Can we predict recurrence of pre-eclampsia or gestational hypertension?

Authors


Prof MA Brown, Department of Renal Medicine, St George Hospital, Kogarah, New South Wales 2217, Australia. Email mbrown@unsw.edu.au

Abstract

Objective  To estimate the rates of recurrence of pre-eclampsia or gestational hypertension in a subsequent pregnancy and to determine factors predictive of recurrence.

Design  Retrospective cohort study.

Setting  St George Public and Private Hospitals, teaching hospitals without neonatal intensive care units.

Participants  A total of 1515 women with a diagnosis of pre-eclampsia or gestational hypertension between 1988 and 1998 were identified from the St George Hypertension in Pregnancy database, a system designed initially for ensuring quality outcomes of hypertensive pregnancies. Of these, 1354 women were followed up, and a further 333 records from women coded as having a normal pregnancy during that period were selected randomly as controls.

Main outcome measures  Likelihood of recurrent pre-eclampsia or gestational hypertension and clinical and routine laboratory factors in the index pregnancy predictive of recurrence of pre-eclampsia or gestational hypertension.

Methods  The index cases from our unit’s database were linked to the matched pregnancy on the State Department of Health database, allowing us to determine whether further pregnancies had occurred at any hospital in the State. The outcome of these pregnancies was determined by review of medical records, using strict criteria for diagnosis of pre-eclampsia or gestational hypertension.

Results  Almost all women with a normal index pregnancy had a further normotensive pregnancy. One in 50 women hypertensive in their index pregnancy had developed essential hypertension by the time of their next pregnancy. Women with pre-eclampsia in their index pregnancy were equally likely to develop either pre-eclampsia or gestational hypertension (approximately 14% each), while women with gestational hypertension were more likely to develop gestational hypertension (26%) rather than pre-eclampsia (6%) in their next pregnancy. Multiparous women with gestational hypertension were more likely than primiparous women to develop pre-eclampsia (11 versus 4%) or gestational hypertension (45 versus 22%) in their next pregnancy. Early gestation at diagnosis in the index pregnancy, multiparity, uric acid levels in the index pregnancy and booking blood pressure parameters in the next pregnancy significantly influenced the likelihood of recurrence, predominantly for gestational hypertension and less so for pre-eclampsia. No value for these parameters was significant enough to be clinically useful as a discriminate value predictive of recurrent pre-eclampsia or gestational hypertension.

Conclusions  Approximately 70% of women with pre-eclampsia or gestational hypertension will have a normotensive next pregnancy. The highest risk group for recurrent hypertension in pregnancy in this study was multiparous women with gestational hypertension. No readily available clinical or laboratory factor in the index pregnancy reliably predicts recurrence of pre-eclampsia.

Introduction

Hypertension occurs in about 10% of all pregnancies,1 about half involving pre-eclampsia, half having gestational hypertension and 1–2% having chronic hypertension with or without pre-eclampsia. Pre-eclampsia imposes a higher risk of fetal death than the other hypertensive disorders, and both pre-eclampsia and gestational hypertension impose a greater risk of fetal growth restriction than in normal pregnancy. Importantly, pre-eclampsia poses an increased risk for maternal morbidity and rarely mortality. Recent data have shown that both pre-eclampsia and gestational hypertension increase the long-term risks of death due to cardiovascular disease,2 making pre-eclampsia a ‘life-long disorder’.3

Although the long-term prognosis is of major importance, the main question after delivery from women who have had pre-eclampsia or gestational hypertension is whether or not this will recur in their next pregnancy. Estimates on this vary, probably influenced by race and other local factors.4–9 Further factors determining the accuracy of these estimates include the precision of both the index pregnancy diagnosis and the diagnosis reported in any subsequent pregnancy. Some studies have reported outcomes only for primiparous women to increase the accuracy of diagnosis of pre-eclampsia in the index pregnancy. While this is a logical approach, multiparous women who have had pre-eclampsia or gestational hypertension also wish to know their likely outcome if they were to have a further pregnancy. Finally, even though the incidence of hypertensive disorders of pregnancy in the next pregnancy has been studied occasionally, few attempts have been made to determine whether specific clinical or laboratory factors in the index pregnancy would have been predictive of the likely outcome in the subsequent pregnancy.

