UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Counseling women with hypertensive disorders of pregnancy

van Oostwaard, M.F.

Publication date2015Document VersionFinal published version

Link to publication

Citation for published version (APA):van Oostwaard, M. F. (2015). Counseling women with hypertensive disorders of pregnancy.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s)and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an opencontent license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, pleaselet the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the materialinaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letterto: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. Youwill be contacted as soon as possible.

Download date:28 Mar 2021

36927 Oostwaard.indd 74 27-10-15 10:16

Chapter 4

Prediction of recurrence of hypertensive

disorders of pregnancy in the term period,

a retrospective cohort study

M.F. van OostwaardJ. Langenveld

E. SchuitK. Wigny

H. Van Susante I. Beune

R. RamaekersD.N.M. Papatsonis

B.W.J. MolW. Ganzevoort

Pregnancy Hypertension: An International Journal of Women’s Cardiovascular Health2014;4:194–202

36927 Oostwaard.indd 75 27-10-15 10:16

Chapter 4

76

Abstract

Objectives: To assess the recurrence risk of term hypertensive disease of pregnancy and to determine which potential risk factors are predictive of recurrence.

Study design: We performed a retrospective cohort study in two secondary and one tertiary care hospitals in the Netherlands. We identified women with a hypertensive disorder in the index pregnancy and delivery after 37 weeks of gestation between January 2000 and December 2002. Data were extracted from medical files and women were approached for additional information on subsequent pregnancies. Adverse outcome was defined as recurrence of a hypertensive disorder in the next subsequent pregnancy.

Main outcome measures: The absolute risk of recurrence and a prediction model containing demographic and clinical factors predictive of recurrence.

Results: We identified 638 women for potential inclusion, of whom 503 could be contacted. Of these women, 312 (62%) had a subsequent pregnancy. Hypertensive disorders recurred in 120 (38%, 95% CI 33–44) women, of whom 15 (5%, 95% CI 3–7) delivered preterm. Women undergoing recurrence were more at risk to develop chronic hypertension after pregnancy (35% versus 16%, OR 2.8, 95% CI 1.5–5.3). Body mass index, non-White European origin, chronic hypertension, maximum diastolic blood pressure, no use of anticonvulsive medication and interpregnancy interval were predictors for recurrence.

Conclusions: Women with hypertensive disorders and term delivery have a substantial chance of recurrence, but a small risk of preterm delivery. A number of predictors for recurrence could be identified and women with a recurrence more often developed chronic hypertension.

36927 Oostwaard.indd 76 27-10-15 10:16

Recurrence of term HDP

77

4

Introduction

Approximately 10% of all pregnancies are complicated by a hypertensive disorder, including gestational hypertension (GH), preeclampsia (PE) and HELLP (hemolysis, elevated liver enzymes and low platelets) syndrome. Ninety percent of women with mild or severe preeclampsia deliver at term (P37 weeks of gestation).1–3 Although less than in early-onset disease, these disorders contribute substantially to both maternal and perinatal outcome.4,5 Late-onset hypertensive disorders do not have the same impact on particular neonatal outcome as do early-onset preeclampsia. Nevertheless, recurrent disease produces more pregnancy complications as preterm delivery, placental abruption and fetal death, than preeclampsia in nulliparous women.6 Several studies have been published on recurrence rates of hypertensive disorders.3,7–12 Most studies concentrate on recurrence rates in women who suffered early-onset severe hypertensive disorders, or are registry-based research lacking information on important risk factors.

The heterogeneity of hypertensive disease of pregnancy, both in cause as in symptomatology, leaves recurrence prediction of these disorders based on a single variable problematic.7 Additionally, prediction models for recurrence have been unsatisfactory because the quality of reporting of studies is generally low and their clinical usefulness or generalizability suboptimal.7 The psychological impact of hypertensive disorders of pregnancy can cause parents to refrain from future pregnancies.13–15 Therefore, accurate knowledge of the risk of recurrence in future pregnancies underlying informed counseling is of the utmost importance.

Our primary aim was to calculate the absolute risk of recurrence of a hypertensive disorder in the next pregnancy after a hypertensive pregnancy with a delivery beyond 37 weeks of gestation. Our secondary aim was to identify independent risk factors for recurrence and to develop a prediction model for recurrence.

36927 Oostwaard.indd 77 27-10-15 10:16

Chapter 4

78

Materials and methods

Study populationWe retrospectively identified patients from electronic databases in three hospitals in the Netherlands: two secondary care centers (Amphia Hospital, Breda and Atrium Medical Centre, Heerlen) and one tertiary care center (Academic Medical Center, Amsterdam). We included consecutive patients who delivered after 37 weeks of gestation of their index pregnancy between January 2000 and December 2002, and were diagnosed with GH, PE or HELLP syndrome.

Women carrying a pregnancy with fetal abnormalities were excluded. Chronic hypertension was not an exclusion criterion. Standard practice for determination of gestational age was ultrasound dating.

We collected demographic data including: age, body mass index (BMI), parity, and cardiovascular risk factors like: smoking, chronic hypertension diagnosed before pregnancy, thrombophilia and family history for cardiovascular disease. Of the index and subsequent pregnancy we collected data on: highest systolic and diastolic blood pressure, use of medication, hospital days and perinatal outcome including gestational age at delivery, birth weight and perinatal death.

