Hypertensive Disorders in Pregnancy and Preterm Delivery and Subsequent Stroke in Asian Women

Gestational hypertension and preeclampsia are hypertensive disorders during pregnancy (HDP).^1,2 Gestational hypertension refers to the onset of hypertension without proteinuria after 20 weeks of gestation, whereas preeclampsia is characterized by the onset of hypertension and proteinuria after 20 weeks of gestation. Gestational hypertension occurs in ≈6% of pregnancies and evolves into preeclampsia in 10% to 20% of cases.^3,4 Preeclampsia is a prevalent life-threatening disorder affecting pregnant women and their fetuses.^5,6 The incidence of preeclampsia range from 3% to 14% of pregnancies worldwide and 5% to 8% in the United States.^1,7 HDP and cardiovascular diseases share many risk factors, including preexisting hypertension, diabetes, obesity, renal disease, collagen vascular disease, antiphospholipid syndrome, advanced age, and endothelial dysfunction.^5,8–12 Such cardiovascular risks that appear during pregnancy may have subsequent health consequences, including stroke.^13–16

Both case-control and cohort studies have reported an association between HDP and later cardiovascular disease.^13–19 However, studies on the association between HDP and the risk of subsequent stroke are limited in Asian populations. In this work, a retrospective cohort study was performed using a set of population-based universal insurance data to investigate the incidence of stroke after delivery among women with HDP, with emphasis on the age-specific risk and the interaction with preterm delivery.

Methods and Materials

Data Sources

This study used data from reimbursement claims of the universal National Health Insurance program, which was launched in March 1995 by incorporating 13 insurance programs in Taiwan. This insurance program provides health care to 99% of the population.²⁰ The National Health Research Institute has been responsible for managing the insurance data. The claims data of 1 million randomly selected subjects from 23 million insured individuals registered from 1996 to 2008 were obtained from the National Health Research Institute. With approval from the National Health Research Institute, the scrambled identification of insured individuals and contracted institutions to linked files were used, including the registry of medical facilities, details of inpatient orders, ambulatory care, as well as dental services and prescriptions. The diagnoses were coded using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM).²¹

Study Sample

From the claims data, we used special delivery codes in the claims file specifically designed by the insurance system to identify the date of delivery of pregnant women. Patients with HDP (ICD-9-CM 642) were then identified from the group of women having their first pregnancies (primigravid) included in the database from January 1, 2000 through December 31, 2004; these women were designated as the HDP cohort. For each HDP case identified, 4 primigravid women without history of HDP were randomly selected, frequency-matched in the same year, and were designated as the non-HDP controls. In total, 174 subjects were excluded; these were the women with histories of stroke or hypertension (ICD-9-CM 401 to 405 or A-code A260 and A269) diagnosed before the index date (the date the subject was selected into the study) or with missing information (age or sex). Then, 1092 women were selected as the HDP cohort and 4715 women were selected as the non-HDP cohort.

Sociodemographic Variables and Comorbidities

The sociodemographic variables used in this study included sex, age, occupation, urbanization of the residential area, and monthly income used for premium estimation. The age of each study subject was measured by the difference between the index date and the date of birth. In accordance with the National Statistics of Regional Standard Classification, all insured persons were grouped into 3 urbanization levels based on population densities (low, moderate, and high).

Moreover, the baseline comorbidity history for each subject was also identified, including diabetes (ICD-9-CM 250 or A-code A181), hyperlipidemia (ICD-9-CM 272.9 or A-code A189), coronary artery disease (CAD, CD-9-CM 410 to 413, 414.0, 414.8, 414.9, or A-code A270 and A279), preterm delivery (ICD-9-CM 644), lupus (ICD-9-CM 710.0), and thrombophilia (ICD-9-CM 286.9).

Statistical Analysis

The study subjects with and without HDP were linked to the registry for inpatient and outpatient claims data to identify those who had stroke develop (ICD-9-CM 430 to 437, 674.0, or A-code A290-294 and A299). Each subject either was followed-up from the index date until December 31, 2008 or was censored. The follow-up time, in person-years, was calculated for each subject until the diagnosis of stroke or until being censored because of death, withdrawal from the insurance system, or loss to follow-up.

The distributions of categorical sociodemographic variables and comorbidities were compared between HDP and non-HDP cohorts, and the differences were examined using a χ² test. Likewise, the incidence densities by sociodemographic variables were calculated for each cohort. The HDP-to-non-HDP rate ratio for stroke with the crude hazard ratio (HR) and 95% CI for each variable were calculated. We used logistic regression analysis to measure OR and the corresponding 95% CI for evaluating the association between each comorbidity and HDP.

Cox proportion hazard regression was used to assess the effects of HDP on the risk of stroke, adjusting for variables that were significantly related to HDP as observed through the χ² analyses. The HR and 95% CI were calculated using the model. Cox proportion hazard regression was also used to examine the interaction between HDP and preterm delivery associated with stroke. However, Kaplan-Meier analysis and the log-rank test were used to estimate the stroke-free proportions for the risk of stroke developing in the HDP and non-HDP cohorts.

