Calcium supplementation to prevent pre-eclampsia: protocol for an individual participant data meta- analysis, network meta-analysis and health economic evaluation

1 RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access Calcium supplementation to prevent preeclampsia: protocol for an individual participant data metaanalysis, network metaanalysis and health economic evaluation Thaís Rocha ,1 John Allotey ,1 Alfredo Palacios ,2 Joshua Peter Vogel ,3 Luc Smits,4 Guillermo Carroli,5 Hema Mistry ,6 Taryn Young,7,8 Zahida P Qureshi,9 Gabriela Cormick ,10 Kym I E Snell,11 Edgardo Abalos,5 Juan- Pablo Pena- Rosas,12 Khalid Saeed Khan ,13 Koiwah Koi Larbi,14 Anna Thorson,12 Mandisa Singata- Madliki,15 George Justus Hofmeyr,16 Meghan Bohren ,17 Richard Riley ,11 Ana Pilar Betran ,12 Shakila Thangaratinam,18,19 On behalf of the International Calcium in Pregnancy (i- CIP) Collaborative Network To cite: RochaT, AlloteyJ, PalaciosA, etal. Calcium supplementation to prevent preeclampsia: protocol for an individual participant data metaanalysis, network metaanalysis and health economic evaluation. BMJ Open 2023;13:e065538. doi:10.1136/ bmjopen-2022-065538 ►Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (http://dx.doi.org/10.1136/ bmjopen-2022-065538). Received 13 June 2022 Accepted 11 April 2023 For numbered affiliations see end of article. Correspondence to Dr John Allotey; j. allotey. 1@ bham. ac. uk Protocol © World Health Organization 2023. Licensee BMJ. ABSTRACT Introduction Low dietary calcium intake is a risk factor for preeclampsia, a major contributor to maternal and perinatal mortality and morbidity worldwide. Calcium supplementation can prevent preeclampsia in women with low dietary calcium. However, the optimal dose and timing of calcium supplementation are not known. We plan to undertake an individual participant data (IPD) metaanalysis of randomised trials to determine the effects of various calcium supplementation regimens in preventing preeclampsia and its complications and rank these by effectiveness. We also aim to evaluate the costeffectiveness of calcium supplementation to prevent preeclampsia. Methods and analysis We will identify randomised trials on calcium supplementation before and during pregnancy by searching major electronic databases including Embase, CINAHL, MEDLINE, CENTRAL, PubMed, Scopus, AMED, LILACS, POPLINE, AIM, IMSEAR, ClinicalTrials. gov and the WHO International Clinical Trials Registry Platform, without language restrictions, from inception to February 2022. Primary researchers of the identified trials will be invited to join the International Calcium in Pregnancy Collaborative Network and share their IPD. We will check each study’s IPD for consistency with the original authors before standardising and harmonising the data. We will perform a series of onestage and twostage IPD randomeffect metaanalyses to obtain the summary intervention effects on preeclampsia with 95% CIs and summary treatment–covariate interactions (maternal risk status, dietary intake, timing of intervention, daily dose of calcium prescribed and total intake of calcium). Heterogeneity will be summarised using tau2, I2 and 95% prediction intervals for effect in a new study. Sensitivity analysis to explore robustness of statistical and clinical assumptions will be carried out. Minor study effects (potential publication bias) will be investigated using funnel plots. A decision analytical model for use in lowincome and middleincome countries will assess the costeffectiveness of calcium supplementation to prevent preeclampsia. Ethics and dissemination No ethical approvals are required. We will store the data in a secure repository in an anonymised format. The results will be published in peerreviewed journals. PROSPERO registration number CRD42021231276. INTRODUCTION Preeclampsia is a pregnancyspecific condition characterised by raised blood pressure and protein in the urine. It is a major cause of maternal and perinatal mortality and morbidity worldwide, contributing to 76 000 STRENGTHS AND LIMITATIONS OF THIS STUDY ⇒The individual participant data (IPD) approach will allow us to explore any differential treatment effect across groups, and model how individuallevel covariates (eg, age, risk of preeclampsia) interact with treatment effect within the same trial to explain variability in outcomes. ⇒By analysing data on the actual amount of calcium taken and adherence to the prescribed regimen, we can explore the doses and frequencies of the clinical benefits of calcium supplementation. ⇒The health economic analysis will inform decisionmakers on current use or future calcium supplementation strategies to prevent preeclampsia based on the efficiency principle. ⇒Limitations include potential unavailability of IPD, which may limit the number of trials included. Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 2RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access maternal and half a million perinatal deaths each year; 99% of these are from lowincome and middleincome countries (LMICs).1–3 Most maternal deaths due to preeclampsia are preventable. Prevention of preeclampsia and its complications is crucial to achieving the healthrelated Sustainable Development Goals,4 and the WHO’s Thirteenth General Programme of Work for universal health coverage.5 Low dietary calcium is a recognised risk factor for preeclampsia.6–8 In LMICs, 80% of pregnant women have a mean calcium intake below the population Institute of Medicine recommended level of 800 mg/day,9 compared with low intake in only about a quarter of pregnant women in highincome countries.10 Calcium supplementation in pregnancy has been shown to reduce the risk of preeclampsia.11 In populations with low dietary calcium intake and in those at high risk of developing preeclampsia, the WHO recommends 1.5–2.0 g per day of oral elemental calcium supplementation during pregnancy to reduce the risk of preeclampsia, although there is no clear recommendation on the timing of initiation.12 A Cochrane review showed that high dose (≥1000 mg per day) of calcium supplementation during pregnancy reduced the risk of preeclampsia (8 trials, 10 678 women: average RR 0.36, 95% CI 0.20 to 0.65; I2=76%). But