Association between migraine, migraine subtype, and adverse pregnancy A systematic review and meta‐analysis

Authors: Anna Steen Hansen, Cecilie Holm Christiansen, Ane Lilleøre Rom, Nina Olsen Nathan, Marie Stampe Emborg, Line Rode, Hanne Kristine Hegaard

Abstract

Introduction

Migraine is one of the most prevalent conditions worldwide. This systematic review aimed to evaluate the association between migraine, its subtypes, and adverse pregnancy outcomes.

Material and Methods

Eligible cohort and retrospective case–control studies were included from PubMed and Embase databases from their inception to May 2024. Adverse pregnancy outcomes of interest were preeclampsia, preterm birth, low birthweight, small for gestational age, and placental abruption. Study quality was assessed using the Newcastle‐Ottawa Scale. Meta‐analyses of the outcomes with their odds ratios (ORs) and adjusted ORs (aOR), including a 95% confidence interval (CI), were performed using RevMan. Outcomes were pooled using random effects models, with separate analyses for cohort and retrospective case–control studies. The protocol was registered with PROSPERO (no. CRD42023404759).

Results

This meta‐analysis included 19 studies (11 cohort and 8 retrospective case–control) encompassing 1 420 690 deliveries. Significant associations were observed between migraine and increased risk of preeclampsia (cohort: aOR 1.28 [95% CI: 1.11–1.47], I ^2^ = 0%), (retrospective case–control: aOR 3.4 [95% CI: 1.81–6.4], I ^2^ = 83%) and preterm birth (cohort: aOR 1.30 [95% CI: 1.17–1.44], I ^2^ = 11%). The meta‐analyses of adjusted data on low birthweight and small for gestational age were inconsistent with respect to statistical significance (cohort: aOR 1.27 [95% CI: 0.89–1.82], I ^2^ = 36% and aOR 1.07 [95% CI: 1.03–1.12], I ^2^ = 0%, respectively). In addition, migraine without aura (MO) (cohort: OR 1.62 [95% CI: 1.30–2.01], I ^2^ = 0%; retrospective case–control: aOR 4.91 [95% CI: 2.78–8.67], I ^2^ = 0%) and migraine with aura (MA) (cohort: OR 2.06 [95% CI: 1–4.27], I ^2^ = 29%) were significantly associated with the risk of preeclampsia. Similarly, MO (cohort: OR 1.28 [95% CI: 1.11–1.49], I ^2^ = 0%) and MA (cohort: OR 1.25 [95% CI: 1.07–1.47], I ^2^ = 0%) were associated with preterm birth risk.

Conclusions

Pregnant women with migraines have a higher risk of preeclampsia and preterm birth compared with those without migraines. Migraine could be associated with an increased risk of low birth weight and small for gestational age. Sub‐analyses indicate an elevated risk of preeclampsia and preterm birth across migraine subtypes. Notably, no previous meta‐analyses have differentiated between migraine subtypes. Additional studies are needed to strengthen these findings.

Key messageMigraine is associated with an increased risk of preeclampsia and preterm birth. It may also be linked to an increased risk of low birth weight and small for gestational age. Sub‐analyses indicate an elevated risk of preeclampsia and preterm birth across migraine subtypes.

INTRODUCTION

Migraine is a neurological condition and is the third most prevalent disease globally, affecting ~15% of the population. It has two major migraine without aura (MO) and migraine with aura (MA), with approximately one‐third of patients with migraine experiencing attacks with aura. ^1^ Migraine is approximately three times more common in women than in men, with its prevalence peaking during the reproductive years (24% of women aged 30–39 years are affected). ^2^ Additionally, the prevalence is estimated at 20% among pregnant women. ^3^ However, significant improvement has been reported in pregnant women with MO, with most experiencing relief by the third trimester. In contrast, fewer pregnant women with MA report improvement or remission. ^4^

