Risk factors for pre‐eclampsia in clinical practice guidelines: Comparison with the evidence

To compare pre‐eclampsia risk factors identified by clinical practice guidelines (CPGs) with risk factors from hierarchical evidence review, to guide pre‐eclampsia prevention.


| I N TRODUC TION
Pre-eclampsia complicates 2%-4% of pregnancies worldwide and its incidence is rising, given the current trends in advanced maternal age of pregnancies and rising body mass. 1 Pre-eclampsia is the hypertensive disorder of pregnancy (HDP) associated with the greatest risk of maternal and fetal morbidity and mortality.As such, a large part of prenatal care is devoted to the detection of preeclampsia, through blood pressure (BP) and proteinuria screening. 2However, as there is currently no approved disease-modifying treatment for pre-eclampsia, current best practice remains the identification of women at risk, the use of preventative therapy, 3 the management of hypertension and other organ manifestations should pre-eclampsia develop and, ultimately, timed birth, as the only intervention that initiates the resolution of this syndrome.
There is international consensus that screening for preeclampsia risk should occur in early pregnancy, to evaluate whether there is an indication for evidence-based preventative measures (e.g.aspirin). 4Whilst adding biochemical markers and ultrasonographic factors to clinical risk factors can double the identification of women who will develop pre-eclampsia before 37 weeks of gestation (i.e.preterm preeclampsia), 5 clinical risk factors remain important for preeclampsia prediction, including risk factors that develop later in pregnancy and mandate enhanced surveillance and timed birth.
Clinical practice guidelines (CPGs) are intended to advise clinicians on high-quality, evidence-based practice.We previously conducted a systematic review of international CPGs for the HDPs, assessing and comparing the quality of CPGs and their recommendations. 6Although almost all current CPGs for pregnancy hypertension list risk factors for pre-eclampsia, the quality of the documents vary, as do the screening recommendations. 6This variability can be difficult to understand, given the limited referencing permissible when guidelines are published in peer-reviewed journals.
As part of the development of a framework of preeclampsia risk factors, 7 we undertook an evidence review of the determinants of pre-eclampsia (Elawad T. A conceptual framework for the determinants of pre-eclampsia.A dissertation submitted in partial fulfilment of the requirements for the degree at the King's College London, Department of Women and Children's Health, Faculty of Life Sciences and Medicine).In this analysis, we sought to compare the risk factors for pre-eclampsia identified in CPGs, and the underlying evidence base.

| Systematic review of CPGs
In a previous systematic review, 17 CPGs were identified for guidance on the diagnosis, evaluation and management of HDPs. 6Full details of our methodology have already been published. 6n brief, we searched online databases (MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Health Technology Assessments, the Database of Abstracts of Reviews of Effects and grey literature) using appropriate keywords and Medical Subject Headings (MeSH), from January 2009 to October 2019, to identify CPGs meeting our eligibility criteria. 6A CPG was defined as an evidence-based document that offered structured advice for healthcare professionals, referenced primary literature and was issued by a professional medical society, government body or similar organisation.Included were CPGs in English, French, Dutch or German that covered the diagnosis, assessment and management of at least one HDP, or were explicit updates to the CPGs described by Gillon et al. 8 Excluded were publications that were adapted only from existing CPGs or were local or regional in scope when there was a relevant national document.
The quality of CPGs was assessed by three independent reviewers (GS, LAM and PvD) using the Appraisal of Guidelines for Research & Evaluation Instrument II (AGREE II) tool, 9 and disagreements were resolved through consensus.AGREE-II has six domains, including rigour of development, the domain that best represents the standard of literature search and the overall quality of evidence used in guideline development.For the 15 CPGs deemed to be clinically useful after AGREE-II assessment, structured tables were used to abstract pre-eclampsia risk factors from recommendations, tables, bullet points or text. 8Summary information about risk factors designated by CPGs as 'major' or 'moderate' have previously been reported; here, this information is presented by risk factor and CPG, along with details of the other risk factors listed and types of sources cited, according to in-text citation.