Against this background, the aims of this study were the following:

  • 1To confirm that women with pre-eclampsia or gestational hypertension have an increased risk of developing these disorders in a subsequent pregnancy compared with women who had a normotensive pregnancy.
  • 2To estimate the rates of pre-eclampsia, gestational hypertension or essential hypertension in the next pregnancy after an index pregnancy complicated by pre-eclampsia or gestational hypertension.
  • 3To determine whether recurrent pre-eclampsia or gestational hypertension differs according to parity in the index pregnancy and whether it can be predicted by interpregnancy interval, gestation at presentation or delivery in the index pregnancy, severity of maternal or fetal outcomes in the index hypertensive pregnancy or by other clinically accessible demographics from the index pregnancy.

The study was approved by the Ethics Committee of New South Wales (NSW) Health Department (2001/09/014 Brown) and the South East Health Human Research Ethics Committee Southern Section (99/67 Brown) as well as each of the participating area health services.10

Methods

A database of women referred for management of hypertensive disorders of pregnancy has been maintained at the St George Hospital Obstetrics and Renal Medicine Departments since 1988. During this time, all cases have been managed jointly by referring obstetricians and only two renal physicians (M.A.B. and G.M.) using a single departmental protocol (available at www.stgrenal.unsw.edu.au). The general policy of the obstetrics department is that women with pre-eclampsia and any woman with gestational hypertension requiring antihypertensive therapy (persistent systolic blood pressure [SBP] ≥ 160 mmHg and/or diastolic blood pressure [DBP] ≥ 90 mmHg) are referred for joint obstetric and physician care. At discharge from hospital, data are recorded, including the diagnosis at delivery, maternal and fetal outcomes and routine laboratory data prior to delivery. This information is then used to track quality indicators within the department and to convey pregnancy records to the woman’s GP and/or physician for follow up as necessary.

Pre-eclampsia is defined according to the criteria of the International Society for the Study of Hypertension in Pregnancy11 using both clinical diagnosis and their research diagnosis (relying on the strict inclusion of proteinuria) to diagnose pre-eclampsia. A clinical diagnosis of pre-eclampsia was made when de novo hypertension was accompanied by proteinuria, i.e. proteinuric pre-eclampsia, or other evidence of multisystem disease (defined below) such as renal insufficiency, elevated liver enzymes, neurological abnormalities such as hyperreflexia with clonus or haematological abnormalities such as thrombocytopenia or haemolysis, i.e. nonproteinuric pre-eclampsia. This means that the definition of clinical pre-eclampsia included women with proteinuria. In this study, we have described the recurrence rates according to both definitions. Gestational hypertension refers to de novo hypertension without any other feature of pre-eclampsia.

The final diagnosis is checked by one of the two renal physicians (M.A.B./G.M.) before entry into this database. Therefore, women entered into the database include women with pre-eclampsia, gestational hypertension, chronic (usually essential) hypertension and women with essential hypertension and superimposed pre-eclampsia. Other relevant data routinely recorded in the index pregnancy include parity, maternal age, twins, maternal diabetes (usually gestational) and gestation at diagnosis and at delivery. Also recorded is evidence of maternal organ dysfunction: increased aspartate transaminase > 50 iu/l, creatinine > 90 μmol/l, hyperuricaemia ≥ 0.35 mmol/l, thrombocytopenia—platelet count < 150 × 109/l, neurological signs (including eclampsia), proteinuria, use of antihypertensive medications prior to delivery, need for convulsion prophylaxis and birthweight and whether small for gestational age (SGA). Severe hypertension was any episode of SBP ≥ 170 mmHg and/or DBP ≥ 110 mmHg.

Hypertension was diagnosed by persistently elevated sphygmomanometer mercury blood pressure (BP) ≥ 140 mmHg systolic and/or ≥90 mmHg diastolic after overnight rest in hospital or as an average BP following a prolonged stay (4–6 hours) in our Day Assessment Unit. Proteinuria is defined as a spot protein/creatinine ratio of ≥30 mg/mmol creatinine12 or, in the early years of this study, a 24-hour urine protein of ≥300 mg/day. In the unusual case of neither of these data being available, proteinuria was only accepted if ≥2+.13

Data

The medical records number and names of women recorded as having pre-eclampsia or gestational hypertension between 1988 and 1998 in this database (referred to as the index pregnancy) were sent to the NSW Department of Health to examine linkage between the index cases and women listed in the NSW Midwives Data Collection (MDC). This latter database is a collection of birth details and pregnancy outcomes of all women delivered in NSW, Australia, since the early 1980s. For the purpose of this study, a further 333 records (based on our expectation of having records for about 660 index pregnancies) from women coded as having a normal pregnancy in our hospital were randomly selected by year of delivery, i.e. about 30 cases per year. Once a subsequent pregnancy in these women was confirmed, they were included in the reference group for likely outcome in subsequent pregnancy for normal index pregnancy.

