cardio sequelae of preeclampsia-eclampsia

Women's Health

Cardiovascular sequelae of preeclampsia/eclampsia:A systematic review and meta-analysesSarah D. McDonald, MD, MSc,a Ann Malinowski, MSc, MD,b Qi Zhou, PhD,c Salim Yusuf, MD, PhD,d,e

and Philip J. Devereaux, MD, PhDc,d Hamilton, Ontario, Canada

Background Preeclampsia affects 3% to 5% of gestations and eclampsia 0.05% to 0.93%, but their subsequentcardiovascular sequelae are unclear. The aim of this study was to determine if women with a history of preeclampsia/eclampsia are at increased risk of long-term cardiovascular sequelae.

Methods From Medline and Embase searches, we included case-control and cohort studies that examined cardiac,cerebrovascular or peripheral arterial disease, or cardiovascular mortality N6 weeks postpartum, in women with and without ahistory of preeclampsia/eclampsia and that controlled for or matched for confounders. Two independent reviewers determinedstudy eligibility and extracted data.

Results Five case-control and 10 cohort studies met eligibility criteria, with a total of 116,175 women with and2,259,576 women without preeclampsia/eclampsia. Most studies focused on women b56 years of age. Relative to womenwith uncomplicated pregnancies, women with a history of preeclampsia/eclampsia had an increased risk of subsequentcardiac disease in both the case-control studies (odds ratio 2.47, 95% CI 1.22-5.01) and the cohort studies (relative risk [RR]2.33, 1.95-2.78), as well as an increased risk of cerebrovascular disease (RR 2.03, 1.54-2.67), peripheral arterial disease(RR 1.87, 0.94-3.73), and cardiovascular mortality (RR 2.29, 1.73-3.04). Meta-regression revealed a graded relationshipbetween the severity of preeclampsia/eclampsia and the risk of cardiac disease (mild: RR 2.00, 1.83-2.19, moderate: RR2.99, 2.51-3.58, severe: RR 5.36, 3.96-7.27, P b .0001).

Conclusions Women with a history of preeclampsia/eclampsia have approximately double the risk of early cardiac,cerebrovascular, and peripheral arterial disease, and cardiovascular mortality. Further research is needed to determine themechanisms underlying these associations and to identify effective prevention strategies. (Am Heart J 2008;156:918-30.)

BackgroundPreeclampsia is one of the most common complications

of pregnancy, affecting 3% to 5% of gestations. It isdefined as the onset of hypertension and proteinuria after20 weeks gestation.1 Eclampsia, the concomitant devel-opment of seizures, is a rare complication of preeclamp-sia in the developed world (with an incidence of 1 in

From the aDivision of Maternal Fetal Medicine, Department of Obstetrics and GynecologyMcMaster University, Hamilton, Canada, bDepartment of Obstetrics and GynecologyMcMaster University, Hamilton, Canada, cDepartment of Clinical Epidemiology andBiostatistics, McMaster University, Hamilton, Canada, dDivision of Cardiology, Deparment of Medicine, McMaster University, Hamilton, Canada, and ePopulation HealtResearch Institute, McMaster University, Hamilton, Canada.This study was supported by the Regional Medical Association Scholarship Fund oHamilton Health Sciences. Dr Yusuf holds the Heart and Stroke Foundation Chair inCardiology. Dr Devereaux is supported by a Canadian Institutes of Health Research NewInvestigator Award.Submitted February 29, 2008; accepted June 26, 2008.Reprint requests: Sarah D. McDonald, MD, MSc, Division of Maternal-Fetal MedicineDepartment of Obstetrics and Gynecology, McMaster University, 1200 Main St WestHSC 3N52B, Hamilton, Ontario, Canada L8N 3Z5.E-mail: [email protected]/$ - see front matter© 2008, Mosby, Inc. All rights reserved.doi:10.1016/j.ahj.2008.06.042

,,

t-h

f

,,

2,000 gestations2) but remains relatively common else-where (0.93%3). Preeclampsia/eclampsia is thought tooriginate from reduced uteroplacental perfusion thatdevelops as a result of abnormal invasion of spiralarterioles.4 Preeclampsia/eclampsia shares many of thesame proposed etiologies as cardiovascular disease (CVD)including diseased arteries5,6 inflammation, and hyper-coagulability.7 Preeclampsia and CVD also share manyrisk factors,8 including obesity,9 insulin resistance,10

dyslipidemia, and endothelial dysfunction.11,12

Although preeclampsia/eclampsia has well-establishedcardiovascular (particularly cerebrovascular) risks at thetime of the affected pregnancy, whether there are long-term cardiovascular consequences is debated.13 Accurateunderstanding of the association between preeclampsia/eclampsia and long-term cardiovascular outcomesrequires a systematic, comprehensive, and unbiasedaccumulation and summary of the available evidence. Weundertook a systematic review of the literature todetermine if women with a history of preeclampsia/eclampsia are at increased risk of cardiac disease,cerebrovascular disease, peripheral vascular disease, orcardiac mortality beyond 6 weeks postpartum comparedto women without a history of preeclampsia/eclampsia.

mailto:[email protected]

http://dx.doi.org/10.1016/j.ahj.2008.06.042

McDonald et al 919American Heart JournalVolume 156, Number 5

MethodsWe followed the MOOSE (Meta-analysis of Observational

Studies in Epidemiology) consensus statement on the conduct ofmeta-analysis of observational studies.14

Eligibility criteriaWe included all case-control and cohort studies that examined

the development of cardiac disease, cerebrovascular disease,peripheral vascular disease, or cardiac mortality occurringN6 weeks postpartum in women with a history of preeclampsia/eclampsia compared with women with unaffected pregnancies.Because we were interested in knowing whether preeclampsia/eclampsia is independently associated with long-term majorcardiovascular events, we required that studies adjusted ormatched for potential confounders. Studies published in anylanguage were eligible for this review. We excluded duplicatepublications and studies involving ≤10 patients.