the quality was graded low due to significant heterogeneity from variations in the underlying risk of preeclampsia.11 Evidence for a lowdose calcium supplement to prevent preeclampsia (<1000 mg/day) is limited.11 Despite countries including calcium in their essential medicines lists, maternal mortality from hypertensive disorders in LMICs remains high.13 14 Optimising calcium intake to prevent preeclampsia is a priority area for the WHO.5 15 The 2018 WHO Guideline Development Group (GDG) highlighted research on the minimal dose and optimal commencement schedule for calcium supplementation as a high research priority.15 It is also not known whether calcium supplementation strategies should target highrisk women only or provide calcium supplements to all pregnant and reproductiveaged women, to confer benefits and be costeffective in preventing preeclampsia. We plan to undertake an individual participant data (IPD) metaanalysis of calcium supplementation to determine the intervention effects on preeclampsia and its complications, assess if the effects vary according to maternal and intervention characteristics, and the costeffectiveness of the different interventions strategies. Objectives Our primary objective is to determine the overall, and differential effects of calcium supplementation (according to maternal and intervention characteristics) on preeclampsia adjusted for cointerventions and baseline maternal calcium status, using an IPD metaanalysis. Our secondary objectives are to: ►Evaluate the effects of calcium supplementation on (1) maternal outcomes such as maternal death, eclampsia, severe maternal morbidity, admission to intensive care unit, haemolysis, elevated liver enzymes, low platelets (HELLP) syndrome and (2) perinatal outcomes such as stillbirth, perinatal death, neonatal death, preterm birth, low Apgar score, small for gestational age baby, and admission and length of stay in the neonatal intensive care unit. ►Produce a rank order of calcium supplementation regimens by effectiveness. ►Develop a decision analytical model to determine the costeffectiveness of different calcium supplementation strategies in an LMIC setting. METHODS AND ANALYSIS Our IPD metaanalytical approach will follow existing methodological guidelines and adhere to the Preferred Reporting Items for Systematic Review and Meta- Analysis of individual participant data (PRISMAIPD) reporting statement.16 The protocol has been registered on the International Prospective Register of Systematic Reviews (PROSPERO; CRD42021231276). Patient and public involvement Women with lived experience of preeclampsia will be involved with this work throughout and have informed the design, outcome selection and reporting. Literature search We will update the search of the 2018 Cochrane review11 until February 2022 to identify new trials that have been published since the last conducted search. This will include searches in databases such as Embase, CINAHL, MEDLINE, CENTRAL, PubMed, Scopus, AMED, LILACS, POPLINE, AIM, IMSEAR, ClinicalTrials. gov and WHO International Clinical Trials Registry Platform, using search strategies adapted from the original Cochrane search, and will include terms for pregnancy such as 'pregnan*’ or ‘wom*’, combined with terms for calcium ‘calcium*’ and randomised trials ‘random*’ or ‘allocation’ (see online supplemental appendix 1). No language restrictions will be applied. Eligibility criteria Any clinical trial with random allocation (individual or cluster) to calcium supplementation (any dose with or without additional supplements or treatments) before or during pregnancy compared with placebo, aspirin or routine care will be eligible for inclusion. Nonrandomised trials and animal studies will be excluded. Outcome measures Study outcomes were informed by the WHO recommendation on calcium supplementation during pregnancy to prevent preeclampsia and its complications,17 and the core outcome set for preeclampsia research.18 The primary outcomes are (1) any onset preeclampsia and (2) earlyonset preeclampsia (diagnosed <34 weeks’ gestation). We will use the authors’ reported definition of preeclampsia. However, suppose the trial IPD Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 3 RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access reports relevant variables. In that case, we will redefine preeclampsia as high blood pressure (defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg after 20 weeks of pregnancy) with significant proteinuria (defined as urine proteincreatinine ratio ≥30 mg/mmol or ≥2+ on dipstick testing or ≥300 mg/24 hours or ≥500 mg per litre). Our secondary outcomes include maternal and offspring complications such as maternal death, eclampsia, severe maternal morbidity (renal, haematological, neurological, hepatic complications), admission to intensive care unit, HELLP syndrome, stillbirth, neonatal death, admission and length of stay in the neonatal intensive care unit, preterm birth or small for gestational age (table 1). We will undertake a subgroup analysis to explore whether the intervention effect is modified by (interacts with) maternal risk status, dietary intake, the timing of intervention, the daily dose of calcium prescribed and total intake of calcium. Study selection At least two researchers will independently select studies using a twostage process. They will first screen the titles and abstracts of studies and then assess the full text of selected studies in detail for eligibility. Disagreement will be resolved via discussion with a third researcher. Data extraction will be done in duplicates. At the study level, extracted data will include country, setting, inclusion and exclusion criteria of participants, intervention, control, primary aim, and definition and assessment of the primary outcome. Establishment of the International Calcium in Pregnancy collaborative network We will contact primary researchers of identified studies via email and invite them to join the collaborative network and share their IPD. To date, seven collaborators have joined the network and shared access to anonymised individual data of 16 111 women (table 2). The network is a global effort to bring together researchers, clinicians