These index cases from our unit’s database were linked to the matched pregnancy on the NSW MDC database. It was then possible to determine whether women in our unit’s database had further pregnancies, using the NSW MDC database. Where no further pregnancy was recorded, it was assumed that no further pregnancy had occurred; the emigration rate from NSW for women in this age group being approximately 3.8% per annum (Australian Bureau of Statistics). This linkage analysis was performed as at 2002, thereby resulting in a follow-up period between 4 and 14 years.

The diagnosis at delivery recorded on the MDC for subsequent pregnancy was not accepted a priori, and all these records were reviewed individually, requiring examination of records at 30 different hospitals throughout NSW, with the majority at our hospitals. The primary outcome measure of the next pregnancy was the maternal outcome, defined using the same criteria as for the index pregnancy, with the potential outcomes being pre-eclampsia, gestational hypertension, essential hypertension, essential hypertension plus pre-eclampsia, or normal pregnancy. For this purpose, pre-eclampsia in the next pregnancy was as defined for the index pregnancy—proteinuric pre-eclampsia in the presence of de novo proteinuria or nonproteinuric pre-eclampsia if hypertension without proteinuria was accompanied by evidence of maternal organ dysfunction. Our analysis focused only on outcomes of the next pregnancy; there were too few further pregnancies to allow a meaningful analysis of the outcomes of later pregnancies in women who had more than one subsequent pregnancy.

Calculations based on the number of women (759) in this database with an outcome pregnancy and a recurrence rate of approximately 15% meant that we achieved study power of 90% for a 95% confidence interval between 3% (requirement 544 women) and 2% (requirement 1224 women) of the point estimate.

Statistical methods

Analysis of the data was performed using the S-plus package. Statistical methods included two-sample t tests and analysis of variance for comparing quantitative patient data across various hypertension subgroups. Comparison of qualitative variables was carried out using tests of proportions or chi-square tests for contingency tables. Logistic regression analysis was used for predicting the probability of various types of hypertension in the outcome pregnancy using clinical measurements obtained at the index and outcome pregnancies. The variables named in the rows of Table 1 were considered as potential predictors for the logistic regression. Smoking was excluded because of the large proportion of missing data, and all forms of diabetes were considered as one for this analysis. Variables were excluded until only those significant at the 5% level were included in the final model. There were some differences in the percent missing for gestational hypertension versus pre-eclampsia cohorts; we tested this further and showed that for all variables for which the percentage of missing data was at least 5%, there was no significant association between outcome diagnosis and tendency for the predictive variable to be missed. Hence, the impact of missing data was unlikely to bias the results of the logistic regression.

Table 1.  Data of index pregnancy for 1354 women with pre-eclampsia or gestational hypertension and 333 women with normal index pregnancy during 1988–98
 Percentage of missing data for cohortPre-eclampsia—clinical (n= 705)Gestational hypertension (n= 649)P value, Pre-eclampsia—clinical versus gestational hypertensionNormal (n= 333)P value; pre-eclampsia—clinical, gestational hypertension and normal groups are compared
  1. ‘Pre-eclampsia—clinical’ is pre-eclampsia diagnosed according to its clinical definition (see text for definition of maternal organ dysfunction). For age, SBP, DBP, gestation at diagnosis and delivery, the data are mean (SD). Significance levels are from analysis of variance procedures. Tests of proportion are based on the appropriate chi-square test for independence. ‘NA’ refers to a test of significance that was not performed due to very low expected cell counts. For diabetes, the significance level refers to a test using three categories: pre-pregnancy diabetes, gestational diabetes and no diabetes. Percentages are determined as the fraction of those for whom information was available.

  2. ‘NS’ refers to a P value that is not significant; i.e. P≥ 0.05.

