methods have been described in full previously.17,18 In line with the institutional review board approval of both hospitals, informed consent was obtained before inclusion. In summary, women were subjected to a temporising management strategy with plasma volume expansion (treatment group) or a strategy without plasma volume expansion (control group). Maternal characteristics at entry into the trial were comparable between randomisation groups, and all women who entered the trial were deemed eligible for a temporising management strategy. In all cases, management consisted of intensive monitoring of fetal (twice-daily fetal nonstress tests, twice-weekly Doppler ultrasound) and maternal conditions (among others, blood pressure at least four times daily and laboratory testing at least twice weekly). Hypertension was managed with fixed protocols (80% of women receiving one or more medications). Blood pressure values were always estimated from three consecutive readings. Although antihypertensive treatment targets differed somewhat between randomisation groups, in practice, diastolic blood pressure after stabilisation in both types of management was 95 mmHg on average. Magnesium sulphate was prescribed for treatment of eclampsia and in women with clinical signs of imminent eclampsia. Severe headache or abdominal pain in association with severe pre-eclampsia or HELLP syndrome was treated with acetaminophen or morphine. Corticosteroids were administered to 71% of women only if delivery was thought to be imminent, based on maternal disease deterioration or fetal Doppler abnormality. All but three mothers of infants that were delivered alive before 33 weeks had been given corticosteroids. Fetal indications for delivery were repeated decelerations or prolonged low variability on fetal heart rate tracings. Maternal indications were therapy-resistant hypertension, pulmonary oedema and recurrent HELLP syndrome.

Table 1.  Definitions of disease and inclusion criteria and definitions of maternal morbidity
  1. CT, computed tomography; MRI, magnetic resonance imaging.

Disease diagnosis
Eclampsia19Generalised convulsions in pregnancy not caused by epilepsy
Severe pre-eclampsia19Diastolic blood pressure of ≥ 110 mmHg and proteinuria (≥0.3 g/24 hours)
HELLP syndrome20Platelet count <100 × 109/l and aspartate aminotransferase ≥70 units/l and lactate dehydrogenase ≥600 units/l
Fetal growth restriction19,21Estimated fetal weight <10th centile at admission/birthweight at delivery
Pregnancy-induced hypertensionDiastolic blood pressure of ≥90 mmHg without significant proteinuria (<0.3 g/24 hours)
Moderate pre-eclampsiaDiastolic blood pressure of ≥90 mmHg but ≤110 mmHg and proteinuria (≥0.3 g/24 hours)
Major maternal morbidity
Placental abruptionClinical/pathology diagnosis of retroplacental haematoma at delivery
Pulmonary oedemaTachypnoea >40/minute, gas diffusion deficit, compatible chest X-ray
Cerebral haemorrhageIntracerebral bleeding diagnosed by CT scan or MRI
Liver haematomaLiver haematoma diagnosed by ultrasound or CT scan
Severe renal insufficiencyUrine output <500 ml/day, serum creatinine >100 mmol/l, creatinine clearance <20 ml/minute
Severe infectious morbidityClinical diagnosis of sepsis with positive blood cultures
Severe thrombotic morbidityPulmonary embolism, catheter-associated thrombosis
EncephalopathyNeurological deficits of central origin
Disseminated intravascular coagulationThe presence of three or more of the following criteria: low platelet count (<100 000 cells/μl), low fibrinogen concentration (<300 mg/dl), presence of D-dimers (≥40 mg/dl), or prolonged prothrombin time (≥14 seconds) and partial thromboplastin time (≥40 seconds)

Amounts of infused fluids were significantly different between randomisation groups, and there was a significant difference in change of haemoglobin count from baseline in the control group versus the treatment group. However, neonatal neurological abnormality at term age (primary outcome), neonatal survival and morbidity and maternal morbidity (secondary outcomes) were similar between groups. As the primary and secondary outcomes were comparable between both groups, we used the combined data for the present exploration.