and epidemiologists (https://www.icipnetwork.com/). A bespoke database will be set up for collaborators to share data. Authors will be allowed to share their data in any format convenient to them. We will consider all variables recorded in the original studies, even those not reported in the publications. Once deposited, the data will be converted to a standardised format, followed by the range and data consistency checking before merging and harmonising. Quality assessment The quality of the IPD from each study will be assessed independently by two researchers. We will use the revised Cochrane tool for assessing the risk of bias in randomised trials (RoB2)19 based on published study characteristics and supplement this with information within the IPD. We will consider six items used in the Cochrane risk of bias tool: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting and other potential sources of bias. We will conduct sensitivity analyses to examine the robustness of statistical and clinical conclusions to inform the inclusion or exclusion of trials considered to be at high risk of bias. Data and integrity checks We will perform integrity checks of IPD received for each trial by evaluating the integrity of randomisation and followup procedures and reviewing the completeness and accuracy of the data.20 Any inconsistencies found (missing data, extreme values, discrepancies between the trial report and the data) will be resolved with the original study authors. The study progress and discrepancies will be recorded. Sample size considerations Formal sample size calculations are not usually undertaken for metaanalyses. A single trial would need 10 847 participants (80% power, 5% error) to detect the interaction OR of 0.62 between lowrisk and highrisk groups, assuming calcium reduces preeclampsia by 20% in a lowrisk group by another 30% in the highrisk population.21 Using power calculations by simulating IPD to match aggregate data (eg, number of participants, events, Table 1 Structured research question Question components Population Pregnant women and women of reproductive age who are not yet pregnant but intending to become pregnant. Intervention Calcium supplementation (with or without additional supplements or treatments) Outcomes Primary outcome Any onset preeclampsia Earlyonset preeclampsia (<34 weeks’ gestation) Secondary outcomes Maternal outcomes Maternal death Eclampsia Severe maternal morbidity (renal, haematological, neurological, hepatic complications) Admission to intensive care unit HELLP syndrome Neonatal outcomes Stillbirth, neonatal death Apgar score <7 after 5 min Admission to the neonatal intensive care unit within 28 days after birth Preterm birth Small for gestational age baby Design of included studies Randomised trials HELLP, haemolysis, elevated liver enzymes, low platelets. Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 4RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access covariate distributions)22 from studies promising their IPD so far (17 526 women) and assuming heterogeneity of 1%–8% in the rates of preeclampsia in the lowrisk group in each trial, we will have over 98% power to detect an interaction OR of 0.62 in our IPD metaanalysis.22 Even when we additionally assume heterogeneity in the overall effect of calcium in the lowrisk group from 0.6 to 0.9, the power will still be 90%, illustrating the large sample size available. We will have similar power for other covariates. Statistical analysis Overall effect We will perform a series of onestage and twostage IPD randomeffect metaanalyses fitted using either frequentist methods (eg, restricted maximum likelihood with CIs derived using Hartung- Knapp correction) or Bayesian methods (eg, with vague or empirically derived prior distributions). In the twostage approach, first, the IPD will be analysed separately for each study to obtain Table 2 List of trials current in the i- CIP network and trials that have agreed to share data (total n=17 526 individuals) Author, year Country Study population risk of PE/start of intervention Intervention Comparator Sample size Data already shared with the i- CIP network Trials currently in iCIP (n=16 111 individuals, 7 trials) (data available already) Villar,49 2006 Argentina, Egypt, India, Peru, South Africa, Vietnam High risk, up to 20 weeks’ gestation 1500 mg calcium carbonate Placebo 8325 Yes Levine,50 1997 USA Low risk, 13–21 weeks’ gestation 2000 mg calcium carbonate Placebo 4589 Yes Belizán,51 1991 Argentina Any risk, 20 weeks’ gestation 2000 mg calcium carbonate Placebo 1194 Yes Ettinger,52 2009 Mexico Low risk, first trimester 1200 mg calcium carbonate Placebo 670 Yes Goldberg,53 2013 Gambia Any risk, 18–20 weeks’ gestation 1500 mg calcium carbonate Placebo 662 Yes Hofmeyr,54 2019 Argentina, South Africa, Zimbabwe High risk, prepregnancy and up to 20 weeks’ gestation 500 mg calcium carbonate Placebo 581 Yes Azami,55 2017 Iran High risk, >20 weeks’ gestation 800 mg calcium carbonate Multivitamin 90 Yes Trials that agreed to share IPD (n=1415 individuals, 7 trials) (data expected to be made available to us) Omotayo,56 2018 Kenya Low risk, 16–30 gestational weeks 1500 mg calcium carbonate 1000 mg calcium carbonate 990 No Asemi,57 2014 Iran Low risk, 16 weeks’ gestation Multivitaminmineral with 250 mg calcium Multivitamin 104 No Karamali,58 2016 Iran High risk, 24–26 weeks’ gestation 1000 mg calcium carbonate, 50 000 IU vitamin D3 Placebo 60 No Samimi,59 2016 Iran High risk, 20 weeks’ gestation 1000 mg calcium carbonate, 50 000 IU vitamin D3 Placebo 60 No Souza,60 2014 Brazil High risk, 20–27 weeks’ gestation 2000 mg calcium carbonate, 100 mg aspirin Placebo 49 No Asemi,61 2015 Iran High risk, 27 weeks’ gestation 800 mg calcium carbonate, 200 mg magnesium, 8 mg zinc, 400 IU vitamin D3 Placebo 46 No Asemi,62 2016 Iran Low risk, 25 weeks’ gestation 500 mg calcium carbonate, 200 IU vitamin D3 Placebo 46 No i- CIP, International Calcium in Pregnancy; PE, preeclampsia. Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 5 RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access relevant aggregate data (eg, a treatment effect estimates and its CI for each study) for each outcome; second, this aggregate data will be combined (pooled) across studies using an appropriate metaanalysis model to produce relevant summary results (eg, a weighted average of the treatment effect). The alternative onestage approach analyses the IPD from all studies in a single step, using a statistical model (eg, a mixed development linear, logistic or Cox regression model) that accounts for the clustering of patients within studies and potential heterogeneity between studies. When the same modelling assumptions and estimation methods are used, onestage and twostage approaches are similar.23 The onestage approach is preferable when rare events are modelled as a more exact likelihood. However, the twostage approach allows more familiar metaanalysis techniques and graphs (eg, forest plots). Therefore, we will perform both onestage and twostage methods and compare any differences.23 Differential effect by subgroups (treatment–covariate interactions) For each outcome, we will examine differences in predefined subgroups to summarise whether the intervention effect is modified by (interacts with) maternal risk status, dietary intake, the timing of intervention, a daily dose of calcium prescribed and total intake of calcium; this analysis will use only withinstudy information to avoid ecological bias from across study information. The onestage analyses will be achieved by centring patientlevel covariates by their mean and including the mean as an additional covariate.24 Nonlinear interactions with continuous covariates (eg, risk status) will be examined using restricted cubic splines.25 IPD network meta-analysis An IPD network metaanalysis will compare and rank intervention effects for the various regimens (and doses), using direct and indirect comparisons while adjusting for covariates that modify treatment effects to alleviate any inconsistency in the network.26 The withinstudy correlation of multiple intervention effects from the same trial will be accounted for (if necessary). A common betweenstudy variance is assumed for all treatment contrasts in the network. We will produce summary (pooled) effect estimates for each treatment contrast (ie, each pair of strategies in the network) with 95% CIs and the borrowing of strength statistics (to reveal the contributions of indirect evidence). Based on the results, the ranking of intervention types will be calculated using resampling methods and quantified by the probabilities of being ranked first, second and last, together with the mean rank and the surface under the cumulative ranking curve. The consistency assumption will be examined for each treatment comparison with direct and indirect evidence (seen as a closedloop within the network plot); this involves estimating the direct and indirect evidence and comparing the two.27 The consistency assumption will also be examined across the whole network using ‘designby- treatment interaction’ models, which allow an overall significance test for inconsistency. If evidence of inconsistency is found, explanations will be sought and resolved by adjusting for covariates that act as effect modifiers using the approach of Donegan et al,28 as identified from the analyses mentioned above. We will display forest plots for each metaanalysis with studyspecific estimates, CIs and weights, alongside the summary (pooled) metaanalysis estimates and a 95% CI. We will translate our findings to the absolute risk prediction scale to help health professionals tailor treatment decisions to an individual’s risk of preeclampsia conditional on their covariates (prognostic factors) and anticipated treatment effects and any interactions.29 Penalisation and shrinkage will alleviate overfitting identified using bootstrapping. Examining potential sources of bias Small study effects (potential publication bias) will be investigated using funnel plots and test for asymmetry if ten or more studies are in a metaanalysis. To examine the impact of studies where IPD were not shared, we will extract aggregate studylevel data (where available) and incorporate them alongside the IPD using the twostage random effect metaanalysis framework. We will also examine the impact of excluding any trials that are not at low risk of bias. Dealing with missing variables A range of strategies will be considered for dealing with missing data in covariates. To analyse randomised trials, mean imputation or the missing indicator method are appropriate to handle missing data in covariates.30 If necessary, we will use multiple imputations for systematically missing variables (considered plausible), which involves borrowing information across studies while allowing for heterogeneity and clustering in a multilevel imputation model.31 Health economic and decision analytical modelling Decision model The costeffectiveness analysis will be designed and analysed following state- of- theart methods and analysis in the economic evaluation of healthcare programmes.32 We will develop a decision tree to determine the costeffectiveness of calcium supplementation regimens during pregnancy for the prevention of preeclampsia. A decision tree is a diagrammatic representation of a decision analysis in which chains of choices are identified, each conditional on a prior choice and with outcomes and probabilities.33 The model structure will be developed based on previous models.34–38 The results of the costeffectiveness analysis will be reported according to the 2022 Consolidated Health Economic Evaluation Reporting Standards statement.39 The main outcome of the model will be the incremental costeffectiveness ratio (ICER). The ICER expresses the additional costs needed to achieve an additional unit of health outcome, that is, the incremental cost per case of Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 6RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access preeclampsia/eclampsia (PE/E) avoided. Mathematically, ICER can be expressed as: Cost 1 −Cost 0 Health benefits1−Health benefits0 Where 1 represents the intervention group, and 0 represents the comparator group. Intervention and comparators The interventions to be evaluated (calcium supplementation regimens), as well as their potential comparators, will be defined according to the parent