All analyses were performed using SAS statistical software (version 9.1 for Windows; SAS Institute); Kaplan-Meier analysis and the log-rank test were performed by R (version 2.11.1 R; Foundation for Statistical Computing). The results were considered statistically significant when 2-tailed P<0.05.

Results

Characteristics of the Study Subjects and Incidence of Stroke

This study consisted of 1092 HDP cases and 4715 non-HDP pregnant women after excluding ineligible subjects. The HDP cohort was older than the non-HDP group (29.5±4.82 years vs 27.8±4.95 years; P<0.001; data not shown). Most subjects were 25 to 29 years old (38.2% in non-HDP, 34.8% in HDP; P=0.048). The degrees of urbanization were similar in both groups (χ²P=0.70). The mean follow-up periods in our study were 6.64±1.57 years in the HDP cohort and 6.40±1.57 years in the non-HDP cohort (data not shown).

Table 1 presents the incidence densities in both cohorts and HDP/non-HDP rate ratios of stroke by sociodemographic status. Occupations and urbanization of residential areas of study subjects were not significantly different between the 2 groups. A V-shape relationship was observed for the incidence of stroke across the age groups in both HDP and non-HDP cohorts. The lowest incidence was 6.2 per 10 000 person-years in women 19 to 24 years of age in the non-HDP cohort and 20.6 per 10 000 person-years in those 25 to 29 years of age in the HDP cohort. An exceptionally higher incidence of 103.1 per 10 000 person-years was noted for HDP patients 15 to 18 years of age. Overall, the stroke incidence was 2.35-fold higher in the HDP than in the non-HDP cohort (30.1 vs 12.8 per 10 000 person-years), with an adjusted HR of 2.04 (95% CI, 1.18–3.51; data not shown).

The Kaplan-Meier survival analysis showed that the stroke-free rate was 2.2% less in the HDP women than in the non-HDP women (log-rank P=0.001; Figure 1). Most events actually occurred within 4 to 8 years of follow-up. Furthermore, the incidence of stroke during puerperium (from delivery to 6 weeks postpartum) was 4.33-fold higher in the HDP cohort than in the non-HDP cohort (142.0 vs 32.8 per 10 000 person-years; data not shown). However, the risk shown in the univariate Cox proportional regression analysis was not significant (HR, 4.32; 95% CI, 0.27–69.1).

• Download figure
• Download PowerPoint

Comorbidities in Study Subjects

Table 2 shows the prevalence of selected morbidities in the study subjects. Patients with HDP were more likely to have diabetes (2.8% vs 0.8%; P<0.0001), hyperlipidemia (1.5% vs 0.5%; P=0.0006), placental abruption (1.9% vs 0.8%; P=0.0009), preterm delivery (31.0% vs 24.1%; P<0.0001), lupus (1.4% vs 0.3%; P<0.0001), and thrombophilia (0.18% vs 0.15%; P=0.79). The logistic regression analysis showed that diabetes mellitus, hyperlipidemia, preterm delivery, placental abruption, and lupus were significantly associated with HDP. The multivariable regression model measured that OR associated with preterm delivery was 1.41 (95% CI, 1.22–1.63).

Table 3 shows the values of comorbidity-adjusted age-specific risk of stroke in both groups that have been measured using the Cox proportional regression analysis. Compared with subjects 19 to 24 years of age in the non-HDP group, the V-shape HR of stroke increased with age in both groups. However, exceptionally higher risk with a large CI range was noted for HDP patients 15 to 18 years of age (HR, 13.4; 95% CI, 1.54–16.7), followed by women aged 35 to 40 years old (HR, 5.56; 95% CI, 1.47–21.0).

Interaction Between HDP and Preterm Delivery

Figure 2 presents the Kaplan-Meier curves for the interaction between HDP and preterm delivery in relation to the occurrence of stroke. Women with HDP and preterm delivery were the most likely to experience a stroke, followed by women with HDP but not preterm delivery; in comparison, women with normal pregnancy were least likely to experience stroke (log-rank P=0.0016). Compared with normal pregnancy, the interaction analysis shows that the risk of stroke was greater for women with only HDP than those with only preterm delivery, and the it was highest for those with both HDP and preterm delivery (P for trend=0.002; Table 4). Women with both HDP and preterm delivery had an adjusted HR of 3.22 (95% CI, 1.48–6.99).

• Download figure
• Download PowerPoint

Discussion

The pathogenesis of preeclampsia likely involves both maternal and fetal factors.^21,22 Abnormal placental vasculature developing in early pregnancy may lead to reduced maternal uteroplacental perfusion. The released anti-angiogenic factors in the maternal circulation could alter maternal systemic endothelial function, leading to hypertension and other related manifestations.^23,24 Gestational hypertension may have the same pathophysiological change as preeclampsia. These factors may partly explain the association between HDP and stroke. HDP is in the cardiovascular risk profile, which may lead to atherosclerosis and act as a clinical marker for poor cardiovascular health.