Migraine is associated with an increased risk of cardiovascular complications, ^5^ and MA specifically increases the risk of ischemic stroke. ^6^ Pregnancy may exacerbate these risks due to compromised vascular compensatory mechanisms. ^7^ Additionally, migraine and its subtypes may share overlapping pathophysiological features with certain adverse pregnancy outcomes such as preeclampsia and preterm birth. ^8^

An umbrella review from 2024, ^9^ including an updated systematic review and meta‐analysis of the Aukes et al. study from 2019, suggests clear evidence of an association between migraine and an increased risk of preeclampsia and preterm birth. However, the authors emphasize that further investigation is needed regarding the association between migraine and other pregnancy complications, such as placental abruption, low birth weight, and small‐for‐gestational‐age infants. ^9^

For the future, the authors of the umbrella review recommend examining the association between migraine subtypes (MO and MA) and pregnancy complications, as the umbrella review did not specifically address the impact of migraine with aura on pregnancy complications. While both migraine subtypes may be associated with an increased risk of pregnancy complications, the association could be stronger for MA due to a closer link to vascular and endothelial dysfunction. ^10^ Notably, no previous meta‐analyses have differentiated between migraine subtypes.

Therefore, this study aimed to conduct a systematic review and meta‐analysis to explore the association between migraine and selected adverse pregnancy outcomes, including preeclampsia, preterm birth, low birth weight, small for gestational age, and placental abruption. Separate analyses were performed for cohort and retrospective case–control studies. Additionally, subgroup analyses were conducted to assess the effects of migraine subtypes (MO and MA).

MATERIAL AND METHODS

This systematic review and meta‐analysis were conducted adhering to the Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) guidelines. ^11^ The review protocol was registered with PROSPERO (no. CRD42023404759).

Data sources

A comprehensive search of databases, including PubMed (Medline) and Embase, was conducted for studies published from their inception until May 18, 2024. The search strategy, initially designed for Embase, was adapted for PubMed. It included index terms and free‐text keywords related to migraine as the exposure and pregnancy as the event. The complete search strategy is available in Appendix S1.

Screening and study selection

We searched for studies investigating pregnancy outcomes in relation to maternal migraine. Inclusion criteria were population‐based studies of pregnant women with migraine, comparing outcomes with those in pregnant women without migraine. The selected outcomes included preeclampsia, preterm birth, low birthweight, small for gestational age, and placental abruption. Studies focusing on populations with selected specific diseases, as well as case reports, reviews, and abstracts (not published data), were excluded. No language restrictions were applied. Abstract and full‐text screening were performed independently by two authors (A.S.H. and C.H.C.), with disagreements resolved through discussion and consensus involving a third party (H.K.H.). COVIDENCE, a web‐based software tool, was used to screen and extract the included articles. ^12^

Data extraction

Data extraction was performed independently by A.S.H. and H.K.H. and cross‐checked by M.S.E. Extracted data included study characteristics (year of publication, population description, country, study period, study design, and sample size), pregnancy outcomes, migraine diagnosis, and unadjusted/adjusted odds ratios (ORs) with covariates. Characteristics of the included cohort and retrospective case–control studies are presented in Tables 1 and 2, respectively.

Quality assessment

Study quality and risk of bias were independently assessed using the Newcastle‐Ottawa Scale (NOS) ^32^ by two authors (A.S.H. and M.S.E.). The NOS evaluates observational studies across three “selection,” “comparability,” and “outcome/exposure” for cohort and retrospective case–control studies. Regarding “comparability,” the studies received one point if they adjusted for age and two points if they adjusted for age and other relevant factors. The domain “selection” refers to the representativeness of cases, controls, and cohorts, while “outcome” represents the quality of the outcomes. A maximum of nine points can be given. Notably, no standardized criteria have been established for defining low, moderate, and high risk of bias with NOS. However, in line with previous systematic reviews, ^33^ , ^34^ we classified studies awarded a full score or only one missing point as low risk of bias, studies with two missing points in different domains as having a moderate risk of bias, and studies missing more than one point in one domain as having a high risk of bias.

