Pregnancy-Related Complications and Incidence of Atrial Fibrillation: A Systematic Review

Pregnancy-related complications are associated with a higher risk of various incident cardiovascular diseases, but their specific potential relationship with incident atrial fibrillation (AF) is less clear. This systematic review summarises the available evidence from observational studies which have examined associations between pregnancy-related complications and the risk of AF. MEDLINE and EMBASE (Ovid) were searched for studies between 1990 to 10 February 2022. Pregnancy-related complications examined included hypertensive disorders of pregnancy (HDP), gestational diabetes, placental abruption, preterm birth, small-for-gestational-age and stillbirth. Study selection, data extraction and quality assessment were completed independently by two reviewers. Narrative synthesis was used to evaluate the results of the included studies. Nine observational studies were included, with eight eligible for narrative synthesis. Sample sizes ranged from 1839 to 2,359,386. Median follow-up ranged from 2 to 36 years. Six studies reported that pregnancy-related complications were associated with a significantly increased risk of incident AF. Hazard ratios (HRs) (95% confidence intervals) for the four studies that evaluated HDP ranged from 1.1 (0.8–1.6) to 1.9 (1.4–2.7). For the four studies that evaluated pre-eclampsia, HRs ranged from 1.2 (0.9–1.6) to 1.9 (1.7–2.2). Current evidence from observational studies suggests pregnancy-related complications are associated with a significantly higher risk of incident AF. However, only a small number of studies examining each pregnancy-related complication were identified, and considerable statistical heterogeneity was observed. Further large-scale prospective studies are required to confirm the association between pregnancy-related complications and incident AF.


Introduction
Pregnancy-related complications such as hypertensive disorders of pregnancy (HDP), preterm birth and fetal growth restriction are interrelated disorders that share risk factors with atrial fibrillation (AF) [1]. Women experiencing pregnancy-related complications are at higher risk of developing AF-related risk factors such as hypertension, diabetes, renal dysfunction, dyslipidaemia and cardiovascular diseases (CVD) compared to women without pregnancy-related complications [2]. Globally, the prevalence of HDP is approximately 10% of all pregnancies [3]. Similarly, the global preterm birth rate has been estimated at 10.6% [4], whereas hyperglycaemia may impact 16.9% of pregnancies [5]. Previous systematic reviews of observational studies have shown that pre-eclampsia and gestational hypertension are associated with a two-to four-fold increased risk of CVD [6][7][8][9][10]. Gestational diabetes nearly doubles the risk of CVD [10][11][12]. Preterm birth has been associated with a two-fold greater risk of developing maternal CVD [13,14].
AF is the most common sustained cardiac arrhythmia and is estimated to impact approximately 33.5 million people globally [15]. Recently, the European Society of Cardiology re-estimated the lifetime risk of AF to be one in three individuals among European populations aged ≥ 55 years [16]. While the incidence and prevalence of AF are higher among males, females with AF have greater morbidity, including a higher risk of stroke [17].
AF and pregnancy-related complications share common pathophysiological mechanisms and risk factors. The incidence of later-life chronic hypertension is increased up to four-fold among women with HDP compared to those with normotensive pregnancies. [6,8] Hypertension is an important risk factor for incident AF [18]. Moreover, pregnancy-related complications are associated with cardiac remodelling and dysfunction, which can persist beyond the postpartum period [19][20][21][22]. These shared common pathophysiological mechanisms and risk factors could help target the prevention, earlier identification and treatment of women at higher risk of developing AF, helping to reduce sex disparities in AF morbidity.
The aim of this systematic review was to evaluate the available evidence examining the association between pregnancy-related complications and incident AF.

Methods
The reporting of this systematic review was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and Synthesis Without Meta-analysis (SWiM) in systematic reviews: reporting guideline [23,24]. The full PRISMA checklist table is reported in the Supplementary Material (Table S1). The protocol was registered on PROSPERO (reference number: CRD42020180058).

