Patient Characteristics and Outcomes After Hemorrhagic Stroke in Pregnancy
Background—Hospitalizations for pregnancy-related stroke are rare but increasing. Hemorrhagic stroke (HS), ie, subarachnoid hemorrhage and intracerebral hemorrhage, is more common than ischemic stroke in pregnant versus nonpregnant women, reflecting different phenotypes or risk factors. We compared stroke risk factors and outcomes in pregnant versus nonpregnant HS in the Get With The Guidelines-Stroke Registry.
Methods and Results—Using medical history or International Classification of Diseases-Ninth Revision codes, we identified 330 pregnant and 10 562 nonpregnant female patients aged 18 to 44 years with HS in Get With The Guidelines-Stroke (2008–2014). Differences in patient and care characteristics were compared by χ2 or Fisher exact test (categorical variables) or Wilcoxon rank-sum (continuous variables) tests. Conditional logistic regression assessed the association of pregnancy with outcomes conditional on categorical age and further adjusted for patient and hospital characteristics. Pregnant versus nonpregnant HS patients were younger with fewer pre-existing stroke risk factors and medications. Pregnant versus nonpregnant subarachnoid hemorrhage patients were less impaired at arrival, and less than half met blood pressure criteria for severe preeclampsia. In-hospital mortality was lower in pregnant versus nonpregnant HS patients: adjusted odds ratios (95% CI) for subarachnoid hemorrhage 0.17 (0.06–0.45) and intracerebral hemorrhage 0.57 (0.34–0.94). Pregnant subarachnoid hemorrhage patients also had a higher likelihood of home discharge (2.60 [1.67–4.06]) and independent ambulation at discharge (2.40 [1.56–3.70]).
Conclusions—Pregnant HS patients are younger and have fewer risk factors than their nonpregnant counterparts, and risk-adjusted in-hospital mortality is lower. Our findings suggest possible differences in underlying disease pathophysiology and challenges to identifying at-risk patients.
WHAT IS KNOWN
Hospitalizations for pregnancy-related stroke are rare but increasing. Hemorrhagic strokes (ie, subarachnoid hemorrhage and intracerebral hemorrhage) are more common than ischemic strokes in pregnant than in nonpregnant women, likely reflecting different phenotypes and risk factors.
Previous analyses of administrative data and small case series reveal that subarachnoid hemorrhage in pregnant women is less often aneurysmal in cause than in nonpregnant women and may be associated with better short-term outcomes.
WHAT THE STUDY ADDS
This analysis of Get With The Guidelines-Stroke, the largest clinically detailed stroke registry, suggests that pregnant female patients admitted with subarachnoid or intracerebral hemorrhage have a decreased risk of in-hospital death when compared with nonpregnant females, even after adjustment for age and other patient characteristics.
Because pregnant women with hemorrhagic stroke may present to the hospital with fewer traditional stroke risk factors than their nonpregnant counterparts, and only moderately elevated blood pressures, it may be challenging to correctly diagnose hemorrhagic stroke and provide prompt and appropriate care.
Additional studies aimed at elucidating the causes and risk factors of hemorrhagic stroke to better target pregnant women at risk should be a high priority.
Although rare, hospitalizations for pregnancy-related stroke are increasing.1,2 Hemorrhagic stroke (HS), which includes both subarachnoid hemorrhage (SAH) and intracerebral hemorrhage (ICH), occurs in approximately 6 per 100 000 deliveries3–6 and accounts for ≤50% of all stroke subtypes in pregnant women when compared with 20% to 30% in nonpregnant women.2,7 This increased proportion of HS versus ischemic stroke (IS) suggests that different phenotypes or underlying risk factors may play a role during pregnancy. Previous analyses of administrative data sets and small case series reveal that pregnancy-related SAH is less likely aneurysmal and may be associated with better short-term outcomes than nonpregnancy-related SAH.5 Similarly, vascular lesions such as cavernous or arteriovenous malformations seem to be present in a relatively small portion of pregnancy-related ICH.8,9 Mortality rates for both SAH and ICH stroke subtypes in pregnancy are still considerable and are reported to be as high as 10% and 20%, respectively.5,6 Furthermore, any degree of disability or death is particularly devastating in the setting of a formerly healthy young mother with a new baby. In an effort to better understand this potentially catastrophic event, we investigated the hypothesis that pregnancy versus nonpregnancy-related HS was associated with a reduced frequency of traditional stroke risk factors, less severe presenting symptoms, and more favorable outcomes by analyzing data in Get With The Guidelines (GWTG)-Stroke, the largest clinically detailed stroke registry.