As overlap of the type of hypertensive disease at inclusion would hamper statistical calculation, we regrouped women by diagnosis at inclusion (see definitions in Table 1). First, all women with HELLP syndrome at admission were selected (irrespective of whether they also fitted the criteria for one of the other groups). Subsequently, from the remaining women, all those with severe pre-eclampsia at admission were selected. The remaining women—those with pregnancy-induced hypertension or moderate pre-eclampsia with severe fetal growth restriction—formed the third group.

The independent trial monitor committee consisted of two gynaecologists (W.M.A. and A.T.J.I.G) and one paediatrician (R.J.B.J.G.) who were not involved in the management of the study. The committee reviewed and classified all cases of major maternal morbidity (criteria in Table 1), which included eclampsia after inclusion and HELLP syndrome recurrence. Recurrent HELLP syndrome was defined by the reappearance of the typical laboratory abnormalities (Table 1) after they had returned to normal values for at least 2 days. Because magnesium sulphate or morphine was only administered in women with clinical symptoms of disease (headache, right upper quadrant abdominal pain), the time of administration of these medications was determined and defined as ‘clinical disease’.

For this analysis, the study endpoints were the following: diagnosis from study entry to discharge, major maternal morbidity and prolongation of pregnancy. Because the aim of this analysis was to evaluate the occurrence of maternal morbidity during temporising management, separate analyses were performed for the period between inclusion and delivery or fetal death and the period thereafter. Although all fetuses at study entry had reassuring fetal monitoring, fetal death was accepted in a restricted number of cases because of very early gestational age and low estimated fetal weight.

For analysis of the incidence patterns of major maternal morbidity and clinical disease, the time interval between inclusion and delivery was divided in time frames. The first time frame covered the first 24 hours of admission. The subsequent time frames were periods of 3 days. Major maternal morbidity, HELLP syndrome or clinical disease were further classified as ‘new’ in the time frame in which they first developed (or reoccurred) or ‘old’ in subsequent time frames if they were still present. Morbidity in the postpartum period was combined in one time frame.

The statistical analysis was performed using Pearson chi-square tests (two sided) and nonparametric Mann–Whitney tests. The association between clinical parameters on inclusion and prolonged temporisation, defined as more than the median number of days of prolongation after admission, was assessed by stepwise backward logistic regression analysis (P [in] 0.05 and P [out] 0.10). The randomisation allocation was forced to remain in the logistic regression model. Other factors entered were centre, use of antihypertensive medication, use of corticosteroid therapy, chronic hypertension, HELLP syndrome, severe pre-eclampsia, fetal growth restriction, the presence of antiphospholipid antibodies, or hereditary thrombophilias, gestational age at inclusion per week increment, estimated fetal weight (per 100 g of increase) or estimated fetal weight ratio (ultrasound-estimated fetal weight divided by the median weight for gestational age), haemoglobin count, creatinine clearance, blood pressure or pulse at inclusion more than the median (7.8 mmol/l, 58 μmol/l, 105 mmHg, 80/minute, respectively) and body mass index (BMI) more than 25 kg/m2. The analysis was performed using SPSS 12.0.2 (SPSS Inc., Chicago, IL, USA). Differences were considered statistically significant at P < 0.05.

Results

Demographical and clinical characteristics at admittance were comparable between the three groups (HELLP syndrome, severe pre-eclampsia, others), except that women with the HELLP syndrome at inclusion were more likely to be Caucasian and nulliparous and less likely to have chronic hypertension or a high BMI. Estimated fetal weight, by definition, was lower in the fetal growth restricted group (Table 2).

Table 2.  Characteristics at entry into the trial, separately presented for inclusion groups
 Total, 216HELLP, 54Severe pre-eclampsia, 80Fetal growth restriction, 82
  • PIH, pregnancy-induced hypertension.

  • *

    PIH or pre-eclampsia.

  • **

    Difference statistically significant at P < 0.05.

  • Continuous data are reported as median (range), categorical data as number (percentage). Estimated fetal weight ratio is estimated fetal weight divided by median weight for gestational age of the reference population.

