Pre-eclampsia is a pregnancy-specific condition associated with hypertension and multiorgan dysfunction such as proteinuria, renal or hepatic impairment and fetal growth restriction.^1–5 It is a heterogeneous disorder with a wide spectrum of multiorgan involvement, which reflects its various pathophysiological pathways. Pre-eclampsia affects between 2% and 8% of pregnancies worldwide⁶ and is a leading cause of both maternal and perinatal morbidity and mortality.^7–10 Each year, 18% of all maternal deaths can be attributed to pre-eclampsia and its complications, with most of these occurring in low- and middle-income countries.^11,12 In the long term, pre-eclampsia is associated with an increased maternal risk of ischaemic heart disease, chronic hypertension, stroke and end-stage renal disease.^13,14 Children from pre-eclamptic pregnancies also have higher risks of cardiovascular diseases,^15,16 mental health disorders and cognitive impairment.^17,18

Two subgroups of pre-eclampsia are well recognised: early-onset, requiring delivery before 34 weeks’ gestation, and late-onset, with delivery occurring at or after 34 weeks’ gestation.^19–21 Early-onset pre-eclampsia is considered to be a pathophysiologically different disease from late-onset pre-eclampsia in the mechanism leading to placental dysfunction and clinical timing during pregnancy.²² Early-onset pre-eclampsia is associated with a considerably higher increased risk of maternal complications, such as a 20-fold higher rate of mortality, than the late-onset type, and early delivery is the only treatment.^23–25 In addition to the prematurity-related complications, the risks of stillbirth and adverse perinatal outcomes are much higher in women with early-onset disease.²⁶

Although the proportion of women with early-onset pre-eclampsia is < 1% of all pregnancies, the complexity of treatment gives rise to high health-care costs.^27,28 Affected women are often admitted to a tertiary care facility, and 30% experience complications that may necessitate management in an intensive care unit.²⁹ Infants usually need prolonged care for the management of complications, including lifelong disabilities, arising as a result of premature delivery. The additional NHS costs incurred in caring for a baby born at or before 28 weeks and a baby born between 28 and 33 weeks are £94,190 and £61,509, respectively.³⁰ The cost to the NHS of caring for preterm babies, linked to neonatal care, such as incubation, and hospital readmissions, has been estimated at £939M annually.³⁰

Late-onset pre-eclampsia, including pre-eclampsia at term, also poses a significant health burden. It accounts for the majority of pre-eclampsia diagnoses in pregnancy. One-fifth of all women with late-onset disease have maternal complications such as HELLP (haemolysis, elevated liver enzymes and low platelet count) syndrome, and more than half of eclamptic seizures occur at term.^28,31,32

Pregnant women who are at high risk of pre-eclampsia require close monitoring and are usually started on prophylactic aspirin in early pregnancy to reduce the risk of development of pre-eclampsia and occurrence of adverse outcomes. Early commencement of this has the potential for maximum benefit,³³ which may be limited to early-onset disease.³⁴ It is important to be able to quantify a woman’s risk of developing pre-eclampsia during the course of pregnancy to help guide clinical decisions and monitoring strategies. The National Institute for Health and Care Excellence prioritises screening for early-onset pre-eclampsia in its research recommendations on antenatal care of women.³⁵

Currently, the assessment of a woman’s risk of developing pre-eclampsia is based mainly on clinical history,³⁶ but such risk-based predictions have been shown to have limited accuracy.³⁷ Risk factors based on clinical characteristics have also been shown to have quantitatively different associations with early- and late-onset pre-eclampsia,²⁶ and, similarly, biochemical and ultrasound markers have variations in their performance in predicting the two types of pre-eclampsia.^37–39 Prediction models incorporating additional tests for biochemical and ultrasound markers may improve the predictive performance of models.^40–42 It is, however, unlikely that a single model will accurately predict both early- and late-onset pre-eclampsia.²⁶

There are more than 60 multivariable prediction models developed to predict pre-eclampsia, using various combinations of clinical, biochemical and ultrasound risk factors.⁴³ Such models and tests for predicting pre-eclampsia have been based on findings from aggregate meta-analysis and primary studies, and none is recommended for use in routine clinical practice. This is because there is an absence of information about the reproducibility of the models or their predictive performance in different settings.

Although interventions such as aspirin have been found to significantly reduce the risk of early-onset pre-eclampsia in women predicted to be at ‘high risk’ of pre-eclampsia using a model, lack of robust information on the accuracy of this model means that we could not rule out potential benefit in women considered to be ‘low risk’. Before they can be used in clinical practice, prediction models need to be appropriately validated in multiple data sets external to that used to develop the model. This often takes many years to accomplish in a primary study, and, as a result, very few models have been externally validated to date.^43–46 Individual studies also often have an insufficient sample size to externally validate the relatively rare but serious condition of early-onset pre-eclampsia.²⁶

Meta-analysis of individual participant data (IPD), whereby the raw participant-level information is obtained and synthesised across multiple data sets, overcomes the limitations above.^47–50 The availability of the raw data substantially increases the sample size beyond what is achievable in a single study, and offers a unique opportunity to evaluate the generalisability of predictive performance of existing models across a range of clinical settings. Using IPD meta-analysis allows the standardisation of predictors and outcome definitions, takes into account the performance of many candidate prognostic variables, directly handles missing predictors and outcomes data, accounts for heterogeneity in baseline risks, and, most importantly, develops, validates and tailors the use of the most accurate prediction models to the appropriate population.

The unmet need for prediction models for pre-eclampsia, particularly early-onset pre-eclampsia, is mainly a result of lack of information on the generalisability of the models and their performances in external cohorts. Hence, before more resources are spent on developing further models, what is needed is external validation of existing models. If existing models’ performances are suboptimal, then further development of new models is warranted with sufficient sample size.