Open access Protocol BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from Pharmacological interventions for the prevention of fetal growth restriction: protocol for a systematic review and network meta-analysis

Alessandra Bettiol,1 Niccolò Lombardi,1 Giada Crescioli,1 Laura Avagliano,2 Alessandro Mugelli,1 Claudia Ravaldi,3,4 Alfredo Vannacci 1,4

To cite: Bettiol A, Abstract Strengths and limitations of this study Lombardi N, Crescioli G, et al. Introduction Fetal growth restriction (FGR) includes Pharmacological interventions different conditions in which a fetus fails to reach for the prevention of fetal ►► Both clinical trials and observational studies will be growth restriction: protocol for a the own full growth, and accounts for 28%–45% of included. systematic review and network non-anomalous stillbirths. The management of FGR is ►► Where possible, results will be synthetised using a meta-analysis. BMJ Open based on the prolongation of pregnancy long enough network meta-analysis, thus allowing to simultane- 2019;9:e029467. doi:10.1136/ for fetal organs to mature while preventing starvation. ously combine both direct and indirect evidence. bmjopen-2019-029467 As for pharmacological management, most guidelines ►► The study team includes clinicians, pharmacologists ►► Prepublication history and recommend treatment with low-dose aspirin and/or with and experts in the field of pregnancy complications additional material for this heparin, although this approach is still controversial and and related therapeutic interventions. paper are available online. To innovative promising therapies are under investigation. view these files, please visit As no firm evidence exists to guide clinicians towards the journal online (http://​dx.​doi.​ Introduction org/10.​ ​1136/bmjopen-​ ​2019-​ the most effective therapeutic intervention, this protocol 029467). describes methods for a systematic review and network Fetal growth restriction (FGR), also known meta-analysis (NetMA) of pharmacological treatments for as intrauterine growth restriction, includes Received 28 January 2019 FGR prevention. different conditions in which a fetus fails to Revised 15 May 2019 Methods and analysis We will search MEDLINE and reach its own growth potential. For many Accepted 21 May 2019 Embase for clinical trials and observational studies years, the most common parameter used http://bmjopen.bmj.com/ performed on gestating women with clinically diagnosed to measure FGR was small for gestational risk of FGR. Experimental interventions will include heparin age (SGA),1 although the two terms are not and low-molecular-weight heparin, acetylsalicylic acid, synonymous. Recently, an international clin- antiplatelet agents, phosphodiesterase type 3 and 5 ical consensus was obtained about the defi- © Author(s) (or their inhibitors, maternal vascular endothelial growth factor nition of FGR2; therefore, works aimed to employer(s)) 2019. Re-use gene therapy, nanoparticles, microRNA, statins, nitric study the risk of growth-related adverse fetal permitted under CC BY-NC. No oxide donors, hydrogen sulphide, proton pump inhibitors, outcome should focus on true FGR. Namely, commercial re-use. See rights melatonin, creatine and N-acetylcysteine, and insulin- on October 2, 2021 by guest. Protected copyright. and permissions. Published by SGA is usually defined by a fetal size <10th BMJ. like growth factors, compared between each other or to centile for a population or customised stan- 1 1Department of Neurosciences, placebo or no treatment. Primary efficacy outcome is FGR. dard. However, this definition includes also Psychology, Drug Research Secondary efficacy outcomes will be preterm birth, fetal or a proportion of constitutionally small but and Child Health, University of neonatal death and neonatal complications. For the safety health fetuses while excluding a group of Florence, Florence, Italy outcome, all adverse events reported in included studies 2 fetuses with biometry >10th percentile but Department of Health Sciences, and experienced by either mothers, fetuses or newborns San Paolo Hospital Medical not meeting their own growth potential. On will be considered. Two review authors will independently School, Università degli Studi di the other hand, the concept of FGR applies Milano, Milano, Italy screen title, abstract and full paper text, and will only to non-healthy fetuses that failed to 3 Department of Health Sciences, independently extract data from included studies. Where reach their growth potential and thus are at University of Florence, Florence, possible and appropriate, for primary and secondary increased risk of adverse outcomes.2 Italy efficacy outcomes, a NetMA will be performed using a 4CiaoLapo - Charity for Healthy According to the definition identified random-effects model within a frequentist framework. 2 Pregnancy, Stillbirth and through a Delphi consensus procedure, early Adverse events will be narratively described. Perinatal Loss Support, Prato, FGR (ie, at gestational age <32 weeks), in the Ethics and dissemination Results will be disseminated Italy absence of congenital anomalies, is defined through a peer-reviewed scientific journal, and by scientific by fetal abdominal circumference (AC)/ Correspondence to congresses and meetings. estimated fetal weight (EFW) <3rd centile or Dr Alfredo Vannacci; PROSPERO registration number CRD42019122831. alfredo.​ ​vannacci@unifi.​ ​it umbilical artery (UA) absent end-diastolic

Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467 1 Open access flow, or by the co-presence of AC (EFW <10th centile meta-analysis (NetMA) of clinical trials and observational BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from and uterine artery pulsatility index (PI) >95th centile, studies conducted on gestating women at diagnosed risk alone or in combination with UA-PI >95th centile. On the of FGR, with the primary objective of estimating the effect other hand, late FGR (ie, at gestational age of 32 weeks of different pharmacological treatments on the incidence of more), is defined, in the absence of congenital anom- of FGR. alies, by AC/EFW <3rd centile or by the co-presence of at Secondary efficacy objectives include the evaluation of least two parameters among (i) AC/EFW <10th centile, the effect of different therapeutic strategies on the inci- (ii) AC/EFW crossing centiles >2 quartiles on growth dence of (i) preterm birth; (ii) fetal or neonatal death centiles or (iii) cerebroplacental ratio <5th centile or and (iii) neonatal complications related to the abnormal UA-PI >95th centile. growth. Safety objective includes the evaluation of adverse Suboptimal fetal growth accounts for 28%–45% of cases events involving treated women, fetuses or newborns. of non-anomalous stillbirth.3 4 Causes of FGR can be fetus-related or mother-related: among the former, there are chromosomal anomalies, genetic syndromes and infections. Among maternal Methods and analysis conditions, the most common are clinically relevant This protocol has been written according to the Preferred Reporting Items for Systematic Review and Meta-Analysis conditions such as autoimmune disorders, clinically rele- 15 vant conditions such as autoimmune disorders, hypox- Protocols guidance. emic conditions (such as severe anaemia, congenital cyanotic heart diseases), cardiovascular diseases (such as Eligibility criteria hypertension) or the exposition to environmental toxins. Studies will be selected according to the eligibility criteria The mechanism leading to FGR involves an abnormal outlined below. trophoblast invasion of the maternal spiral arteries during pregnancy, which results in an incomplete remodelling of Study designs these vessels and in the persistence of a high-resistance We will consider for inclusion both clinical trials and and low-flow uteroplacental circulation, which on its observational cohort studies, either prospective or retro- turn determines and insufficient gaseous and nutrient spective. Observational cross-sectional studies or case– exchange for optimal fetal growth.5 control studies will be excluded. Similarly, we will exclude This results in a cascade of events, including reduced reviews and meta-analyses, letters to the editor, case placental perfusion, imbalance in angiogenic factors, reports, case series and expert opinions. and in vascular endothelial growth factor (VEGF) and placental growth factor, and may lead to placen- Participants ta-mediated complications of pregnancy such as FGR, We will include studies performed on gestating women at http://bmjopen.bmj.com/ preeclampsia, placental abruption and late pregnancy diagnosed risk of FGR, including16 the following: 6 7 loss. ►► History of late pregnancy loss or FGR. The management of FGR is based on the prolongation ►► History of recurrent pregnancy loss (define by three of pregnancy long enough for fetal organs to mature or more consecutive first trimester spontaneous while avoiding irreversible fetus’ sufferance.8 As for phar- abortions). macological management, most guidelines recommend ►► Hypertensive diseases. treatment with low-dose aspirin—in preference by gesta- ►► Pre-eclampsia (in the current or previous tional week 16—for prevention of FGR,9–12 although pregnancies). on October 2, 2021 by guest. Protected copyright. 13 this approach is not universally accepted. The use of ►► Diabetes mellitus. heparin is also controversial: the Canadian guideline ►► Congenital cyanotic heart diseases. recommends that heparin should be offered in selected ►► Restrictive pneumopathies. 9 women, although recent evidence indicates that enoxa- ►► Severe renal diseases. parin is not effective in preventing FGR in women with ►► Autoimmune diseases. previous severe or early-onset FGR, or in those with ►► Hereditary or acquired thrombophilia. thrombophilia, and can therefore not be recommended ►► Severe anaemia. for this purpose.1 14 Furthermore, several other promising Other risk factors for FGR reported by the clinicians as therapies are currently under investigation.1 present in the cohort of women in the retrieved studies To date, given that no firm evidence exists to guide clini- will be evaluated case by case for possible inclusion. cians towards the most effective therapeutic intervention, No restriction on maternal age will be applied. planned early birth is recommended and offered once a We will include only studies performed on women fetus reaches a viable gestational age and size. However, without or not carrying fetuses with congenital anoma- preterm birth is further associated with a consistent risk lies, including abnormal karyotype and/or genetic anom- of morbidity and mortality.1 alies, and malformations. We will also include only studies In light of the above-mentioned lack in knowledge, performed on women without intrauterine infections we aim to perform a systematic review and network and/or without history of drug or alcohol abuse.