Demographical characteristics
Maternal age (years)30 (18–41)30 (22–41)31 (20–40)29 (18–39)
Multiparity65 (30)11 (20)**31 (39)23 (28)
Multipara with history of PIH*38 (58)5 (45)19 (61)14 (61)
PIH* in first-degree relatives73 (34)22 (41)26 (33)25 (31)
BMI > 25 kg/m294 (44)16 (30)**43 (54)35 (43)
Chronic hypertension69 (32)11 (20)**38 (48)20 (24)
Caucasian156 (72)49 (91)**48 (60)59 (72)
Clinical characteristics at inclusion
Gestational age (weeks.days)30.0 (24.1–33.6)30.3 (24.1–33.6)30.3 (24.2–33.5)29.0 (24.3–33.2)
Estimated fetal weight (g)1101 (298–2401)1214 (427–2226)1231 (518–2401)886 (298–1721)**
Estimated fetal weight ratio0.72 (0.32–1.17)0.77 (0.48–1.03)0.80 (0.41–1.17)0.65 (0.32–0.84)**
Systolic blood pressure (mmHg)160 (100–220)160 (100–200)170 (140–220)150 (120–180)
Diastolic blood pressure (mmHg)105 (75–140)100 (75–125)110 (90–140)95 (90–115)
Heart rate (beats/minute)80 (44–110)75 (56–104)80 (56–110)76 (44–108)
Haemoglobin count (mmol/l)7.8 (6.0–9.9)7.8 (6.2–9.4)7.6 (6.0–9.8)7.9 (6.0–9.9)
Eclampsia before inclusion5 (2.3)3 (5.6)2 (2.5)0 (—)
Randomised to treatment group111 (51)27 (50)43 (54)41 (50)

At inclusion, 56 women (26%) matched more than one diagnosis. If women were reclassified by diagnosis at discharge, 171 women (79%) matched more than one diagnosis. Figure 1 demonstrates the difference in diagnostic subgroup classification in the study population at study entry and at discharge. At study entry, 25% had a diagnosis of HELLP syndrome. This proportion had increased to 43% (93/216) by the time of discharge. Similarly for severe pre-eclampsia, the proportion increased from 45% (98/216) at study entry to 74% (159/216) at discharge. Initially, 58% (125/216) of participants had a diagnosis of fetal growth restriction (ultrasound assessment), but this had increased to 92% (198/216) by the time of delivery (birthweight).

Figure 1.

Diagnoses on inclusion (left) and total up to discharge (right). Absolute numbers of women in each subgroup presented, with number of cases of eclampsia in brackets. PIH, pregnancy-induced hypertension; FGR, fetal growth restriction.

Fifty-four women (25%) had HELLP syndrome on admission (Figures 1 and 2, Table 3). In 14 of these women (26%), the HELLP syndrome did not improve before delivery. It improved in the remaining 40 women, although 10 of these women (10/40 = 25%) had a recurrence before delivery. The duration of pregnancy prolongation in the nonimproving group (median 2 days; range 0–6) was significantly shorter than that in the remainder. This, in turn, was comparable with the other inclusion groups (median 11 days [2–44]). In the total study population, the incidence of new or recurrent HELLP syndrome episodes was 5.5% (range 1.9–8.7%) per time frame of 3 days during the first 3 weeks after inclusion. At 3 weeks after inclusion, 34 women (16%) remained undelivered, and of this group, one woman developed HELLP syndrome at 30 days after inclusion (Figure 2A). Nine women (4.2%) had their first HELLP episode postpartum, and three women had recurrent HELLP postpartum. These numbers were similar between the three inclusion groups.

Figure 2.

HELLP syndrome (A), major maternal morbidity (B) and clinical disease (C) (defined as the clinical need for the administration of morphine or magnesium sulphate) presented for the day of admission and thereafter for 3-day periods until delivery and subdivided in fractions of new-onset symptoms, persistent ‘old’ disease, absence of disease and delivered women. The vertical scale is proportional.