Data were extracted from medical files. Information on subsequent pregnancies was gained first through medical files. A subsequent pregnancy was defined as an on-going pregnancy beyond 16 weeks of gestation. If this information was not available in the records of the institution, we tried to contact the individual patients. If the patient reported that she had had an on-going subsequent pregnancy, a written informed consent was obtained, after which her gynecologist or family doctor was contacted for additional data. All contacted patients were asked if they had developed chronic hypertension at time of data collection. All ethically available public resources were consulted if patients were lost to follow-up.

The outcome was: hypertensive disorders of pregnancy, a composite of PE, superimposed PE, GH and HELLP syndrome. Preeclampsia was defined as hypertension (diastolic blood pressure ≥90 mmHg or systolic blood pressure ≥140 mmHg on two occasions, 4–5 h apart) in combination with proteinuria (defined as 1+ (0.3 g/l) or more on proteinuria dipstick test, a protein/creatinine ratio of 30 mg/mmol or more in a random sample or an urine protein excretion of 300 mg or more per 24 h.) after 20 weeks of pregnancy.16 Women with hypertension without proteinuria, or a significant rise in blood pressure (a rise of ≥ 30 mmHg) if known chronic hypertension, were considered to have GH. Complementing the definition of GH with a significant rise of 30 mmHg or more

36927 Oostwaard.indd 78 27-10-15 10:16

Recurrence of term HDP

79

4

in blood pressure was made to emphasize a distinction between chronic hypertension and a pathological process related to pregnancy. Chronic hypertension was defi ned as a history of preconceptional hypertension or hypertension detected in the fi rst half of pregnancy. Superimposed preeclampsia includes de novo proteinuria, or a sudden increase in proteinuria if already present, in a woman with chronic hypertension.16 HELLP syndrome was defi ned by hemolysis (elevated lactate dehydrogenase (LDH) levels ≥600 U/L), elevated liver enzymes by levels of aspartate transaminase (ASAT) or alanine transferase (ALAT) ≥70 U/L and low platelets <100,000/mm.17 SGA was defi ned as birth weight below the 10th percentile, according to the ACOG practice bulletin18 and adjusted for gestational age based on a local reference population.

Statistical analysisWe expressed continuous variables as mean with standard deviation (SD) or median with interquartile range (IQR) if not normally distributed. Differences in baseline characteristics or outcomes between groups were tested with parametric (unpaired t-test) or non-parametric (Mann–Whitney-U test) tests as appropriate. Categorical variables were compared with Chi square tests. p values less than 0.05 were considered to indicate statistical signifi cance.

Missing data are often not missing completely at random, but rather selectively missing. Removing the women with missing values (complete case analysis) could result in a loss of precision and biased study results.19–21 Imputation of selectively missing values can reduce bias and allows for the inclusion of all women in the analysis.

Therefore, we multiple imputed these missing values 10 times. The imputation models included all the candidate predictors and the outcome. The imputation was performed in the total database (N = 638), after which only those women with a subsequent pregnancy were selected (N = 312).

For each of the candidate predictors the odds ratio for the chance of recurrence was calculated. Univariable and multivariable logistic regression analysis was performed with recurrence of hypertensive disease (yes/no) as an outcome measure. Maternal age, BMI, blood pressures and interpregnancy interval were analyzed as continuous variables. Linearity of their association with the outcome was assessed using cubic splines.22 Although pre-selection based on the univariable analysis is in general not recommended23, we performed a pre-selection for the multivariable analysis based on the p-value (<0.20) in the univariable analysis in order to retain a reasonable number of events per variable to fi t the multivariable model.24 The combination of predictors that best predicts recurrence was obtained with a backward selection procedure using the

36927 Oostwaard.indd 79 27-10-15 10:16

Chapter 4

80

Aikaike Information Criterion (AIC), which is similar to selection based on a p-value of 0.157. As imputed data sets differ from each other, predictor selection was performed in each imputation set separately. For inclusion in the final prediction model, we used the majority method, i.e. predictors were included in the final model if selected in at least five out of ten imputed data sets.25 The regression coefficients and standard errors of these final predictors were combined from the ten data sets using Rubin’s rules to come to the final prediction model.26

With each model development there is a chance of overfitting, meaning that the model too strongly fits to the data on which it was developed and consequently may perform poorly when externally validated. To assess the degree of overfitting or optimism, we internally validated the model using bootstrapping techniques. One hundred bootstrap samples were drawn from the original data set with replacement, allowing for multiple sampling of the same individual. Within each sample the entire modeling process described above was repeated. This yielded a shrinkage factor, with which the regression coefficients of the predictors were multiplied (uniformly shrunken) to correct the model for optimism and overfitting.27

Overall performance of the prediction model was assessed with the 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, which is for a dichotomous outcome similar to an area under the Receiver Operating Characteristic curve (AUC). A model with a c-statistic of 0.50 has no discriminative power at all (comparable to a coin flip), and a c-statistic of 1.0 reflects perfect discrimination. The predicted probabilities were compared with the observed frequencies of the recurrence of an hypertensive disorder using calibration plots to assess the calibration of the model.27,28

Results

Population We identified 638 patients who delivered after 37 weeks and were diagnosed with a hypertensive disorder. After exclusion of 135 (21%) who were lost to follow-up, 503 women were included. Of these, 191 (38%) did not become pregnant again and 312 (62%) had a subsequent pregnancy after a median of 2.7 years (IQR: 2). Of the 191 women who refrained from a subsequent pregnancy, 25 (3.9%) made known that this was due to

36927 Oostwaard.indd 80 27-10-15 10:16

Recurrence of term HDP

81

4

perceived risk. The mean follow-up time since delivery of the index pregnancy was 11 years (SD 10 months). The study profi le is outlined in Fig. 1.

