Prediction of recurrence of hypertensive disorders of pregnancy between 34 and 37 weeks of gestation: a retrospective cohort study


Dr MF van Oostwaard, Erasmus Medical Centre – Sophia, Department of Obstetrics and Gynaecology, PO Box 2060, 3000 CB, Rotterdam, the Netherlands. Email


Please cite this paper as: van Oostwaard M, Langenveld J, Bijloo R, Wong K, Scholten I, Loix S, Hukkelhoven C, Vergouwe Y, Papatsonis D, Mol B, Ganzevoort W. Prediction of recurrence of hypertensive disorders of pregnancy between 34 and 37 weeks of gestation: a retrospective cohort study. BJOG 2012;119:840–847.

Objective  To assess the recurrence risk of late-preterm hypertensive disease of pregnancy, and to determine whether potential risk factors are predictive.

Design  Retrospective cohort study.

Setting  Three secondary and three tertiary care hospitals in the Netherlands.

Population  We identified women with a hypertensive disorder in the index pregnancy and delivery at 34–37 weeks of gestation, between January 2000 and December 2002.

Methods  Data were extracted from medical files and women were approached for additional information on subsequent pregnancies. An adverse outcome was defined as the recurrence of a hypertensive disorder in the next subsequent pregnancy.

Main outcome measures  Absolute risk of recurrence and a prediction model containing demographic and clinical factors predictive for adverse outcome.

Results  We identified 425 women who matched the criteria, of whom 351 could be contacted. Of these women, 189 (54%) had had a subsequent pregnancy. Hypertensive disorders recurred in 96 (51%, 95% CI 43–58%) women, of whom 17 (9%, 95% CI 5–14%) delivered again before 37 weeks of gestation. Chronic hypertension and maternal age were the strongest predictors for recurrence. Women undergoing recurrence had a nine-fold chance of developing chronic hypertension (37% versus 6%, OR 8.7, 95% CI 3.3–23).

Conclusions  Women with hypertensive disorders and late-preterm delivery have a 50% chance of recurrence, but only a 9% chance of recurrence resulting in delivery before 37 weeks of gestation. Women with chronic hypertension are prone to develop recurrence, and women with a recurrence more often developed chronic hypertension.


Hypertensive disorders of pregnancy, including pre-eclampsia (PE) and HELLP (haemolysis, elevated liver enzymes and low platelets) syndrome affect 2–7% of pregnancies. Late-preterm onset (e.g. between 34 and 37 week of gestation) represents 5–17% of these hypertensive disorders.1–3 Hypertensive disorders contribute substantially to both maternal and perinatal outcome worldwide, and are the leading cause of maternal death in the Netherlands.4,5

As hypertensive disorders in pregnancy are heterogeneous conditions, predicting the recurrence of these disorders based on a single variable has been disappointing.6 Additionally, the quality of reporting of studies and their clinical usefulness or generalisability have been suboptimal, making prediction models unsatisfactory until now.6 Hypertensive disorders of pregnancy may have significant psychological impact on women and their partners, and many parents refrain from future pregnancies because of their earlier experiences.7–9 Therefore, informed counselling about future pregnancies is important.

For years, recurrence rates of 65% were cited, based on the most quoted study of Sibai et al.10 Recently, more studies have been published on recurrence rates of hypertensive disorders.3,11–14 Most studies concentrate on recurrence rates in women who suffered early-onset severe hypertensive disorders or on recurrence rates in women who had pre-eclampsia at term. There have been no studies published on the recurrence of late-preterm onset hypertensive disorders.11 There may be a different underlying pathophysiological mechanism, different risk factors and recurrence risks compared with term or early-onset PE, so data from either of these groups should not be extrapolated to the late-preterm gestational ages.12 For this reason we decided to explore this group, as it comprises a significant population in the non-academic setting. Our primary aim was to calculate the absolute risk of recurrence of a hypertensive disorder in the next subsequent pregnancy after delivery between 34 and 37 weeks of gestation because of a hypertensive disorder. Our secondary aim was to identify independent risk factors for recurrence using multivariate analysis, and to build a prediction model.