The GWTG-Stroke registry is a voluntary quality improvement program and patient education tool that captures patients with acute stroke using a Web-based, patient management system with decision support. The data coordinating center for GWTG is Quintiles Real-World & Late Phase Research (Cambridge, MA), and the Duke Clinical Research Institute serves as the statistical analytic center. The Duke University Medical Center Institutional Review Board approved all related analyses. A detailed description of the GWTG methodology has been previously reported.10
Case Identification and Data Abstraction
Patient-level data from 1991 GWTG hospitals participating in the GWTG program were analyzed from 2008 to 2013. Acute stroke diagnoses were confirmed by chart review, and patient data were then collected and entered by trained hospital personnel. Pregnancy, our exposure of interest, is defined in GWTG-Stroke as a woman who is pregnant or <6 weeks postpartum at the time of the index event.11
The study population consisted of pregnant and nonpregnant women between the ages of 18 to 44 years old who were discharged with a primary diagnosis of IS or HS (ie, SAH or ICH). The 10 892 women with confirmed SAH or ICH were confirmed as pregnant (330) or nonpregnant (10 562) by a combination of medical history and International Classification of Diseases-Ninth Revision codes. Admission for stroke in pregnancy often resulted in a principle International Classification of Diseases-Ninth Revision code that was not modified by pregnancy; however, some pregnant women with SAH or ICH were assigned codes specific for pregnancy-related cerebrovascular events (671.5X, 674.0X). In this way, we identified pregnant women with SAH or ICH if (1) a clinical diagnosis of SAH or ICH occurred in combination with a medical history of pregnancy in GWTG or (2) a clinical diagnosis of SAH or ICH occurred in combination with a principle International Classification of Diseases-Ninth Revision discharge diagnosis code for pregnancy-related stroke (671.5X or 674.0X). We determined that SAH or ICH in a nonpregnant woman was present if the clinical diagnosis occurred in the absence of a medical history of pregnancy in GWTG and the absence of an International Classification of Diseases-Ninth Revision discharge code for pregnancy-related stroke (671.5X or 674.0X). The same methodology was applied in determining eligible pregnancy-related IS in GWTG-Stroke.
Among a total of 2 637 488 patient encounters in 1991 participating hospital sites, patients were excluded if admitted before 2008 (n=605 118; 107 sites), male sex (n=960 025; 9 sites), age not 18 to 44 years (n=1 017 163; 239 sites), nonpregnant IS (n=24 303; 1277 sites), diagnosis of transient ischemic attack or unspecified stroke (n=10 510; 78 sites), missing pregnancy information (n=6278; 62 sites); were discharged/transferred to another acute care facility (n=1856) and left against medical advice (n=817); or had missing discharge designation information (n=188; 20 sites; Figure I in the Data Supplement).
Patient and hospital characteristics, as well as care delivery and discharge outcomes, are reported for 4 different cohorts: (1) pregnant versus nonpregnant patients with SAH; (2) pregnant versus nonpregnant patients with ICH; (3) pregnant versus nonpregnant patients with any HS; and (4) pregnant patients with HS versus IS. We report proportions for categorical variables and medians with interquartile range or means with standard deviations for continuous variables. The short-term outcomes of stroke analyzed were in-hospital mortality, the proportion of patients discharged to home, independent ambulation at discharge, and hospital length of stay.
Differences in these characteristics were compared using χ2 tests or Fisher exact test when appropriate for categorical row variables and Wilcoxon rank-sum tests for continuous row variables. All variables had <5% missing data unless otherwise specified.
Conditional logistic regression was used to assess the association of pregnancy with outcomes, conditioning on categorical (nominal) age by year. Multivariable models further adjusted for race (white versus other), medical history of atrial fibrillation/flutter, previous stroke/transient ischemic attack, coronary artery disease/previous myocardial infarct, carotid stenosis, diabetes mellitus, peripheral vascular disease, hypertension, dyslipidemia, smoking, region, hospital type (teaching/nonteaching), number of beds, annual ICH and SAH volume, rural location, and The Joint Commission primary stroke center status. Additional analyses assessed sensitivity to our analytic approach by evaluating matched cohorts. The first sensitivity approach matched 6:1 on age alone and the second matched 3:1 on age, medical histories of diabetes mellitus, and hypertension. In these analyses, conditional logistic regression used the matched group as conditional strata. These analyses are included as Tables I to IV in the Data Supplement.