| Evidence review for pre-eclampsia risk factors
We used the methods of Hiatt et al. to develop a comprehensive model for the determinants of pre-eclampsia. 10 A broad group of experts in pre-eclampsia was assembled from the Epidemiology Working Group of the PREgnancy Care Integrating translational Science, Everywhere (PRECISE) Network. 7A working model of determinants of preeclampsia was expanded from variables found to have significant associations with pre-eclampsia by pooled results in umbrella reviews (i.e.systematic reviews of systematic reviews). 11,12

| Literature search
The search strategy was developed in consultation with a clinical librarian (HE) at the British Medical Association.In brief, Medline (Ovid) was searched from January 2010 to January 2021, using keywords covering all potential determinants of pre-eclampsia.The highest level of evidence supporting a relationship between a risk factor and preeclampsia was identified in a hierarchical fashion.Umbrella reviews were sought that focused on pre-eclampsia, and only if none were identified were the keywords broadened to identify any studies in pregnancy.If no relevant umbrella reviews were identified, then the process was repeated to identify relevant systematic reviews.If no systematic reviews were identified or identified for all risk factors of interest, then large observational studies (including secondary analyses of trials) were sought, searching individually for relevant risk factors.Observational studies with at least 1000 participants were targeted, as described by Bartsch et al., 13 to be more representative of the general population and to have sufficient statistical power to assess less prevalent, but potentially important, risk factors. 14Smaller observational studies, case reports or series, qualitative reviews and editorials were excluded.(For details, including keywords, see Table S3.)

| Data extraction
Titles and abstracts of articles were screened to assess eligibility.Potentially eligible studies underwent full-text review.Data abstracted were general study characteristics, strength of association between each risk factor and pre-eclampsia (estimated as relative risk (RR), odds ratio (OR) or diagnostic OR (DOR), and reported, adjusted where possible, or calculated from the prevalence of pre-eclampsia among women with and without the risk factor), and the characteristics necessary to assess study quality.Subcategories of a potential risk factor were also considered, such as a body mass index (BMI) categorised as overweight or obese.
As described by Hiatt et al., 10 the strength of association between risk factors and the outcome of interest (preeclampsia) was evaluated as definite, probable, possible or not significant. 15The evaluation was based on point estimates, extracted as reported or calculated from primary data using previously published cut-offs (Table 1). 10,16If a study reported outcomes as proportions, a RR was calculated as a simple ratio between those with the risk factor of interest and those without.Results of the I 2 statistic were also extracted (or calculated from the Q statistic) to reflect heterogeneity.RR and OR were used interchangeably for the model, as pre-eclampsia occurs in <10% of the unexposed population, making the OR a reasonable approximation of the RR. 17 Recommendations prepared by Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) were used to assess the quality of the evidence, as high, moderate, low or very low.A crossdisciplinary team (M-LV, KP, TE, CEL, MW-K, MV, JF, RS, HDM) adapted GRADE criteria through consensus into a standardised process for this pregnancy project, to minimise discrepancies between reviewers. 18,19Table 1 shows that as a starting point, umbrella or systematic reviews were considered to be of high quality and observational studies were considered to be of low quality. 20However, the final quality rating for each methodology could be modified based on additional characteristics: decreased, based on study limitations (risk of bias), important inconsistency, indirectness, imprecise data or publication bias; or increased, based on large effect size or dose-response gradient.One reviewer (TE) assessed the quality of the evidence using these GRADE criteria, and any uncertainty was resolved by discussion and consensus reached with a second and third reviewer (CEL and RS).

| Comparison of CPG risk factors with the literature
A descriptive comparison of pre-eclampsia risk factors was made between those identified in CPGs and those identified from the literature search.The strength of association with pre-eclampsia and the quality of the underlying evidence were assigned and compared with the CPG, with an overall designation of risk factors as 'major' or 'moderate'.Risk factors are presented according to traditional history taking, as demographics and social determinants of health, past history, family history and current pregnancy.