Study population

We retrospectively identified women from electronic databases in six hospitals in the Netherlands: three secondary care centres (Amphia Hospital, Breda; Deventer Hospital, Deventer; and Kennemer Gasthuis, Haarlem) and three tertiary care centres (Maastricht University Medical Centre, Maastricht; Academic Medical Centre, Amsterdam; and Maxima Medical Centre, Veldhoven). We consecutively included women who delivered between 34 and 37 weeks of gestation in their index pregnancy between January 2000 and December 2002, and who were diagnosed with gestational hypertension (GH), pre-eclampsia (PE), HELLP syndrome and/or delivered a small for gestational age (SGA) baby. Women carrying a pregnancy with fetal abnormalities were excluded, but chronic hypertension was not an exclusion criterion. The standard practice for determining gestational age was ultrasound dating, or if this had not been performed a reliable first date of the last menstrual cycle was used.

We collected demographic data, including: age, body mass index, parity and cardiovascular risk factors, such as smoking, chronic hypertension diagnosed before pregnancy, thrombophilia and family history for cardiovascular disease. For index and subsequent pregnancies we collected the following data: highest systolic and diastolic blood pressure; use of medication; number of days spent in hospital; and perinatal outcome, including gestational age at delivery, birthweight and perinatal death.

Data were extracted from the medical files. Information on subsequent pregnancies was gained primarily through the medical files. A subsequent pregnancy was defined as a continuing pregnancy beyond 16 weeks of gestation. If such information was not available in the records of the institution, individual women were contacted. If the patient reported that she had had a continuing subsequent pregnancy, her gynaecologist or family doctor were contacted for data after written informed consent. All ethically available public resources were consulted if women were lost to follow-up.

Hypertensive disorders of pregnancy include PE, superimposed PE, GH, HELLP syndrome and/or delivery of an SGA child. Pre-eclampsia was defined as hypertension (diastolic blood pressure ≥90 mmHg or systolic blood pressure ≥140 mmHg on two occasions, 4–5 hours apart) in combination with proteinuria (defined as 1+ [0.3 g/l] or more on a proteinuria dipstick test, a protein/creatinine ratio of 30 mg/mmol or more in a random sample, or a urine protein excretion of 300 mg or more per 24 hours) after 20 weeks of pregnancy.15 Women with hypertension without proteinuria, or with a significant rise in blood pressure with known chronic hypertension, were considered to have GH. GH associated with a significant rise of 30 mmHg or more in blood pressure was used to emphasize the distinction between chronic hypertension and a pathological process related to pregnancy. Chronic hypertension was defined as the presence or history of preconceptional hypertension or hypertension detected in the first half of pregnancy. Superimposed pre-eclampsia is indicated by newly detected proteinuria, or by a sudden increase in proteinuria if already present, in a woman with chronic hypertension.15 HELLP syndrome was defined by haemolysis (elevated lactate dehydrogenase, LDH ≥ 600 iu/l), elevated liver enzymes, indicated by levels of aspartate transaminase (ASAT) or alanine transferase (ALAT) ≥ 70 iu/l, and a low platelet count <100 000/mm.16 SGA was defined as a birthweight below the tenth percentile, according to the American College of Obstetricians and Gynecologists (ACOG) practice bulletin,17 and adjusted for gestational age based on a local reference population.

Statistical analysis

We expressed continuous variables as means with standard deviations (SDs) or, if not normally distributed, medians with interquartile ranges (IQRs). Differences in baseline characteristics or outcomes between groups were tested with parametric (unpaired Student’s t test) or non-parametric (Mann–Whitney U-test) tests, as appropriate. Categorical variables were compared with chi-square tests. P values <0.05 were considered to indicate statistical significance.

For each of the candidate predictors the odds ratio for the chance of recurrence was calculated. Univariate and multivariate logistic regression analyses were performed, with recurrence of hypertensive disease (yes/no) as the outcome measure. Aikaike’s information criterion (AIC) was used to select the strongest predictors for recurrence in a backward-selection procedure.