Table 3.  Morbidity after inclusion, separately presented for groups with diagnosis at admission
 Total, 216HELLP, 54Severe pre-eclampsia, 80Fetal growth restriction, 82
  • *

    Some women had more than one case of morbidity.

  • **

    Others versus HELLP or severe pre-eclampsia, P < 0.05.

  • Data are presented as n (%) or median (range) as appropriate.

Maternal outcomes
HELLP persistent until delivery14 (6)14 (26)00
HELLP after inclusion before delivery or fetal death (primary/recurrent)*43 (20)10 (19)18 (23)15 (18)
HELLP after delivery or fetal death (primary/recurrent)*12 (6)4 (7)2 (3)6 (7)
Major maternal morbidity26 (12)8 (15)8 (10)10 (13)
 Eclampsia after inclusion*4202
 Pulmonary oedema*8341
 Placental abruption*5104
 Liver haematoma*1001
 Severe infectious morbidity*3021
 Severe thrombotic morbidity*2011
 Encephalopathy*5212
 Cerebral haemorrhage0000
 Severe renal insufficiency0000
 Disseminated intravascular coagulation0000
 Other*2110
Clinical disease**91 (42)32 (59)38 (48)21 (26)
Fetal death20 (9)5 (9)6 (8)9 (11)
Prolongation of pregnancy
Days from inclusion to delivery or fetal death8.2 (0.1–44)8.4 (0.5–44)8.2 (0.1–44)8.2 (0.5–42)
Days from later/recurrent HELLP to delivery or fetal death2.8 (0.3–26)3.0 (0.3–22)1.5 (0.5–26)4.0 (0.5–19)

Twenty-six women had major maternal morbidity (12%; Figure 2) unrelated to diagnosis at admission. These are set out in Table 3. The two cases of ‘other’ major morbidity were a woman with peripartum cardiomyopathy and another with a hypocalcaemic delirium that developed within hours after the start of magnesium sulphate therapy (standard dose, magnesium concentration in the therapeutic range).22 There were no maternal deaths.

Of the women in the ‘HELLP group’, 14 (26%) were delivered without improvement shortly after inclusion (median 2 days; range 0.5–6). The maternal condition was the indication for delivery in four women; in two women, delivery was induced after fetal death, and five women in this group also had major morbidity. The remaining 40 (74%) women with HELLP improved after inclusion and had a median interval between inclusion and delivery of 12 days (range 2–44). In this group, ten women had a recurrent HELLP syndrome before delivery (without major morbidity) and three women developed major morbidity during temporising treatment.

The five women who were included after eclamptic convulsions had been stabilised before randomisation and deemed fit for further prolongation of pregnancy, initially to allow corticosteroid treatment for fetal lung maturation. Prolongation of pregnancy was 4 days (range 1–18). Additional morbidity was observed in only one woman who developed temporary visual disturbances after 18 days, and pregnancy was terminated.

In Tables 4 and 5, the individual cases with major maternal morbidity are described. Of those with major maternal morbidity before delivery, the onset occurred less than 24 hours before delivery in 15 of 20 women (75%). Of the remaining five women, fetal death had occurred in three women, and temporising management was undertaken to stabilise the maternal condition before delivery. In another woman, eclampsia had occurred, but after stabilisation, the maternal and fetal conditions allowed delay of delivery to administer corticosteroid therapy. In the last case, mild pulmonary oedema was diagnosed with shortness of breath and a compatible chest X-ray. However, oxygen saturations were good with oxygen therapy, and the woman was managed conservatively for 7 days, at which time caesarean section was performed.