In the 312 women with a subsequent pregnancy, 120 women (38%, 95% CI 33–44) had recurrence of a hypertensive disorder in the subsequent pregnancy, of whom 15 women (5%, 95% CI 3–7) had recurrence of a hypertensive disorder and delivered before 37 weeks of gestation. Ten women with recurrence delivered before 34 weeks (9%, 95% CI 4–14). The subsequent pregnancy was uneventful regarding hypertensive disorders in 192 women (62%, 95% CI 56–67).

Chapter 3

Chapter 4

F i g ur e 1 . F l o w ch a rt o f co n s e cu ti v e ly i nc lu d e d pa ti e n ts w i th a h i s to ry o f a d e li v e ry be tw ee n 34 an d 37 w ee k s d u e t o a h y pe rt e ns iv e d is o rd e r an d s ub s e qu en t pr e gn an ci es .

Lost to follow up N = 79 (19%)

No subsequent pregnancy Perceived risk: N = 36 (8%) Other reasons: N = 121 (28%)

Included patients N = 425

Subsequent pregnancy N = 189

No recurrence: 93 (49%) Recurrence: 96 (51%)

├ < 37 wks: 17 (9%) └ ≥ 37 wks: 79 (42%)

Lost to follow up N = 135 (21%)

No subsequent pregnancy Perceived risk: N = 25 (4%) Other reasons: N = 166 (26%)

Included patients N = 638

Subsequent pregnancy N = 312

No recurrence: 192 (62%) Recurrence: 120 (38%)

├ < 37 wks: 15 (5%) └ ≥ 37 wks: 105 (33%)

N = 503

F i g ur e 1 . F l o w ch a rt o f th e 6 38 c o ns e c ut i ve l y i n cl ud e d pat i e nt s w ith a h i st o ry o f a h y pe rt e ns iv e d is o rd e r, w h o d e liv e r ed af te r 37 w e ek s , a nd su b se q ue n t pr eg n an cie s .

Figure 1. Flowchart of the 638 consecutively included patients with a history of a hypertensive disorder, who delivered after 37 weeks, and subsequent pregnancies.

To assess the presence of selection bias, we compared baseline characteristics between women with a subsequent pregnancy to women who did not become pregnant or were lost to follow-up. These characteristics are shown in Table 1. Women who had a next pregnancy were signifi cantly younger (29 versus 33 years), more often nulliparous (85% versus 47%), and were less likely to smoke (9% versus 18%), to be White European (74% versus 85%) or to have a family history of coronary disease (63% versus 79%), than women without a subsequent pregnancy. The index pregnancy in women with a subsequent pregnancy was signifi cantly less often complicated by multiple pregnancy (1% versus 4%), SGA (18% versus 26%) or GH (58% versus 68%), and the maximum diastolic blood pressure was signifi cantly 1 mmHg higher (101 versus 100 mmHg) compared to the group without a subsequent pregnancy.

36927 Oostwaard.indd 81 27-10-15 10:16

Chapter 4

82

For women who decided to refrain from a subsequent pregnancy due to perceived risk, we evaluated pregnancy characteristics to see if they were different from the entire group (Table 2). In the index pregnancy, these 25 women were on average 2 years older (32 versus 30 years), used antihypertensive and anticonvulsive medication more often (28% versus 12% and 28% versus 3%, respectively). Hospitalization was longer (5 versus 3 days) and it was complicated more often by eclampsia (12% versus 1%) and a SGA child (40% versus 19%).

Table 1. Baseline clinical characteristics and index pregnancy parameters of the 638 included patients subdivided by category; with a subsequent pregnancy, lost to follow-up and without a subsequent pregnancy.

Demographic characteristics Patients with a subsequent

pregnancy(n = 312)

Patients without a subsequent pregnancy

(n = 191)

P* Patients who were lost to

follow-up(n = 135)

P**

Age at delivery (years)SmokingEthnicity:

White EuropeanCaribbeanAsianSub-Saharan AfricaMiddle East

Body mass index (kg/m2)Chronic hypertension before pregnancyFamily history of coronary disease

29 (4) 27 (9%)

187 (74%) 32 (13%) 9 (4%) 18 (7%) 6 (2%) 25 (22-30)

21 (7%)

196 (63%)

33 (5) 35 (18%) 154 (85%) 15 (8%) 7 (4%) 3 (2%) 2 (1%) 26 (22-31)

19 (10%)

151 (79%)

< .01< .01

< .01.369.628.022.446.168

.196

< .01

30 (6) 17 (13%)

19 (29%) 22 (33%) 1 (2%) 15 (23%) 6 (9%) 26 (23-31)

8 (6%)

55 (41%)

.187< .01

< .01.072.159.047.130.766

.751

< .01Pregnancy characteristics of index pregnancyMultiple pregnancyNullipareousMaximum systolic bloodpressure (mmHg)Maximum diastolic blood pressure (mmHg)Antihypertensive medicationGestational age at delivery (days)Birth weight (grams)Hospital daysSmall for Gestational Age§