Search strategyMedline (1966-January 2006) and Embase (1980-January 2006)

were searched with the help of an experienced librarian usingseparate comprehensive search strategies for each database. Wealso reviewed the reference lists of all eligible studies.

Study selectionTwo independent reviewers (S.D.M. and A.M.) screened the

title and abstract of all citations identified in the search. The full-text article was retrieved if either reviewer considered thecitation potentially relevant. The same 2 reviewers indepen-dently evaluated the eligibility of the full-text article of allcitations selected in the screening process, and the ϕ value (ie,chance independent agreement) was 0.83. Disagreements wereresolved through a consensus process whereby reviewersdiscussed the reasoning behind their decisions, and in all cases,one reviewer realized they had made an error. An independentthird adjudicator (P.J.D.) was available if uncertainty ordisagreement existed.

Outcome measuresOur primary outcome was a composite of cardiac disease

(ischemic heart disease, coronary artery disease, myocardialinfarction, congestive heart failure, and death from any ofthese). Secondary outcomes included cerebrovascular disease(including mortality), peripheral arterial disease, and cardiovas-cular mortality (including death from cardiac, cerebrovascular,or peripheral arterial disease, although no studies reported ondeath from the latter). We required that outcomes occurred atleast 6 weeks postpartum and obtained information fromauthors where necessary.

Data abstractionTwo reviewers (S.D.M. and A.M.) independently abstracted

the following data from full-text articles: country of origin,period of study, study design, characteristics of cases andcontrols, and duration of follow-up. A pilot data extractionform was created and modified before collection of final data.Disagreements were resolved through the consensus processdescribed above. Authors were contacted for further infor-mation where necessary.

Quality assessmentTwo reviewers (S.D.M. and A.M.) independently performed

the quality assessment and abstracted the following factors:sampling process, completeness of follow-up, the likelihood ofdetecting preeclampsia/eclampsia (in the case-control studies)or CVD (in the cohort studies), and confounders for which therewas adjustment or matching. A priori, it was judged thatadjustment for the following variables was appropriate: age andother traditional cardiovascular risk factors (hypertension,hyperlipidemia, diabetes or impaired glucose tolerance, familyhistory of CVD, and smoking). Disagreements were resolvedusing the above-described consensus process. The 2 reviewershad a raw agreement of 73 of the 77 adjustment/matchingvariables, with an associated ϕ value of 0.45.

Data analysis and synthesisFor the case-control studies, we computed the pooled odds

ratio, and for the cohort studies, the pooled relative risk of anoutcome in the women with preeclampsia/eclampsia relative tothe women with uncomplicated pregnancies. We pooledadjusted study results using a random effects model.15 Forindividual studies that included data on N1 severity ofpreeclampsia (such as mild, moderate, and severe), the riskestimates for individual severities were pooled to create acomposite risk estimate for all severities of preeclampsia foreach particular study.We calculated an I

2 value as a measure of heterogeneity forour pooled analyses. An I

2 value represents the percentage oftotal variation across studies due to heterogeneity rather thanchance, and we considered an I2 value ≤25% as low.16 Beforethe analysis, we specified several hypotheses to explainheterogeneity (ie, an I2 value N25%) in the direction andmagnitude of effect among studies. We hypothesized that theeffect size might differ based on the duration of follow-up inthe cohort studies (N20 years vs ≤20 years), the data source(information directly from the physician or patient vs othersource), the completeness of patient follow-up (N80% vs≤80%), adequacy of adjustment (≥2 traditional cardiac riskfactors vs b2), or the severity of the preeclamptic episode(mild, moderate, or severe). Given the large number ofanalyses (20), we set a threshold for significance of P b .01 todiminish the likelihood that a significant finding would occurby chance alone.Data from the studies that presented results based on the

severity of preeclampsia/eclampsia were included in a meta-regression analysis examining the graded relationship betweenthe severity of preeclampsia/eclampsia and the risk of cardiacdisease. We classified the severity of the preeclamptic/eclampticepisode as follows: “mild” represented uncomplicated pree-clampsia, “moderate” represented preeclampsia complicated byeither seizures (eclampsia) or poor fetal growth, and “severe”represented preeclampsia/eclampsia complicated by pretermdelivery and/or fetal death. In the meta-regression model, thelog-transformed RR of CVD from the severity levels was treatedas the dependent variable, and the severity levels of pre-eclampsia/eclampsia were used as the independent variable.The analysis was weighted by the inverse of the variance of thedependent variable.We performed post hoc sensitivity analyses excluding Funai

et al17 and Kestenbaum et al18 as they reported compositeoutcomes involving N1 of our prespecified outcomes, and

Figure 1

Study process.

920 McDonald et alAmerican Heart Journal

November 2008

they were unable to supply data for the individual outcomeswhen requested.For the meta-analysis, we used Review Manger (RevMan 4.2,

The Cochrane Collaboration, Oxford, United Kingdom), and forthe meta-regression analysis, we used SAS 9.1 (SAS Institute,Cary, NC). The Hamilton Health Sciences Research Ethics Boardapproved the study.