Table 4.  Individual characteristics of women with major maternal morbidity before delivery
Case numberGestational age at inclusion (weeks)Gestational age at morbidity (weeks)Gestational age at delivery (weeks)DiagnosisManagement
732 + 234 + 334 + 3Placental abruptionEmergency caesarean section
1727 + 628 + 329 + 3Pulmonary oedemaConservative
2830 + 330 + 530 + 5EclampsiaEmergency caesarean section
4031 + 532 + 533 + 0EclampsiaInduction of labour after stabilisation
4328 + 028 + 228 + 3Placental abruption (leading to fetal death)Delivery
5028 + 229 + 029 + 0Infected thrombotic tip of deep venous lineDelivery
5228 + 029 + 129 + 2Pulmonary oedemaDelivery
5530 + 331 + 331 + 3Pulmonary oedemaDelivery
5828 + 630 + 130 + 1Pulmonary oedemaDelivery, short period of ventilation
7431 + 231 + 431 + 4Hypocalcaemic deliriumDelivery, supplementation
9732 + 633 + 033 + 0Placental abruptionEmergency caesarean section
11829 + 031 + 131 + 2Pulmonary oedemaDelivery
12727 + 130 + 430 + 4Placental abruptionEmergency caesarean section
12829 + 530 + 130 + 1Pulmonary oedemaDelivery, short period of ventilation
13127 + 127 + 527 + 6EncephalopathyDelivery
14230 + 030 + 630 + 6Placental abruptionEmergency caesarean section
16425 + 026 + 226 + 6Pulmonary oedema after fetal deathConservative, delivery
17324 + 126 + 327 + 1Eclampsia and encephalopathy after fetal deathMagnesium sulphate, delivery
19128 + 030 + 430 + 4EncephalopathyMagnesium sulphate, delivery
20224 + 526 + 126 + 5Infected deep venous line, adult respiratory distress syndromeAntibiotics, ventilation, delivery after fetal death
Table 5.  Individual characteristics of women with major maternal morbidity after delivery
Case numberGestational age at inclusion (weeks)Gestational age at delivery (weeks)Days after deliveryDiagnosisManagement
2431 + 233 + 4+ 2Pulmonary oedemaConservative
5630 + 230 + 5+ 3Eclampsia and encephalopathyMagnesium sulphate
7930 + 631 + 5+ 2CardiomyopathySupportive care
10528 + 629 + 3+ 2, + 14Subcapsular liver haematoma pulmonary embolism from deep vein thrombosisAnticoagulant therapy
16029 + 029 + 3+ 1EncephalopathyMagnesium sulphate
17426 + 330 + 0+ 14UrosepsisAntibiotics

The incidence of major maternal morbidity was 4.2% at 2–4 days after inclusion and 1.7% (range 0–2%) thereafter per time frame of 3 days (Figure 2B). In 6 of the 26 women with major maternal morbidity or eclampsia (23%), the onset was after delivery.

All but three mothers who were delivered before 32 weeks had been given corticosteroids for fetal lung maturation. No difference in incidence of major morbidity (including HELLP syndrome) was observed between mothers who had been treated with corticosteroids at admission (12/94 = 13%), mothers who received corticosteroids later (8/59 = 14%) or mothers who were not given corticosteroids (6/63 = 10%).

During the first day after inclusion, 34 women (15.7%) were treated with magnesium sulphate or morphine (clinical disease, Figure 2). In the second time frame, 66 undelivered women (30.5%) received this medication, which was newly prescribed in 32 women (14.9%). In the third time frame, these figures were 41 (24.6%) and 9 (5.4%), respectively. Thereafter, the mean percentage of undelivered women who were newly treated with magnesium or morphine was 3.4% (range 0–5.6%) for each time frame during the remaining period until delivery (Figure 2C).