study’s ‘IPD metaanalysis’. Target population The decision model will be applied to a hypothetical population of pregnant women and women of reproductive age who are not yet pregnant but intend to become pregnant, regardless of their risk for preeclampsia and their daily calcium intake. Other populations considered will be pregnant women with a high risk of preeclampsia and pregnant women with low calcium intake. Study perspective The study will be conducted from the public healthcare system perspective using IPD estimates for Argentina and published literature. Measurement of effectiveness The health benefits will be measured as cases of PE/E avoided, lifeyears (LYs) gained and disabilityadjusted LYs (DALYs) avoided. For women, we will estimate the LY gained by subtracting the life expectancy from the mean age of an eclampsia patient, while for newborn LY gained will be considered as the average life expectancy in the country. We will use disability weights from the global burden of diseases and countryspecific lifeexpectancy tables for Argentina.40 41 Results will be presented as cost per case of PE/E avoided, cost per LY gained and cost per DALYs averted. Estimating resources and costs The analysis will also include two main cost categories: 1. Costs of implementing the interventions (calcium acquisition costs, etc). 2. Costs associated with using healthcare services by individuals in both the intervention and comparator groups (hospital stay costs in different complexity of care, laboratory tests, among others). The costs of health events will be estimated for both mother and children using the microcosting method.42 Time horizon The time horizon will be from pre or early pregnancy until the discharge of mother and child from the hospital. Discount rate Since all costs and PE/E cases will occur within the first year, no discounting will be applied to either cost and PE/E cases. For LY and DALYs, a 3% discount rate will be used in accordance with Bill and Melinda Gates Foundation Reference Case guidelines for LMIC.43 Currency, date, conversions The costs of implementing the intervention and those associated with the use of healthcare services by individuals will be valued in local currency and then converted to US dollars using international market exchange rates and international dollars through the purchasing power parity conversion factor published by the World Bank database.44 Cost-effectiveness threshold To define whether the intervention is costeffective, as the hypothesis is that calcium supplementation will not be ‘better and cost saving’ than placebo, it will be necessary to establish a decision rule, defined as a willingness- to- pay value for the outcome of interest will be used as a threshold. Despite previous use and recommendations of higher thresholds, such as the WHO’s recommendation of up to three times the gross domestic product (GDP) per DALY,45 we will adopt a more stringent threshold consistent with recent studies: 1 times GDP per capita per DALY or QALY.46 47 That is, if for a given intervention the ICER lies above this threshold, then it will be deemed too expensive in relation to its added benefit and thus not costeffective, whereas if the ICER lies below this threshold, the intervention will be judged costeffective and a ‘good buy’. The GDP per capita will be obtained from the World Bank database.44 Sensitivity analysis Sensitivity analysis will be used to report and assess the level of confidence (or uncertainty) that may be associated with the key model parameters (calcium efficacy, etc). A tornado diagram (deterministic sensitivity analysis) will be generated to plot univariate variations in ICER due to defined variations in key parameters. Probabilistic sensitivity analysis will additionally be performed using 2000 Monte Carlo simulations. We simultaneously sampled from the distributions of each input parameter in each simulation to estimate the ‘probability’ of the intervention being costeffective at different thresholds. Ethics and dissemination The current project involves a metaanalysis of anonymised datasets. No ethical approvals are needed for this project. Guidance on participant data storage and management will be adhered to. The dataset is not open access. Findings will be published in peerreviewed journals, presented at UK national and international conferences, shared with policymakers and international organisations, and disseminated to women and their families through links with patient groups and relevant charities. DISCUSSION We propose an IPD metaanalysis of randomised trials to evaluate the effects of calcium supplementation in Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 7 RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access preventing preeclampsia, its complications, and other maternal and fetal–neonatal complications. We will also use an IPD network metaanalysis to compare and rank intervention effects for the various calcium regimens (and doses). In addition, we will assess the costeffectiveness of calcium supplementation to prevent preeclampsia using a modelbased economic evaluation for use in LMIC. The 2018 GDG update reported that calcium supplementation is likely to increase equity. Universal calcium supplementation is expected to prevent 21 500 maternal deaths each year and reduce maternal DALYs by 620 000.48 However, the dose and timing of choice for optimal calcium supplementation to prevent preeclampsia are not yet known. With access to IPD containing over 15 000 participants, our IPD metaanalysis will have a larger sample size than any individual study trying to identify if a particular subgroup benefits the most from calcium supplementation and determine the effects on rare but important outcomes of earlyonset preeclampsia (delivery <34 weeks’ gestation), stillbirth and perinatal deaths, and complications such as HELLP syndrome. By accessing the data on the actual timing of commencement of the intervention, the amount of calcium taken by individual women and their adherence, we can determine if there is an interaction between the effect of calcium treatment and the exact dose taken by the woman. We can then tailor recommendations to the individual conditional