Primiparous (%)0.57774NS62<0.0001
Smoker (%)491317NS18NS
Age (years)029 (5)29 (5)NS27 (5)<0.0001
First-trimester SBP (mmHg)5118 (11)120 (12)<0.05112 (11)<0.0001
First-trimester DBP (mmHg)571 (8)73 (8)<0.0167 (7)<0.0001
Pre-pregnancy diabetes (%)201.40.5NS0<0.01
Gestational diabetes (%)6.25.02.4
Gestation at diagnosis (weeks)036 (3)37 (3)<0.001NA<0.0001
Gestation at delivery (weeks)037 (3)38 (2)<0.00139 (2)<0.0001
SGA (%)2267<0.0019<0.0001
Proteinuria (%)2630<0.0010NA
Liver disease (%)29210.6<0.0010NA
Hyperuricaemia (%)297351<0.0010NA
Platelets < 150 × 109/l (%)25300.1<0.0016NA
Platelets < 100 × 109/l (%)2560<0.0010NA
Renal impairment (%)29230.4<0.001 NA
Neurological features (%)0.5210.2<0.0010NA
Severe hypertension (%)0.54717<0.0010NA
Antihypertensive use at delivery (%)0.57464<0.0010NA

The main analyses were the following:

  • 1The incidence of the primary outcome measure according to the diagnosis in the index pregnancy, i.e. pre-eclampsia or gestational hypertension.
  • 2A comparison of baseline data at delivery in the index pregnancy between women who were known to have had subsequent pregnancies and those for whom no further pregnancy records were noted. This analysis was included to ensure that women for whom no further pregnancy was recorded were not a different group (apart from expected differences in parity) than those who had subsequent pregnancies recorded.
  • 3Analysis of the incidence of the primary outcome measures according to subgroups of parity in the index pregnancy.
  • 4Logistic regression analysis to determine the predictive value of factors recorded at delivery in the index pregnancy (as above) as well as booking BP in the next pregnancy and interpregnancy interval for the primary outcome measures (pre-eclampsia or gestational hypertension) in the next pregnancy. Separate models were fitted for those whose index pregnancy was gestational hypertension and pre-eclampsia. Initially, all available predictors were used in building the model as described above. Variables which were not significant at the 5% level were successively eliminated.

Results

The tracking of patient data is shown in Figure 1. A total of 1515 women with pre-eclampsia or gestational hypertension were identified from our unit’s database. One hundred and twenty one (8%) cases were removed before submission to the MDC database due to incorrect coding on review or insufficient outcome data recorded on the database. A further 31 (2%) cases were removed because index pregnancy did not appear in the MDC database and therefore could not be linked to subsequent pregnancies. About 1363 cases were analysed in the MDC database, and data for further pregnancies were available for 759 (56%) of these women. Approximately one-third of these women had only this one further pregnancy. Thirty-six (2.6%) women delivered prior to 32 weeks of gestation in their index pregnancy and 11 (1.4%) prior to 32 weeks in their subsequent pregnancy. The outcome analyses in this study apply to the 759 pregnancies which followed the index (hypertensive) pregnancy and to the 333 pregnancies which followed the normal index pregnancies (controls).

Figure 1.

Flow chart of cases analysed in study.

Table 1 shows that majority of women in all groups were primigravidas. Normal women were slightly younger at their index pregnancy than women with gestational hypertension or pre-eclampsia, and these women had significantly lower first-trimester BP. Women with pre-eclampsia or gestational hypertension had a higher incidence of developing gestational diabetes (P= 0.003) and were delivered earlier than normal pregnant women. Women with pre-eclampsia, but not those with gestational hypertension, had higher rates of SGA babies (26 versus 9%, P < 0.0001).

Table 2 confirms that overall women for whom no further pregnancy records were available (assumed to have not had further pregnancies) generally had similar demographics and pregnancy outcomes in their index pregnancies as women who had further pregnancies, apart from being older and having a greater proportion of multiparas (P < 0.001). The rates of subsequent pregnancies in women with pre-eclampsia (55%) and gestational hypertension (58%) were similar, P= 0.31.

Table 2.  Comparison of data of the index pregnancy between women who had further pregnancies and those who had no further pregnancies recorded
FactorPre-eclampsia—clinicalPre-eclampsia—proteinuriaGestational hypertension
Further pregnancyNo more pregnancyP valueFurther pregnancyNo more pregnancyP valueFurther pregnancyNo more pregnancyP value
  1. ‘Pre-eclampsia—clinical’ is pre-eclampsia diagnosed according to its clinical definition (see text).

  2. ‘NS’ refers to a P value that is not significant; i.e. P≥ 0.05.