ResultsDescription of studiesFour hundred ninety nonduplicate titles and abstracts

were identified in our search (Figure 1). Based on ourscreening process, 66 citations were selected to undergo

full-text article review. The most common reasons forexclusion were study design and failure to report anycardiovascular outcomes.Five case-control studies19-23 and 10 cohort stu-

dies17,18,24-31 were included, involving a total of 118,990women with a history of preeclampsia/eclampsia and2,259,576 women with unaffected pregnancies (Table I).Earlier studies typically did not define preeclampsia (alsocalled toxemia), whereas later studies used the standarddefinition of hypertension during pregnancy with pro-teinuria or edema or both. The studies originated mainlyfrom North America and Europe, and all were in thedeveloped world. Importantly, most of the studiesreported outcomes in relatively young women. Thelargest study, by Ray et al30 in 2005, observed women fora median of 8 years after delivery and found a mean age of38 years at first hospitalization for CVD or revasculariza-tion. Ten of the remaining studies also focused onrelatively young women, b56 years of age.Considering the quality of the studies, most sampling

methods were acceptable, most studies had completefollow-up, and there was probably little bias in thedetection of CVD in the cohort studies although therewas the potential for recall bias of preeclampsia/eclampsia in the case-control studies (Table II). The largerstudies typically adjusted for more confounders than thesmaller ones.Relative to women with uncomplicated pregnancies,

women with a history of preeclampsia/eclampsia had anincreased risk of subsequent cardiac disease, ourprimary outcome, in both the 4 case-control studies(odds ratio 2.47, 95% CI 1.22-5.01) (Figure 2) and the 10cohort studies (RR 2.33, 95% CI 1.95-2.78) (Figure 3).The single eligible case-control study that examined therisk of cerebrovascular disease (Thorogood23) reportedan increased risk of 2.6 (95% CI 1.5-4.3), in keepingwith the pooled estimate in the 6 cohort studies (whichincluded death from cerebrovascular disease, RR 2.03,95% CI 1.54-2.67) (Figure 4). The cohort studies alsodemonstrated that preeclamptic/eclamptic women hada trend toward an increased risk of subsequentperipheral arterial disease (3 cohort studies, RR 1.87,95% CI 0.94-3.73) (Figure 5) and a significant increase incardiovascular mortality (5 cohort studies, RR 2.29, 95%CI 1.73-3.04) (Figure 6).The heterogeneity in the pooled risk estimates of our

outcomes was moderate, ranging from an I2 test =

35.7% (for cardiac disease in the case-control studies) to66.3% (for peripheral arterial disease). The observedheterogeneity was not explained by our a priori–defined hypotheses (follow-up for N20 years vs≤20 years, N80% follow-up vs ≤80%, informationdirectly from the physician/patient vs other source, andadjustment of ≥2 traditional cardiac risk factors vs b2).For instance, the risk estimates of disease afterpreeclampsia were similar for studies that adjusted for

Table I. Characteristics of included studies

Type of study Author, year CountryNo. of cases/ exposed topreeclampsia/eclampsia

Case-control Mann et al,19 ⁎ 1976 England 77Rosenberg et al,22 1983 USA 255

Croft and Hannaford,21 ‡ 1989 England 39

Thorogood et al,23 ⁎ 1992 England and Wales 71

Haukkamaa et al,20 2004 Finland 141

Total for case-control studies 583Cohort Jonsdottir et al,25 1995 Iceland 203

Hannaford et al,24 ⁎, ‡ 1997 England 3000 ⁎

Irgens et al,29 ⁎ 2001 Norway 24 155

Smith et al,28 ⁎ 2001 Scotland 18 487

Kestenbaum et al,18 ⁎ 2003 USA 20 552

Wilson et al,27 2003 Scotland 1043

Funai et al,17 2005 Jerusalem 1055

Kaaja et al,26 ⁎ 2005 Finland 397

Ray et al,30 ⁎ 2005 Canada 36 982

Wikstrom et al,31 2005 Sweden 12 533

Total for cohort studies 118 407Total for case-control and cohort studies 118 990

MI, Myocardial infarction; N/A, not available; CVA, cerebrovascular disease; IHD, ischemic heart disease or coronary artery disease; PVD, peripheral vascular disease; pt-yr,patient-years.⁎Authors provided information at our request, including where necessary, data on women who were N6 weeks postpartum.†Approximate calculation based on figures in article.‡Both used information from Royal College of General Practitioners (RCGP) oral contraceptive study, however, different outcomes.


No. of controls/unexposed

Outcome(s); averageage at outcome (y)

Average durationof follow-up (y);

range (where stated)No. of events or

rate (cohort studies)

207 Nonfatal MI; N/A; included b45 N/A Not applicable802 First nonfatal MI;

44 in preeclamptics,42 in controls (median);included b50

N/A Not applicable

54 First MI; b40 (16%);b55 (78%)

19 Not applicable

770 CVA; N/A; averageage not available, ⁎included 16-39

N/A Not applicable

99 IHD; 52 in preeclamptics,55 in controls (median);included b66

30 (median); 24-36 Not applicable

19327340 † IHD death; 64 (mean) 42 (mean); 1-60 39 deaths in preeclamptics,

11 in eclamptics, 17 in unexposed18 451 ⁎ Total IHD

(first episode of acute MI,other acute IHD, chronic IHD),CVA, PVD; N/A

25-26 (not stated ifmean/median)

0.61/1000 pt-yr in preeclamptics,0.35/1000 in unexposed

602 117 CV death, CVA death; N/A 13 (median); 0-25 † 143 deaths in preeclamptics≥37 wk (= 6.6/1000),41 in preeclamptics deliveredat 16-36 wk (= 15.5/1000)

84 487 IHD admission ordeath; 38-39 ⁎ (median)

15-19 N/A

132 069 ⁎ first acute CV event(MI, CVA, revascularization);36 (mean); included 15-45

7.8 (mean) 118 total, in severe preeclamptics 11(= 29.4/100 000 pt-yr),in mild 24 (= 18.4/100 000 pt-yr)

796 Admission or deathfrom CVA, IHD(includes heart failure);60 (mean)

(mean/median not stated) 10-48 IHD admissions in preeclampticsand in unexposed, 48 and 38,respectively; 26 and 10 IHD deaths,respectively; 16 and 5 CVA deaths,34 and 13 CVA admissions,9 and 3 diagnoses ofintermittent claudication

36 858 CV death (including CVA,sudden death); 53 (median)