on dose and adherence. Furthermore, our IPD metaanalysis will allow us to tailor calcium treatment strategies considering treatment effects on individuallevel factors (including prognostic factors and treatment–covariate interactions). We can model prognostic factors to predict a women’s preeclampsia risk better, conditional on prognostic factors and the expected response to calcium treatment. Thus, we will combine baseline risk and treatment response information to guide treatment decisions based on individuallevel information. The WHO GDG also highlighted an overall lack of information on the costeffectiveness of calcium supplementation in LMICs, which is crucial to plan implementation. Therefore, we will evaluate the costeffectiveness of different calcium supplementation strategies in the LMICs context. To facilitate the adoption of the economic model, we will provide the model in an openaccess format. Other researchers can input their countryspecific epidemiological and cost data to determine the costeffectiveness estimates for their countries. Potential limitations of this study include our inability to obtain IPD from all identified trials due to no contact with original study author, willingness to share raw data or because access to primary data is no longer available. These will be clearly reported as part of our PRISMA flow diagram and a sensitivity analysis to examine the impact of non- IPD studies will be carried out by incorporating these with the IPD studies. There may also be variations in how variables are reported in the shared IPD, which may limit our ability to assess whether the intervention effect is modified by these individuallevel covariates. We will minimise the above limitation through robust data cleaning and harmonisation procedures. The findings of this IPD metaanalysis and costeffectiveness analysis will directly inform guidelines and policymakers in LMICs. The results will assist healthcare managers, other healthcare service providers and policymakers make informed decisions regarding the ongoing use of calcium or future calcium supplementation strategies to prevent preeclampsia based on the efficiency principle. Author affiliations 1WHO Collaborating Centre for Global Women’s Health, Institute of Metabolism and Systems Research, University of Birmingham, Birmingham, UK 2Health Economics, Institute for Clinical Effectiveness and Health Policy, Buenos Aires, Argentina 3Maternal, Child and Adolescent Health Program, Burnet Institute, Melbourne, Victoria, Australia 4Department of Epidemiology, Care and Public Health Research Institute, Maastricht University, Maastricht, The Netherlands 5Centro Rosarino de Estudios Perinatales (CREP), Rosario, Argentina 6Warwick Evidence, University of Warwick, Coventry, UK 7Centre for Evidence- Based Health Care, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa 8South African Cochrane Centre, South African Medical Research Council, Cape Town, South Africa 9Department of Obstetrics and Gynecology, University of Nairobi, Nairobi, Kenya 10Department of Health Technology Assessment and Health Economics, Institute for Clinical Effectiveness and Health Policy, Buenos Aires, Argentina 11Centre for Prognosis Research, School of Medicine, Keele University, Keele, UK 12Reproductive Health and Research, World Health Organization, Geneva, Switzerland 13Public Health, University of Granada Faculty of Medicine, Granada, Spain 14Action on Preeclampsia (APEC), Accra, Ghana 15Effective Care Research Unit (ECRU), East London Hospital Complex, East London, South Africa 16Obstetrics and Gynaecology, Frere Hospital, East London, South Africa 17Centre for Health Equity, University of Melbourne School of Population and Global Health, Carlton, Victoria, Australia 18WHO Collaborating Centre for Global Women’s Health, Institute of Metabolism and Systems Research, University of Birmingham College of Medical and Dental Sciences, Birmingham, UK 19Birmingham Women’s and Children’s Hospitals NHS Foundation Trust, Birmingham, UK Twitter Thaís Rocha @Thais_P_Rocha, Alfredo Palacios @AlfrePalacios13, Joshua Peter Vogel @josh_vogel, Taryn Young @TarynYoung3 and Khalid Saeed Khan @ Profkkhan Collaborators Helen Moraa; University of Nairobi, Rana Zahroh; University of Melbourne Contributors ST and JA planned the study. TR wrote the initial draft of the protocol manuscript with additional input writing input from AP, JPV, LS, KIES, EA, J- PP- R, KSK, KKL, AT, MS- M, RR, GJH, GCa, APB, HM, MB, TY, ZPQ and GCo. TR and JA designed the tables. All authors contributed to the drafts and final version of the manuscript. JA and ST are guarantors. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. Funding The UKRI Medical Research Council supports this work—Global Maternal and Neonatal Health grant number MR/T010185/1. This work is also funded by the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Sexual and Reproductive Health and Research (SRH), WHO. JPV is supported by the NHMRC Investigator grant. Disclaimer The author is a staff member of the World Health Organization. The author alone is responsible for the views expressed in this publication and they Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 8RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access do not necessarily represent the views, decisions or policies of the World Health Organization. Competing interests None declared. Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details. Patient consent for publication Not applicable. Provenance and peer review Not commissioned; externally peer reviewed. Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peerreviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Open access This is an open access article distributed under the terms of the Creative Commons Attribution IGO License (CC BY 3.0 IGO), which permits use, distribution,and reproduction in any medium, provided the original work is properly cited. In any reproduction of this article there should not be any suggestion that WHO or this article endorse any specific organization or products. The use of the WHO logo is not permitted. This notice should be preserved along with the article’s original URL. ORCID iDs ThaísRocha http://orcid.org/0000-0003-0113-6877 JohnAllotey http://orcid.org/0000-0003-4134-6246 AlfredoPalacios http://orcid.org/0000-0001-7684-0880 Joshua PeterVogel http://orcid.org/0000-0002-3214-7096 HemaMistry http://orcid.org/0000-0002-5023-1160 GabrielaCormick http://orcid.org/0000-0001-7958-7358 Khalid SaeedKhan http://orcid.org/0000-0001-5084-7312 MeghanBohren http://orcid.org/0000-0002-4179-4682 RichardRiley http://orcid.org/0000-0001-8699-0735 Ana PilarBetran http://orcid.org/0000-0002-5631-5883 REFERENCES 1 von Dadelszen P, Magee LA. Preventing deaths due to the hypertensive disorders of pregnancy. Best Pract Res Clin Obstet Gynaecol 2016;36:83–102. 2 Khan KS, Wojdyla D, Say L, etal. Who analysis of causes of maternal death: a systematic review. Lancet 2006;367:1066–74. 3 Clark SL, Belfort MA, Dildy GA, etal. Maternal death in the 21st century: causes, prevention, and relationship to cesarean delivery. Am J Obstet Gynecol 2008;199:36. 4 UNDP. What are the sustainable development goals? Goal 3, good health and wellbeing [United Nations Development Programme]. 2022. Available: https://www.undp.org/sustainable-development- goals#good-health 5 World Health Organisation. Thirteenth General programme of work 2019- 2023. 2019. Available: https://apps.who.int/iris/bitstream/ handle/10665/324775/WHO-PRP-18.1-eng.pdf [Accessed 23 May 2022]. 6 Tittsler RP, Pederson CS, Snell EE, etal. Symposium on the lactic acid bacteria. Bacteriol Rev 1952;16:227–60. 7 Belizán JM, Villar J, Repke J. The relationship between calcium intake and pregnancyinduced hypertension: up- to- date evidence. Am J Obstet Gynecol 1988;158:898–902. 8 Repke JT, Villar J. Pregnancyinduced hypertension and low birth weight: the role of calcium. Am J Clin Nutr 1991;54:237S–241S. 9 Institute of Medicine of the National Academies. Dietary reference intakes for calcium and vitamin D. Report brief November 2010; 2011. 10 Cormick G, Betrán AP, Romero IB, etal. Global inequities in dietary calcium intake during pregnancy: a systematic review and metaanalysis. BJOG 2019;126:444–56. 11 Hofmeyr GJ, Lawrie TA, Atallah ÁN, etal. Calcium supplementation during pregnancy for preventing hypertensive disorders and related problems. Cochrane Database Syst Rev 2018;10:CD001059. 12 World Health Organization. Calcium supplementation during pregnancy for prevention of preeclampsia and its complications; 2018. 13 Kassebaum NJ, Bertozzi- Villa A, Coggeshall MS, etal. Global, regional, and national levels and causes of maternal mortality during 1990- 2013: a systematic analysis for the global burden of disease study 2013. Lancet 2014;384:980–1004. 14 WHO U, UNFPA, World Bank Group and the United Nations Population Division. Trends in maternal mortality 2000 to 2017: estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division; 2019. 15 WHO Guidelines Approved by the Guidelines Review Committee. WHO recommendation: calcium supplementation during pregnancy for the prevention of preeclampsia and its complications. Geneva World Health Organization; 2018. 16 Stewart LA, Clarke M, Rovers M, etal. Preferred reporting items for systematic review and metaanalyses of individual participant data: the prismaipd statement. JAMA 2015;313:1657–65. 17 WHO recommendation: calcium supplementation during pregnancy for the prevention of preeclampsia and its complications; 2018. 18 Duffy J, Cairns AE, Richards- Doran D, etal. A core outcome set for preeclampsia research: an international consensus development study. BJOG 2020;127:1516–26. 19 Sterne JAC, Savović J, Page MJ, etal. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019;366:l4898. 20 Alfirevic Z, Kellie FJ, Stewart F, etal. Identifying and handling potentially untrustworthy trials in pregnancy and childbirth Cochrane reviews. 2021. Available: https://pregnancy.cochrane.org/news/ identifying-and-handling-potentially-untrustworthy-trials-pregnancy- and-childbirth-cochrane [Accessed 27 May 2022]. 21 Demidenko E. Sample size and optimal design for logistic regression with binary interaction. Stat Med 2008;27:36–46. 22 Ensor J, Burke DL, Snell KIE, etal. Simulationbased power calculations for planning a twostage individual participant data metaanalysis. BMC Med Res Methodol 2018;18:41. 23 Burke DL, Ensor J, Riley RD. Metaanalysis using individual participant data: onestage and twostage approaches, and why they may differ. Stat Med 2017;36:855–75. 24 Hua H, Burke DL, Crowther MJ, etal. Onestage individual participant data metaanalysis models: estimation of treatmentcovariate interactions must avoid ecological bias by separating out withintrial and acrosstrial information. Stat Med 2017;36:772–89. 25 Riley RD, Debray TPA, Fisher D, etal. Individual participant data metaanalysis to examine interactions between treatment effect and participantlevel covariates: statistical recommendations for conduct and planning. Stat Med 2020;39:2115–37. 26 Riley RD, Jackson D, Salanti G, etal. Multivariate and network metaanalysis of multiple outcomes and multiple treatments: rationale, concepts, and examples. BMJ 2017;358:j3932. 27 Dias S, Welton NJ, Caldwell DM, etal. Checking consistency in mixed treatment comparison metaanalysis. Stat Med 2010;29:932–44. 28 Donegan S, Williamson P, D’Alessandro U, etal. Combining individual patient data and aggregate data in mixed treatment comparison metaanalysis: individual patient data may be beneficial if only for a subset of trials. Stat Med 2013;32:914–30. 29 Kent DM, Steyerberg E, van Klaveren D. Personalized evidence based medicine: predictive approaches to heterogeneous treatment effects. BMJ 2018;363:k4245. 30 Sullivan TR, White IR, Salter AB, etal. Should multiple imputation be the method of choice for handling missing data in randomized trials? Stat Methods Med Res 2018;27:2610–26. 31 Quartagno M, Grund S, Carpenter J. Jomo: a flexible package for twolevel joint modelling multiple imputation. The R Journal 2019;11:205. 32 Drummond MF, Sculpher MJ, Claxton K, etal. Methods for the Economic Evaluation of Health Care Programmes, 4th ed. Oxford: Oxford University Press, 2015. 33 Culyer AJ. The dictionary of health economics, 2nd ed. Chelthenham, UK: Edward Elgar, 2010. 