A strong association between HDP and subsequent stroke was found; in addition, preterm delivery enhanced the risk. This finding is in accordance with other studies. A Norwegian follow-up study found that women with histories of preeclampsia and preterm delivery had a 5.08-fold higher risk of cerebrovascular mortality than women who were free of preeclampsia and delivered at term.¹⁴ Wilson et al¹⁶ have demonstrated a 3.6-fold elevated risk of mortality from cerebrovascular disease in women with preeclampsia. In a case-controlled study, Brown et al¹⁷ have reported that women with preeclampsia had a relative risk of 1.63 for ischemic stroke developing. The results of the meta-analysis conducted by Bellamy et al¹³ demonstrated that the relative risk of future stroke for preeclampsia is 1.81 (95% CI, 1.54–2.67). Another meta-analysis by McDonald et al¹⁵ have shown that women with a history of preeclampsia have an increased risk of subsequent stroke (relative risk, 2.03; 95% CI, 1.54–2.67).

Hypertensive disorders during pregnancy are complicated by placental abruption, preterm delivery, or fetal death.^25–29 In our analysis, HDP (independent of diabetes and hyperlipidemia) was an important risk factor associated with stroke. Previous data analysis showed that women with gestational hypertension had a slightly increased risk of stroke developing, consistent with other studies.^13,16 Thus, in our study, the reported association was likely diluted by gestational hypertension. However, gestational hypertension is as important as preeclampsia as a factor that can induce subsequent stroke.¹⁸ The present study further demonstrated that preterm delivery further increased the HR of stroke to 3.22 for women with HDP. The association between stroke and preterm delivery alone was not significant in the multivariable analysis. However, the preterm delivery did increase the stroke risk for 126% in women with HDP. This finding is consistent with previous studies that also demonstrated that preterm delivery further increases the risk of stroke in women with HDP.^14,18 Based on the results of the Denmark registry-based study, women with preeclampsia and preterm delivery are at an elevated risk for higher subsequent cardiovascular disorders, such as hypertension, ischemic heart disease and congestive heart failure, and type 2 diabetes mellitus.¹⁸ However, placental abruption, a risk factor of preterm delivery,²⁷ was not a significant risk factor for stroke in our analysis (data not shown).

Pregnant adolescents and older women with HDP had a higher HR of future stroke than women aged 25 to 29 years. Adolescent pregnancy has not yet been proven as a risk factor for HDP.⁷ Because the sample size of teen pregnancy was small in this study, a high incidence of subsequent stroke in teens with HDP may be attributable to chance. To the best of our knowledge, this is the first report of this type, although it warrants a further study. It is well-known that adolescents are at increased risk for adverse pregnancy outcomes, such as low-birth-weight newborns, infant deaths, and premature death in later life.^30,31 Biological immaturity and socioeconomic factors might account for the poorer outcome among pregnant adolescents.

This study has several limitations. Adolescent pregnancy is relatively rare in Taiwan, and the sample size in the HDP group was small. With a large CI and HR attributable to only a single stroke case identified, random error should be considered for the highest stroke incidence among all age groups. However, adolescent pregnancy in the non-HDP group also had increased risk of stroke with a greater sample size. The multivariate analysis demonstrated that this V-shape age-specific association did not occur by chance. Second, the National Health Research Institute database provided limited information on sociodemographic characteristics, with information unavailable on marital status, educational level, body mass index, smoking habit, and laboratory data. These variables could not be adjusted in the analysis. However, we were able to use occupation, income, and residential area for adjustment. These socioeconomic characteristics had no significant association. Third, some information on chronic conditions, such as hyperlipidemia, was unavailable for few individuals. However, this situation happened in both groups. Fourth, the matched cohort study design may benefit this study with convenience and efficiency, but there is a possibility of overestimation of association between HDP and stroke. To reduce this possibility, we excluded women with preexisting hypertension before the pregnancy. Finally, stroke and other diseases were identified by ICD-9-CM codes. However, patients with stroke in Taiwan are generally cared for at larger hospitals with adequate diagnoses and the majority of pregnant women receive adequate prenatal cares. These codes were reviewed and validated by auditors of medical records for the insurance system to insure the accuracy of the claims.

Conclusions

In conclusion, our study results can be generalized to apply to pregnant women in Taiwan for the association between HDP and stroke because we used representative population data. Pregnant women with HDP are at ≈2-fold increased risk for future stroke. The risk increases further for women with both HDP and preterm delivery. This potential interaction is worthwhile to consider in prevention strategies. Moreover, adolescents with HDP may have a higher risk of subsequent stroke than older pregnant women with HDP. This association remains unclear. Further prospective studies regarding pregnancy disorders in teen mothers are needed.

Sources of Funding

This study was supported in part by the Executive Yuan National Science Council (grant number NSC 98-2621-M-039-001), Clinical Trial and Research Center of Excellence of the Department of Health (grant number DOH99-TD-B-111-004), Taiwan Department of Health Cancer Research Center of Excellence (grant number DOH99-TD-C-111-005), and the China Medical University Hospital (grant number 1MS1).

Disclosures

None.

Footnotes