The model was validated internally with bootstrapping procedures. Models, developed on this study population, may be over fitted, indicating that high predictions are too high and low predictions are too low. A bootstrap (re-sampling) procedure was performed to assess the level of optimism. One hundred bootstrap samples were drawn with replacement, and models were fitted in each sample. The mean difference in performance between the bootstrap samples and the original data was used to assess optimism. The bootstrap procedure provided a shrinkage factor, with which the regression coefficients of the predictors were multiplied (shrunken) to prevent the model giving too extreme predictions for women with hypertensive disorders in pregnancy.

The overall performance of the prediction model was assessed with R2. This performance measure indicates the percentage of the total variation in women with and without recurrence that can be explained by the predictors in the model. The discriminative ability of the model, being the ability of the model to distinguish women with recurrence from women without recurrence, was assessed with the c statistic. A model with a c statistic of 0.50 has no discriminative power at all (comparable with a coin flip), and a c statistic of 1.0 reflects perfect discrimination.

Missing data are often not completely random, but are rather selectively omitted. Simply deleting the women with missing values (so-called complete case analysis) would thus result in invalid study results. The imputation of selectively omitted values can reduce bias and allows for including all women in the analysis. Therefore, we imputed these missing values ten times after completion of the selection. The imputation models included all the candidate predictors, and also plurality, ethnicity, education, family history of cardiovascular disease, family history of hypertension, family history of diabetes mellitus, family history of pre-eclampsia, smoking, medical anamnesis, systolic blood pressure, time interval between index pregnancy and second pregnancy, and the outcome variable. Analyses were performed in each of the ten multiple-imputed data sets. Estimates from the ten data sets were then combined into one overall estimate and variance, according to Rubin’s rules.


We identified 425 women who delivered between 34 and 37 weeks of gestation because of a hypertensive disorder. Of these 425 women, 79 (19%) were lost to follow-up, 157 (37%) did not become pregnant again and 189 (44%) had a subsequent pregnancy after a median of 2.6 years (IQR 1.9–3.5 years). Of the 157 women who refrained from a subsequent pregnancy, 36 (23%) made it known that this was because of the perceived risk. The median follow-up time since delivery of the index pregnancy was 8.5 years (IQR 7.8–9.3 years). The study profile is outlined in Figure 1.

Figure 1.

 Flowchart of women with a history of delivery between 34 and 37 weeks of gestation, because of a hypertensive disorder, and subsequent pregnancies included in the study.

To test for selection bias, we compared baseline characteristics between women with a subsequent pregnancy and women who did not become pregnant or were lost to follow-up. These characteristics are shown in Table 1. Women who conceived again were 3 years younger (29 versus 32 years of age) and smoked less often (11% versus 20%). They were also more likely to be white (92% versus 82%), nulliparous at index (89% versus 49%) and more likely to have HELLP syndrome in the index pregnancy (31% versus 20%), and were less likely to have had a multiple pregnancy (6% versus 28%).

Table 1.   Baseline clinical characteristics after the index pregnancy of the 425 included women, subdivided by category: with a subsequent pregnancy; lost to follow-up; and without a subsequent pregnancy
Demographic characteristicsWomen with a subsequent pregnancy
(n = 189)
Women without a subsequent pregnancy
(n = 157)
Women who were lost to follow-up
(n = 79)
  1. Values that are statistically significant are indicated in bold.

  2. *Data are presented as means (SDs); body mass index and hospital days are medians (IQRs).

  3. **Significant differences only for the subgroup ‘lost to follow-up’.

  4. ***Significant difference only for the subgroup ‘with a subsequent pregnancy’.

  5. ****Percentages sum up to more than 100% because of overlapping of disorders.