Search Strategy
Ovid MEDLINE(R) and Ovid EMBASE were searched to identify eligible studies from 1990 to 10 February 2022. The full search details are described in the Supplementary Material (Tables S2 and S3). The search strategy combined pregnancy-related complications, cardiovascular diseases and AF. In addition, AF keywords with wild cards were used to capture wording variation.
The search was restricted to human studies, articles published in English and observational study designs, including cross-sectional, case-control, and cohort studies.
Studies identified through the searches were imported to EndNote (version X9) and duplicate records were removed. De-duplicated results were exported to Rayyan QCRI for screening [25]. Two reviewers (TA and AA) independently screened the titles and abstracts according to the inclusion/exclusion criteria. The full-texts of articles identified as potentially relevant at the title and abstract stage were retrieved in full-text and assessed independently by two reviewers (TA and AA). Disagreements were resolved through discussion and consulting with co-authors (SLH and DAL). Reference lists of included studies and relevant reviews were also screened to identify any additional relevant articles.

Inclusion and Exclusion Criteria
Observational studies including cross-sectional, case-control and cohort studies which examined associations between pregnancy-related complications and incidence of AF after delivery were eligible for inclusion. Any study which included women who had a diagnosis of AF prior to their first pregnancy or women who developed AF during their index pregnancy or labour period was excluded. Authors of studies that reported AF as part of composite CVD were contacted to provide relevant data regarding AF separately. Reviews, conference posters or abstracts, editorials and commentaries were excluded.

Data Extraction
Data were extracted independently by two reviewers (TA and AA) from the selected articles using a pre-prepared standardised data extraction form and any disagreements were resolved by discussion. The data were extracted into a spreadsheet collating: article information including author names, publication year, country, funding sources and conflicts of interest; population and comparator data including exclusion criteria, age, co-morbidities, prevalence and type of all pregnancy-related complications reported; methodology including methods used to identify AF, study period, type of exposure, follow-up time, list of variables adjusted for in analyses; results including incidence of AF among exposed and non-exposed, and adjusted and unadjusted point estimates.

Quality Assessment
Two reviewers (TA and AA) independently assessed the risk of bias in the included studies using the Newcastle-Ottawa Scale (NOS) [26].

Statistical Analysis
Meta-analysis was not conducted due to heterogeneity in the included studies, such as variation in exposure definitions, age of the cohorts and follow-up time. Therefore, a narrative synthesis using summarised effect estimates was conducted to describe the evidence.

Participants
The number of participants included in the eight cohort studies ranged from [28]

Data Sources
The included studies utilised data from Canada [27,31], the USA [28,32], Denmark, [33] Korea [35] and the UK [29,30,34]. The two studies conducted in Canada used routinely collected health data from Ontario [31] and Quebec [27]. Both studies carried out in the USA used the same cohort from the Rochester Epidemiology Project [28,32], while the studies conducted in Denmark [33] and Korea [35] utilised routinely collected data at a nationwide level. The studies conducted in the UK used the UK Biobank data [29] and data from CALIBER [30].
HDP was defined as any diagnosis using an ICD code or self-reported at recruitment Diagnosis for any previously mentioned conditions based on ICD-9&10 codes from inpatient and outpatient records, death registrations, and primary care diagnoses HDP was defined by a validated electronic diagnostic algorithm and using HICDA codes. Defined using ICD-9 and ICD-10.

Follow-Up
Follow-up durations ranged from a median of 2 [34]

Adjustment for Potential Confounding Factors
Six studies adjusted for factors such as socioeconomic status, maternal age and diabetes [27,30,31,[33][34][35]. Two studies adjusted for a smaller group of risk factors: the first adjusted for age at enrolment and race [29], and the second adjusted for education, smoking and obesity [28]. Table 2 summarises the risk factors adjusted for each individual study.

Assessment of Study Quality
All included studies were rated as good quality according to the Newcastle-Ottawa Scale, with total scores ranging from 6-9 (Table 1 and Table S5).

Pregnancy-Related Complications and Incident Atrial Fibrillation
Six of the eight included studies reported a significant association between pregnancyrelated complications and incident AF [27,28,30,31,33,35]. The associations were attenuated but remained statistically significant after adjustment for potential confounders (Table 2). Two studies reported no association between pregnancy complications and incident AF [29,34]. The event rate for incident AF among women with pregnancy-related complications was in the range of 0.1%-1% for all cohort studies [27,[29][30][31] except for one study, which was 38.0% [28].

Preterm Birth
One study examined the association between preterm birth and incident AF [27]. The risk of developing AF among women who had a previous preterm birth was significantly higher than among women who had not experienced a preterm birth, with HR 1.4 (1.3-1.6) (see Table 2 and Figure A1) [27].

Gestational Diabetes
One study evaluated the role of gestational diabetes and incident AF and revealed a higher risk of AF among women with gestational diabetes compared to women without gestational diabetes, with HR 1.4 (1.1-1.7) (see Table 2 and Figure A1) [33].

