In this study, G-HA was associated with stroke after delivery in women. In addition, there seemed to be differences in the effects and interactions of G-HA and Hx-HA depending on the type of subsequent stroke. Therefore, by investigating the history of headache disorder before pregnancy and closely monitoring the G-HA occurrence during pregnancy, we expect to be able to identify high-risk groups for each type of stroke. This may help pregnant women establish a primary prevention plan of subsequent stroke after delivery.

We used Korean NHIS data in this study. The majority of Koreans use Korean NHIS because of its reasonable price and wide coverage [24]. In addition, since the physician has no choice but to enter the disease code and treatment code into the NHIS database after treatment in order to receive support for reimbursement of medical items, the NIHS database contains complete personal medical history until the occurrence of clinical events (e.g., stroke) or death of each Korean citizen [23, 24]. Thus, the Korean NHIS database minimized data loss and enabled complete follow-up during the observation period of our study population [23]. In other words, it is real-world data of pregnant women nationwide.

Our data showed a clear association between G-HA and stroke after delivery. However, we also found the relationship between G-HA and stroke occurrence should be interpreted differently depending on the stroke type. Hemorrhagic stroke showed a relatively clear association with G-HA. Both G-HA and Hx-HA were associated with a high risk of hemorrhagic stroke after delivery, but there was no interaction between the two on the occurrence of stroke. In other words, although the [G-HA ( +) & Hx-HA( +)] group showed the highest cumulative incidence rate of hemorrhagic stroke or ICH in Fig. 3, it means that G-HA was not related to Hx-HA. One might think that since secondary headaches accompanying hemorrhagic stroke were diagnosed as G-HA, a close correlation between the two was observed. However, in this study, we analyzed the association between headaches “during pregnancy” and subsequent stroke occurrence “after childbirth,” so there is a temporal difference between the two. Additionally, as shown in Additional file 1: Table S1, this hypothesis is rejected because there was no hemorrhagic stroke that occurred within 1 month after G-HA diagnosis. Initially, the authors assumed that the close association between G-HA and hemorrhagic stroke was the result of hemorrhagic complication immediately after delivery due to eclampsia/preeclampsia [28, 29]. However, even when further analysis excluded women with gestational hypertension, G-HA was still closely associated with hemorrhagic stroke (Additional file 2: Table S2). Additionally, among a total of 565 subsequent hemorrhagic stroke events, only 4 cases in the G-HA ( +) group occurred in the post-partum period (Additional file 3: Table S3). Therefore, we believe that the influence of this type of pathological mechanism is not likely to be significant. Primary headaches usually improve during the first trimester of pregnancy [6, 17]. Of course, even so, most cases of G-HA discovered during pregnancy will be primary headache disorder. However, some of these headaches may be secondary to organic lesions (e.g., vasculitis, cavernous malformation, and moyamoya disease) [30]. Therefore, in terms of hemorrhagic stroke after delivery, we may recommend brain magnetic resonance imaging to women who developed G-HA during pregnancy to identify organic lesions that may cause hemorrhagic stroke, regardless of previous history of primary headache disorders. Additionally, in our data, G-HA showed a close association with ICH, while it did not show statistical significance with SAH. Therefore, it can be also interpreted that G-HA may be associated with a long-standing hypertensive condition that persists even after delivery.

On the other hand, G-HA showed a limited association with subsequent ischemic stroke. Women with both G-HA and Hx-HA had a nearly fivefold higher risk of ischemic stroke compared with women with neither headache type. However, the association between the two disappeared when considering the interaction with Hx-HA. In addition, the [G-HA( +) & Hx-HA(-)] group did not show a statistically significant difference in risk of ischemic stroke from the [G-HA(-) & Hx-HA(-)] group. These results may indicate that the persistence of previous headache disorders even during pregnancy is more important than the occurrence of headaches during pregnancy itself in the occurrence of ischemic stroke after delivery. In fact, women with Hx-HA without G-HA also showed a significantly higher risk of ischemic stroke compared to the [G-HA(-) & Hx-HA(-)] group. History of primary headache disorders, especially migraine, is a well-known risk factor for stroke [31]. The vasospasm that occurs during migraine attack reduces cerebral blood flow, which can lead to an ischemic stroke [11,12,13]. In addition, in migraine patients, platelet aggregation is activated, various prothrombotic markers are elevated, and paradoxical embolism through the patent foramen ovale is prevalent [7, 10, 13, 15]. Usually, migraines improve in the first trimester of pregnancy, but if they persist during pregnancy, it can mean intractable migraines with severe vasospasm. Therefore, in terms of ischemic stroke, it seems important to establish a primary prevention plan by closely monitoring whether G-HA persists during pregnancy in pregnant women with a history of primary headache disorder.

In our data, the prevalence of stroke after delivery is very low at 0.12%. Considering that the median age of our study population is 31 years and the observation period is a maximum of 9 years, it can be seen that most of our data represent a young age stroke prevalence of < 40 years. Therefore, 0.12% is considered a reasonable prevalence. In fact, it is consistent with the Korea Stroke Epidemiology Report reported in 2018 and the data available from the National Statistical Office (http://kosis.kr). However, because these women are young, they will suffer from a disability for a very long time. Therefore, we think that paying attention to the primary prevention of stroke after delivery is sufficiently medically and sociologically meaningful.

Our study has several limitations. First, although it used the prospectively collected Korean NHIS database, this is a retrospective cohort study. Thus, our results confirm the association between G-HA and stroke after pregnancy, but do not imply a causality. Second, due to the retrospective nature of the study, we were unable to learn in detail about the headache symptoms of G-HA experienced by participants (e.g., migraine-like or tension-type-like). Third, the prevalence of history of headache disorder in our cohort was estimated to be lower than generally known. Of course, it is true that the prevalence of primary headache disorder has been reported to be lower in the Korean population than in the Caucasian population. However, since our cohort included medical records from 2011 to 2020 for women who gave birth between 2012 and 2013, the impact of only having a washout period of up to 2 years to evaluate past medical history may be more significant. This low prevalence may underestimate the influence of history of headache disorder. Fourth, due to the characteristics of Korea, most of the study population are Asians. Therefore, additional validation studies are needed to apply our results to other races. Fifth, because our study determines prevalence based on the disease code entered by the attending physician, people who had a previous history of headache disorder but did not visit the hospital may be classified as G-HA if they were first detected due to pregnancy. Finally, because most of the women included in the analysis were young, the accompanying vascular risk factors were small. Therefore, the possibility that the effect of G-HA on stroke was relatively overestimated should be considered because the effect of these comorbidities was small.