Age at delivery (years)* 29 (4) 32 (5) 31 (5) <0.001
Smoking 21 (11%) 31 (20%) 17 (22%) <0.001
White 143 (92%) 118 (82%) 27 (63%) <0.001
Body mass index (kg/m2)*24 (21–27)23 (21–26)24 (21–28)0.799
Chronic hypertension before pregnancy17 (9%)19 (12%)6 (8%)0.504
Thrombophilia (tested in 78, 18%)14 (8%)4 (8%)1 (5%)0.665
Family history of coronary disease**111 (59%)108 (75%) 21 (49%) 0.004
Pregnancy characteristics of index pregnancy
Multiple pregnancy*** 11 (6%) 44 (28%) 21 (28%) <0.001
Nullipareous 169 (89%) 77 (49%) 53 (72%) <0.001
Maximum systolic blood pressure* (mmHg*)159 (23)159 (22)158 (25)0.990
Maximum diastolic blood pressure* (mmHg*)103 (14)103 (13)104 (16)0.989
Gestational age at delivery (days)*251 (6)250 (6)250 (6)0.079
Birth weight (grams)*2259 (526)2156 (490)2294 (541)0.060
Hospital days*9 (6–14)9 (5–15)10 (6–15)0.766
Small for gestational age****74 (39%)73 (47%)34 (43%)0.387
Gestational hypertension**,****62 (33%)56 (36%) 43 (54%) 0.003
Preeclampsia****97 (51%)77 (49%)30 (38%)0.130
Eclampsia****7 (4%)4 (2.5%)4 (5%)0.604
HELLP syndrome**** 58 (31%) 32 (20%) 14 (18%) 0.026

For women who decided to refrain from a subsequent pregnancy because of the perceived risk, we evaluated pregnancy characteristics to see if they were more at risk than the entire group. Women who refrained from pregnancy had had a higher diastolic blood pressure (108 [SD 11] versus 102 [SD 14] mmHg), used antihypertensive medication more often (58% versus 42%), were in hospital for longer (12 [IQR 9–15] versus 9 [IQR 5–14] days) but had a smaller level of proteinuria (1303 versus 2653 mg/24 hours) in the index pregnancy (Table 2).

Table 2.   Characteristics of women who refrained from a subsequent pregnancy because of the perceived risk
 Entire group
n = 425
No subsequent pregnancy because of perceived risk
n = 36
  1. Values that are statistically significant are indicated in bold.

  2. *Data are presented as means (SDs); hospital days as median (IQR).

Age at delivery (years)*29 (4)31 (4)0.292
Gestational age at delivery (days)*252 (6)245 (7)0.077
Birthweight (grams)*2256 (525)2170 (513)0.486
Maximum systolic blood pressure (mmHg)*158 (23)163 (21)0.262
Maximum diastolic blood pressure (mmHg)* 102 (14) 108 (11) 0.028
Maximum proteinuria (mg/24 hours)* 2653 (4267) 1303 (1901) 0.007
HELLP98 (23%)13 (36%)0.071
Eclampsia15 (4%)1 (3%)0.632
Hospital days 9 (5–14) 12 (9–15) 0.047
Anticonvulsive medication73 (14%)10 (28%)0.650
Antihypertensive medication 178 (42%) 21 (58%) 0.028

The mean gestational age at delivery in the index pregnancy was 36 weeks (SD 6 days), and the mean birthweight was 2256 g (SD 525 g). In the subsequent pregnancy, the mean gestational age at delivery was 39 weeks (SD 18 days) and the mean birthweight was 3171 g (SD 728 g). In the index pregnancy, 76 women had a multiple gestation (18%), versus 11 women (6%) in the subsequent pregnancy. Perinatal death with various aetiologies occurred in eight (2%) women in the index pregnancy, and in three women (1%) in the next pregnancy.

Of the 189 women with a subsequent pregnancy, 96 women (51%, 95% CI 43–58%) had a recurrence of a hypertensive disorder in the subsequent pregnancy, 17 of whom (9%, 95% CI 5–14%) had a recurrence of a hypertensive disorder and delivered before 37 weeks of gestation. Eight women with recurrence delivered before 34 weeks of gestation (4%, 95% CI 2–8%). The subsequent pregnancy was uneventful regarding hypertensive disorders in 93 women (49%, 95% CI 20–33%). For the 83 women requiring antihypertensive and/or anticonvulsive medication during the index pregnancy, a hypertensive disorder reoccurred in 45 women (51%). This was not statistically different compared with women without these medications (51% versus 50%, P = 0.405).