Primiparous (%)9061<0.0019061<0.0018658<0.001
Smoker (%)1214NS1112NS2114NS
Age (years)28 (5)31 (6)<0.00128 (5)31 (6)<0.00128 (5)30 (6)<0.001
Booking SBP (mmHg)118 (11)118 (11)NS118 (11)117 (11)NS120 (12)119 (12)NS
Booking DBP (mmHg)71 (8)71 (8)NS71 (7)71 (8)NS73 (8)73 (8)NS
Diabetes pre-pregnancy (%)0.52.5NS0.82.4NS0.50.4NS
Gestation at diagnosis (weeks)36 (3)35 (4)NS36 (4)35 (4)NS37 (3)37 (3)NS
Gestation at delivery (weeks)37 (3)37 (3)NS37 (3)36 (3)NS38 (2)38 (2)<0.01
SGA (%)2625NS2423NS78NS
Proteinuria (%)6166NS10010000NS
Liver disease (%)2221NS1817NS01NS
Hyperuricaemia (%)7373NS7273NS5152NS
Platelets < 150 × 109/l (%)2932NS1930<0.0100NS
Platelets < 100 × 109/l (%)57NS47NS00
Renal impairment (%)2421NS2117NS00NS
Neurological features (%)2121NS2219NS00NS
Severe hypertension (%)4648NS5455NS1422<0.01
Antihypertensive use at delivery (%)7473NS7877NS6267NS

Figure 2 shows the likelihood of developing pre-eclampsia or gestational hypertension in the next pregnancy. Whereas almost all women with a normal index pregnancy had a further normotensive pregnancy, approximately 30% of women hypertensive in their index pregnancy had subsequent pregnancies complicated by hypertension, and 1 in 50 women had developed essential hypertension by the time of their next pregnancy. Women who had pre-eclampsia in their next pregnancy had an intrauterine growth restriction rate of 19%, and 58% were delivered preterm. The respective rates for women with gestational hypertension in their next pregnancy were 23 and 24%. Women with pre-eclampsia in their index pregnancy were equally likely to develop either pre-eclampsia or gestational hypertension (approximately 15% each), while women with gestational hypertension were more likely to develop gestational hypertension again (25%) rather than pre-eclampsia (5%) in their next pregnancy (P < 0.001).

Figure 2.

Outcome in next pregnancy according to diagnosis in index pregnancy. Pre-eclampsia is considered further as nonproteinuric pre-eclampsia and proteinuric pre-eclampsia (see text for details).

When women with pre-eclampsia in their index pregnancy were further considered as either proteinuric pre-eclampsia or nonproteinuric pre-eclampsia, we demonstrated that both these groups had the same rate of gestational hypertension in their outcome pregnancies (13%), while the nonproteinuric pre-eclampsia index group had a higher proportion of pre-eclampsia cases (21 versus 11%) in the outcome pregnancy (P= 0.04).

Table 3 shows that the major influence of parity in the index hypertensive pregnancy was within the group who had gestational hypertension in their index pregnancy. Multiparous women in this group were much more likely than primiparous women to develop pre-eclampsia (12 versus 4%) or gestational hypertension again (49 versus 22%), and only 39% had a normotensive next pregnancy (P < 0.0001). Of the women with gestational hypertension in their first pregnancy, 2.2% had developed essential hypertension prior to their next pregnancy. Although clinically significant, the sample size was too small to conduct a formal hypothesis test, and therefore, only data from the larger groups are represented.

Table 3.  Diagnosis in next pregnancy according to diagnosis and parity in index pregnancy
  Index pregnancy
Pre-eclampsia—clinicalPre-eclampsia—proteinuriaGestational hypertension
Primiparas (n= 354)Multiparas (n= 38)Primiparas (n= 215)Multiparas (n= 24)Primiparas (n= 314)Multiparas (n= 53)
  1. EH, essential hypertension.

  2. The P value reflects comparison across pre-eclampsia, gestational hypertension and normal groups.

  3. EH and EH + pre-eclampsia groups were not included in analysis due to small sample size.

Next pregnancyPre-eclampsia (%)14191017411
Gestational hypertension (%)121413122245
EH (%)202024
EH + pre-eclampsia (%)030404
Normal (%)716876677236
P value0.710.54<0.000001

Table 4 lists the factors determined by logistic regression analysis to be significant predictors (P < 0.05) of gestational hypertension or pre-eclampsia relative to a normal outcome in a subsequent pregnancy.