Gestational hypertension§

Preeclampsia§

Eclampsia§

HELLP syndrome§

3 (1%) 266 (85%)

155 (17)

101 (9) 46 (15%)

278 (10) 3255 (574%) 3 (2-5) 57 (18%) 182 (58%) 128 (41%) 3 (1%) 22 (7%)

7 (4%) 90 (47%)

156 (20)

100 (8) 32 (17%)

277 (10) 3276 (633) 3 (1-6) 50 (26%) 129 (68%) 62 (33%) 4 (2%) 6 (3%)

.033< .01

.769

.044.546

.230

.704

.278

.035

.039

.054

.293

.063

4 (3%) 86 (64%)

154 (19)

101 (10) 17 (13%)

276 (10) 3285 (586) 3 (1-5) 17 (13%) 96 (71%) 39 (29%) 0 (0%) 1 (1%)

.126< .01

.314

.534

.548

.144

.623

.272

.138

.011

.015

.253< .01

Continuous data are presented as means (SD); Body mass index and Hospital days in median (IQR)* p value for the difference between women with versus without subsequent pregnancy** p value for the difference between women with subsequent pregnancy versus lost to follow up.*** oral or intravenous antihypertensive medication or magnesiumsulfate.§ percentages sum up to more than 100% because of overlapping of disorders

36927 Oostwaard.indd 82 27-10-15 10:16

Recurrence of term HDP

83

4

The mean gestational age at delivery in the index pregnancy was 39 weeks and 4 days (SD: 10 days) and the mean birth weight 3257 g (SD: 594 g). In the subsequent pregnancy, mean gestational age at delivery was 39 weeks (SD: 23 days) and mean birth weight 3380 g (SD: 827 g). In the index pregnancy 14 women had a multiple gestation (2%), versus 3 (1%) in the subsequent pregnancy. Perinatal death with various etiologies occurred in 6 (1%) women in the index pregnancy and in 9 (3%) in the next pregnancy.

Severity of recurrent hypertensive disease For the 120 women who had recurrence of a hypertensive disorder in the next pregnancy, we compared the clinical characteristics of the index and subsequent pregnancy. The means of the highest systolic and diastolic blood pressures were lower (151 versus 160 mmHg, p-value .024 and 98 versus 104 mmHg, p-value < .01, respectively) in the subsequent pregnancy. The maximum amount of proteinuria was also signifi cantly lower (135 versus 300 g, p-value .01). Gestational age at delivery, antihypertensive (0% versus 3%, p-value 1.00) and anticonvulsive medication (2% versus 0%, p-value .250) and hospital time (3 versus 4 days, p-value .167) were similar between pregnancies.

Table 2. Characteristics of women who refrained from a subsequent pregnancy due to perceived risk.

Entire groupN = 613

No subsequent pregnancy due to

perceived riskN = 25

P

Age at delivery (years)* Gestational age at delivery (days)*Birth weight (grams)* Max systolic BP (mmHg)* Max diastolic BP (mmHg)*Max proteinuria (mg/24h)*Small for Gestational Age HELLPEclampsia Hospital days Anticonvulsive medicationAntihypertensive medication

30 (5) 278 (10) 3278 (591) 155 (18) 101 (9) 300 (1000) 114 (19%) 27 (4%) 4 (1%) 3 (3) 19 (3%) 73 (12%)

32 (4) 273 (9) 3022 (637) 161 (23) 103 (8) 300 (1530) 10 (40%) 2 (8%) 3 (12%) 5 (5) 7 (28%) 7 (28%)

.013.05

.035

.207

.234

.501.008.398

<.001.003

<.001.017

*Data are presented as means (SD); Proteinuria and hospital days in median (IQR)

Recurrence of individual syndromeFor the 312 women with a subsequent pregnancy, we compared the recurrence of hypertensive disease for any of the individual syndromes (Table 3). Compared to the absence of that specifi c hypertensive syndrome, SGA signifi cantly recurred more often as SGA again (23% versus 7%, p-value <.001), PE as PE (16% versus 6%, p-value .002)

36927 Oostwaard.indd 83 27-10-15 10:16

Chapter 4

84

and HELLP as HELLP (10% versus 1%, p-value .001). GH recurred less often as PE than if GH was not present (6% versus 16%, p-value .003).

Table 3. Analysis of each individual syndrome on recurrence in the 312 women with a subsequent pregnancy.

Subsequent pregnancyIndex pregnancyN = 312

Type of syndrome With syndrome in index pregnancy n/N (%) *

Without syndrome in index pregnancy n/N (%) *

p-value

Small for Gestational Age N = 57 (18%)

Any SGA GH PE HELLP

22/57 (39) 12/52 (23) 13/55 (24) 4/55 (7) 0/54 (0)

98/255 (38) 17/251 (7) 60/245 (24) 26/249 (10) 4/249 (2)

.982<.001.894.476.349

Gestational hypertensionN = 182 (58%)

Any SGA GH PE HELLP

67/182 (37) 18/177 (10) 45/176 (26) 10/178 (6) 2/178 (1)

53/130 (41) 11/126 (9) 28/124 (23) 20/126 (16) 2/125 (2)

.480.675.553.003.721

Preeclampsia N = 128 (41%)

Any SGA GH PE HELLP

53/128 (41) 11/124 (9) 28/122 (23) 20/122 (16) 2/123 (2)

67/184 (36) 18/179 (10) 45/178 (24) 10/180 (6) 2/180 (1)

.373

.731.645.002.700

HELLP syndrome N = 22 (7%)

Any SGA GH PE HELLP

5/22 (23) 0/21 (0) 2/21 (10) 3/21 (14) 2/21 (10)

115/290 (40) 29/282 (10) 71/279 (25) 27/283 (10) 2/282 (1)

.116

.123.102.482.001

Percentages of the specific types of syndromes may add up to more than 100% because of overlapping of syndromes.* recurrence of the hypertensive syndromes, in the women with and without the specific syndrome in the index pregnancy, as indicated at the start of each row.