Statistical analyses

Meta‐analyses were conducted for outcomes reported in two or more studies. Unadjusted ORs and adjusted ORs (aORs) were analyzed separately. When unadjusted ORs were not provided, they were calculated from raw data. Cohort and retrospective case–control studies were analyzed independently. Meta‐analyses were performed using RevMan (version 5.4) with random effects models to account for the anticipated high heterogeneity, as found in previous studies. ^35^ Heterogeneity was assessed using the I ^2^ statistic, with values >50% considered to indicate substantial heterogeneity, as described in the Cochrane Handbook for Systematic Reviews of Interventions. ^36^ The level of statistical significance was set at p < 0.05.

RESULTS

We identified 4789 references, including 1664 from PubMed and 3125 from Embase. After removing duplicates, 3844 references remained for title and abstract screening. Of these, 3776 were excluded, leaving 68 articles for full‐text review. Finally, 19 studies met the inclusion criteria and were included in this review (Figure 1). A detailed list of excluded articles and their references is provided in Appendix S2.

The review included 19 studies, comprising 11 cohort studies ^13^ , ^14^ , ^15^ , ^16^ , ^17^ , ^18^ , ^19^ , ^20^ , ^21^ , ^23^ , ^24^ and 8 retrospective case–control studies. ^8^ , ^25^ , ^26^ , ^27^ , ^28^ , ^29^ , ^30^ , ^31^ The studies were conducted between 1980 and 2022 and included populations from 13 different countries, involving 1 420 690 deliveries. Migraine subtype was investigated in seven studies. ^8^ , ^19^ , ^21^ , ^23^ , ^27^ , ^28^ , ^30^

Migraine diagnosis was based on the International Classification of Disease (ICD)‐9 (or ICD‐8/10) ^24^ codes, ^15^ the International Classification of Headache Disorders (ICDH)‐2 (or ICDH‐1 ^28^ /ICDH‐3 ^27^ ) criteria, ^8^ , ^19^ , ^30^ , ^31^ self‐reported diagnoses, ^14^ , ^20^ , ^29^ self‐reported physician's diagnoses, ^16^ , ^17^ , ^23^ , ^25^ , ^26^ validated questionnaires, ^21^ and hospital or medication records. ^13^ , ^18^

The outcomes were defined as preeclampsia as elevated blood pressure ≥140/90 mmHg and proteinuria (≥30 mg/dL or 1+ on more than two urine sticks assessed more than 4 h apart), low birthweight as infant birthweight ≤2500 g, preterm birth as ≤37 gestational weeks, and small for gestational age as infant birthweight ≤10th percentile (Tables 1 and 2).

Covariate adjustment varied between studies. Nearly, all were adjusted for maternal age (except two studies). ^20^ , ^29^ Moreover, all except seven studies adjusted for parity ^8^ , ^13^ , ^19^ , ^20^ , ^25^ , ^27^ , ^29^ and all except five adjusted for smoking. ^13^ , ^15^ , ^25^ , ^26^ , ^30^ Similarly, seven studies adjusted for hypertension, ^15^ , ^16^ , ^19^ , ^20^ , ^23^ , ^24^ and eight studies adjusted for body mass index or obesity. ^16^ , ^18^ , ^23^ , ^25^ , ^26^ , ^27^ , ^30^ Table 1 presents the results of the various adjustments. An unadjusted OR was calculated from raw data in eight of the 19 studies. ^13^ , ^16^ , ^17^ , ^18^ , ^21^ , ^23^ , ^24^ , ^28^

Preeclampsia

Eight cohort studies ^13^ , ^15^ , ^17^ , ^18^ , ^19^ , ^20^ , ^23^ , ^24^ and six retrospective case–control studies ^25^ , ^26^ , ^28^ , ^29^ , ^30^ , ^31^ examined the association between migraine and preeclampsia. All eight cohort studies (n = 1 412 684 deliveries) provided raw data or unadjusted OR. The meta‐analysis showed a statistically significant OR of 1.35 (95% CI: 1.18–1.56, I ^2^ = 64%) (Figure 2). The risk of bias was considered high for one study, ^23^ moderate for two studies, ^13^ , ^20^ and low for five studies. ^15^ , ^17^ , ^18^ , ^24^ , ^28^ Three cohort studies (n = 42 289 deliveries) provided adjusted ORs. The meta‐analysis showed a statistically significant aOR of 1.28 (95% CI: 1.11–1.47, I ^2^ = 0%) (Figure 2). Among these three studies, the risk of bias was considered moderate for one study, ^20^ and low for two studies. ^15^ , ^17^