Fetal deaths occurred in 20 cases (9%). In 17 cases, the parents, neonatologists and obstetricians had refrained from intervention based on a poor prognosis because of a low gestational age and estimated fetal weight at the time fetal distress became apparent. In three cases, fetal death had been unexpected, in one case because of placental abruption and in two cases because of sudden placental insufficiency. Adverse perinatal outcomes and their associations are described elsewhere.14

Median pregnancy prolongation in the trial population was 8.2 days (range 0.1–44) and was comparable between the subgroups. In logistic regression analysis, prolongation of >7 days was positively associated with higher estimated fetal weight and higher haemoglobin on admission, while it was negatively associated with higher gestational age, antihypertensive treatment and higher serum creatinine on admission and with allocation to plasma volume expansion strategy (Table 6). Sixty-five percent of all cases had a correct prediction by the model (59% ≤7 days of interval and 71% >7 days of interval). The receiver operating characteristic curve had poor predictive power with an area under the curve of 0.72 (95% CI 0.65–0.79) (Figure 3). Adverse perinatal outcomes and their associations are described in more detail elsewhere.14

Table 6.  Parameters on inclusion that contributed significantly to a model predicting an interval of >7 days between inclusion and delivery (or fetal death), compared with a shorter interval, calculated by logistic regression analysis. Treatment allocation was forced to remain in the model.
 OR95% CI
  1. CI, confidence interval; OR, odds ratio.

Plasma volume expansion0.550.31–0.99
Gestational age per 1-week increase0.650.52–0.81
Estimated fetal weight per 100-g increase1.301.14–1.49
Antihypertensive treatment0.450.25–0.81
Haemoglobin ≥ 7.8 mmol/l1.820.99–3.36
Serum creatinine ≥ 58 μmol/l0.560.31–1.02
Figure 3.

Receiver operating characteristic graph of prediction model for prolongation of pregnancy >7 days.

Discussion

In the literature, HELLP syndrome is frequently associated with increased maternal morbidity and mortality.2,3 In this study, however, 74% of the women with HELLP syndrome improved after inclusion and had a low risk of major complications, which did not differ from the other inclusion groups. All three inclusion groups (HELLP syndrome, severe pre-eclampsia or fetal growth restriction with pregnancy-induced hypertension) had a similar rate of major complications, although the median prolongation of pregnancy was of a similar duration. A similar pattern was observed for the incidence of administration of morphine and/or magnesium sulphate, a proxy for clinical subjective disease. Overall risks were in accordance with most publications, although these publications did not specify the time of occurrence and the association with the duration of prolongation.4,12,23

Classification of disease on admission was conspicuously absent from the model predicting duration of prolongation of pregnancy. Apparently, a larger fetal weight, when adjusted for gestational age, allowed a longer period of prolongation, possibly because of the clinician’s assessment of a reduced fetal risk. The negative effect of higher gestational age was probably because of an increasing preference for delivery at a higher gestational age. The opposite effects of higher haemoglobin and serum creatinine concentrations might be related to oxygen transfer and severity of disease. The negative effect of plasma volume expansion in this respect was detected in univariate analysis in an earlier publication.18 The presence of thrombophilias has been linked to the development of severe maternal morbidity14 but was not associated with the length of prolongation of pregnancy. Overall, the performance of the model is poor and not clinically useful for selection of women who will benefit from temporising management.

During the median of 8.2 days of pregnancy prolongation and during the period after delivery, 26 women had an episode of major morbidity or eclampsia and three unexpected fetal deaths occurred. Whether maternal morbidity would have been prevented by a strategy that is aimed at immediate delivery after stabilisation of maternal disease is uncertain.13 The three unexpected fetal deaths could have been prevented through immediate delivery, but this should be balanced with the increased gestational age in the entire group that, we speculate, may have halved neonatal morbidity and mortality.7–9,24 In a separate analysis of this cohort, with each increasing week of gestational age, the odds ratio was 0.42 (95% CI 0.33–0.54) for adverse infant outcome (perinatal death, severe neonatal morbidity).14 The incidence of accepted perinatal death is higher than that in other randomised studies of temporising management, largely because of the fact that other studies included women only after 28 weeks,8,9 whereas this study included women from 24 weeks.