Composite Pregnancy Complications
One study examined the association of composite pregnancy complications such as maternal placental syndrome and incident AF, reporting a higher risk of incident AF in women with a history of maternal placental syndrome who had a higher risk of incident AF compared to women without HR 1.5 (1.1-2.0) (see Table 2 and Figure A1) [31].

Discussion
This systematic review identified eight cohort studies [27][28][29][30][31][33][34][35] and one casecontrol study [32] examining the association between pregnancy-related complications and incident AF. Overall, our findings indicate that pregnancy complications, particularly pre-eclampsia, may be associated with an increased risk of incident AF; however, there was considerable heterogeneity between studies, which may have resulted from inconsistency in the definition of the pregnancy-related complications and/or variation in the follow-up time. Moreover, there are other potential explanations for this heterogeneity, such as timing of data collection due to changes in pregnancy care, exposure definition, medical advances and postpartum follow-up. Therefore, these results need to be interpreted with caution.
Previous systematic reviews have investigated the association between pregnancyrelated complications and CVD, with most focusing on either a composite of CVD outcomes, myocardial infarction, stroke, heart failure or CVD mortality with pre-eclampsia [6][7][8][9], gestational diabetes [11,12], preterm birth [13,14,36], miscarriage [37] and various pregnancyrelated complications. [10] Further, one review evaluated the incidence of AF during pregnancy. [38] Our review extends and clarifies this existing knowledge by focusing on associations between pregnancy-related complications and incident AF after pregnancy.
Our findings, which showed an overall increased risk of AF, were in line with previous reviews. [6,9,10,12,13] A recent review of reviews found a significant association between pre-eclampsia and a two-fold increased risk of ischemic heart disease and a composite CVD outcome, a four-fold increased risk for heart failure and a 1.5-1.9-fold increased risk of stroke [39].
There were some differences in the methods used to identify AF between the studies, particularly between countries. For example, in the UK, read codes are used by general practitioners to capture primary care transactions with greater detail/completeness of clinical coding than most other parts of the world. Further, undetected or incorrectly diagnosed AF may abate the validity of the assessment of the outcome. This is demonstrated in one nested case-control study where 147 incident AF cases were identified using ICD-9 and Mayo-adapted HICDA codes, but after a chart review, 36 cases were excluded, and 2 cases were added [32]. Furthermore, one study enrolled women from the UK biobank with previous live birth history at the baseline visit (2006-2010) [29]. The documentation of incident AF was during follow-up from the baseline visit, while the birth history could have been up to 30 years before the baseline visit [29]. The reported mean age at enrolment was approximately 52 years for women with HDP compared to 57 years for women without HDP [29]. The time from first birth to baseline visit was 23 years for women with HDP and 31 years and women without HDP. This gap between birth and UK biobank baseline visits creates an immortal time interval because women who developed incident AF in that period were counted as having prevalent AF by design and excluded.
The event rate for incident AF among women with pregnancy-related complications in Garovic et al. [28] was 38% compared to around 1% for all other cohort studies [27,[29][30][31]. The higher incidence of AF reported by Garovic et al. [28] is likely due to a longer follow-up time, with a median of 36 years. This highlights the importance of longer follow-up times in evaluating incident AF. Furthermore, Ray et al. [31] found the mean age of onset of AF, heart failure or ventricular dysrhythmia was 38 years, while the mean age at delivery was 29 years. Additionally, the divergence in the cumulative probability of AF hospitalisation increased with time [31]. Therefore, it is possible that studies with shorter follow-up times or studies that include younger women may underestimate incident AF as these women may not have had enough time to develop AF. Hence, these results indicate that studies with longer follow-up times (>10 years) are preferable when examining the association between pregnancy-related complications and incident AF.
Regarding baseline characteristics such as hypertension and diabetes, there was variability in the definition and inclusion of these conditions between studies which led to large variability in the estimated prevalence between studies. It has commonly been assumed that higher cardiovascular risk during or after pregnancy among women with pregnancy-related complications is largely due to shared pre-pregnancy cardiovascular risk factors [40,41]. However, higher cardiovascular risk profiles later in life in part explain the association between pregnancy-related complications and CVD [42,43].
Peripartum cardiomyopathy is an uncommon form of cardiomyopathy that develops during the later stages of pregnancy or shortly after giving birth [44]. In a matched retrospective cohort, AF was one of the most common incident CVD among women with peripartum cardiomyopathy one year after delivery [45]. Moreover, previous systematic reviews have reported that the prevalence of pregnancy-related complications, such as HDP was 22% (95% CI; 16%-28%) [46] and the prevalence of AF was 5% (1%-11%) [47] among women with peripartum cardiomyopathy. However, a recent study showed that pregnancy-related complications were associated with higher risk of CVD independently from peripartum cardiomyopathy [48]. Another recent study that utilised data from the National Inpatient Sample database, highlighted that women with peripartum cardiomyopathy and AF had a higher rate of in-hospital mortality (4% vs. 0.7%, p = 0.02) and greater use of mechanical ventilation (14% vs. 6.7%, p = 0.044), when compared to those without AF [49]. These findings emphasise the significance of early detection and management of AF in women with peripartum cardiomyopathy in order to improve their outcomes and reduce healthcare resource utilisation [49].
A recent review highlighted that numerous guidelines have recognised HDP as a predictor of CVD [50]. Indeed, the 2019 International Federation of Gynaecology and Obstetrics suggested that pregnancy-related complications, including HDP, gestational diabetes mellitus, foetal growth restriction, preterm birth, recurrent pregnancy loss and placental abruption, be acknowledged as predictors of CVD [51]. However, the recent 2020 European Society of Cardiology (ESC) guidelines for the diagnosis and management of AF did not highlight the effect of pregnancy-related complications on the risk of incident AF [16]. This may be due to the small number of previously published studies that have evaluated the association between pregnancy-related complications and incident AF, as demonstrated in this review.
The potential association between pregnancy-related complications and incident AF likely involves multiple mechanisms. First, risk factors such as metabolic syndrome, which are common to multiple outcomes, may drive incident AF [52]. Second, cardiac remodelling and fibrosis, which occur with pregnancy-related complications, can lead to long-term impaired cardiac function [22,[53][54][55] and increase the risk of incident AF. Another possible explanation may be that pregnancy-related complications increase the risk of incident AF risk factors, such as hypertension, an unfavourable lipid profile and diabetes [56][57][58]. Finally, genetic factors may play a role in the observed associations, as a recent study found that single nucleotide polymorphisms (SNP) rs2200733 near the PITX2 gene was associated with pre-eclampsia and AF [59].