Six retrospective case–control studies (n = 2841 deliveries) provided raw data or OR. The meta‐analysis showed a statistically significant OR of 3.3 (95% CI: 2.22–4.90, I ^2^ = 83%) (Figure 3). The risk of bias was considered high for three studies, ^25^ , ^26^ , ^31^ moderate for two studies, ^29^ , ^30^ and low for one study. ^28^ Four retrospective case–control studies (n = 2511 deliveries) provided adjusted ORs. The meta‐analysis showed a statistically significant aOR of 3.4 (95% CI: 1.81–6.40, I ^2^ = 83%) (Figure 3). The risk of bias was considered high for two studies ^25^ , ^26^ and moderate for two studies. ^29^ , ^30^

Preterm birth

Nine cohort studies ^13^ , ^14^ , ^15^ , ^16^ , ^19^ , ^20^ , ^21^ , ^23^ , ^24^ examined the association between migraine and preterm birth. All nine cohort studies (n = 363 730 deliveries) provided raw data or unadjusted ORs. The meta‐analysis showed a statistically significant OR of 1.26 (95% CI: 1.16–1.36, I ^2^ = 39%) (Figure 4). The risk of bias was considered high for three studies, ^14^ , ^21^ , ^23^ moderate for two studies, ^13^ , ^20^ and low for four studies. ^15^ , ^16^ , ^19^ , ^24^ Three cohort studies (n = 39 601 deliveries) provided adjusted ORs. The meta‐analysis showed a statistically significant aOR of 1.30 (95% CI: 1.17–1.44, I ^2^ = 11%) (Figure 4). The risk of bias was considered moderate for one study, ^20^ and low for two studies. ^15^ , ^19^

Low birthweight

A total of five cohort studies ^13^ , ^15^ , ^19^ , ^23^ , ^24^ examined the association between migraine and low birthweight. All five cohort studies (n = 350 022 deliveries) provided raw data or unadjusted OR. The meta‐analysis showed statistically significant results, with an OR of 1.19 (95% CI: 1.15–1.23, I ^2^ = 0%) (Figure 5). The risk of bias was considered high for one study, ^23^ moderate for one study, ^13^ and low for three studies. ^15^ , ^19^ , ^24^ Two cohort studies (n = 30 151 deliveries) provided adjusted ORs. The meta‐analysis showed no statistically significant association, with an aOR of 1.27 (95% CI: 0.89–1.82, I ^2^ = 36%) (Figure 5). The risk of bias was considered low for both studies. ^15^ , ^19^

Small for gestational age

Five cohort studies ^15^ , ^19^ , ^20^ , ^21^ , ^24^ examined the association between migraine and small for gestational age. All five cohort studies (n = 291 156 deliveries) provided raw data or unadjusted OR. The meta‐analysis did not show a statistically significant association with an OR of 1.04 (95% CI: 0.95–1.14, I ^2^ = 52%) (Figure 6). The risk of bias was considered high for one study, ^21^ moderate for one study, ^20^ and low for three studies. ^15^ , ^19^ , ^24^ Three cohort studies (n = 39 601 deliveries) provided adjusted ORs. The meta‐analysis showed a statistically significant aOR of 1.07 (95% CI: 1.03–1.12, I ^2^ = 0%) (Figure 6). The risk of bias was considered moderate for one study ^20^ and low for two studies. ^15^ , ^19^