Chronic hypertension after pregnancyThe follow-up time since the last pregnancy was 2 months to 13 years. Chronic hypertension was reported by 131 women. We had in total missing data on 199 patients (31%). Of 100 women with recurrence, 35 (35%) developed chronic hypertension, compared to 24 of 149 (16%) women without recurrence (OR 2.8, 95% CI 1.5–5.3). If they subsequently delivered preterm due to recurrence, the odds ratio to develop chronic hypertension after recurrence was 2.4 (5 of 12 women compared to 54 of the 237 women without recurrence before 37 weeks of gestation; 95% CI 0.6–9.0). Of women who refrained from a subsequent pregnancy 12 of 24 (50%) developed chronic hypertension.

36927 Oostwaard.indd 84 27-10-15 10:16

Recurrence of term HDP

85

4

Prediction modelThe distribution and univariable analysis of candidate predictors of women with and without recurrence is shown in Table 4. Gestational age was not assessed as a predictor, since it was one of the selection criteria. White European origin was protective for recurrence in this cohort. A BMI over 30 kg/m2, chronic hypertension before pregnancy, interpregnancy interval, multiparity at index, maximum systolic and diastolic blood pressure and use of antihypertensive medication in the index pregnancy were signifi cantly related to reoccurrence of a hypertensive disorder in the next pregnancy.

Finally, multivariable analysis of the association between these signifi cant predictors and recurrence is also shown in Table 4. The shrinkage factor, estimated with the bootstrapping procedure, was 0.73. Systolic blood pressure was not included in the multivariable analysis to prevent collinearity due to its obvious correlation with diastolic blood pressure. Parity, HELLP syndrome and antihypertensive medication in the index pregnancy were no longer associated with recurrence. The demographic predictors BMI and chronic hypertension before pregnancy were positively related to recurrence hypertensive disorders of pregnancy and respectively demonstrated odds ratios of 1.61 (95% CI 0.90–2,94) and 2.02 (95% CI 0.73–5.60). White European origin was negatively related to recurrence, with an odds ratio of 0.68 (95% CI 0.39–1.20). Index pregnancy characteristics maximum diastolic blood pressure (OR 1.03, 95% CI 1.00–1.02) and interpregnancy interval (OR 1.01, 95% CI 1.00–1.02) showed very faint positive relations to recurrence of hypertensive disorders of pregnancy. Use of anticonvulsive medication during the index pregnancy was negatively related (OR 0.31, 95% CI 0.06–1.73). All the variables with a signifi cant association with recurrence were included in the prediction model. The R2 of the model was 15.2% (95% CI 13.6–16.9%) and the c-statistic was 0.69 (95% CI 0.63–0.75). The range of predicted recurrence if this model was used in our current study population was 9–88% (mean predicted recurrence: 39%). The calibration plot presented in Fig. 2, shows a good agreement between predicted and observed recurrence.

36927 Oostwaard.indd 85 27-10-15 10:16

Chapter 4

86

Tabl

e 4. D

istrib

utio

n of

and

mul

tivar

iabl

e ana

lysis

bet

wee

n ca

ndid

ate p

redi

ctor

s and

recu

rrenc

e of h

yper

tens

ive d

isord

ers i

n th

e sub

sequ

ent p

regn

ancy

.

Cand

idat

e pre

dict

or

Recu

rren

ceN

= 12

0N

o re

curr

ence

N =

192

Univ

aria

ble a

naly

sisM

ultiv

aria

ble a

naly

sis*

% M

issin

g***

OR95

% C

Ip

β**

OR95

% C

Ip

Dem

ogra

phic

char

acte

ristic

sAg

e at d

eliv

ery

< 25

year

s

25-3

5 yea

rs

> 35

year

s

95

(79%

) 1

6

(13%

)

9

(8%

)

14

0

(73%

)

31

(16%

)

21

(11%

)

0.76 ref

0.63

0.39

– 1.

47re

f0.

28 –

1.44

0.41 ref

0.27

0% 0% 0%Bo

dy m

ass i

ndex

> 30

kg/m

2Ch

roni

c hyp

erte

nsio

n be

fore

pr

egna

ncy

Smok

ing

Whi

te Eu

rope

an

40

(3

3%)

14

(1

2%)

10

(9

%)

78

(6

5%)

39

(2

0%)

7

(4%

)

18

(9%

)

150

(7

8%)

1.99

3.49

0.90

0.52

1,15 -

3,45

1.37 –

8.9

20.

40 –

2.0

60.

31 –

0.90

<0.0

1

<0.0

10.

810.

01

0.48

0.70

-0.3

8

1.61

2.02

0.68

0.90

– 2,

94

0.73

– 5.

