Associations between adverse pregnancy outcomes and cognitive impairment and dementia: a systematic review and meta-analysis Authors: Eliza C Miller, Patrick Conley, Mohammad Alirezaei, Katrin Wolfova, Mitzi M Gonzales, Zaldy S Tan, Sarah E Tom, Lynn M Yee, Adam M Brickman, Natalie A Bello Published: 2024 ABSTRACT SummaryBackgroundIndividuals with adverse pregnancy outcomes have an increased risk of cerebrovascular disease, but the association between adverse pregnancy outcomes and cognitive impairment and dementia is less well established. We aimed to synthesise, combine, and assess the growing body of data examining the associations between adverse pregnancy outcomes and mild cognitive impairment and dementia in parous women.MethodsIn this systematic review and meta-analysis, we searched PubMed (MEDLINE), Web of Science, and Embase from database inception up to July 18, 2024, with no language restrictions, for observational studies or clinical trials that reported mild cognitive impairment or dementia as outcomes and included female individuals or women who had an adverse pregnancy outcome, including hypertensive disorders of pregnancy, gestational diabetes, stillbirth, fetal growth restriction, preterm birth, or placental abruption. We excluded studies of men, nulliparous women, women with pre-pregnancy conditions associated with impaired cognition, and studies examining cognitive impairment within 6 months of pregnancy. Database searches were supplemented by manual review of the reference lists of included studies. If studies met eligibility criteria but did not have sufficient data for meta-analysis (ie, did not report a summary statistic or a hazard ratio [HR] for outcome estimation), they were included in the systematic review and excluded from the meta-analysis. After removing duplicates, two investigators independently screened titles and abstracts using Covidence software, with potentially eligible studies undergoing full-text review by the same reviewers, with further review by a third reviewer and disagreements resolved by discussion and group consensus. Study quality was assessed and summary statistics extracted by two reviewers independently. The primary outcomes of our study were mild cognitive impairment, all-cause dementia, Alzheimer’s disease, and vascular dementia. Heterogeneity was measured using the Q test and I2 statistic, and we used random-effects models with inverse-variance weighting to assess the association between adverse pregnancy outcome and primary outcomes with sufficient meta-analysable data via pooled adjusted HRs and 95% CIs. The study protocol was registered with PROSPERO, CRD42023453511.FindingsOf 11 251 publications identified, 15 studies (including 7 347 202 participants) met inclusion criteria for the systematic review, and 11 studies (6 263 431 participants) had sufficient data for meta-analysis. A history of any adverse pregnancy outcome was associated with higher risk of all-cause dementia (adjusted HR 1·32 [95% CI 1·17–1·49]; I2= 80%), Alzheimer’s disease (1·26 [1·04–1·53]; I2=63%), and vascular dementia (1·94 [1·70–2·21]; I2=0%). A history of any hypertensive disorder of pregnancy was significantly associated with all-cause dementia (1·32 [1·11–1·57]; I2=74%) and vascular dementia (1·78 [1·46–2·17]; I2=0%), but not Alzheimer’s disease (1·24 [0·98–1·57]; I2=66%). Stillbirth was not significantly associated with higher risk of all-cause dementia (1·26 [95% CI 0·93–1·71]; I2=62%). In individual studies, similar effect directions were observed for preterm birth and fetal growth restriction, but data were insufficient for meta-analysis.InterpretationGiven their increased risk of dementia, women with a history of adverse pregnancy outcomes should be evaluated for additional dementia risk factors and monitored closely for any signs of cognitive decline. Furthermore, to obtain more reliable findings, future studies should measure both exposures and outcomes prospectively and objectively.FundingNational Institutes of Health, National Institute of Neurological Disorders and Stroke, National Institute on Aging, and National Heart, Lung and Blood Institute. Introduction Adverse pregnancy outcomes are associated with increased risk and earlier onset of cerebrovascular disease and its risk factors, such as hypertension and metabolic syndrome.1 Given the strong association between cerebrovascular disease and dementia,2 a similar association between a history of adverse pregnancy outcomes and future dementia risk has been hypothesised.3-5 Female individuals have higher lifetime risk of dementia, more years lived with dementia-related disability, and higher dementia-related mortality than do male individuals.6-8 Understanding the role of adverse pregnancy outcomes—a sex-specific risk factor for cerebrovascular disease—is of high interest in the effort to reduce the global burden of disease from cardiovascular causes.9 This understanding is particularly important nowadays because the prevalence of multiple adverse pregnancy outcomes, including gestational hypertension,10 pre-eclampsia,10,11 and gestational diabetes,12-14 is increasing, and the prevalence of other adverse pregnancy outcomes, such as preterm birth and stillbirth, has not shown measurable decline.15,16 