Strengths and Limitations
This review has several strengths. First, a comprehensive search strategy was used and included CVD in general as an outcome rather than limiting the searches to studies of AF alone. The review methodology was comprehensive, with screening and data extraction independently conducted by two reviewers. However, there are some limitations. First, only nine studies were eligible for inclusion. Second, the exposure definitions varied between studies which could increase the heterogeneity between studies. Third, some of the studies could not control for common confounding factors due to insufficient recording in routine healthcare record data sources. Fourth, the follow-up time varied from less than 1 year to 38 years, and only three studies had an adequate long-term follow-up; hence, the incidence of AF was low. Fifth, most studies included young pregnant women (<30 years old) with short follow-up duration; therefore, the risk of incident AF may be underestimated. Sixth, AF misclassification is possible resulting from utilising codes to identify AF. Seventh, the inability to determine whether the increased incidence of AF is related to the spontaneous onset of labour or iatrogenic preterm delivery among women who have experienced preterm birth. Finally, of the nine included studies, eight were from Western countries, which may limit the generalisability of the findings.
Large-scale, prospective cohort studies with long-term follow-up (>10 years), are needed to expose the relationship between pregnancy-related complications and incident AF. Future studies should also aim to investigate the temporal relationships between pregnancy-related complications, AF and peripartum cardiomyopathy.

Conclusions
Pregnancy-related complications (pre-eclampsia, hypertensive disorders of pregnancy, preterm birth and maternal placental syndrome) could be associated with a higher risk of incident AF. However, the small number of included studies and the significant heterogeneity in those studies suggest that further research is required.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/jcm12041316/s1, Table S1: PRISMA checklist; Table S2: Search strategy for Ovid MEDLINE(R) from 1990 to April 06, 2020, and updated on 10 February 2022; Table S3: Search strategy for Ovid Embase from 1990 to April 06, 2020, and updated on 10 February 2022; Table S4: Baseline characteristics of the participants in the included studies and Table S5: Risk of Bias Assessment in included Studies using Newcastle-Ottawa Scale.