60

0.39

– 1.2

0

0.09

0.18

0.17

7% 0% 3% 21%

Char

acte

ristic

s of i

ndex

preg

nanc

yM

ultip

areo

us

Max

imum

syst

olic

bp (m

mH

g)M

axim

um d

iast

olic

bp (m

mH

g)Us

e of a

ntih

yper

tens

ive

med

icatio

nUs

e of a

ntico

nvul

sive

med

icatio

nD

iagn

osis

HEL

LP sy

ndro

me

Inte

rpre

gnan

cy in

terv

al

(mon

ths)

21

(11%

) 15

0

± 17

104

± 9

22

(18%

)

2

(2%

)

5 (4

%)

44

± 2

9

25

(2

1%)

13

3 ± 1

1

101

± 9

18

(9

%)

9

(5%

)

17

(9%

)

35

± 2

1

2.14

1.02

1.04

2.17

0.35

0.45

1.02

1.14 –

4.0

31.0

1 – 1.

061.0

1 – 1.

06

1.11 –

4.2

4

0.07

– 1.6

20.

16 –

1.25

1.01 –

1.06

0.02

<0.0

1<0

.01

0.02

0.18

0.12

<0.0

1

0.03

-1.16 0.01

1.03

0.31

1.01

1.0 –

1.02

0.06

– 1.7

3

1.00

– 1.0

2

0.05

0.18

0.02

0% 16%

13%

0% 0% 0% 2%* T

he in

terc

ept o

f the

mod

el is

: -3.

1939

** ß

-coe

ffici

ents

wer

e mul

tiplie

d by

a sh

rinka

ge fa

ctor

of 0

.73 t

o im

prov

e pre

dict

ion

in fu

ture

pat

ient

s **

* im

puta

tion

was

use

d to

com

pens

ate f

or th

ese m

issin

g va

lues

Pred

icto

rs w

ith a

p-v

alue

bel

ow 0

.20

in th

e uni

varia

ble a

naly

sis a

re sh

own

in b

old

36927 Oostwaard.indd 86 27-10-15 10:16

Recurrence of term HDP

87

4

Figure 2. Calibration plot of the prediction model with 95% Confi dence intervals and histogram.

Discussion

Main fi ndingsIn this study, term onset hypertensive disorders are explored for their recurrence rates, which was previously unstudied. We found that 62% of the women had a subsequent pregnancy. Hypertensive disorders recurred in 120 (38%) women, of whom 15 (5%) subsequently delivered before 37 weeks. BMI, non-White European origin, chronic hypertension before pregnancy, maximum blood pressure, use of anticonvulsive medication and interpregnancy interval were statistical signifi cant predictors for recurrence.

Strengths and limitationsStrengths 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) and type of hypertensive disorder.

Unfortunately, our cohort does not comprise a control group. For that reason, we could not compare the risk of developing hypertensive disorders in a subsequent pregnancy to women with an uncomplicated fi rst pregnancy. Also, a lost to follow up rate of 21% limits the interpretation of our fi ndings. Furthermore, the addition of the signifi cant rise of 30 mmHg or more in blood pressure to the defi nition of gestational hypertension,

      

36927 Oostwaard.indd 87 27-10-15 10:16

Chapter 4

88

is debatable. We do think that women with chronic hypertension can experience a pregnancy related pathologic process, without developing proteinuria.

InterpretationThe differences between the women with or without a subsequent pregnancy or who were lost to follow-up are considered not clinically significant and most likely not due to the events in the index pregnancy. A younger woman with less children is understandably more likely to engage in another pregnancy.

The 135 (21%) women lost to follow-up, mostly could not be traced after having moved. Of the women, who refrained from a subsequent pregnancy, 25 (13%) indicated this was due to the fear of recurrence. This concurs with other reports of the psychological impact of hypertensive disorders.14 The recurrence rate in women with a subsequent pregnancy may not necessarily extrapolate to this group. The comparison between this group and the other women in this cohort (Table 2), does suggest a more serious clinical syndrome in the index pregnancy. Information on how these women have been counseled is unavailable. Nevertheless, these data suggest that after a term pregnancy, the clinical syndrome influences women’s choice for having a next pregnancy.

The recurrence of hypertensive disorders leading to prematurity, after an index pregnancy with a term delivery was 5%. While this percentage is relatively low, the overall recurrence rate of hypertensive disorders irrespective of gestational age in the subsequent pregnancy was 38%. If a hypertensive disorder of pregnancy recurred, in general the disorder was milder in the subsequent pregnancy, regarding maximal blood pressure and proteinuria. None of the specific types of hypertensive diseases were more related to a recurrence than the other types, but they all seem to have a tendency to recur in the same type as before.

There have been other studies describing recurrence rates of hypertensive disease, but only a few studies report specific on recurrence after a term delivery. Two registry based studies9,10 demonstrated recurrence rates of 14% and 13% respectively, after a previous preeclampsia with a term delivery. Evaluation of risk factors is limited in registry based research, but the women were mostly under 27 years of age. Lykke et al.9 excluded women with cardiovascular disease and in the cohort by Mostello et al.10 chronic hypertension was very rare (4%). This could explain the lower recurrence rate reported by these publications. Using data of a Collaborative Perinatal Project, Zang et al.11 calculated a similar recurrence rate of 15% after a term GH or PE. Inclusion of this trial has been decades ago and comprised a group of women at young age (11–25 years of age at the first pregnancy) and low BMI (21 kg/m2), accounting for lower recurrences.