Table 4.  Odds ratios, P values and 95% confidence intervals for factors predictive of recurrence of pre-eclampsia or gestational hypertension
 Index pregnancy gestational hypertensionIndex pregnancy pre-eclampsia
Next pregnancy gestational hypertensionNext pregnancy pre-eclampsiaNext pregnancy gestational hypertensionNext pregnancy pre-eclampsia
Multiparas3.70, P= 0.0024 (1.78–7.64)7.57, P= 0.0008 (2.31–24.78)
Gestation at diagnosis in index pregnancy0.86, P= 0.0021 (0.78–0.95)0.82, P= 0.016 (0.71,0.96)
Uric acid level in index pregnancy3.59, P= 0.029 (1.14–11.35)
SBP at booking in next pregnancy1.05, P= 0.00065 (1.02–1.09)
DBP at booking in next pregnancy1.06, P= 0.00041 (1.02–1.10)1.08, P= 0.0009 (1.03–1.13)

We found that for women with a diagnosis of gestational hypertension in their index pregnancy, only multiparity, early gestation at diagnosis of hypertension and higher DBP at booking in their next pregnancy were predictive of recurrence of gestational hypertension (Table 4). For mothers who were multiparous in the index pregnancy, the odds of gestational hypertension in the subsequent outcome pregnancy increased by approximately 3.7 times relative to those for mothers who were primiparous in the index pregnancy.

Factors predictive of subsequent pregnancy being complicated by pre-eclampsia for those women with gestational hypertension in index pregnancy were multiparity, early gestation at diagnosis and uric acid level in their index pregnancy.

For women with a diagnosis of pre-eclampsia in their index pregnancy, only higher DBP at booking in their next pregnancy was predictive of recurrent pre-eclampsia and higher SBP at booking in the next pregnancy the sole predictor of gestational hypertension. Unfortunately, no specific value for these clinical factors could be determined so as to be used as a discriminant value that reliably predicted recurrent pre-eclampsia or gestational hypertension.

Discussion

This study provides clinicians who manage hypertensive pregnant women with guidelines as to what they can inform the women about the likely outcome in their next pregnancy. Overall, about 65–70% of women with pre-eclampsia or gestational hypertension will have a normotensive next pregnancy. The high-risk group for recurrent hypertension in pregnancy identified in this study was multiparous women with gestational hypertension. Fortunately, this group of women generally have good maternal and fetal outcomes,14 but they are known to have an increased risk of hypertension in later life15 and death due to stroke16 and cardiovascular causes.17 This therefore becomes an important observation in our study.

Perhaps related to this issue, it is apparent from Table 1 that women with pre-eclampsia or gestational hypertension commenced their pregnancy with BPs slightly but significantly higher than women who went on to have a normotensive pregnancy. This would not have been clinically apparent as the first-trimester mean BPs of the women destined for pre-eclampsia or gestational hypertension were well within the normal range. Table 1 also demonstrates that women with gestational hypertension or pre-eclampsia had higher rates of diabetes during pregnancy. This related partly to higher rates of type I or type II diabetes due to the study design, which excluded such women from the ‘normal’ cohort, but these groups also had the expected higher incidence of gestational diabetes. The relatively benign implications for the fetus in gestational hypertension are also apparent in that the SGA rate was the same as in women with normal pregnancies and significantly lower than in women with pre-eclampsia. A further point of interest from Table 1 is that while more women with pre-eclampsia had hyperuricaemia, it is important to note that at least half the women with gestational hypertension also had elevated plasma uric acid, suggesting that this is a poor discriminatory test between pre-eclampsia and gestational hypertension. Finally, while almost one in five women with gestational hypertension had episodes of severe hypertension (≥170/110 mmHg), this was a lower rate than in women with pre-eclampsia, almost half of whom had at least one episode of severe hypertension. Taken together, the major findings from the data in Table 1 are that gestational hypertension is a fairly benign disorder for the fetus but has moderate hypertensive risks for the mother and that women with either pre-eclampsia or gestational hypertension commence their pregnancy with BPs slightly higher than those destined for a normotensive pregnancy, but this difference is such that it will not be clinically apparent when the woman is first seen in her pregnancy.

A further public health observation is that 2–3% of this cohort developed essential hypertension in their 30s prior to their next pregnancy. This was not apparent in women who had normotensive index pregnancies and underscores the importance of ensuring that all women with hypertension in pregnancy are considered at future cardiovascular risk and therefore require good attention to healthy lifestyle and regular follow up with healthcare professionals.