30 (median); 24-36 41 CV deaths in preeclamptics;269 in unexposed

3162 IHD in the last 12 mo;48 in preeclamptics,46 in unexposed (mean);included 25-64

28 (not stated if mean/median) 2.9% rate of atherosclerosis duringlast 12 mo in preeclamptics(n = 12), 0.7% in unexposed (n = 22)

989 283 † Hospitalization for orrevascularization forCV event (CAD, CVA, PVD);38 (median); included 14-50

8.7 (median) 305 in patients with MPSincluding preeclampsia =500/million pt-yr

383 081 Fatal or nonfatal IHD;48 (median), 33-75 range

19-28 (mean/mediannot stated)

8.6 IHD/10 000 pt-yr after mildpreeclampsia, 12.8 for severepreeclampsia (4.1 in unexposed),123 mild preeclamptics,53 severe, 2306 unexposed

2 257 6442 259 576


November 2008

Table II. Quality assessment of included studies

Type ofstudy Author Data source Sampling process

Case-control Mann et al19 Questionnaire (i) with interviewer or if refused,mailed (ii) to patient or (iii) general practitioner

All married cases and random selection of controls fromhospital diagnostic indexes from 3 regions in England(West Metropolitan, Oxford, and Wessex), 1968-72

Rosenberg et al22 Hospital interview using standard questionnaire Refusal rate of 19% and 32%, respectively, in cases andhospital controls from 155 hospitals in 3 geographicregions (greater Boston, New York, and Delaware Valley)—only easily accessible cases included, however, controlsdrawn from similar area

Croft andHannaford21

Medical records Random selection of controls; cases and controls fromsame population for RCGP oral contraception study,1968-69 in United Kingdom

Thorogood et al23 General practitioners interviewed about past medicaland obstetric histories using structured questionnaire;clinical notes, computed tomographic scan reports,discharge letters, and postmortem reports reviewed

All female CV deaths aged 16-39 y; randomselection of controls from same general practitioner ascase in England and Wales, 1986-88

Haukkamaa et al20 History from self-administered questionnairesconfirmed through hospital records

Controls from private gynecology clinics for annualcheckups; cases from University center in Helsinki,Finland, 1996-2000

Cohort Jonsdottir et al25 Hospital and autopsy records All women from National Hospital birth cohort inReykjavik, Iceland, 1931-47

Hannaford et al24 General practitioners Original cohort inclusion for RCGP oral contraceptionstudy was ‘broad,’ 1968-69 in United Kingdom

Irgens et al29 Birth and death registries All births N16 wk in Norway, 1967-92

Smith et al28 Birth and death registries All singleton first births in Scotland, 1981-85

Kestenbaum et al18 Birth and death registries All live singleton births 1987-97 in Washington state;random selection of controls

Wilson et al27 Aberdeen maternity databank All preeclamptics; random selection of controls fromAberdeen, Scotland, 1951-70

Funai et al17 Data from labor ward logbooks (Jerusalem perinatalstudy) with linkage to death registry(population-based cohort)

All births in the region of Jerusalem, Israel from 1964-76

Kaaja et al26 FINRISK study (population survey studyingCV risk factors) abstraction of data from laborward logbooks

Random selection of Finnish population in 2002;71.2%-79.2% participation rate across the region

Ray et al30 Birth and death registries; hospitalization database All hospital deliveries in Ontario, Canada, 1990-2004

Wikstrom et al31 Birth and death registries; hospitalization database All hospital deliveries in Sweden from 1973-86 (N99% ofall births in Sweden)

RCGP, Royal College of General Practitioners; HRT, hormone replacement therapy; DM, diabetes mellitus; GDM, gestational diabetes mellitus; ICD, International Classification ofDiseases; IDDM, insulin dependent diabetes mellitus; HTN, hypertension; BMI, body mass index.


≥2 traditional risk factors compared to those thatadjusted for b2, respectively (RR of cardiac disease 2.37,95% CI 1.90-2.97 vs 2.23, 95% CI 1.57-3.18 [P = .73], RRof cerebrovascular disease 2.08, 95% CI 1.51-2.86 vs

2.37, 95% CI.33-4.23 [P = .70], RR of peripheral arterialdisease 1.94, 95% CI 0.80-4.72 vs 1.56, 95% CI 0.42-5.84[P = .79], and RR of cardiovascular mortality 3.07, 95%CI 2.18-4.33 vs 2.01, 95% CI 1.55-2.61 [P = .06]).

Completeness offollow-up

Detection of preeclampsia/eclampsiain case-control studies and CVD

in cohort studiesCovariates that were adjusted

for or matched

3 cases and 33 controls could notbe traced and were excluded

By questionnaire—potential recall bias Age, hypertension, smoking, hospital,marital status, oral contraceptives, year of admission

No information By hospital interview–potential recall bias Age, hypertension, smoking, geographic locationof hospital, oral contraceptives

No information Medical records reviewed for a history ofpreeclampsia at baseline, before MI

Age, hypertension, smoking, social class,oral contraceptives, hysterectomy, or HRT use

No information Focus of study oral contraceptives, therefore,biased recall of preeclampsia less likely

Age ± 5 y, marital status

10 cases and 3 controls with nomaternity data were excluded

Diagnosis of atherosclerosis by angiography Age, DM, hypertension, dyslipidemia, smoking,obesity, hormone-replacement therapy,interval from menarche to menopause ≥37 y

99.5% traced Hospital and autopsy records reviewed byinvestigator using standardized definitions ofcardiac disease, unaware of pregnancy diagnosis

Age, year of death

~75% of original cohort lostto follow-up

Baseline checklist of medical problems;general practitioners then reported new illnesses,medications, cause of death on ongoing basis

Age, smoking, social class

No information ICD codes used to determine cause of deathin registry

Age at delivery, year of delivery

95.6% complete data ICD codes used to determine cause of deathin registry

Age at delivery, essential hypertension,socioeconomic class, maternal height,lowest birth weight quintile, preterm delivery