Missing values for categorical variables were imputed to the most likely category (eg, white race or no for comorbidities). Hospital characteristics were not imputed. Variables with >10% to 15% missing were excluded from the model. The colinearity between all covariates in the model was assessed using variance inflation factors. As a variance inflation factor >5 may be indicative of colinearity issues, the correlation between variables with variance inflation factor >5 and the other covariates were examined. All tests are 2 tailed with a level of statistical significance at P<0.05. All statistical analyses were performed using SAS version 9.3 software (SAS Institute Inc., Cary, NC).
SAH and ICH Among Pregnant Versus Nonpregnant Patients
During the study period, there were 5897 patients with SAH included from 615 participating study sites, of whom 152 (2.6%) were pregnant and 5745 (97.4%) were nonpregnant; and 4995 patients with ICH included from 814 study sites, of whom 178 (3.6%) were pregnant and 4817 (96.4%) were nonpregnant. Pregnant patients with SAH or ICH were younger than the nonpregnant patients with SAH or ICH, respectively, and shared many other characteristics that were different from their nonpregnant comparators (Table 1). In addition to being more often insured by Medicaid, they were significantly less likely to have traditional HS risk factors, including hypertension, dyslipidemia, or smoking or to be on antihypertensive or antiplatelet/anticoagulant medications (Table 1). Pregnant versus nonpregnant patients with ICH were less likely to have diabetes mellitus or previous stroke/transient ischemic attack.
Pregnant versus nonpregnant SAH patients were also less likely to have an altered level of consciousness and have no neurological symptoms at initial examination (defined as no neurological signs or symptoms present on arrival or when the first neurological examination was performed by the documenting physician). Presenting median poststroke systolic blood pressure (SBP; 144 versus 136 mm Hg) and diastolic blood pressure (DBP; 83 versus 80 mm Hg) were significantly higher for the pregnant SAH patients, whereas no significant difference was observed in pregnant versus the nonpregnant ICH patients (Table 1). In the sensitivity analyses matched on age, history of diabetes mellitus, and hypertension, both the pregnant SAH and ICH patients had significantly higher presenting median poststroke blood pressures than the nonpregnant SAH and ICH patients (Table I in the Data Supplement). About short-term outcomes, in-hospital mortality was significantly lower in the pregnant versus the nonpregnant patients with SAH (2.6% versus 14.4%) or ICH (10.1% versus 19.6%; Table 1). Differences in favor of better outcomes for pregnant patients persisted in multivariable regression analysis: adjusted odds ratios (aORs) for in-hospital mortality for SAH: (aOR, 0.17 [95% confidence interval, 0.06–0.45]) and ICH (aOR, 0.57 [0.34–0.94]; Table 2) and in the sensitivity analysis matched on age, history of diabetes mellitus, and hypertension: for SAH (aOR, 0.10 [0.03–0.35]) and ICH (aOR 0.51 [0.27–0.95]; Table II in the Data Supplement). Pregnant SAH patients also had a significantly higher likelihood of home discharge (aOR, 2.60 [1.67–4.06]) and independent ambulation at discharge (aOR, 2.40 [1.56–3.70]) in the multivariable but not in the sensitivity analysis. There was no difference between groups for prolonged hospital length of stay (Table 2).
HS Overall in Pregnant Versus Nonpregnant Patients
In a separate analysis, we combined SAH and ICH subtypes to analyze trends in HS overall. Of the 10 892 combined HS patients, 330 (3.0%) were pregnant (Table 3). Pregnant versus nonpregnant HS patients were younger, more likely to have Medicaid insurance, and to be black. Pregnant HS patients were less likely to have pre-existing stroke risk factors, such as hypertension, smoking, previous stroke/transient ischemic attack, or diabetes mellitus, and were less likely to be on medications such as antihypertensives, cholesterol-reducers or antiplatelet/anticoagulants (Table 3). Neither initial poststroke median SBP nor median DBP differed significantly between groups, and both were below the threshold for severe preeclampsia (SBP≥160 mm Hg or DBP≥110 mm Hg) and stage 2 hypertension (SBP≥160 mm Hg or DBP≥100 mm Hg) and almost half were below the threshold for preeclampsia (SBP≥140 mm Hg or DBP≥90 mm Hg; Table 3). Although the pregnant patients were more often treated in hospitals with a higher annual number of HS admissions, there were no significant differences between groups in hospital size, rural location, or teaching status (Table 3).