36927 Oostwaard.indd 88 27-10-15 10:16

Recurrence of term HDP

89

4

Unfortunately, Zang and Mostello only investigated women with (severe) preeclampsia, whereas our cohort included HELLP syndrome, SGA and GH as well. Ganzevoort et al.29 stated that it is reasonable to regard these diseases as dynamic variations of one syndrome. HELLP syndrome seems to recur less often, but SGA recurs in 20–24%.10,30,31 There are almost no data available on recurrence of GH. In our cohort there were 176 patients with GH. The occurrence of hypertensive disorders in their subsequent pregnancy is 38% (Table 3) and even more often if GH was accompanied by SGA (32 women, 47%). GH comprising more than half of our cohort might explain the much higher overall recurrence rate in our cohort.

If we had excluded women with chronic hypertension from this cohort, we would have included 272 normotensive women with a subsequent delivery. Recurrence and preterm delivery would have occurred in 10 patients (4%) and recurrence at any gestational age in 96 (35%) previously normotensive women. Excluding chronic hypertension would have had limited influence on the recurrence rate in our cohort.

Before this study, we conducted a study on recurrence of hypertensive disorders in the late preterm period.12 In the current study we excluded SGA in normotensive pregnancies, because of its controversy. The recurrence rate in this previous study was higher (51%). This concurs with the general concept that a hypertensive syndrome occurring earlier in pregnancy is a more serious disorder, with corresponding more severe underlying pathophysiology, also leading to reoccurrence in the next pregnancy.

Prediction modelEven though individual odds ratios are low, the prediction model does create predicted probabilities of risk between 9% and 88%, meaning that the model can detect women who have a very low risk of recurrence (9%), up to a risk of 88%, which is substantially higher than that of the overall study population (38%). Using this model, these women can be counseled accordingly to their individual risk estimation. The performance and discriminative ability of the prediction model is reasonable, which is in line with several other studies, although inclusion criteria or endpoints do not concur. Some of the predictors in our prediction model show a signifi cant, but small association with the outcome. Nevertheless, together they predict recurrence reasonably, as could be expected from the multifactorial theories on preeclampsia. Anticonvulsive medication, which we believed to indicate a more serious condition in the index pregnancy, was unexpectedly protective for recurrence. Causes for unexpected fi ndings can be various.32 Just chance, or possibly an unknown confounder could influence the effect of anticonvulsive therapy. Furthermore, women who received anticonvulsive therapy

36927 Oostwaard.indd 89 27-10-15 10:16

Chapter 4

90

possibly refrain from further pregnancies more often, causing selection bias in our model. It is advised to keep these unexpected predictors in the model.32

Finally it is important to take into account the changes in management in the past decade. Induction of labor is advised earlier in pregnancy, which may cause patients with hypertension to not reach the term period.

In comparison to our previous cohort12, it is surprising that maternal age did not predict recurrence in this cohort. The overview of prediction tests for preeclampsia provided by Sep et al.7 shows that BMI, multiparity and chronic hypertension have appeared as predictors in other research. Diastolic blood pressure8 and longer interpregnancy intervals33 were also predictive of the previous research. Ethnicity has been associated with the occurrence of preeclampsia34. Adding more variables to the model would increase the performance. In order to do so, the study population should be larger.

ConclusionPreconception care and counseling is a growing aspect of obstetric care. It can help couples to 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 counseling can be improved, but more individually tailored prediction models are needed.

36927 Oostwaard.indd 90 27-10-15 10:16

Recurrence of term HDP

91

4

References1 Sibai B, Dekker G, Kupferminc M. Pre-eclampsia. Lancet 2005;365:785–99.2 Sibai BM. Diagnosis and management of gestational hypertension and preeclampsia. Obstet Gynecol

2003;102:181–92.3 Hauth JC, Ewell MG, Levine RL, Esterlitz JR, Sibai BM, Curet LB. Pregnancy outcomes in healthy nulliparous

women who subsequently developed hypertension. Obstet Gynecol 2000;95:24–8.4 Khan KS, Wojdyla D, Say L, Gülmezoglu AM, Van Look PF. WHO analysis of causes of maternal death: a

systematic review. Lancet 2006;367:1066–74.5 Schutte JM, Schuitemaker NW, van Roosmalen J, Steegers EAP. Dutch maternal mortality committee.

Substandard care in maternal mortality due to hypertensive disease in pregnancy in The Netherlands. BJOG 2008;115:732–6.

6 Hnat MD, Sibai BM, Caritis S, Hauth J, Lindheimer MD, MacPherson C, et al. Perinatal outcome in women with recurrent preeclampsia compared with women who develop preeclampsia as nulliparas. Am J Obstet Gynecol 2002;186(3):422–6.

7 Sep S, Smits L, Prins M, Peeters L. Prediction tests for recurrent hypertensive disease in pregnancy, a systematic review. Hypertens Pregnancy 2010;29(2):206–30.

8 Langenveld J, Buttinger A, van der Post J, Wolf H, Mol B, Ganzevoort W. Recurrence risk and prediction of a delivery under 34 weeks of gestation after a history of a severe hypertensive disorder. BJOG 2011;118(5):589–95.

9 Lykke JA, Paidas MJ, Langhoff-Roos J. Recurring complications in second pregnancy. Obstet Gynecol 2009;113:1217–24.

10 Mostello D, Kallogjeri D, Tungsiripat R, Leet T. Recurrence of preeclampsia: effects of gestational age at delivery of the fi rst pregnancy, body mass index, paternity, and interval between births. Am J Obstet Gynecol 2008;199:55–7.