| CPGs
The 15 CPGs previously identified by AGREE-II as 'clinically useful' were included in this analysis,  as described in the prior systematic review (Table S2). 6 In rief, most CPGs (n = 13) were national in scope and produced by professional societies.,[35][36][37][38][39]42,44 All but the Brazilian guideline (i.e.14/15 CPGs) listed risk factors for pre-eclampsia.6 Just over half of the CPGs (8/14) stratified risk factors into levels of importance.When listed as 'major'/'high' and 'moderate' risk factors (n = 6; NED, IRL, European Society of Cardiology (ESC), 26 American College of Obstetricians and Gynecologists (USA), [30][31][32]  Inconsistency was defined as variation between studies (heterogeneity), indirectness whether the paper answered the question we aimed to answer; imprecision defined according to the confidence interval of the summary estimates, publication bias, as a tendency towards publication of studies that showed positive results, and magnitude of effect, as determined by the RR.

T A B L E 1 Strength of association between risk factors and pre-eclampsia based on point estimates of various summary measures
National Institute for Health and Care Excellence (UK), Polish Society of Hypertension (POL)), aspirin was recommended for one 'major' risk factor or at least two 'moderate' factors.Other CPGs presented lists of risk factors to identify 'increased risk'; sometimes highlighting factors associated with a particularly high risk, designated here as 'major' (n = 2; Society of Obstetricians and Gynaecologists of Canada (CAN), 28,29 Ministry of Health, New Zealand (NZL)), or otherwise presenting a list with no associated strength of association (n = 6; World Health Organization (WHO), Society of Obstetric Medicine of Australia and New Zealand (SOMANZ), 43 French Society of Hypertension (FRA), 34 La Société Tunisienne de Gynécologie Obstétrique (TUN), International Society for the Study of Hypertension in Pregnancy (ISSHP), 27 and German Society of Gynecology and Obstetrics (DEU) 40 ).
The CPGs varied with regards to the provision of in-text citations for risk factors.Three CPGs cited no such supporting literature (WHO, IRL and ESC), which when provided, was not necessarily linked with the risk factors cited.

| Evidence
Eighty pre-eclampsia risk factors were identified.Two, proposed by one CPG each, were not considered further because they were considered both vague and covered by individual conditions already included as risk factors: any 'prior adverse pregnancy outcome' and any 'placental insufficiency in obstetric history'.
Table 2 presents the 78 risk factors for pre-eclampsia, according to their strength of association and quality of evidence, and whether they are generally evident in early pregnancy (n = 60 white table cells and n = 4 footnoted for a lack of evidence), or become evident only as pregnancy progresses (n = 8, blue table cells, and n = 6 footnoted for a lack of evidence), recognising that there are some additional factors that could be both, such as anxiety or anaemia.First, ten 'major' and 11 'moderate' risk factors were designated by the CPGs, two of which were both 'major' and 'moderate' risk factors (i.e.multiple pregnancy and ART), and all of which can be identified in early pregnancy.Second, the strength of association and quality of evidence for risk factors were not closely aligned.For risk factors designated as 'major' by CPGs (in bold), associations ranged from definite to possible and the quality of evidence ranged from moderate to very low.For risk factors designated as 'moderate' by the CPGs, (in italics), associations ranged from definite to none and quality of evidence ranged from high to very low.
Table 3 shows that the 78 risk factors evaluated were derived from demographics and social determinants of health (n = 8), past medical (n = 27), obstetric (n = 10) and family (n = 5) histories, and conditions arising early or later during the current pregnancy (n = 28).The strength of association and quality of evidence are presented along with the CPGs that endorsed them.

| Definite associations
There were eight risk factors with definite associations with pre-eclampsia (shown in dark green, Table 3): demographics (adolescence); past medical history (obesity, chronic hypertension, pre-gestational diabetes mellitus (DM), considered as type-1 and type-2 DM separately, and severe anaemia); past obstetric history (prior pre-eclampsia); and current pregnancy (fetal trisomy 13).
Obesity (i.e.BMI ≥ 30 kg/m 2 ) was the only risk factor with a 'definite' association with pre-eclampsia based on highquality evidence (n = 14 CPGs).No CPG, even those that highlighted only a subgroup with BMI ≥ 35 mg/kg 2 (NED, IRL, TUN, NZL, ESC, UK, POL), endorsed obesity as a 'major' risk factor, whereas 6/14 regarded it as a 'moderate' risk factor.