Comparison with previous studies

Surprisingly few studies have addressed this issue to date. The likelihood of recurrence in other studies ranged from 13 to 55%.7–9,18–20 Sibai et al. showed that early-onset severe pre-eclampsia (in two studies of 125 and 406 women) was highly likely to recur (in two-thirds), and the time of onset in the next pregnancy was equally divided between second and third trimesters and term pregnancy. No factors clearly predicted this recurrence, although the high recurrence rate appeared to be related to early onset in the index pregnancy or black race. Our study was not selected according to gestation at onset of pre-eclampsia or gestational hypertension, but this factor was found to be a significant predictor of subsequent gestational hypertension or pre-eclampsia only in our gestational hypertension population. This discrepancy between our findings and those of Sibai et al. may be in part explained by significantly different populations studied. In Sibai’s et al. earlier study of prognosis of pre-eclampsia in primigravid women, the average age of subjects was approximately 18 years, and 86% of women were black. In his subsequent study reviewing long-term prognosis of second-trimester pre-eclampsia, the average gestation at diagnosis for the women was 25 weeks. These women are quite different from our study population where the average age was approximately 30 years, and the average gestation at diagnosis of pre-eclampsia in the index pregnancy was approximately 36 weeks. It may therefore be that our results are only applicable to the relatively ‘mild’ group of women in this study, only a small proportion (2.6%) of whom were delivered before 32 weeks.

In their study of approximately 5000 women whose first pregnancy was complicated by hypertension, Campbell et al.4 found that approximately 62% of women with proteinuric pre-eclampsia had a normotensive second pregnancy (7.5% having recurrent pre-eclampsia and 29% developing mild pre-eclampsia in their second pregnancy), women with ‘mild pre-eclampsia’ (nonproteinuric hypertension in pregnancy, similar to gestational hypertension) in their first pregnancy had a 69% chance of normotensive second pregnancy (with 2% having pre-eclampsia and 29% having recurrent mild pre-eclampsia) and 90% of women who had normotensive first pregnancies went on to have normotensive second pregnancies.

The recent pre-eclampsia community guideline (PRECOG) study21 described a number of factors that result in increased risk of pre-eclampsia occurring in any pregnancy. They recognised the importance of history of pre-eclampsia as a risk factor for moderate, severe and early-onset pre-eclampsia in subsequent pregnancy, in addition to adverse perinatal outcomes associated with preterm delivery. They report a relative risk of developing pre-eclampsia of 7.19 (5.85–8.83) if women have a history of pre-eclampsia; however, beyond this observation, the paper did not look further at factors predictive of a recurrence of pre-eclampsia.

While much attention has been directed towards predicting pre-eclampsia in first pregnancies, few studies have addressed systematically what, if any, factors were predictive of recurrent pre-eclampsia or gestational hypertension. Factors such as gestational diabetes mellitus in the index pregnancy,18 early gestation at diagnosis of hypertension,7 SGA infant in index pregnancy7 and higher DBP in second pregnancy7 have all been reported as possible risk factors.

Hjartardottir et al.22 investigated the recurrence of hypertension in second pregnancy. They reported much higher rates of recurrence than the rate our study demonstrates (overall risk of recurrence 73% [range 58–94%]). This may reflect population differences between our studies. While Hjartardottir et al. studied women with greater than or equal to two singleton births with hypertension confirmed in first pregnancy (similar to our study), women were selected according to family clusters on the Iceland Genealogy Database. This is in contrast to our heterogeneous population of women.

Mostello et al.23 conducted a study aiming to identify risk factors for pre-eclampsia in parous women. Although not entirely focusing on recurrent pre-eclampsia, they found that the risk of recurrent pre-eclampsia in the second pregnancy was inversely proportional to the gestational age at delivery of the first pregnancy.

Campbell et al.4 reported that for pre-eclampsia (proteinuric) pregnancies that ended before 37 weeks, the risk of recurrent pre-eclampsia increased from 7 to 13%; however, there was no impact of duration of pregnancy for either normotensive or mild pre-eclampsia first pregnancies on subsequent pregnancies.

We have found that early gestation at diagnosis in index pregnancy, multiparity, uric acid levels in the index pregnancy and booking BP parameters for the next pregnancy may significantly influence the likelihood of recurrence of gestational hypertension, but surprisingly, no parameters in the index pregnancy reliably predicted recurrent pre-eclampsia (Table 4). Although these findings for women with gestational hypertension are hard to interpret without discriminant values (e.g. precise BP threshold above which risk substantially increases), they do provide a framework enabling clinicians to stratify a higher risk to these women in postpartum or preconception counselling.