The past decade has seen rapid growth of research investigating adverse pregnancy outcomes and their long-term effects on maternal cognition.17-23 Multiple observational studies, including several meta-analyses, investigating the association between hypertensive disorders of pregnancy and dementia risk, have identified a moderate association with all-cause dementia and up to three times increased risk of vascular dementia.24-28 However, to our knowledge, few studies have investigated associations between dementia risk and other adverse pregnancy outcomes, such as gestational diabetes, preterm birth, fetal growth restriction, placental abruption, and stillbirth.29-33 Additionally, major differences exist in internal and external validity among individual epidemiological studies, include variations in study designs, exposure and outcome measurements, sample size, follow-up, and confounder adjustments. For example, prospective cohorts provide stronger temporal evidence but often have loss to follow-up. Conversely, retrospective studies allow for larger sample sizes but can introduce selection bias. Adverse pregnancy outcomes are defined differently across studies, with some using clinical diagnoses29 and others relying on self-reported data.34-37 Confounder adjustments vary across studies, as do sample size and duration of follow-up. Country-level differences in health care might also contribute to differential recording of exposures and outcomes. These important differences have probably contributed to inconsistent study results. Furthermore, although an association between a history of adverse pregnancy outcomes and risk of Alzheimer’s disease in later life has been hypothesised, few epidemiological studies have investigated this association.29,30,36,38-41 Therefore, we undertook a systematic review and meta-analysis of all previously published studies evaluating the association between a comprehensive range of adverse pregnancy outcomes and cognitive impairment or dementia of any cause in later life (ie, more than 6 months after pregnancy), to synthesise the available evidence, help resolve previous inconsistencies, and identify areas in need of further study. Methods Search strategy and selection criteria In this systematic review and meta-analysis, we constructed a comprehensive search strategy with the assistance of a professional research librarian and systematically searched PubMed (MEDLINE), Web of Science, and Embase using Boolean logic and Medical Subject Heading terms. Complete search strings are in the appendix (pp 2-5). We searched for publications between database inception (earliest 1974) and July 18, 2024, with no language restrictions. Database searches were supplemented by manual review of the reference lists of included studies. Eligible studies were required to be case–control, cohort, registry, or randomised trials that reported mild cognitive impairment or dementia as outcomes in female individuals or women of any age who had previously had an adverse pregnancy outcome during a pregnancy that lasted more than 20 weeks. Any adverse pregnancy outcome was defined as a history of any one or more of the following: gestational hypertension, pre-eclampsia (including haemolysis, elevated liver enzymes, low platelets [HELLP] syndrome) or eclampsia (including chronic hypertension with superimposed pre-eclampsia), gestational diabetes, stillbirth (also known as intrauterine fetal demise), delivery of a small-for-gestational-age infant, fetal growth restriction, preterm birth (including medically indicated and spontaneous), and placental abruption. Studies were excluded if they focused on men, nulliparous women, or women with an underlying diagnosis of intellectual disability or pre-existing dementia, a history of stroke, or other conditions associated with cognitive impairment diagnosed before the adverse pregnancy outcome. Studies were also excluded if they focused on the development of dementia or cognitive impairment within 6 months of the index pregnancy, or if a more recent publication contained the same population and outcomes in an updated analysis. If the study met inclusion criteria for the systematic review, but did not have sufficient data for meta-analysis (eg, hazard ratios [HRs] for outcome estimations) or did not report a summary statistic, it was included in the systematic review but excluded from the meta-analysis. Using Covidence systematic review software (Veritas Health Innovation, Melbourne, VIC, Australia), duplicates were removed and two investigators (PC and MA) independently screened titles and abstracts. Studies identified for potential inclusion underwent full-text review verification independently by two investigators (PC and MA) to assess overall eligibility on the basis of the inclusion and exclusion criteria. If either reviewer judged a study as potentially relevant, a third reviewer (NAB or ECM) assessed the study, and disagreements were resolved by discussion and group consensus. First author name, study cohort location, year of publication and years of the study, study design, sample size, and summary estimates of descriptive characteristics of participants including age, race or ethnicity (or both if available), education level, parity, and maternal