| Probable associations
The majority of associations (n = 39) with pre-eclampsia were probable (shown in medium green, Table 3).
High-quality evidence supported three risk factors.Overweight (i.e.BMI = 25.0-29.9kg/m 2 ) and stage-1 hypertension (defined as systolic BP 130-139 mmHg and/ or diastolic BP 80-89 mmHg at booking or <20 weeks of gestation) 47 were endorsed by few CPGs (i.e., n = 2 and 3,  respectively), and none were designated as 'major' or 'moderate' risk factors.No CPGs endorsed prehypertension at booking as a risk factor.Moderate-quality evidence supported six risk factors.Antiphospholipid antibody syndrome (APAS, n = 12 CPGs, 8/12 'major') was highly endorsed and family history of pre-eclampsia in mother or sister (n = 5, 1/5 'major' and 3/5 'moderate') was also endorsed by the CPGs.Other risk factors were endorsed by one CPG each (i.e.obstructive sleep apnoea, smoking and any infection in the index pregnancy).No CPG endorsed prior stillbirth.
Very low-quality evidence supported five risk factors, including the well-endorsed ART (n = 7 CPGs, 1/7 'major' and 1/7 'moderate'); oocyte donation, specified in 3/7 of the CPGs that specified ART, was listed as both a 'major' and 'moderate' risk factor in different guidelines.

| Possible associations
There were 13 possible associations with pre-eclampsia (shown in very light green, Table 3).Only evidence of moderate quality supported urinary tract infection in the index pregnancy (n = 1 CPG).Low-quality evidence supported six risk factors, including 'prior HDP', endorsed by n = 4 CPGs, with all listing this as a 'major' risk factor.Very low-quality evidence supported six risk factors, including an interpregnancy interval of ≥10 years, which was endorsed by many CPGs (n = 9) and frequently listed as a 'moderate' risk factor (in 6/9).

| Not significant
According to our methodology, no association could be demonstrated for eight risk factors, all based on very low-quality evidence (Table 3).Three were endorsed by a single CPG: prior small-for-gestational-age (SGA) infant (as 'moderate'), fetal trisomy 18 and vaginal bleeding in early pregnancy.
A summary of risk factors with a demonstrated association with pre-eclampsia are presented in Table 4.

| Summary of findings
The CPG-recommended pre-eclampsia risk factors are not well aligned with the published evidence.'Major' risk factors usually have definite to probable associations with preeclampsia, based on moderate-to very low-quality evidence, with two exceptions.'Prior HDP' has a possible association, based on low-quality evidence.'Autoimmune disease' has no supporting evidence but includes conditions for which there is low-quality evidence (e.g.rheumatoid arthritis).'Moderate' risk factors in general have weaker relationships with pre-eclampsia, based on lower quality evidence, but maternal obesity is a notable exception.
Indeed, obesity is the strongest evidence-informed preeclampsia risk factor, having a definite association with preeclampsia, based on high-quality evidence.Also, there are other evidence-informed risk factors that are listed as neither 'major' nor 'moderate' in the guidelines, particularly maternal overweight and stage-1 hypertension or prehypertension at booking, based on high-quality evidence.
A number of pre-eclampsia risk factors are of particular relevance to low-and middle-income countries (LMICs).Some factors have definite (i.e.adolescence or severe anaemia) or probable (i.e.sickle cell disease or anaemia) associations with pre-eclampsia, yet only adolescence is listed, and then only by the WHO.Although no association with preeclamspia is demonstrable for other risk factors (i.e.HIV, tuberculosis and malaria), the quality of evidence is very low.
The CPGs focus on pre-eclampsia risk factors identified in early pregnancy to guide low-dose aspirin therapy.However, there are additional, well-supported risk factors that become evident as pregnancy progresses and influence investigations, maternal-fetal surveillance and/or timed birth.Examples include common conditions in pregnancy, like anaemia (particularly severe anaemia), infections, gestational weight gain and gestational DM.