Loss of N20% of cohort (but similarbaseline characteristics of traced andnot traced women)

ICD codes used to determine reasonfor hospitalization in registry

Age, GDM, smoking, race, type of insurance,parity, cesarean delivery, delivery year(ruled out chronic HTN)

65.5% of original cohort traced usingprobability matching; traced womenwere older, higher BMI, highersocioeconomic status

ICD codes used to determine reasonfor hospitalization in registry

Age at delivery, social class

92% traced ICD codes used to determine causeof death in registry

Age, IDDM, GDM, social class, education,year of delivery, infant birthweight b2500 g

No information All data self-reported Age, DM, impaired glucose tolerance,hypertension, increased cholesterol, BMI,parity, mother's MI status, age at delivery

No information ICD codes used cause of hospitalization,death in registries; databases have a sensitivityof 55%-88% for a diagnosis of coronaryartery disease and specificity of 92%-98%

Age, DM, GDM, hypertension, dyslipidemia,smoking (not well coded in database),obesity, income quintile, rural residence,drug dependence, renal disease, multiples,systemic lupus erythematosus, length of stay

No information; before 1987,(when follow-up for IHD began due tolimited coverage in hospitalizationregister before 1987), 2.8% had emigrated,0.35% had died

ICD codes used cause of hospitalization,death in registries; the positive predictivevalue of ICD coding for preeclampsia duringstudy period was 68% (because of a missingsecond record of proteinuria)

Age, socioeconomic level, category of hospitalin which first child born (ruled out diabetes,essential hypertension)


November 2008

In further attempt to explain heterogeneity, post hocanalyses were performed separating fatal from nonfatalcardiac disease (RR 2.53, 95% CI 2.01-3.17, I2 test =6.9% and RR 1.90, 95% CI 1.28-2.82, I2 test = 77.3%,

respectively) and similarly separating fatal from nonfatalcerebrovascular disease (RR 2.26, 95% CI 1.11-4.57,I2 test = 74.4% and RR 1.93, 95% CI 1.52-2.45, I2 test =23.9%, respectively).

Figure 2

Forest plot of the risk of cardiac disease after preeclampsia/eclampsia in case-control studies. Cardiac disease was defined as ischemic heardisease, coronary artery disease, myocardial infarction, congestive heart failure, and death from any of these. Sizes of data markers indicate theweights of each study in the analysis. Random indicates that the random effects model was used for statistical pooling.

Figure 3

Forest plot of the risk of cardiac disease after preeclampsia/eclampsia in cohort studies. Cardiac disease was defined as ischemic heart disease,coronary artery disease, myocardial infarction, congestive heart failure, and death from any of these. For studies that reported risk estimates fordifferent severities of preeclampsia, data were pooled across all levels of severity of preeclampsia (personal communication from Dr Joel Ray, MD,MSc [February 2007] for estimates of risk for women with preeclampsia and from Dr Rolv Lie, MD [April 2007] for events occurring at least 6weeks postpartum). Sizes of data markers indicate the weights of each study in the analysis. Random indicates that the random effects model wasused for statistical pooling.


Effect of severe forms of preeclampsiaMeta-regression revealed a graded relationship

between the severity of preeclampsia/eclampsia andthe risk of our primary outcome, cardiac disease asfollows: mild preeclampsia, RR 2.00 (95% CI 1.83-2.19); moderate preeclampsia, RR 2.99 (95% CI 2.51-3.58); and severe preeclampsia, RR 5.36 (95% CI 3.96-7.27), P b .0001 (Table III, Figure 7). The I

2 valuessuggested homogeneity of data across each of thecategories of risk (0% for each category mild,moderate, and severe).Similarly, the risk of cerebrovascular mortality was

higher after preeclampsia complicated by preterm

t

delivery (hazard ratio [HR] 5.08, 95% CI 2.09-12.55)than preeclampsia with delivery at term (HR 0.98, 95%CI 0.50-1.91) in study of Irgens et al,29 P = .004. Therewas a nonsignificant trend in the study of Kestenbaumet al18 toward an increased risk of cerebrovasculardisease after severe preeclampsia (HR 3.3, 95% CI 1.7-6.5) compared to mild preeclampsia (HR 2.2, 95% CI1.3-3.6), P = .34.Post hoc sensitivity analyses excluding studies that

presented data as composite outcomes including N1 ofour individual outcomes (Funai et al,17 2005 andKestenbaum et al,18 2003) yielded similar results as thepooled estimates including them. For instance, the

Figure 4

Forest plot of the risk of cerebrovascular disease after preeclampsia/eclampsia in cohort studies. Cerebrovascular disease was defined as includingmortality from cerebrovascular disease. For studies that reported risk estimates for different severities of preeclampsia, data were pooled across alllevels of severity of preeclampsia (personal communication from Dr Joel Ray, MD, MSc [February 2007] for estimates of risk for women withpreeclampsia and from Dr Rolv Lie, MD [April 2007] for events occurring at least 6 weeks postpartum). Sizes of data markers indicate the weightsof each study in the analysis. Random indicates that the random effects model was used for statistical pooling.

Figure 5

Forest plot of the risk of peripheral arterial disease after preeclampsia/eclampsia in cohort studies. Sizes of data markers indicate the weights ofeach study in the analysis. Random indicates that the random effects model was used for statistical pooling (personal communication from Dr JoelRay, MD, MSc [February 2007] for estimates of risk for women with preeclampsia).


November 2008

pooled RR including and then excluding Funai et al17

(2005) and Kestenbaum et al18 (2003) for cardiac disease(including death) were 2.34 (95% CI 1.91-2.87) and 2.19(95% CI 1.72-2.79), P = .70, respectively, and forcerebrovascular disease (including death), were 2.03(95% CI 1.54-2.67) and 1.69 (95% 1.32-2.17), P = .33,respectively.