birth year, were extracted from each eligible study; we did not contact study authors if data were missing. For each included study, two investigators (PC and MA) manually extracted the summary estimates (hazard ratios [HR] and 95% CIs). Both unadjusted and adjusted point estimates were extracted where available; when multiple adjusted models were available, the most fully adjusted model estimate was extracted. In cases of disagreement, a third investigator (NAB) confirmed the estimates. This study did not require ethical approval. This systematic literature review and meta-analysis was prospectively registered with PROSPERO, CRD42023453511, and we report the results according to the PRISMA guidelines.42 Data analysis The primary outcomes of our study were mild cognitive impairment, all-cause dementia, Alzheimer’s disease, and vascular dementia. We defined mild cognitive impairment as meeting all of the following criteria: subjective cognitive complaint, objective cognitive impairment, preserved general cognitive function, and maintained activities of daily living; consistent with both the diagnostic criteria proposed by Petersen et al and National Institute on Aging-Alzheimer’s Association criteria.43,44 We included studies where the diagnosis of dementia met the criteria defined by the ICD (any applicable edition during the study period) or other valid diagnostic standards such as the Diagnostic and Statistical Manual of Mental Disorders, 5th edition. Due to multiple exposures and outcomes, we created several separate models to meta-analyse the associations between adverse pregnancy outcomes and the outcomes of interest. The primary models examined all adverse pregnancy outcomes and all-cause dementia, Alzheimer’s disease, and vascular dementia. Separate models examined associations between individual adverse pregnancy outcomes and all-cause dementia, Alzheimer’s disease, and vascular dementia, when possible. When multiple effect sizes were presented and a composite effect size was unavailable, we aggregated the effect sizes of the individual studies into one.45 Aggregation was only required for three studies,29,40,46 because all other studies provided a composite HR. We used a log transformation of the HRs and their corresponding 95% CI in the construction of the pooled estimates and forest plots, which were then back-transformed. The presence of heterogeneity was measured using the Q-test and the extent of heterogeneity was measured with the I2 statistic. Models used random effects and inverse-variance weighting to produce pooled unadjusted and adjusted HRs and 95% CIs, with overall p values calculated using the Wald test. We used a funnel plot and Egger’s test with the standard weighted regression model to assess for possible publication bias on original, non-aggregated HRs. Two investigators (PC and MA) also independently assessed the quality of the included observational cohort and cross-sectional studies using the National Heart, Lung, and Blood Institute’s Quality Assessment Tool.47 This tool evaluates studies on the basis of 14 criteria that examine key methodological aspects, such as the clarity of the research objectives, definition of the study population, measurement of exposures and outcomes, management of confounding variables, and adequacy of statistical analyses. Each study was systematically reviewed against these criteria to ensure a comprehensive assessment of methodological rigour and potential risk of bias (appendix pp 18-32). We did all data analyses in R (version 2024.04.2+764) using the metafor package. A two-sided p value of less than 0·05 was considered the threshold for statistical significance. Role of the funding source The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Results Of the 11 251 publications identified through our search, 15 studies (including 7 347 202 total individual participants) met inclusion criteria for the systematic review, and 11 studies (6 263 431 participants) had sufficient data for meta-analysis (figure 1). Studies that underwent full-text review and were excluded are listed, together with the reasons for exclusion, in the appendix (pp 6-12). Individual study characteristics for the studies included in the systematic review, including number of participants, are listed in table 1. All included studies were published between 2015 and 2024, and comprised one case–control study,34 12 cohort studies,29-31,35,36,38-40,46,48-50 and two Mendelian randomisation studies.32,33 Five studies were performed in US populations; one study was done in cohorts from Australia, the Netherlands, Sweden, and the UK; and nine studies were done in European countries (Denmark, Finland, France, the Netherlands, Sweden, and the UK). Additional details of the included studies, such as method of exposure and outcome ascertainment, length of follow-up, and adjustment factors, are listed in the appendix (pp 13-17). Effect estimates for all exposures by cognitive outcome assessed are shown in table 2, including hypertensive disorders of pregnancy individually and collectively (11 studies),29,33-36,38-40,46,48,49 pre-eclampsia when reported separately from other hypertensive disorders (seven studies),29,34,38-40,46,49 preterm birth (one study),29 stillbirth (three