Placental abruption

Two cohort studies ^21^ , ^24^ and one retrospective case–control study ^8^ examined the association between migraine and placental abruption. One cohort study did not report cases of placental abruption among women with migraine, so no OR was calculated. ^21^ Consequently, only one cohort study remained, providing unadjusted data with an OR of 1.15 (95% CI: 0.91–1.45). ^24^ The retrospective case–control study showed an unadjusted OR of 2.20 (95% CI: 1.29–3.76) and an adjusted aOR of 2.14 (95% CI: 1.22–3.75). The risk of bias was considered low for this study. ^8^

Migraine subtypes and outcomes

Two cohort studies ^19^ , ^23^ and three retrospective case–control studies ^27^ , ^28^ , ^30^ examined the association between migraine subtypes and preeclampsia. The two cohort studies provided raw data for women with MO. For women with MO, the pooled OR was statistically significant at 1.62 (95% CI: 1.30–2.01, I ^2^ = 0%) (Appendix S3; Figure S1). Both cohort studies also provided raw data for women with MA. Similarly, for women with MA, the pooled OR was statistically significant at 2.06 (95% CI: 1.00–4.27, I ^2^ = 29%) (Appendix S3; Figure S2).

Three retrospective case–control studies provided raw data or OR for women with MO. Retrospective case–control studies showed higher risks for MO, with OR at 4.91 (95% CI: 4.14–5.82, I ^2^ = 0%) (Appendix S3; Figure S3). Only one of the retrospective case–control studies provided data on MA, showing a significantly increased risk of preeclampsia for women with MA, with an OR of 3.2 (95% CI: 1.9–5.4). ^30^ Additionally, adjusted data were provided by two retrospective case–control studies for women with MO. The meta‐analysis showed a significant association with an aOR of 4.91 (95% CI: 2.78–8.67, I ^2^ = 0%) (Appendix S3; Figure S3). ^27^ , ^30^ Only one study provided adjusted data on women with MA, which showed an aOR of 3.7 (95% CI: 2.1–6.4) compared with women without migraine. ^30^

Two cohort studies ^21^ , ^23^ examined the association between migraine subtypes and preterm birth. The cohort studies provided raw data or ORs for women with MO and MA. The meta‐analysis on women with MO showed a statistically significant association, with an OR of 1.28 (95% CI: 1.11–1.49, I ^2^ = 0%) (Appendix S3; Figure S4). The meta‐analysis on women with MA also showed a significant association with increased risk, with an OR of 1.25 (95% CI: 1.07–1.47, I ^2^ = 0%) (Appendix S3; Figure S5).

One cohort study ^23^ examined the association between migraine subtypes and low birth weight. The study provided raw data and adjusted analyses. For women with MO, the OR was reported as 1.04 (95% CI: 0.85–1.28) and the adjusted risk ratio (aRR) was 0.93 (95% CI: 0.75–1.15). For women with MA, the OR was 1.16 (95% CI: 0.94–1.43) and the adjusted risk ratio (aRR) was 1.06 (95% CI: 0.87–1.31).

Another cohort study ^21^ examined the association between migraine subtypes and small for gestational age. The study provided raw data only. For women with MO, the OR was 0.75 (95% CI: 0.21–2.64). For women with MA, the OR was 6.02 (95% CI: 2.22–16.32).

A retrospective case–control ^8^ study examined the association between migraine subtypes and placental abruption. The study provided unadjusted and adjusted odds ratios. For women with MO, the OR was 2.21 (95% CI: 1.07–4.53) and the aOR 2.11 (95% CI: 1.00–4.45). For women with MA, the OR was 2.20 (95% CI: 1.04–4.66) and the aOR 1.59 (95% CI: 0.70–3.62). A cohort study ^21^ did also investigate migraine subtypes and placental abruption, but no cases of the outcome were reported among women with migraine, and therefore no OR was calculated.