Discussion

We determined that women with a history of pre-eclampsia/eclampsia have approximately double the riskof subsequent cardiac disease compared to women withuncomplicated pregnancies even after controlling formany confounders. Similarly increased risks of long-termcerebrovascular and peripheral arterial diseases andcardiovascular mortality among women with preeclamp-

sia/eclampsia were also found. The meta-regression maysuggest a “dose-response,” with higher risks of CVDnoted in women with a history of more severepreeclampsia, particularly those who experienced pre-term delivery, poor fetal growth, or fetal death. Theresults of this meta-analysis suggest that preeclampsia/eclampsia is as important a risk factor as many traditionalcardiac risk factors. Importantly, these women are at riskfor early onset CVD, with 11 of the 14 included studiesdemonstrating events in relatively young women,b56 years of age.Although a recent meta-analysis by Bellamy et al32

found similar results as we did in cardiac disease andcerebrovascular disease and mortality, our study addsnew information by (i) establishing an associationbetween preeclampsia and peripheral arterial disease(the consistency of the magnitude and direction of the

Figure 6

Forest plot of the risk of cardiovascular mortality after preeclampsia/eclampsia in cohort studies. Cardiovascular death was defined as includingdeath from cardiac disease (from ischemic heart disease, coronary artery disease, myocardial infarction, congestive heart failure), cerebrovasculardisease, and peripheral arterial disease (although no studies actually reported this latter outcome). For studies that reported risk estimates fordifferent severities of preeclampsia, data were pooled across all levels of severity of preeclampsia (personal communication from Dr Rolv Lie, MD[April 2007] for events occurring at least 6 weeks postpartum). Sizes of data markers indicate the weights of each study in the analysis. Randomindicates that the random effects model was used for statistical pooling.

Table III. Study data examining risk of subsequent cardiac disease or death with increasing severity of preeclampsia (see Figure 7)

Author ExposureSeverity level ⁎ of preeclampsia

for meta-regression Adjusted risk †

Jonsdottir et al25 Preeclampsia Mild 1.90 (1.05-3.52)Eclampsia Moderate 2.61 (1.11-6.12)

Smith et al28 Preeclampsia Mild 2.0 (1.5-2.5)Preeclampsia + poor fetal growth Moderate 3.3 (2.2-4.9)Preeclampsia + preterm delivery Severe 4.5 (2.7-7.4)Preeclampsia + pretermdelivery + poor fetal growth

Severe 7.0 (3.3-14.5)

Kestenbaum et al18 Preeclampsia (“mild”) Mild 2.2 (1.3-3.6)Preeclampsia (“severe”) Moderate 3.3 (1.7-6.5)

Irgens et al29 Preeclampsia Mild 1.69 (1.03-2.75)Preeclampsia + preterm delivery Severe 7.52 (3.85-14.66)

Ray et al30 Preeclampsia Mild 2.1 (1.8-2.4)Preeclampsia + poor fetal growth Moderate 3.1 (2.2-4.5)Preeclampsia + fetal death Severe 4.4 (2.4-7.9)

Wikstrom et al31 Preeclampsia Mild 1.9 (1.6-2.2)Preeclampsia (“severe”) Moderate 2.8 (2.2-3.7)

⁎As defined in Methods section.†Original studies all reported Hazard Ratios except Jonsdottir (RR) andWikstrom (incidence rate ratios) that were converted to RR (personal communication from Dr Rolv Lie, MD [April2007] for events occurring at least 6 wk postpartum and from Dr Joel Ray, MD, MSc [February 2007] for estimates of risk for women with preeclampsia).


risk across all forms of CVD strengthens the argumentof an association with preeclampsia), (ii) involvingmore data because of the inclusion of 2 additionalcohort studies (Jonsdottir et al,25 1995 and Kaaja et al,26

2005) as well as case-control studies (although case-control studies are subject to recall bias, the evidencefor a relationship between preeclampsia and CVD isstrengthened by the comparable degree of associationacross various study types), (iii) including a qualityassessment that is particularly important in observa-tional meta-analyses, and (iv) formally exploring the

association between preeclampsia of varying severity(mild, moderate, and severe) and subsequent cardiacdisease using a meta-regression.The increased risk of CVD in women with a history of

preeclampsia/eclampsia may be due to shared riskfactors,33 damage from the preeclampsia/eclampsia, orboth. Preeclampsia/eclampsia and CVD share an abnor-mal metabolism with increased insulin resistance, oftentermed (dys)metabolic syndrome.34 Awoman's utero-placental circulation may be affected by a dysmetabolicstate during her youth (leading to preeclampsia) in

Figure 7

Graph of the risk of cardiac disease according to severity of preeclampsia. Graph of the risk of cardiac disease (including death) afterpreeclampsia of mild, moderate, or severe severity. “Mild” represents uncomplicated preeclampsia, “moderate” represents preeclampsiacomplicated by either eclampsia or poor fetal growth, and “severe” represents preeclampsia complicated by preterm delivery and/or fetal death(personal communication from Dr Joel Ray, MD, MSc [February 2007] for estimates of risk for women with preeclampsia and from Dr Rolv Lie,MD [April 2007] for events occurring at least 6 weeks postpartum).