studies),29,30,50 placental abruption (one study),29 fetal growth restriction (one study),29 and gestational diabetes (three studies).29,31,32 Outcomes reported included all-cause dementia (12 studies),29-33,35,36,38,39,46,49,50 Alzheimer’s disease (nine studies),29,30,32-34,36,38-40 vascular dementia (five studies),29,30,33,38,39 other or unspecified dementia (four studies),29,30,38,39 and mild cognitive impairment (one study).48 Study quality was rated fair to good, with good agreement between raters (appendix pp 18-32).47 Risk of publication bias was low (appendix p 33). Hypertensive disorders of pregnancy (ie, chronic hypertension, gestational hypertension, pre-eclampsia, eclampsia, and HELLP syndrome) showed the most consistent increase in risk across studies and outcomes, with effect sizes ranging from a non-significant increase in risk to a four-fold higher risk of Alzheimer’s disease in those with a history of pre-eclampsia in one US study (table 2).40 Elevated risk of vascular dementia was seen for preterm birth in one study (adjusted HR 1·71 [95% CI 1·16–2·52]).29 No other significant effects were noted for individual adverse pregnancy outcomes for any of the outcomes, although gestational diabetes,29 fetal growth restriction,29 and stillbirth30 all had non-significant increases in vascular dementia risk (table 2). Only one cohort study,48 including 40 women with a history of pre-eclampsia and 40 without, included mild cognitive impairment as an outcome. Although this study showed that more women in the exposed group had a diagnosis of mild cognitive impairment or dementia (diagnosed by expert consensus), the effects were not significant.48 Two Mendelian randomisation studies32,33 evaluated the effects of gestational diabetes and hypertensive disorders of pregnancy on the long-term incidence of dementia, respectively. The studies found no significant causal association between gestational diabetes or hypertensive disorders of pregnancy and the risk of developing dementia.32,33 11 studies (including 6 263 431 participants) reported sufficient data to be included in the meta-analysis.29-31,35,36,38-40,46,49,50 Due to only one study reporting mild cognitive impairment as an outcome, this outcome could not be meta-analysed. Five individual pregnancy exposures (gestational diabetes, stillbirth, history of any hypertensive disorder of pregnancy, gestational hypertension, and pre-eclampsia or eclampsia) had data from multiple studies, permitting meta-analysis.29-31,35,36,38-40,49,50 Results of the meta-analysis showed that a history of any adverse pregnancy outcome was associated with a higher risk of all-cause dementia (unadjusted HR 1·54 [95% CI 1·29–1·83], I2=85%; adjusted HR 1·32 [95% CI 1·17–1·49], I2=80%; figure 2A). Any adverse pregnancy outcome history was also associated with an increased risk of Alzheimer’s disease (unadjusted HR 1·87 [95% 1·00–3·49], I2=83%; adjusted HR 1·26 [95% CI 1·04–1·53], I2=63%; figure 2B), and vascular dementia (adjusted HR 1·94, [95% CI 1·70–2·21], I2=0%; figure 2C). Stillbirth was not significantly associated with higher risk of all-cause dementia (adjusted HR 1·26 [95% CI 0·93–1·71], I2=62%; figure 3). Similarly, gestational diabetes was not significantly associated with higher risk of all-cause dementia (adjusted HR 1·14 [95% CI 0·54–2·41], I2=79%; appendix pp 37, 41). A history of any hypertensive disorder of pregnancy was associated with all-cause (adjusted HR 1·32 [95% CI 1·11–1·57], I2=74%) and vascular dementia (1·78 [1·46–2·17], I2=0%), but was not significant for Alzheimer’s disease (1·24 [0·98–1·57], I2=66%; figure 4A-C). When gestational hypertension and pre-eclampsia or eclampsia were analysed separately, effect sizes were similar to when assessed together as any hypertensive disorders, although the association between gestational hypertension and all-cause dementia was not significant, and neither preeclampsia nor gestational hypertension showed significant associations with Alzheimer’s disease(appendix pp 38-40). Where available, forest plots for unadjusted estimates are shown in the appendix (pp 34-37). Discussion We found that a history of any adverse pregnancy outcome was associated with a moderately increased risk of later-life all-cause dementia and Alzheimer’s disease, and nearly double the risk of vascular dementia. For individual adverse pregnancy outcomes, we found significant associations between hypertensive disorders of pregnancy and all-cause and vascular dementia, but not Alzheimer’s disease, and no association between stillbirth and all-cause dementia. Our results suggest a consistent association between a history of adverse pregnancy outcomes and risk of dementia. These results align with those of previous meta-analyses of the association between hypertensive disorders of pregnancy and cognitive impairment or risk of dementia.24-28 The association of adverse pregnancy outcomes other than hypertensive disorders of pregnancy with risk of dementia is less well established, although a consistent association of non-hypertensive adverse pregnancy outcomes with future cerebrovascular disease risk has been demonstrated in multiple studies.37,51-53 In our systematic review, some individual studies found that preterm birth, gestational