Pregnancy-Related Hemorrhagic Versus IS
There were 10 892 patients with HS from 1429 sites, 330 of whom were pregnant, and 24 641 women with IS from 1466 sites, 338 of whom were pregnant. Forty-nine (49% of all) pregnancy-related strokes were because of HS, and 51% were because of IS. Pregnant HS versus IS patients were more often black and were less likely to have traditional stroke risk factors, including previous stroke, diabetes mellitus, dyslipidemia, or be on previous antihypertensive or antiplatelet/anticoagulants (Table 4). Initial poststroke blood pressures were significantly higher for the HS versus IS pregnant patients (SBP, 143 versus 127 mm Hg and DBP, 84 versus 78 mm Hg) but were frequently below the threshold for severe preeclampsia and stage 2 hypertension (Table 4).
Our analysis of the largest clinically detailed stroke registry suggests that pregnant female patients with either SAH or ICH have a decreased risk of death when compared with nonpregnant females within GWTG-Stroke, after accounting for age and other patient characteristics. Important differences emerged between the pregnant versus the nonpregnant HS patients, including the presence of fewer traditional stroke risk factors on admission, less severe symptoms on hospital arrival, and better short-term outcomes. These differences persisted in the sensitivity analyses matched for age and for the stroke and preeclampsia risk factors of hypertension and diabetes mellitus. Our findings support previous investigations of peripartum SAH that report lower mortality rates than in nonpregnant patients with SAH (10.3% versus 18.3%), perhaps related to a decreased incidence of aneurysmal cause.5
In our investigation, pregnant women with ICH also had lower in-hospital mortality than a nonpregnant cohort. Although the pregnant patients seemed to have a healthier clinical prestroke profile, their higher rates of favorable outcomes persisted even after a multivariable analysis conditional on age and adjusted for key factors. Data from a 2006 survey of 1582 Japanese facilities (38 cases of pregnancy-related ICH)12 provide some insight into possible causes. The Japanese general population is known to have higher rates of nontraumatic intracranial hemorrhage than their Western counterparts, likely because of a higher incidence of uncontrolled hypertension and small-vessel disease.12,13 In this survey, 42% of the pregnant Japanese women with ICH had no pre-existing disorder, 29% had vascular lesions (arteriovenous malformation or Moyamoya disease), and 39.5% had preeclampsia or the Hemolysis, Elevated Liver enzymes, and Low Platelet counts (HELLP) syndrome (not mutually exclusive).12 A second Japanese survey (97 pregnancy-related HS) found that 32% were because of an unknown cause, 36% to other vascular lesions, 16.5% to cerebral aneurysms, 12% to obstetric complications (eg, hypertensive disorders of pregnancy), and 3% to other cerebrovascular disorders.8 The Japanese and other population-based studies have demonstrated an increased risk of both HS and IS in women with preeclampsia.1,2,5,6,14,15 The GWTG-Stroke registry does not capture information about preeclampsia, so we could not discern which pregnant patients had this disorder. We did observe that although pregnant HS patients were less likely to have pre-existing hypertension documented in the medical record than the nonpregnant patients, the first recorded blood pressures on hospital arrival were significantly higher in the pregnant versus in the nonpregnant SAH and ICH patients. In both groups, almost half the patients had median blood pressure readings well below the threshold for severe preeclampsia and stage 2 hypertension.16,17 It is possible that severe hypertension in these pregnant women existed but was not captured, and there were missed opportunities for pre-emptive therapy. Alternatively, it may be that for patients with preeclampsia, the disruption of cerebral autoregulation or endothelial dysfunction creates vulnerability to cerebrovascular injury even at blood pressures that are in the higher end of the recommended target range (BP, 130–155/80–105 mm Hg).17–19
There are physiological and hemodynamic changes in normal pregnancy, including vascular remodeling from the placenta, increased filling and cardiac output with hypercoagulability, or, in some cases, of postpartum hemorrhage, hypocoagulability, which may predispose these patients to HS.17,20,21 Prenatal care may also impact the phenotype of pregnancy-related HS. Although GWTG-Stroke offers the largest, clinically detailed analysis of pregnancy-related HS, it has limitations. The observational study design precludes conclusions on causality. The accuracy of analyses depends on the integrity of documentation at the time of stroke and review of medical chart data. Site-level patient enrollment bias may affect these findings. In addition, because participation in the GTWG program is voluntary, generalizability may be limited. Although there was substantial clinical detail captured, additional information on pregnancy-specific (eg, hypertensive disorders of pregnancy), stroke-relevant variables and long-term maternal and fetal outcomes are not included in this registry. Details on the possible cause of HS subtypes were neither available nor additional risk factors (eg, substance abuse) beyond the traditional stroke risk factors.
