Materials and Methods
This study was conducted following the Declaration of Helsinki. The
study protocol was also approved by the Institutional Review Board (IRB)
of Dalin Tzu Chi Hospital of Buddhist Tzu Chi Medical Foundation
(approval number, B10402022). The IRB absolved the requirement for
written informed consent due to no direct contact with individual
patients from this de-identified database.
We used the Taiwan National Health Insurance Research Database (NHIRD)
to analyze the incidence rate of stroke in women with
pre-eclampsia/eclampsia and compared it to those without
pre-eclampsia/eclampsia. Taiwan NHIRD contains all the records of
diagnosis and treatment of approximately 99% of people from inpatient,
outpatient, and emergency
departments22. The data
collection of pregnant women from the Taiwan NHIRD ranged from 2000 to
2017, and it was included in this study for statistical analysis. The
data included were evaluated by the National Health Insurance
Administration (NHIA) quarterly expert reviews on every 50 to 100
ambulatory and inpatient claims filed by each medical
institution23. False
diagnostic reports are subject to severe penalties from the
NHIA24.
Records of pregnant women in this database were collected and
categorized into two groups women, those with and without
pre-eclampsia/eclampsia. Based on the International Classification of
Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes for
pregnant women, the codes were 650, 651, 652, and 653, while those for
pregnant women with pre-eclampsia/eclampsia were 642.4, 642.5, 642.6,
and 642.7. The primary outcomes included: codes for hemorrhagic stroke
being 430, 431, 432, and codes for ischemic stroke being 433, 434, 435,
436, and 437.
Between 2001 and 2017, 1,384,427 pregnant women with delivery were
registered in the Taiwan NHIRD. We included all pregnant women in Taiwan
with different socioeconomic statuses, living areas (such as
metropolises to rural), and hospital levels. The sample size exceeded
million people. Figure 1 shows our study’s flow diagram. We excluded
38,707 cases with missing confounders and 566 cases with stroke history.
In addition, we only included women with the delivery age between 18 to
45 years. Finally, we enrolled 1,338,334 cases in this study, divided
into groups of normal delivery with (N=8,077) and without (N=1,316,550)
pre-eclampsia/eclampsia.
In this study, we used exact matching method to control covariates, that
is, these two cohorts had the same age, same distribution for
comorbidities, and socioeconomic status (all p= 1.000). After 1:4 exact
matching, 6,053 cases were selected in the pre-eclampsia/eclampsia group
and 24,212 in the non-pre-eclampsia/eclampsia group. The follow-up time
in this study was from 2000 to 2017. We divided the 17 years follow up
into short (0-1, 1-3, 3-5 years), intermediate (5-10), and long (10-15
years), to further differentiate the risks at different intervals.
Covariates included were age, season, cesarean section or normal
spontaneous delivery, multiple gestations, hospital levels, and
comorbidities. The hospital level was included in the analysis, account
for different care qualities during pregnancy at different hospital
levels. The comorbidities included hypertension, gestational diabetes
mellitus (GDM), anemia, and delivery conditions such as antepartum
hemorrhage (APH) and postpartum hemorrhage (PPH). These young women, age
18 to 45, had no comorbidities, such as chronic kidney disease, heart
failure, hypercholesterolemia, peripheral vascular disease, heart valve
disorders, after matching. Socioeconomic variables, including geographic
region, urbanization level, and monthly income-based insurance premiums
were analyzed to reduce bias resulting from lifestyle.