diabetes, stillbirth, and fetal growth restriction showed the same direction of effect towards higher risk of either all-cause (gestational diabetes, stillbirth) or vascular dementia (preterm birth, gestational diabetes, fetal growth restriction). However, in our meta-analysis we did not find a significant association between stillbirth or gestational diabetes and all-cause dementia, and data were insufficient to meta-analyse fetal growth restriction or preterm birth as individual exposures. Notably, we found a pronounced inconsistency between the Mendelian randomisation studies32,33 and the observational studies.29,31,36,38-40,46,49 Although the non-significant results from the Mendelian randomisation studies should be confirmed with future research, several limitations within these two types of studies could explain the discrepancy. In Mendelian randomisation studies, pleiotropy of genetic variants and their association with protective factors for dementia, lack of accounting for gene-environment interactions, and a weak association between genetic factors and the exposures might push the results toward the null. However, a higher likelihood of measurement bias and uncontrolled confounding factors in observational studies than in Mendelian randomisation studies might lead to overestimation of effects. The connection between a wide range of adverse pregnancy outcomes and vascular dementia is consistent with multiple prospective and retrospective studies1 that found increased incidence of cardiovascular risk factors and disease in individuals who had adverse pregnancy outcomes of multiple types. Although a causal pathway has not been identified, adverse pregnancy outcomes that are mediated by the placenta share many common pathophysiological features with cardiovascular disease, in particular hypertension, including angiogenic imbalance, vascular endothelial dysfunction, and systemic inflammation. Imbalance of angiogenic factors (ie, soluble fms-like tyrosine kinase-1 [known as sFlt-1] and placental growth factor [PlGF]) during pregnancy is a hallmark of placental dysfunction that has a central role in the pathophysiology of adverse pregnancy outcomes mediated by the placenta, including hypertensive disorders of pregnancy,54 fetal growth restriction, placental abruption, and stillbirth.55 Angiogenic imbalance and endothelial dysfunction are similarly implicated in the pathophysiology of vascular cognitive impairment and dementia,56 and elevated PlGF is now considered a biomarker for vascular cognitive impairment in later life.57 Additionally, shared risk factors for both adverse pregnancy outcomes and vascular dementia, such as pre-pregnancy obesity, chronic hypertension, and smoking, might partly explain the association between adverse pregnancy outcomes and vascular dementia, although the studies included in our meta-analysis adjusted for these and other potential confounding factors. The pathways that might link adverse pregnancy outcomes with increased risk of Alzheimer’s disease are unclear. Alzheimer’s disease is the most common cause of dementia,58 and vascular risk factors are hypothesised to have a role in its development and progression, and so the association we identified might be partly driven by underlying vascular contributions to Alzheimer’s pathophysiology or clinical presentation. Individuals diagnosed clinically with Alzheimer’s disease typically have evidence of cerebrovascular disease as well,59 and so the associations with Alzheimer’s disease that we observed might reflect the additive effects of vascular disease on the development of dementia in individuals with Alzheimer’s disease. The term vascular dementia encompasses a broad array of pathologies and is used to indicate all forms of dementia associated with cerebrovascular diseases, including single or multiple cerebral infarcts and more broad cerebrovascular processes, such as hypertensive cerebral small vessel disease or cerebral amyloid angiopathy.2,60 However, retrospective studies might have relied predominantly on diagnoses that were based on older concepts of vascular or so-called multi-infarct dementia, which required a history of or radiographic evidence of stroke.61 Thus, the more robust associations between adverse pregnancy outcomes and vascular dementia seen in retrospective studies might be skewed towards individuals with clinical stroke, which has been demonstrated to occur earlier in life in those with a history of adverse pregnancy outcomes.62 However, adverse pregnancy outcomes might also share non-vascular pathways with Alzheimer’s disease. For example, genetic and epigenetic changes in the STOX1 gene are associated with both pre-eclampsia and late onset Alzheimer’s disease,63 and dysregulation of prototype proteolytic enzymes involved in amyloid precursor protein processing has been shown to be involved in the pathophysiology of pre-eclampsia.64 Chronic inflammation might provide another mechanistic link between adverse pregnancy outcomes and Alzheimer’s disease.65-67 Furthermore, as well as being a robust marker of vascular cognitive impairment, PlGF is positively correlated with Alzheimer’s pathology, including higher Braak neurofibrillary tangle stage and increased Alzheimer’s type neuropathological