Despite increasing incidence of stroke in pregnant and young adults,7 stroke remains a relatively uncommon yet morbid event in pregnancy. Our study provides the largest, most clinically detailed series of these women and suggests that pregnant SAH patients fare better than nonpregnant SAH patients in terms of in-hospital mortality, discharge to home and independent ambulation. Pregnant ICH patients had lower odds of in-hospital mortality than their nonpregnant counterparts. Because these pregnant SAH and ICH patients may present to the hospital with fewer traditional stroke risk factors, and only moderately elevated blood pressures, it may be more challenging to diagnose stroke and provide prompt and appropriate care. Additional studies aimed at elucidating the causes and risk factors to better target pregnant women at risk for SAH and ICH should be a high priority.
Sources of Funding
This project was funded by a Get With The Guidelines (GWTG)-Stroke Young Investigator Database Research Seed Grant. The GWTG-Stroke program is provided by the American Heart Association (AHA)/American Stroke Association. GWTG-Stroke is sponsored, in part, by Medtronic and has been funded in the past through support from Boeringher-Ingelheim, Merck, Bristol-Myers Squib/Sanofi Pharmaceutical Partnership, Janseen Pharmaceutical Companies of Johnson & Johnson and the AHA Pharmaceutical Roundtable.
The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. Dr Schwamm serves as a volunteer for the American Heart Association as chair of the stroke clinical workgroup for Get with the Guideline-Stroke and chair of the Healthcare Accreditation Science Committee. He serves as the PI of an NINDS-funded SPOTRIAS network trial, MR WITNESS, which is a phase 2 safety study of alteplase delivered in an extended time window with MR-guided patient selection (NCT01282242). The study is funded primarily by NINDS, and alteplase is provided by Genentech to MGH for distribution to sites, as well as modest per patient supplemental site payments. Genentech has no control over study design, analysis, or publication. He reports receiving significant research support from the Patient-Centered Outcomes Research Institute, serving as a stroke systems consultant to the Massachusetts Department of Public Health, and serving as a scientific consultant regarding trial design and conduct to Lundbeck (international steering committee, DIAS3, 4 trial); and Penumbra (data and safety monitoring committee, Separator 3D trial). The other authors report no conflicts.
The Data Supplement is available at http://circoutcomes.ahajournals.org/lookup/suppl/doi:10.1161/CIRCOUTCOMES.115.002242/-/DC1.
- Received August 17, 2015.
- Accepted September 23, 2015.
- © 2015 American Heart Association, Inc.
- Kuklina EV,
- Tong X,
- Bansil P,
- George MG,
- Callaghan WM
- Sharshar T,
- Lamy C,
- Mas JL
- Takahashi JC,
- Iihara K,
- Ishii A,
- Watanabe E,
- Ikeda T,
- Miyamoto S
- Schwamm LH,
- Fonarow GC,
- Reeves MJ,
- Pan W,
- Frankel MR,
- Smith EE,
- Ellrodt G,
- Cannon CP,
- Liang L,
- Peterson E,
- Labresh KA
- 11.↵Child health USA 2011. U.S. Department of Health and Human Services, Health Resources and Services Administration, Maternal and Child Health Bureau. 2011.
- Yoshimatsu J,
- Ikeda T,
- Katsuragi S,
- Minematsu K,
- Toyoda K,
- Nagatsuka K,
- Naritomi H,
- Miyamoto S,
- Iihara K,
- Yamamoto H,
- Ohno Y
- Kubo M,
- Kiyohara Y,
- Kato I,
- Tanizaki Y,
- Arima H,
- Tanaka K,
- Nakamura H,
- Okubo K,
- Iida M
- Bushnell C,
- McCullough LD,
- Awad IA,
- Chireau MV,
- Fedder WN,
- Furie KL,
- Howard VJ,
- Lichtman JH,
- Lisabeth LD,
- Piña IL,
- Reeves MJ,
- Rexrode KM,
- Saposnik G,
- Singh V,
- Towfighi A,
- Vaccarino V,
- Walters MR
- Bushnell C,
- Chireau M