November 2008

advance ofmanifestations in her coronary circulation lateron in life.28,30 Several decades ago, Sibai et al35 suggestedthat pregnancy is a screen for later hypertension anddiabetes. This meta-analysis supports the idea thatpregnancy is “a stress test for life”36 and the developmentof CVD. It is also possible that preeclampsia/eclampsiamay damage the vascular endothelium or the kidneys.Preeclampsia/eclampsia is associated with endothelialdysfunction.37 Microalbuminuria, an early sign of renaldisease38,39 that is positively correlated with the presenceof ischemic heart disease and atherosclerotic riskfactors,40 is significantly increased in women with ahistory of preeclampsia.41,42 A 20% incidence of micro-albuminuria compared with 2% in controls (P b .05) wasfound in one study within 7 years of delivery.41

Limitations of this systematic review include residualconfounding and potential confounders that were notexplored in most of the original studies (such as chronicrenal disease and obesity) or at all (such as sedentarylifestyle). Several authors were unable to provide requesteddata. Some of the potentially highest risk women wereexcluded from some studies. For instance, Smith et al28

excluded women with fetal deaths or those whose infantsdelivered before 24weeks of gestation andweighed b500g.Strengths of the study include the large number of

patients, the consistency of the findings across all forms of

CVD, and the exploration of heterogeneity based on theseverity of preeclamptic episode with meta-regression.In conclusion, even after controlling for many tradi-

tional cardiovascular risk factors, women who have hadpreeclampsia/eclampsia are at substantially increasedrisk of CVD. With 3% to 5%30 of pregnancies complicatedby preeclampsia and with an average of 128.9 millionbirths per year,43 approximately 3.87 to 6.45 millionpregnancies per year are affected with preeclampsiaworldwide. Although the prevalence of all forms of CVDin women aged 45 to 54 may be approximately 2%,44 wefound that women with a history of preeclampsia had anRR of 2.33 (95% CI 1.95-2.78). Hence, there may beapproximately 150,815 to 360,920 additional women inthe world per year with CVD after preeclampsia.However, these numbers are likely underestimates asthey do not include peripheral arterial disease orasymptomatic disease, and moreover, the developingworld has not only higher rates of preeclampsia but alsoburgeoning rates of CVD.45,46 Further research is neededto understand the etiology of the increased CV risk, tothen explore if interventions can prevent these seriouscomplications of preeclampsia/eclampsia and how pre-eclampsia/eclampsia might be incorporated into CV riskassessment along with other new risk factors such aschronic renal disease and abdominal obesity. Research in


the prevention of preeclampsia/eclampsia is also critical.In the meantime, education of patients and health careproviders about this increased CV risk is an importantfirst step. Reducing modifiable risk factors such asimproving diet, exercising, and eliminating smoking is alogical goal for all patients and particularly those who areat increased CV risk such as women with a history ofpreeclampsia/eclampsia.

We gratefully acknowledge the additional data and

clarification provided by Drs. Joel Ray, P. Hannaford,

L. Irgens, R. Kaaja, B. Kestenbaum, J. Mann, M.

Thorogood, L. Rosenberg, G.C. Smith, and N. Poulter.

We are indebted to our translators including Dr. Arne

Ohlsson and Shizuka Aparcero.

References1. Roberts JM, Cooper DW. Pathogenesis and genetics of pre-

eclampsia. Lancet 2001;357:53-6.2. Douglas KA, Redman CW. Eclampsia in the United Kingdom. BMJ

1994;309:1395-400.3. Geographic variation in the incidence of hypertension in pregnancy.

World Health Organization International Collaborative Study ofHypertensive Disorders of Pregnancy. Am J Obstet Gynecol 1988;158:80-3.

4. Granger JP, Alexander BT, Llinas MT, et al. Pathophysiology ofhypertension during preeclampsia linking placental ischemia withendothelial dysfunction. Hypertension 2001;38:718-22.

5. Zeek PM, Assali NS. Vascular changes in the decidua associated witheclamptogenic toxemia of pregnancy. Am J Clin Pathol 1950;20:1099-109.

6. Brosens JJ, Pijnenborg R, Brosens IA. The myometrial junctional zonespiral arteries in normal and abnormal pregnancies: a review of theliterature. Am J Obstet Gynecol 2002;187:1416-23.

7. Said J, Dekker G. Pre-eclampsia and thrombophilia. Best Pract ResClin Obstet Gynaecol 2003;17:441-58.

8. Ray JG. Dysmetabolic syndrome, placenta-mediated disease andfuture risk of cardiovascular disease. Fetal Matern Med Rev 2004;15:1-16.

9. O'Brien TE, Ray JG, Chan WS. Maternal body mass index and therisk of preeclampsia: a systematic overview. Epidemiology 2003;14:368-74.

10. Innes KE, Wimsatt JH, McDuffie R. Relative glucose tolerance andsubsequent development of hypertension in pregnancy. ObstetGynecol 2001;97:905-10.

11. Roberts JM. Endothelial dysfunction in preeclampsia. Semin ReprodEndocrinol 1998;16:5-15.

12. Chambers JC, Fusi L,Malik IS, et al.Associationofmaternal endothelialdysfunction with preeclampsia. JAMA 2001;285:1607-12.

13. Chesley LC. Remote prognosis after eclampsia. Perspect NephrolHypertens 1976;5:31-40.

14. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observationalstudies in epidemiology: a proposal for reporting. Meta-analysis OfObservational Studies in Epidemiology (MOOSE) group. JAMA2000;283:2008-12.

15. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control ClinTrials 1986;7:177-88.

16. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency inmeta-analyses. BMJ 2003;327:557-60.

17. Funai EF, Friedlander Y, Paltiel O, et al. Long-term mortality afterpreeclampsia. Epidemiology 2005;16:206-15.

18. Kestenbaum B, Seliger SL, Easterling TR, et al. Cardiovascular andthromboembolic events following hypertensive pregnancy. Am JKidney Dis 2003;42:982-9.

19. Mann JI, Doll R, Thorogood M, et al. Risk factors for myocardialinfarction in young women. Br J Prev Soc Med 1976;30(2):94-100.

20. Haukkamaa L, Salminen M, Laivuori H, et al. Risk for subsequentcoronary artery disease after preeclampsia. Am J Cardiol 2004;93:805-8.

21. Croft P, Hannaford PC. Risk factors for acute myocardial infarction inwomen: evidence from the Royal College of General Practitioners'oral contraception study. BMJ 1989;298:165-8.

22. Rosenberg L, Miller DR, Kaufman DW, et al. Myocardial infarction inwomen under 50 years of age. JAMA 1983;250:2801-6.