For the 96 women who had a recurrence of a hypertensive disorder in the next pregnancy, we compared the clinical characteristics of the index and subsequent pregnancies. The means of the highest systolic and diastolic blood pressures were lower (160 versus 148 mmHg, P < 0.001, and 105 versus 96 mmHg, P < 0.001, respectively) in the subsequent pregnancy. In addition, the mean level of proteinuria was less (3403 versus 808 mg/24 hours, P = 0.006), which diagnosed 47 (49%) women with pre-eclampsia in the index pregnancy and 22 (23%) in the subsequent pregnancy. The mean duration of admittance at the hospital was shorter (12 versus 6 days, P < 0.001). Furthermore, anticonvulsive medication was used less often: in 18 (19%) versus seven women (7%), P < 0.019. Whereas the use of antihypertensive medication was comparable: 41 (43%) versus 31 women (32%), P = 0.100.

For the 189 women with a subsequent pregnancy we compared the recurrence of any of the individual syndromes (Table 3). None of the individual syndromes are more related to overall recurrence than the others, but they all seem to have a tendency to recur in the same type of hypertensive disorder as before.

Table 3.   Analysis of each individual syndrome on recurrence in the 189 women with a subsequent pregnancy
 Index pregnancy (%)Recurrence in the subsequent pregnancy P
  1. Values that are statistically significant are indicated in bold.

  2. Percentages sum up to more than 100% because of the overlapping of syndromes.

Small for gestational age74 (39)41 (55%) P = 0.371 SGA: 23 (31%) <0.001
GH: 21 (28%)0.292
PE: 7 (9%)0.295
HELLP: 3 (4%)0.571
Gestational hypertension62 (33)35 (56%) P = 0.128SGA: 9 (15%)0.277
GH: 27 (44%) 0.005
PE: 3 (5%) 0.032
HELLP: 0 (0%)0.060
Pre-eclampsia97 (51)47 (48%) P = 0.509SGA: 12 (12%)0.110
GH: 29 (30%)0.504
PE: 18 (19%) 0.002
HELLP: 6 (6%)0.066
HELLP syndrome58 (31)27 (47%) P = 0.438SGA: 6 (10%)0.084
GH: 15 (26%)0.180
PE: 12 (21%) 0.011
HELLP: 6 (10%) 0.004

Prediction model

The distribution and univariate analysis of candidate predictors of women with and without recurrence is shown in Table 4. Gestational age was not assessed as a predictor, as it was one of the selection criteria. Only the predictors ‘chronic hypertension before pregnancy’ and ‘maternal age at delivery’ were significantly related to recurrence. Finally, multivariable analysis of the association between chronic hypertension and age at delivery and recurrence is also shown in Table 4. The shrinkage factor, estimated with the bootstrapping procedure, was 0.96. Age at delivery has an effect, with an odds ratio of 1.1 per year (95% CI 1.02–1.2). Chronic hypertension demonstrated an odds ratio of 7.9 (95% CI 2.6–24) on the recurrence of hypertensive disorders of pregnancy. The R2 of the model was 0.17 and the c statistic was 0.71 (95% CI 0.64–0.78).

Table 4.   Distribution and univariate analysis between candidate predictors and the recurrence of hypertensive disorders in the subsequent pregnancy
Candidate predictorRecurrence, n = 96No recurrence, n = 93Univariable analysisMultivariable analysis*% missing
OR95% CIOR95% CI
  1. Values that are statistically significant are indicated in bold.

  2. *Beta coefficients to calculate the ORs were multiplied by a shrinkage factor of 0.96 to improve prediction in future women.

  3. **Diabetes mellitus, manifested cardiovascular disease, thrombophilia and/or kidney disease.

Demographic characteristics
Age at delivery (years) 30 ± 428 ± 4 1.1 1.03–1.2 1.1 1.02–1.2 0
Body mass index > 30 kg/m2 14 (17%)7 (9.2%)1.80.69–4.415
Chronic hypertension before pregnancy 28 (29%)4 (4.4%) 9.2 3.1–27 7.9 2.6–24 0.5
Cardiovascular risk factors other than hypertension**  7 (7%)3 (3%)1.40.27–6.90
Characteristics of index pregnancy
Primiparous 82 (85%)87 (94%)0.400.15–1.10
Maximum diastolic blood pressure (mmHg)104 ± 13102 ± 151.00.99–1.042
Diagnosis of small for gestational age 41 (43%)34 (37%)1.30.72–2.30
Diagnosis of HELLP syndrome 24 (25%)28 (30%)0.790.41–1.51
Diagnosis of pre-eclampsia 48 (50%)51 (55%)0.820.46–1.50

The use of antihypertensive medication before the subsequent pregnancy was not included in the model, because of limited degrees of freedom, but did reach a significant difference regarding recurrence. Thirteen (86%) of the 15 women on medication developed recurrence, compared with 83 women (48%) not using medication (OR 7.1, 95% CI 1.5–47).