11 Zhang J, Troendle JF, Levine RJ. Risks of hypertensive disorders in the second pregnancy. Paediatr Perinat Epidemiol 2001;15:226–31.

12 van Oostwaard MF, Langenveld J, Bijloo R, Wong KM, Scholten I, Loix S, et al. Prediction of recurrence of hypertensive disorders of pregnancy between 34 and 37 weeks of gestation: a retrospective cohort study. BJOG 2012;119(7):840–7.

13 van Pampus MG, Wolf H, Mayruhu G, Treffers PE, Bleker OP. Long-term follow-up in patients with a history of (H)ELLP syndrome. Hypertens Pregnancy 2001;20:15–23.

14 Rep A, Ganzevoort W, Bonsel GJ, Wolf H, de Vries JI. Psychosocial impact of early-onset hypertensive disorders and related complications in pregnancy. Am. J. Obstet. Gynecol. 2007;197(2):158e1–6.

15 Schaaf JM, Bruinse HW, van der Leeuw-Harmsen L, Groeneveld E, Koopman C, Franx A, et al. Reproductive outcome after early-onset pre-eclampsia. Hum Reprod 2011;26:391–7.

16 ACOG practice bulletin. Diagnosis and management of preeclampsia and eclampsia. Number 33, January 2002. American College of Obstetricians and Gynecologists. Int Gynaecol Obstet 2002;77(1):67–75.

17 Sibai BM. The HELLP syndrome (hemolysis, elevated liver enzymes, and low platelets): much ado about nothing? Am J Obstet Gynecol 1990;162(2):311–6.

18 ACOG practice bulletin. Intrauterine growth restriction. Number 12, January 2000. American College of Obstetricians and Gynecologists. Int Gynaecol Obstet 2001;72(1):85–96.

19 Donders AR, van der Heijden GJ, Stijnen T, Moons KG. Review: a gentle introduction to imputation of missing values. J. Clin. Epidemiol. 2006;59:1087–91.

20 Greenland S, Finkle WD. A critical look at methods for handling missing covariates in epidemiologic regression analyses. Am. J. Epidemiol. 1995;142:1255–64.

21 van der Heijden GJ, Donders AR, Stijnen T, Moons KG. Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: a clinical example. J Clin Epidemiol 2006;59:1102–9.

22 Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361–87.

36927 Oostwaard.indd 91 27-10-15 10:16

Chapter 4

92

23 Sun GW, Shook TL, Kay GL. Inappropriate use of bivariable analysis to screen risk factors for use in multivari-able analysis. J Clin Epidemiol 1996;49:907–16.

24 Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373–9.

25 Vergouwe Y, Royston P, Moons KG, Altman DG. Development and validation of a prediction model with missing predictor data: a practical approach. J Clin Epidemiol 2010;63:205–14.

26 Rubin DB. Multiple imputation for non response in surveys. New York, NY: Wiley; 1987.27 Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating

assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361–87.28 Miller ME, Hui SL, Tierney WM. Validation techniques for logisticregression models. Stat Med 1991;10:1213–

26.29 Ganzevoort W, Rep A, Bonsel GJ, de Vries JIP, Wolf H. For the PETRA investigators. Dynamics and incidence

patterns of maternal complications in early-onset hypertension of pregnancy. BJOG 2007;114:741–50.30 Ananth CV, Kaminsky L, Getahun D, Kirby RS, Vintzileos AM. Recurrence of fetal growth restriction in

singleton and twin gestations. J Matern Fetal Neonatal Med 2009;22:654–61.31 Sibai BM, Ramadan MK, Chari RS, Friedman SA. Pregnancies complicated by HELLP syndrome (hemolysis,

elevated liver enzymes, and low platelets): subsequent pregnancy outcome and long-term prognosis. Am J Obstet Gynecol 1995;172(1):125–9.

32 Schuit E, Groenwold RH, Harrell Jr FE, de Kort WL, Kwee A, Mol BW, et al. Unexpected predictor-outcome associations in clinical prediction research: causes and solutions. CMAJ 2013;185(10): E499–505.

33 Conde-Agudelo A, Belizan JM, Norton MH, Rosas-Bermúdez A. Effect of the interpregnancy interval on perinatal outcomes in Latin America. Obstet Gynecol 2005;106:359–66.

34 Caughey AB, Stotland NE, Washington AE, Escobar GJ. Maternal ethnicity, paternal ethnicity, and parental ethnic discordance: predictors of preeclampsia. Obstet Gynecol 2005;106(1):156–61.202.

36927 Oostwaard.indd 92 27-10-15 10:16

Recurrence of term HDP

93

4

Contribution to AuthorshipMiriam van Oostwaard and Josje Langenveld are the main authors, having designed the study, performed part of the data collection and did most of the writing of the article. Irene Beune, Hilde van Susante, Roos Ramaekers and Kiki Wigny carried out most of the data collection in the various institutes and reviewed the article. Ewoud Schuit performed the statistical tests, wrote the ‘statistics’ section and reviewed the article. Dimitri Papatsonis, Ben Willem Mol and Wessel Ganzevoort contributed to the design of the study, assisted and supervised with data collection and reviewed the article.

36927 Oostwaard.indd 93 27-10-15 10:16