23. Thorogood M, Mann J, Murphy M, et al. Fatal stroke and use of oralcontraceptives: findings from a case-control study. Am J Epidemiol1992;136:35-45.

24. Hannaford P, Ferry S, Hirsch S. Cardiovascular sequelae of toxaemiaof pregnancy. Heart (British Cardiac Society) 1997;77:154-8.

25. Jonsdottir LS, Arngrimsson R, Geirsson RT, et al. Death rates fromischemic heart disease in women with a history of hypertension inpregnancy. Acta Obstet Gynecol Scand 1995;74:772-6.

26. Kaaja R, Kinnunen T, Luoto R. Regional differences in the prevalenceof pre-eclampsia in relation to the risk factors for coronary arterydisease in women in Finland. Eur Heart J 2005;26:44-50.

27. Wilson BJ, Watson MS, Prescott GJ, et al. Hypertensive diseases ofpregnancy and risk of hypertension and stroke in later life: resultsfrom cohort study. BMJ 2003;326:845.

28. Smith GC, Pell JP, Walsh D. Pregnancy complications and maternalrisk of ischaemic heart disease: a retrospective cohort study of129,290 births. Lancet 2001;357:2002-6.

29. Irgens HU, Reisaeter L, Irgens LM, et al. Long term mortality of mothersand fathers after pre-eclampsia: population based cohort study. BMJ2001;323:1213-7.

30. Ray JG, Vermeulen MJ, Schull MJ, et al. Cardiovascular health aftermaternal placental syndromes (CHAMPS): population-based retro-spective cohort study. Lancet 2005;366:1797-803.

31. Wikstrom AK, Haglund B, Olovsson M, et al. The risk of maternalischaemic heart disease after gestational hypertensive disease. BJOG2005;112:1486-91.

32. Bellamy L, Casas JP, Hingorani AD, et al. Pre-eclampsia and risk ofcardiovascular disease and cancer in later life: systematic review andmeta-analysis. BMJ 2007;335:974.

33. Magnussen EB, Vatten LJ, Lund-Nilsen TI, et al. Prepregnancycardiovascular risk factors as predictors of pre-eclampsia: populationbased cohort study. BMJ 2007;335:978.

34. Rodie VA, Freeman DJ, Sattar N, et al. Pre-eclampsia andcardiovascular disease: metabolic syndrome of pregnancy. [Review][187 refs] Atherosclerosis 2004;175:189-202.

35. Sibai BM, el-Nazer A, Gonzalez-Ruiz A. Severe preeclampsia-eclampsia in young primigravid women: subsequent pregnancyoutcome and remote prognosis. Am J Obstet Gynecol 1986;155:1011-6.

36. Williams D. Pregnancy: a stress test for life. Curr Opin ObstetGynecol 2003;15:465-71.

37. Roberts JM, Taylor RN, Goldfien A. Endothelial cell activation as apathogenetic factor in preeclampsia. Semin Perinatol 1991;15:86-93.

38. Friedman S, Rabinerson D, Bar J, et al. Microalbuminuriafollowing gestational diabetes. Acta Obstet Gynecol Scand 1995;74:356-60.


November 2008

39. Agrawal B, Berger A, Wolf K, et al. Microalbuminuria screening byreagent strip predicts cardiovascular risk in hypertension. J Hypertens1996;14:223-8.

40. Jensen JS, Feldt-Rasmussen B, Strandgaard S, et al. Arterialhypertension, microalbuminuria, and risk of ischemic heart disease.Hypertension 2000;35:898-903.

41. Nisell H, Lintu H, Lunell NO, et al. Blood pressure and renal functionseven years after pregnancy complicated by hypertension. Br J ObstetGynaecol 1995;102:876-81.

42. North RA, Simmons D, Barnfather D, et al. What happens to womenwith preeclampsia? Microalbuminuria and hypertension followingpreeclampsia. Aust N Z J Obstet Gynaecol 1996;36:233-8.

ON THE MOVE

Don’t miss a single issue of the journal! To ensure prompt servand complete the form below.

Please send your change of address notification at least 6 weregret we cannot guarantee replacement of issues missed bec

JOURNAL TITLE:Fill in the title of the journal here. _____________________

OLD ADDRESS: N

Affix the address label from a recent issue of the journal here. C

N

A

C

COPY AND MAIL THIS FORM TO: OR FAX TOElsevier Subscription Customer Service 341-523-517011380 Westline Industrial Drive,St. Louis, MO 63146

Send us your new address at least 6 we

43. The World Factbook 2008 (CIA). Available at: https://www.cia.gov/library/publications/the-world-factbook/index.html. Lastaccessed February, 2008.

44. Percentage who have experienced cardiovascular conditions (everand recently) by sex and age, 2003, England. London: Department ofHealth (2004) Health Survey for England 2003.

45. Reddy KS, Yusuf S. Emerging epidemic of cardiovascular disease indeveloping countries. Circulation 1998;97:596-601.

46. Yusuf S, Hawken S, Ounpuu S, et al. Effect of potentially modifiablerisk factors associated with myocardial infarction in 52 countries(the INTERHEART study): case-control study. Lancet 2004;364:937-52.

?

ice when you change your address, please photocopy

eks before your move to ensure continued service. Weause of late notification.

_____________________________________

EW ADDRESS:

learly print your new address here.

ame ____________________________________________________

ddress __________________________________________________

ity/State/ZIP _____________________________________________

: OR PHONE: OR E-MAIL:1-800-654-2452 JournalCustomerService-Outside the U.S., call [email protected]

eks ahead

https://www.cia.gov/library/publications/the-world-factbook/index.html

https://www.cia.gov/library/publications/the-world-factbook/index.html

Reproduced with permission of the copyright owner. Further reproduction prohibited withoutpermission.

cardio sequelae of preeclampsia-eclampsia

Documents