It is of interest, evident in Table 2, that indices of severity of either pre-eclampsia or gestational hypertension had little influence on whether women decided to carry another pregnancy. Rather, the main determinants of the decision to have another pregnancy were as expected, namely older age and having already had more than one child.

A point worth noting in this study is that the groups of pre-eclampsia and gestational hypertension may not be as independent as they first appear. Women with gestational hypertension at term are likely to be delivered. It is possible, therefore, that women with gestational hypertension in their index pregnancy who developed pre-eclampsia in their subsequent pregnancy were in fact destined for pre-eclampsia in their index pregnancy, but delivery was effected prior to pre-eclampsia emerging. A further point worth noting is that women who had clinical features of pre-eclampsia in the index pregnancy without proteinuria had a significantly higher rate of developing pre-eclampsia in the subsequent pregnancy (21 versus 11%, P= 0.037) than those who had proteinuric pre-eclampsia in the index pregnancy. It is apparent from Figure 2 that the main reason for this finding is the higher recurrence of pre-eclampsia with proteinuria (9 versus 5%) but a lower rate of pre-eclampsia without significant proteinuria (2 versus 16%) in those who had proteinuric pre-eclampsia in the index pregnancy. This highlights the benefits in clinical practice of using additional clinical criteria above proteinuria alone for the diagnosis of pre-eclampsia to highlight a group at higher risk of developing abnormalities of kidney, liver, neurologic or haematological functions in their subsequent pregnancy.

Strengths and limitations

The potential weakness of our study is the possibility that some of the 43% of women for whom no further pregnancy data existed in NSW actually had further pregnancies. We believe this to be an unlikely event for several reasons. First, the MDC is designed to capture outcome data on all pregnancies in NSW. Our study included an internal check on this aspect in that only 31 of 1394 (2%) cases we had managed had no records held in the MDC database. Second, the Australian Bureau of Statistics data showed that the emigration rate from the state of NSW for women in this age group is less than 4%, so it is unlikely that a large number of further pregnancies in this cohort of women had occurred outside the state. Third, those without records of a further pregnancy were older and had a higher proportion of multiparas in the index pregnancy than in women who had known further pregnancies, suggesting that they had decided not to have further pregnancies having already had more than one child. Finally, the average number of children per woman of reproductive age in NSW is approximately 1.76 (Australian Bureau of Statistics).

The main strengths of our study are the large cohort, the detailed assessment of every case note to ensure accurate diagnoses and management in the index pregnancy by only two physicians using a uniform management protocol which standardised the index pregnancy data more than in units where management varies according to clinicians’ different practices.

Conclusions

This study has been conducted with careful attention to precision of diagnosis of the hypertensive disorder of pregnancy. Data from this large cohort show that pre-eclampsia or gestational hypertension recur in almost one-third of pregnancies, the highest risk being for multiparous women with gestational hypertension which portends hypertension in two-thirds of subsequent pregnancies. While some factors such as early gestation at diagnosis and high uric cid levels in women with gestational hypertension alert clinicians to the possibility of pre-eclampsia in the next pregnancy, no factors in the index pregnancy in women with pre-eclampsia reliably predict pre-eclampsia in the next pregnancy.

The implications of this study for the practising clinician are as follows. If there has been no hypertension in a prior pregnancy, this is very reassuring that hypertension is unlikely to develop in the next pregnancy. If some form of hypertension has occurred, even the relatively benign condition of gestational hypertension, then there is a small but significant risk of developing pre-eclampsia in the next pregnancy. Women with such a history should be seen more often in a subsequent pregnancy with a view to taking more seriously observations such as increments in BP (even when not into a frankly hypertensive range), isolated proteinuria developing in the absence of hypertension and slowing of fetal growth.

At the least, clinicians can use these data to assist answering that very common question postpartum: ‘What is the likelihood I will get high BP again in my next pregnancy?’

Acknowledgements

We thank Lucy Bowyer and Lisa McHugh for their assistance with data collection, Lee Taylor and Kim Lim from NSW Health MDC for their role in linkage analysis of pregnancies of the cohort and the Medical Records Departments of the many hospitals across NSW who assisted in this study.

Ancillary