| Comparison with current literature
To our knowledge, this is the first evidence-informed comparison of pre-eclampsia risk factors with those endorsed by CPGs.Deserving of specific mention is the only 'possible' association between pre-eclampsia and 'prior HDP'; this risk factor was cited as a 'major' risk factor by four CPGs, whereas the others cited 'prior pre-eclampsia' as a major risk factor, and for that, there is a definite relationship.
Although we demonstrated a lack of close alignment between the guideline risk factors and the evidence, it was not usually possible to understand why this was the case.Guidelines usually cite one reference in support of all the risk factors with the relative recognised by 'major' or 'moderate' designations, without further citations.Very few CPGs included a broad array of higher-order evidence, such as systematic reviews and large observational studies, as in our analysis; the most highly cited systematic review was over 15 years old. 45No CPG cited umbrella reviews that could have been incorporated into the 2019 guidelines. 11,12It is common for CPGs to cite other guidelines, often with little or no citation of primary evidence for risk factors, even when the CPGs had high scores on rigour of development.All of this contributes to the sense that although there has been much focus on quality rating scales for guidelines, further improvement is necessary before CPGs will effectively translate evidence into practice in the field of pregnancy hypertension.Pre-eclampsia risk assessment, through counting 'major' or 'moderate' risk factors, detects fewer cases of preterm pre-eclampsia than a multivariable approach. 5,88Also, the most important risk factors identified by the CPGs are not aligned with the published prediction models, 89 which most commonly identify the following risk factors as important: BMI (19/40 models), uterine artery pulsatility index (17/40), angiogenic markers (16/40 for both placental growth factor (PlGF) and pregnancy-associated plasma protein A (PAPP-A)), ethnicity (14/40) and BP (12/40).The absence of angiogenic imbalance as a risk factor for pre-eclampsia in the CPGs is notable.Also, 'major' CPG risk factors were not as well supported in these models: prior pre-eclampsia (9/40 models), chronic hypertension (2/40), pre-gestational diabetes (0 but 2/40 included fasting blood glucose), CKD (0 although 1/40 included serum creatinine), SLE (0), APAS (0), ART (6/40), multiple pregnancy (0) and prior HDP (0). 89lthough some may regard universal aspirin administration as preferable to a reconsideration of pre-eclampsia risk screening, this is debated.Aspirin compliance is suboptimal, among even women identified as being at high risk, 90 and pregnant women are averse to taking medication in pregnancy, particularly when small risks have been identified. 91Also, the universal administration of aspirin would not address the prevention of term pre-eclampsia or risk factors that require alternative approaches (e.g.exercise for a sedentary lifestyle).
Given that screening for pre-eclampsia risk should be implemented for all pregnant women, a recent systematic review emphasised the importance of the 'population attributable risk', related not only to strength of association and quality of evidence for the risk factor and pre-eclampsia, but also to how commonly the risk factor occurs, and whether its relationship with pre-eclampsia is modifiable. 13For example, addressing a risk factor with a strong association with pre-eclampsia but low population prevalence (e.g.APAS) will have little impact on pre-eclampsia incidence at the population level; this is more likely to be affected by addressing a more common risk factor (e.g.overweight), even if the association with pre-eclampsia is not as strong.