34 Meads CA, Cnossen JS, Meher S, etal. Methods of prediction and prevention of preeclampsia: systematic reviews of accuracy and effectiveness literature with economic modelling. Health Technol Assess 2008;12:iii–iv, 35 Chicaíza- Becerra LA, García- Molina M, Oviedo- Ariza SP, etal. Cost effectiveness of calcium supplement in reducing preeclampsiarelated maternal mortality in Colombia. Rev Salud Publica (Bogota) 2016;18:300–10. 36 Feldhaus I, LeFevre AE, Rai C, etal. Optimizing treatment for the prevention of preeclampsia/eclampsia in Nepal: is calcium supplementation during pregnancy costeffective? Cost Eff Resour Alloc 2016;14:13. Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from 9 RochaT, etal. BMJ Open 2023;13:e065538. doi:10.1136/bmjopen-2022-065538 Open access 37 Meertens LJE, Scheepers HCJ, Willemse JPMM, etal. Should women be advised to use calcium supplements during pregnancy? A decision analysis. Matern Child Nutr 2018;14:e12479. 38 Memirie ST, Tolla MT, Desalegn D, etal. A costeffectiveness analysis of maternal and neonatal health interventions in Ethiopia. Health Policy Plan 2019;34:289–97. 39 Husereau D, Drummond M, Augustovski F, etal. Consolidated health economic evaluation reporting standards 2022 (cheers 2022) statement: updated reporting guidance for health economic evaluations. BMJ 2022;376:e067975. 40 Mathers CD, Sadana R, Salomon JA, etal. Estimates of Dale for 191 countries: methods and results. World Health Organization; 2000. 41 Health statistics and information systems. Metrics: disabilityadjusted life year (DALY). Quantifying the burden of disease from mortality and morbidity. n.d. Available: https://www.who.int/ healthinfo/global_burden_disease/metrics_daly/en/ 42 Mogyorosy Z, Smith P. The main methodological issues in costing health care services: a literature review. 2005. Available: https://www. york.ac.uk/che/pdf/rp7.pdf 43 Bill and Melinda Gates Foundation NI, the Health Intervention and Technology Assessment Program (Thailand), and the University of York, Centre for Health Economics. Bill and melinda gates Foundation methods for economic evaluation Project (MEEP). 2014. Available: http://www.idsihealth.org/wp-content/uploads/2016/05/ Gates-Reference-case-what-it-is-how-to-use-it.pdf 44 World Bank. Purchasing power parities and the size of World economies: results from the 2017 international comparison program. 2020. Available: https://openknowledge.worldbank.org/handle/ 10986/33623 45 Health WCoMa. Macroeconomics and health: investing in health for economic development/report of the Commission on macroeconomics and health. 2001. Available: https://apps.who.int/ iris/handle/10665/42435 46 Ochalek J, Lomas J, Claxton K. Estimating health opportunity costs in lowincome and middleincome countries: a novel approach and evidence from crosscountry data. BMJ Glob Health 2018;3:e000964. 47 Woods B, Revill P, Sculpher M, etal. Countrylevel costeffectiveness thresholds: initial estimates and the need for further research. Value Health 2016;19:929–35. 48 Bhutta ZA, Ahmed T, Black RE, etal. What works? Interventions for maternal and child undernutrition and survival. Lancet 2008;371:417–40. 49 Villar J. World health organization calcium supplementation for the prevention of Preeclampsia trial group. World health organization randomized trial of calcium supplementation among low calcium intake pregnant women. Am J Obstet Gynecol 2006:639–49. 50 Levine RJ, Hauth JC, Curet LB, etal. Trial of calcium to prevent preeclampsia. N Engl J Med 1997;337:69–76. 51 Belizán JM, Villar J, Gonzalez L, etal. Calcium supplementation to prevent hypertensive disorders of pregnancy. N Engl J Med 1991;325:1399–405. 52 Ettinger AS, Lamadrid- Figueroa H, Téllez- Rojo MM, etal. Effect of calcium supplementation on blood lead levels in pregnancy: a randomized placebocontrolled trial. Environ Health Perspect 2009;117:26–31. 53 Goldberg GR, Jarjou LMA, Cole TJ, etal. Randomized, placebocontrolled, calcium supplementation trial in pregnant Gambian women accustomed to a low calcium intake: effects on maternal blood pressure and infant growth. Am J Clin Nutr 2013;98:972–82. 54 Hofmeyr GJ, Betrán AP, Singata- Madliki M, etal. Prepregnancy and early pregnancy calcium supplementation among women at high risk of preeclampsia: a multicentre, doubleblind, randomised, placebocontrolled trial. Lancet 2019;393:330–9. 55 Azami M, Azadi T, Farhang S, etal. The effects of multi mineralvitamin D and vitamins (c+e) supplementation in the prevention of preeclampsia: an RCT. Int J Reprod Biomed 2017;15:273–8. 56 Omotayo MO, Dickin KL, Chapleau GM, etal. Cluster- Randomized noninferiority trial to compare supplement consumption and adherence to different dosing regimens for antenatal calcium and ironfolic acid supplementation to prevent preeclampsia and anaemia: rationale and design of the micronutrient initiative study. J Public Health Res 2015;4:582. 57 Asemi Z, Samimi M, Tabassi Z, etal. Multivitamin versus multivitaminmineral supplementation and pregnancy outcomes: a singleblind randomized clinical trial. Int J Prev Med 2014;5:439–46. 58 Karamali M, Asemi Z, Ahmadi- Dastjerdi M, etal. Calcium plus vitamin D supplementation affects pregnancy outcomes in gestational diabetes: randomized, doubleblind, placebocontrolled trialexpression of concern. Public Health Nutr 2021;24:4369. 59 Samimi M, Kashi M, Foroozanfard F, etal. The effects of vitamin D plus calcium supplementation on metabolic profiles, biomarkers of inflammation, oxidative stress and pregnancy outcomes in pregnant women at risk for preeclampsia. J Hum Nutr Diet 2016;29:505–15. 60 Souza EV, Torloni MR, Atallah AN, etal. Aspirin plus calcium supplementation to prevent superimposed preeclampsia: a randomized trial. Braz J Med Biol Res 2014;47:419–25. 61 Asemi Z, Esmaillzadeh A. The effect of multi mineralvitamin D supplementation on pregnancy outcomes in pregnant women at risk for preeclampsia. Int J Prev Med 2015;6:62. 62 Asemi Z, Samimi M, Siavashani MA, etal. Calcium- Vitamin D cosupplementation affects metabolic profiles, but not pregnancy outcomes, in healthy pregnant women. Int J Prev Med 2016;7:49. Protected by copyright. on July 11, 2023 at Granada/Medicina/CC Salud PO Box 750.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065538 on 11 May 2023. Downloaded from