All women contacted were asked if they had developed chronic hypertension at the time of data collection (0.4–8.6 years after the last pregnancy). Sixty-four women reported the development of chronic hypertension. In the women with a subsequent pregnancy, we were missing data for a total of 28 women. Of 83 women with recurrence, 31 (37%) developed chronic hypertension, compared with five of 78 (6%) women without recurrence (OR 8.7, 95% CI 3.3–23). If they also delivered preterm because of the recurrence, the odds ratio of developing chronic hypertension after recurrence was 4.8 (eight of 15 women compared with 28 of 146 women without recurrence before 37 weeks of gestation; 95% CI 1.7–14). Of women who refrained from a subsequent pregnancy, 23 of 86 (26%) developed chronic hypertension.


This cohort study explores the recurrence risks of hypertensive disorders of pregnancy in the late-preterm period; as such, it fills a blank in the literature. We found that half of the women had a subsequent pregnancy. Hypertensive disorders recurred in 96 (51%) women, of whom 17 (9%) delivered again before 37 weeks of gestation. Chronic hypertension and maternal age were statistically significant predictors for recurrence.

In this study, late-preterm onset hypertensive disorders are explored for their recurrence rates, which was previously unstudied. Strengths of the present study are the reasonable cohort size, the information on women who did not become pregnant again to account for sources of bias, an adequate follow-up time, the variability in settings (both academic and non-academic medical centres) and type of hypertensive disorder.

Our study is too small to construct a prediction model on the recurrence of hypertensive disorders in the late-preterm period in the subsequent pregnancy. Also, our cohort doesn’t include a control group. For this reason, we could not compare the risk of developing hypertensive disorders in a subsequent pregnancy with women who had an uncomplicated first pregnancy. The inclusion of SGA in normotensive pregnancies is perhaps controversial, but appropriate if considered as a continuum of hypertensive disorders of pregnancy.18 Although this assumption may have clinical relevance, it must be acknowledged that a diagnosis of SGA will included some appropriately grown babies and other pathologies that are not related to the hypertensive disease spectrum. The effect of SGA on recurrence was separately assessed within the multivariate prediction model. Using a definition of SGA below the tenth percentile was decided upon in view of the ACOG guideline, and its widespread use in other research. Although the fifth percentile would provide much cleaner data, in clinical practice the distinction between normal and restricted growth cannot be made using a percentile, and neither can its origin.

The differences between the women with or without a subsequent pregnancy, or who were lost to follow-up, are not clinically significant. If a woman is older, multiparous or has had a multiple pregnancy, it is obviously more likely that she considers her family to be complete and chooses not to engage in a subsequent pregnancy.

Ninety-seven (19%) women were lost to follow-up, mostly because they could not be traced after having moved. Twenty-three percent of women who refrained from a subsequent pregnancy indicated that this was because of their fear of recurrence. This agrees with other reports of the psychological impact of hypertensive disorders.8 Differences in this group may or may not put these women at greater risk, and the recurrence rate in women with a subsequent pregnancy may not necessarily extrapolate to this group. It is possible that counseling could be improved in this group of women using more accurate data; however, there is no information on how these women have been counseled to date.

The recurrence of hypertensive disorders leading to a delivery before 37 weeks of gestation, after an index pregnancy with a delivery between 34 and 37 weeks of gestation, was 9%. Although this percentage is relatively low, the overall recurrence rate of hypertensive disorders irrespective of gestational age in the subsequent pregnancy was 51%. If a hypertensive disorder of pregnancy recurred, in general the disorder was milder in the subsequent pregnancy.