| Strengths and limitations
The strengths of this paper include the comprehensive search strategies to identify CPGs and evidence for individual risk factors, 6 and the use of published methodology to evaluate the strength of association and the quality of the evidence. 10e offer a unique perspective on gaps between practice recommendations and evidence-informed risk factors, even within guidelines rated as being of high quality.We have distinguished between risk factors evident in early pregnancy and those that emerge as pregnancy progresses; this pragmatic and comprehensive approach acknowledges that pre-eclampsia risk may evolve and that the risk of adverse outcomes can be mitigated by close surveillance and Abbreviations: ART, assisted reproductive technologies; BMI, body mass index; BP, blood pressure; dBP, diastolic blood pressure; DM, diabetes mellitus; FGR, fetal growth restriction; GDM, gestational diabetes mellitus; HDP, hypertensive disorder of pregnancy; HIV, human immunodeficiency virus; IVF, in vitro fertilisation; NS, not significant; sBP, systolic blood pressure; SGA, small for gestational age.a Strength of association was assessed according to relative risk and odds ratio criteria in Table 1.
b Quality of evidence was assessed according to GRADE criteria, detailed in Table S3.
c Socio-economic status was based on income.
d Abnormal uterine artery Doppler included bilateral notching, or an increased pulsatility or resistance index persisting beyond 24 weeks gestational age. e The association between barrier contraception and pre-eclampsia was observed among nulliparous women.f This assessment was based on a large observational study (retrospective cohort study) excluded from a systematic review that was restricted to case-control studies and had a far smaller number of women (approximately 500) with methamphetamine exposure. 87timed birth, either to minimise risk of complications pre-eclampsia develops or to prevent pre-eclampsia from developing at term gestational age.
The limitations of our analysis include that international CPGs are almost exclusively from high-income countries, so it is unsurprising that they may not address risk factors of unique or particular importance to LMICs (e.g.malaria or seasonality).Despite following published methodology, 10 we restricted our search to Medline, to focus on a peer-reviewed, curated collection of citations of articles in journals approved and indexed to include MeSH terms.We excluded small observational studies (<1000 participants), in which some risk factors have been identified, from our evidence; the quality of this evidence may be improved by future systematic reviews or large studies.Finally, although we used strength of association criteria for RR and OR interchangeably, the low incidence of pre-eclampsia (2%-4% of pregnancies) means that the use of OR is unlikely to have exaggerated the association.

| CONCLUSION
][13] With the availability of multivariable prediction models in early and later pregnancy, digital health technologies for data processing and an awareness that pre-eclampsia risk may evolve as pregnancy progresses, we are well placed to refresh our strategy to identify, throughout pregnancy, the women at increased risk of pre-eclampsia, and to modify their likelihood of pre-eclampsia and/or pre-eclampsia adverse outcomes, accordingly.

AU T HOR C ON T R I BU T ION S
The project was conceived by LAM, PvD, VF, TE and GS.All authors contributed to the writing and critical review of the article, and approved the final version for publication.The PRECISE Network contributed to the study design and provided constructive challenge as part of the development of a suite of conceptual frameworks for placental disease.were based on low-quality evidence.Factors that are not shaded were based on evidence of very low quality.The following factors endorsed by CPGs are excluded, as there was no rigorous evidence identified to evaluate their association with pre-eclampsia: 'autoimmune disease' as a group, elevated prepregnancy triglycerides, family history of early-onset CVD, gestational hypertension, FGR, fetal triploidy, hyperplacentation (not otherwise specified), fetal hydrops, gestational trophoblastic disease and cocaine use.
Based on evidence of very low quality, the following factors were not supported as being associated with pre-eclampsia: prior SGA infant, vaginal bleeding in early (current) pregnancy, fetal trisomy 18, thalassemia, HIV, tuberculosis, anxiety, malaria (current pregnancy). b Abnormal uterine artery Doppler included bilateral notching, or an increased pulsatility or resistance index persisting beyond 24 weeks gestational age.

T A B L E 2
Matrix of risk factors for pre-eclampsia, according to strength of association and quality of evidence a
Family history in mother or sister Maternal age >40 years Excessive weight gain Race/ethnicity: Sub-Saharan African, South Asian, S A PPROVA L Approval for the PRECISE study was obtained in King's College London (Ref: HR-17/18-7855), Aga Khan University Hosptial (Ref: 2018/REC-740, The Gambia Government/The Medical Research Council, The Gambia Joint committee (Ref: SCC 1619), and the Mozambique Ministry of Health, National Bioethics Committee for Health (545/CNBS/18).

F
U N DI NG I N FOR M AT ION This work was supported by funding from a UK Research and Innovation Global Challenges Research Fund (GCRF) GROW award (MR/P027938/1).
Abbreviations: BMI, body mass index; CVD, cardiovascular disease; DM, diabetes mellitus; HDP, hypertensive disorder of pregnancy; HIV, human immunodeficiency virus; SGA, small for gestational age.a Risk factors in the darkest shading were based on high-quality evidence.Factors in moderate shading were based on moderate-quality evidence.Factors in light shading