Diagnostic criteria define the subclasses within the clinical spectrum of hypertensive disorders. These subclasses are not necessarily pathophysiologically distinct. In our opinion, the overlap of diagnostic labels within women, which even increases during disease progression, supports this notion. This is further stressed by the lack of association of the labels with baseline characteristics and outcomes. The clinical distinction between clinically labelled categories is therefore artificial as it suggests a knowledge of causality that does not exist in the literature. To some extent, therefore, it is irrelevant for clinical practice. In our view, trials and cohort studies should include women with all subclasses of disease that are deemed severe (Table 1). However, subgroup analyses and careful phenotype description remain important, as the quest for understanding the underlying pathophysiology continues.

It is difficult to find other studies that address the dynamic nature of the disease. Cohort studies are retrospective in nature and label cases in hindsight, whereas prospective trial populations label patients at inclusion and describe the onset of new diagnoses irrespective of timing. The study presented here allows a unique view of the dynamics of hypertensive disorders of pregnancy and the time-dependent onset of complications during a temporising management strategy.

It is uncommon to perform temporising management in women with eclampsia. However, in our experience, after successful stabilisation at low gestational ages, intervention may be safely postponed in individual cases to allow for infant maturation. There is a comparable discussion about whether temporising management is appropriate in HELLP syndrome. Several cohort studies, however, have shown that it may be feasible.4,12,13,23

This study has showed that complications from early-onset hypertensive disorders of pregnancy occur in a more or less steady rate during prolongation of pregnancy and that prediction of complications or duration of prolongation is not feasible. These findings may support advocates of a management strategy that aims to stabilise and deliver as it may prevent some maternal complications and unintentional fetal deaths. However, as all maternal complications were reversible, and prolongation of pregnancy possibly diminishes neonatal morbidity and mortality, it remains uncertain which management option benefits the mothers and their infants most. In our opinion, the results of our study provide sufficient uncertainty in this respect to design a randomised trial of temporising management versus stabilisation and delivery.

Conclusion

Pre-eclampsia is a highly dynamic disease, with extensive overlap of subtypes of the syndrome. The separate presentation of different subgroups of disease should be used with caution. Prolongation of pregnancy in early-onset hypertensive disorders of pregnancy by adhering to a careful temporising strategy allows onset of new HELLP syndrome episodes, and reversible major maternal morbidity, but may improve overall perinatal healthy survival.

Funding

The Dutch Health Care Insurance Board funded the study (grant number OG98-021).

Appendix

PETRA is the acronym for Pre-eclampsia Eclampsia TRial Amsterdam. We are indebted to the participants of the trial.

PETRA investigators

Otto P Bleker MD, PhD; Kees Boer MD, PhD; Gouke J Bonsel MD, MPH, PhD; Willem PF Fetter MD, PhD; Wessel Ganzevoort MD; Herman P van Geijn MD, PhD; Laila de Groot PT, PhD; Johanna HK Joosten MSc; Joke H Kok MD PhD; Joris AM van der Post MD, PhD; Annelies Rep MD; Petra EM van Schie PT; Veronique AM Schaaf PT; Loekie van Sonderen MD; Johanna IP De Vries MD, PhD; Aleid G Van Wassenaer MD, PhD; and Hans Wolf MD, PhD.

Trial monitoring committee

Willem M Ankum MD, PhD; Reinoud JBJ Gemke MD, PhD; and Attie TJI Go MD.

Department of Obstetrics and Gynaecology (H.P.vG., J.H.K.J., A.R., J.I.P.DV., A.T.J.I.G.), Department of Neonatology (W.P.F.F.), Department of Movement Sciences (L.dG., J.I.P.DV.), Department of Physiotherapy (P.E.M.vS., V.A.M.S.), Department of Pediatrics (R.J.B.J.G.), VU University Medical Centre, Amsterdam, The Netherlands; Department of Obstetrics and Gynaecology (O.P.B., K.B., W.G., J.A.M.vdP., H.W., W.M.A.), Department of Social Medicine – Public Health Epidemiology (G.J.B.), Department of Neonatology (J.H.K., L.vS.), Academic Medical Centre, Amsterdam, The Netherlands.

Ancillary