DISCUSSION

This systematic review and meta‐analysis revealed a statistically significant elevated risk of preeclampsia and preterm birth among pregnant women with migraine compared with those without migraine. An important novel aspect of this study is that we examine the association between each subtype of migraine, migraine without aura (MO) and migraine with aura (MA), and adverse maternal and fetal outcomes. This has not previously been investigated in a systematic review and meta‐analysis. Our sub‐analyses showed a higher risk of preeclampsia and preterm birth among women with both MO and MA compared with women without migraine. However, there still are too few studies that have included migraine subtypes for this meta‐analysis to be conclusive. Therefore, future studies should investigate whether women with MO and MA have the same risk of adverse maternal and fetal outcomes, as the symptom burden during pregnancy is different in the two subtypes, and the risk of stroke is higher in women with MA than in those with MO. ^37^

As there are too few studies investigating placental abruption, it was not possible to conduct a meta‐analysis. More studies on women with migraine and abruption placenta are needed in the future.

Furthermore, the meta‐analyses evaluating the risk of low birthweight and small for gestational age were inconsistent with respect to statistical significance.

The findings of this review and meta‐analysis align with those of the 2019 meta‐analysis by Aukes et al. ^35^ and the 2024 umbrella review by Phillips et al. ^9^ All analyses reveal an increased association between migraine, preeclampsia, and preterm delivery.

We observed much higher OR, aOR, and heterogeneity in our meta‐analysis based on case–control studies (case–control OR of 3.30 [95% CI 2.22–4.90, I ^2^ = 83%], case–control aOR of 3.40 [95% CI 1.81–6.40, I ^2^ = 83%]) compared with cohort studies (cohort OR of 1.35 [95% CI 1.18–1.56, I ^2^ = 64%], cohort aOR of 1.28 [95% CI 1.11–1.47, I ^2^ = 0%]). This is the rationale behind our decision to conduct meta‐analyses separately for the different study designs, as we find this approach more transparent. In addition, including lower‐quality studies (case–control) alongside higher‐quality ones (cohort) might dilute the reliability of the overall findings. This is in line with the overall findings in the umbrella review by Philips et al., ^9^ which also conducted sensitivity analyses restricted to study design.

In this study, we examine the association between maternal migraine and low birth weight, as well as small for gestational age. This is important because both low birth weight and small for gestational age are associated with an increased risk of short‐ and long‐term morbidity. ^38^ , ^39^ Based on cohort studies, we found that maternal migraine was associated with significantly increased odds of low birth weight in pooled meta‐analyses (OR of 1.19 [95% CI 1.15–1.23]), while studies providing adjusted odds ratios reported a non‐significant increase in odds (aOR of 1.27 [95% CI 0.89–1.82]). Slightly higher odds were reported for small for gestational age in unadjusted data from cohort studies (OR 1.04 [95% CI 0.95–1.14]), and cohort studies providing adjusted odds ratios (aOR of 1.07 [95% CI 1.03–1.12]).

Our findings align with the results of the umbrella study by Philips et al. ^9^ However, it is notable that there are still very few cohort studies in this field, and even fewer when pooled meta‐analyses are conducted based on adjusted OR, which is likely the most valid approach due to the risk of residual confounding The associations between migraines and low birth weight and small for gestational age remain unclear, and more studies are needed.

Our sub‐analyses showed a higher risk of preeclampsia and preterm birth among women with both MO and MA compared with women without migraine. However, the limited number of studies investigating migraine subtypes prevents conclusive results. Therefore, future studies should investigate whether women with MO and MA have the same risk of adverse maternal and fetal outcomes, as these subtypes differ in symptom burden during pregnancy and stroke risk, which is higher in women with MA than in those with MO. ^37^

The observed associations between migraine and adverse pregnancy outcomes, including preeclampsia and preterm birth, could be explained biologically by overlapping pathogeneses. Migraine, preeclampsia, placental abruption, and small for gestational age are thought to be associated with increased platelet activation. ^40^ , ^41^ Moreover, migraine has been linked to endothelial dysfunction, ^42^ a key component of preeclampsia pathophysiology. ^43^ Endothelial dysfunction is believed to contribute to systemic inflammation, thrombosis, and impaired vascular reactivity. ^42^ Elevated levels of C‐reactive protein, a putative marker of systemic inflammation, have been reported in individuals with migraines, ^44^ and this marker has also been implicated in preterm birth ^45^ and preeclampsia. ^46^ Additionally, reduced proangiogenic factors in individuals with migraines may contribute to endothelial dysfunction and vascular disorders such as preeclampsia. ^47^ Future research is needed regarding the mechanisms underlying the associations between migraines and adverse pregnancy outcomes. Moreover, clinical intervention studies examining the use of low‐dose aspirin could be relevant for women with migraine before 16 weeks gestation to mitigate the elevated risk of preeclampsia and preterm birth. ^35^ , ^48^