Mostello et al.13 found an overall recurrence rate of 22% in a subset of the study of women with pre-eclampsia delivering between 33 and 36 weeks of gestation in the first pregnancy. Mostello, however, only included women with pre-eclampsia, which may explain the difference with our study. Two recent population-based studies show recurrence rates (at any gestational age) of 15% and 7% in women who have had pre-eclampsia in one previous pregnancy.12,14 These studies also show that preterm delivery, cardiovascular risk factors, pre-existing renal disease and more preceding pregnancies complicated by the disorder were associated with a higher chance of recurrence. The recurrence rate was lower in women with multiple gestations in the index pregnancy in these studies. Unfortunately, the aforementioned studies only investigated women with (severe) pre-eclampsia, whereas our cohort also included HELLP syndrome, SGA and GH. Ganzevoort et al.18 stated that it is reasonable to regard these diseases as dynamic variations of one syndrome. Several studies report that HELLP syndrome recurs less often, but that SGA recurs in 20–24% of women.13,19,20 There are almost no data available on the recurrence of GH. In our cohort there were 62 women with GH. The occurrence of hypertensive disorders in their subsequent pregnancy was 56% (Table 3). If GH was accompanied by SGA (17 women), the recurrence was 65%. A hypertensive disorder recurred in three out of seven women (43%) with GH, associated with signs of HELLP syndrome (clinically defined as PE). Comprising one-third of our cohort, GH might explain the much higher overall recurrence rate recorded here.

The low recurrence rate recorded by McDonald et al.14 (7%) can be explained to some extent by the exclusion of women with chronic hypertension before pregnancy. If the same method had been used in this cohort, we would have included 156 normotensive women with a subsequent delivery. Recurrence of hypertensive disorders and preterm delivery occurred in ten women (6.4%), and recurrence at any gestational age occurred in 68 women (44%).

The performance and discriminative ability of the prediction model is reasonable, which is in line with several other studies, although inclusion criteria or end points do not agree. Sep et al.6 created an overview of prediction models on the recurrence of hypertensive disorders. Almost all studies focus on recurrence of just one disease, and there are only two models that identified chronic hypertension as a predictor.21,22 Both studies focus on the recurrence of hypertensive disorders after an initial pregnancy with HELLP syndrome. Only one report provides odds ratios: women with chronic hypertension compared with normotensive women have an odds ratio of 1.8–3.7 of developing pre-eclampsia or HELLP, after an initial pregnancy with HELLP syndrome and delivery before 28 weeks of gestation.22 In our study chronic hypertension shows a greater effect on recurrence. Perhaps chronic hypertension has more effect on recurrence after hypertensive disorders other than HELLP, but it is impossible to compare these outcomes when such different inclusion criteria are used.

Finally, it is important to take into account the changes in management in the past decade. Induction of labour is advised earlier in pregnancy, which may cause the prediction model to slightly overestimate the recurrence rate, because more women with milder disease will now deliver before 37 weeks of gestation. The shrinkage factor does not correct for this.


Preconception care and counselling is a growing aspect of obstetric care. Couples can make a better informed choice about a subsequent pregnancy and, if needed, monitoring can be intensified in the subsequent pregnancy. Using the results of this study, individual risk counselling can be improved, but more individually tailored prediction models are needed.

Disclosure of interests

None of the authors have a conflict of interest to report.

Contribution to authorship

MvO and JL are the main authors, having designed the study, performed part of the data collection and written most of the article. RB, KMW, IS and SL all carried out most of the data collection in the various institutes and reviewed the article. CH, together with YV, performed the statistical tests, wrote the ‘statistics’ section and reviewed the article. DP, BWM and WG contributed to the design of the study, assisted and supervised with data collection and reviewed the article.

Details of ethics approval

The ethical advisory board considered this study not to be subject to the Medical Research Involving Human Subjects Act (WMO). This decision was based on the fact that the women in this cohort did not have to undergo any examinations, nor did they have to visit the hospital. Therefore, the board approved the study.


The preparative research was also made possible by a stipend that was granted by the Dutch perinatology and maternal disease working party and a trust fund granted by the Erasmus University, Rotterdam.