change.68 Additionally, levels of a variety of maternal hormones during pregnancy can vary considerably and influence pregnancy outcomes;69-71 because of the known association between sex hormones and cognitive outcomes,72,73 shared hormonal pathways could be implicated in the association between adverse pregnancy outcomes and cognitive outcomes in later life. Our study has limitations. We found significant heterogeneity among most included studies. Published data for adverse pregnancy outcomes other than hypertensive disorders of pregnancy were scarce, precluding meta-analysis of many of these exposures. Additionally, given the known strong associations between adverse pregnancy outcomes and cardiovascular death,74 survivor bias might have attenuated the observed associations between adverse pregnancy outcomes and dementia, which typically occurs later in life; hence, our results might have been biased towards the null. Although three studies used competing risk models,30,38,39 they did not consider death a competing risk; however, a few of the included studies did take competing risks for dementia subtypes into account.30,38,39 As with all meta-analyses, publication bias could have skewed our results, although our funnel plot did not show evidence of this. All studies were done in US or European populations, potentially limiting the generalisability of our results to other settings, in particular low-income and middle-income countries. Lack of precision in the definitions of mild cognitive impairment made most studies that included cognitive impairment as an outcome unsuitable for inclusion in this systematic review. Adverse pregnancy outcomes are a sex-specific exposure, which pregnancy-capable people of any gender might experience; however, gender and sex were not reported separately in any of the studies included, so the potential modifying effects of gender could not be assessed. In addition, diagnostic criteria for various adverse pregnancy outcomes such as gestational hypertension and pre-eclampsia have changed over time. Because a broad array of adverse pregnancy outcomes were included in our primary exposure of any adverse pregnancy outcome, the nuances of changes in definitions are unlikely to have substantially influenced our main results, but they might have affected the results of analyses for individual pregnancy outcomes. Furthermore, many of the studies included in our systematic review used registry data for ascertainment of exposures and outcomes, rather than prospective adjudication of dementia diagnoses, which might have led to misclassification of dementia subtypes or underdiagnosis. Reliability of exposure and outcome ascertainment probably varied considerably depending on the source. Some studies31,34-36,50 used personal recall for exposure ascertainment, which has been shown to have high specificity but poor sensitivity,75-77 while others used more reliable manual review of medical records.32,39,40,48,49 The quality of registry-based or ICD data for pregnancy exposures also varies across countries and years. Dementia diagnoses from registries or ICD codes might also have high positive predictive value but lower sensitivity.78,79 These limitations would be expected to bias our results to the null. Finally, because two studies did not have composite adverse pregnancy outcomes available for the main analysis,29,46 we aggregated the available HRs, which might have obscured individual exposure effects. Although we set a higher correlation coefficient during aggregation, which helps account for the independence of the aggregated exposures, this might have decreased the precision of our effect estimates. In this systematic review and meta-analysis, we synthesised the growing body of literature supporting a link between adverse pregnancy outcomes and cognitive decline and dementia and illustrate the many remaining knowledge gaps and areas for future investigation. Most longitudinal ageing studies, although rigorous in their cognitive assessments and dementia diagnoses, do not include reproductive histories and comprehensive reporting of pregnancy complications. To rectify this oversight, ongoing longitudinal cohort studies of cognitive ageing should include a reproductive history among the multitude of variables collected, and researchers conducting longitudinal studies focused on pregnancy should consider partnership with ageing experts as their cohorts age. Ideally, common data elements for obstetric history should be used to allow for consistency across studies and reduce heterogeneity. In conclusion, in our comprehensive systematic review and meta-analysis, which included a broad variety of pregnancy exposures in addition to hypertensive disorders of pregnancy, we found that individuals with a history of any adverse pregnancy outcome had moderately increased risk of all-cause dementia and Alzheimer’s disease, and approximately double the risk of vascular dementia. Given the increasing number of effective therapies for early-stage dementia,80 these results suggest a need for early cognitive screening in those with a history of these pregnancy exposures to identify cognitive impairment and dementia in the early and potentially treatable stages. Supplementary Material