Similar risk factors could, in part, explain the association between migraine and adverse pregnancy outcomes. For instance, migraine has been associated with an elevated prevalence of cardiovascular risk factors, such as chronic hypertension and obesity. ^23^ In addition, hypertension and obesity are considered high and moderate risk factors for preeclampsia. ^49^ Maternal hypertension has also been associated with a higher risk of preterm birth and placental abruption. ^50^

Moreover, there is a high risk to confound migraines with preeclampsia symptoms during pregnancy due to overlapping clinical features, including severe headaches, visual disturbances, nausea, and vomiting complicating their differentiation. Migraines typically manifest as throbbing, unilateral headaches, often accompanied by aura, while preeclampsia‐related headaches are generally bilateral, constant, and unrelenting. ^51^ , ^52^ However, preeclampsia is usually distinguished by elevated blood pressure, proteinuria, and abnormal laboratory parameters, such as elevated liver enzymes or thrombocytopenia. In addition, preeclampsia is defined as occurring after the 20th week of gestation. ^52^

The strengths of this systematic review include adherence to PRISMA guidelines, the inclusion of a large sample size (1 420 690 deliveries), and a focus on migraine subtypes, which have not been explored in previous reviews. Many included studies were register‐based or multicenter studies, increasing the generalization of the results.

However, this review has some limitations. Given that migraine is the exposure of interest, no one has been able to conduct randomized controlled trials, necessitating the inclusion of only observational studies. The included studies reported varied ratios which led to the calculation of OR from raw data in 8 of the 19 studies, with the inclusion of a large proportion of unadjusted ORs in our analyses. Although conclusions based on aORs are preferred, these meta‐analyses were based on many fewer studies. Moreover, none of our included studies adjusted for migraine medications as their use is poorly reported, underlying the unaccountability in reviews. Although high‐, middle‐, and low‐income countries across five continents were represented in this review, the overrepresentation of high‐income countries limits the generalization of findings to low‐income settings.

The choice of the most suitable model (fixed or random effects) for our meta‐analyses should be considered. The option of a random effects model for meta‐analyses, as recommended by the Cochrane Handbook, accounts for expected heterogeneity among studies with varying population sizes, diagnostic methods, and adjustments. ^36^ Separate analyses for retrospective case–control and cohort studies, as well as adjusted and unadjusted ORs, provide a more comprehensive examination of data and facilitate a nuanced understanding of the results.

CONCLUSION

This systematic review and meta‐analysis identified significant associations between migraine in pregnant women and the risks of preeclampsia and preterm birth. Sub‐analyses further demonstrated an increased risk of preeclampsia and preterm birth regardless of migraine type. Additionally, adjusted data revealed a significant association between migraine and small for gestational age. However, adjusted analyses for low birthweight showed no significant association, and a meta‐analysis on placental abruption was not feasible due to insufficient studies. This systematic review emphasizes that more studies are still needed for placental abruption, small for gestational age, and low birthweight, as well as investigation of the relationship between all outcomes and migraine subtype.

AUTHOR CONTRIBUTIONS

All authors were responsible for the conception and design of this study. Anna Steen Hansen conducted data collection, extraction, and management, performed the analyses, and drafted the initial manuscript, which was reviewed and revised. Cecilie Holm Christiansen and Marie Stampe Emborg conducted data collection, extraction, and management. Hanne Kristine Hegaard conducted data collection, extraction, and management and performed the analyses. All authors approved the final draft.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information