2 Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467 Open access BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from No other restriction on maternal clinical conditions ►► Neonatal complications, including necrotising enter- will be applied. ocolitis, respiratory distress syndrome, bronchopul- monary dysplasia, intraventricular haemorrhage, Interventions retinopathy of prematurity, periventricular leucoma- Based on current experimental treatments for FGR,1 we lakia and other conditions related to the abnormal will consider the following therapeutic interventions: growth and reported in the considered studies. ►► Heparin or low-molecular-weight heparin. In studies reporting the primary outcome, also data ►► Acetylsalicylic acid. related to placental lesions (ie, fetal and maternal vascular 18 ►► Other antiplatelet agents (including ditazole, cloric- malperfusion) will be retrieved. romen, clopidogrel, ticlopidine, dipyridamole, carba- As for safety outcomes, we will consider any side effect salate, epoprostenol, indobufen, iloprost, aloxiprin, experienced by treated women or by fetuses or newborns eptifibatide, tirofiban, triflusal, beraprost, trepros- in included studies; side effects will be defined based on tinil, prasugrel, cilostazol, ticagrelor, cangrelor, vora- authors’ definitions. paxar and selexipag). In addition, in studies reporting the primary outcome, ►► Phosphodiesterase type 5 inhibitors. all maternal adverse events (such as pre-eclampsia, ►► Phosphodiesterase type 3 inhibitors. placental abruption, etc) will be evaluated. ►► Maternal VEGF gene therapy. ►► Nanoparticles. Timing ►► MicroRNA. There will be no timing restriction. ►► Statins. Setting ►► Nitric oxide donors. ►► Hydrogen sulfide. There will be no restriction by type of setting. ►► Proton pump inhibitors. Language ►► Melatonin. We will include only articles written in the English ►► Creatine. language. ►► N-acetylcysteine. ►► Insulin-like growth factor 1 and 2 (IGF1 and IGF2). Information sources and search strategy Studies in which patients were co-treated with more Electronic searches will be performed in the databases than one above-mentioned treatment will be included, as MEDLINE and Embase. well. The MEDLINE search strategy is reported in online Additional pharmacological interventions, not listed supplementary material 1. above and detected by screening of retrieved references The MEDLINE search strategy will be adapted to the or in the bibliographies of evaluated studies, will be syntax and subject headings of the Embase. http://bmjopen.bmj.com/ further considered; if pertinent, search strategies will be Records will be retrieved on the same day from all updated and related references will be retrieved. sources. There will be no restriction in the pregnancy week of The search strategy will be updated towards the beginning of the therapeutic intervention. end of the review, after being validated to ensure that the MEDLINE strategy retrieves a high proportion of Comparators eligible studies found through any means and indexed We will consider studies comparing the effect of the in MEDLINE. above-mentioned interventions versus placebo or no on October 2, 2021 by guest. Protected copyright. treatment (considered together), or versus active control, Study records that is, a second treatment listed above. Data management Retrieved records will be managed using the software Outcomes Endnote. We will include only studies evaluating the primary effi- cacy outcome, that is, FGR. As the definition of FGR is Selection process based on biometric measures not always reported in Two review authors will independently screen the details in published studies, we will accept the diagnosis extracted records. The two review authors will inde- of FGR provided by the authors of the studies. pendently identify studies for inclusion by screening In studies evaluating the above-mentioned primary titles and abstracts yielded by search, eliminating those outcome, also the following secondary efficacy outcomes deemed irrelevant. We will retrieve full-text articles for will be considered (when reported): all references that at least one of the review authors will ►► Preterm birth, defined as a delivery before completing identify for potential inclusion. We will select studies for 37 weeks of gestation. inclusion on the basis of review of full-text articles. We will ►► Fetal or neonatal death, including the events related resolve discrepancies through discussion. to early or late pregnancy loss, and perinatal and early Neither of the review authors will be blind to the journal neonatal death.17 titles or to the study authors or institutions.

Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467 3 Open access

Data collection number of women experiencing any event out of the total BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from Two review authors will independently extract data from number of treated patients. the included studies. Data abstracted will include demographic information, Risk of bias methodology, intervention details, all reported clinically Two review authors will independently assess the included relevant conditions and outcomes. Data will be extracted studies for bias. To assess the risk of bias of included at the trial arm level, rather than the individual patient randomised controlled trials, we will follow the Cochrane level. We will resolve discrepancies between authors Handbook for Systematic Reviews of Interventions.20 through discussion. Specifically, we will assess the risk of bias in included trials for the following domains: selection (random sequence Data items generation; allocation concealment); performance Extracted data will include the name of the study authors (blinding of participants and personnel); detection and year of publication, the study design and character- (blinding of outcome); attrition (incomplete outcome istics (including single or double blinding and randomis- data); reporting (selective reporting); other unclear bias. To assess the risk of bias of observational studies, we will ation), the country in which participants were recruited, 21 and eventual funding sources. follow the Newcastle-Ottawa Quality Assessment Scale. As for the population, we will extract the women’s age, Specifically, for included cohort studies, we will consider the specific risk factors for FGR and other clinically rele- the following domains: selection (representativeness of vant comorbidities. the exposed cohort; selection of the non-exposed cohort; As for the intervention and the comparator, we will ascertainment of exposure; absence of outcome of interest at start of study); comparability; outcome (assessment of extract the active principle of the experimental interven- outcome; appropriate length of follow-up; adequacy of tion, its route of administration, the treatment dosage, follow-up of cohorts). the gestation week at time of treatment beginning and For each domain in the two tools, we will describe the duration of treatment. the procedures undertaken for each study, including We will extract the number of randomised participants, verbatim quotes. A judgement as to the possible risk of the number of participants included in the analysis, the bias on each domain will be made from the extracted number of participants with events for binary outcomes information, rating from ‘low risk’ to ‘high risk’. and the reported definition of outcomes, if appropriate. The judgements will be made independently by two Whenever possible, we will use results from an intention- review authors; disagreement will be resolved first by to-treat analysis. discussion and then by consulting a third author. We will compute graphic representations of potential

Outcomes and prioritisation bias within included studies, using the software RevMan http://bmjopen.bmj.com/ The primary outcomes will be the number of women V.5.3 (Review Manager 5.3). experiencing FGR out of the total number of treated patients. Data synthesis The secondary outcomes will be the number of cases of If studies are sufficiently homogeneous in terms of design preterm birth and of fetal or neonatal death, out of the and comparator, we will synthesise results using a NetMA, total number of treated patients. For neonatal compli- thus allowing to simultaneously combine evidence from cations, we will consider the number of newborns with both direct head-to-head comparisons and indirect a given complication out of the total number of treated evidence, that is, interventions compared through a on October 2, 2021 by guest. Protected copyright. cases. common comparator.22 23 For efficacy outcomes, where Odds Ratio (ORs) or relative risks and related Confidence Intervals (CIs) are Measures of treatment effect reported, these will be transformed to absolute numbers. All considered outcomes are based on dichotomous data. Placental characteristics will be considered out of the For all considered efficacy outcomes, we will perform a total number of cases for whom placental evaluation was NetMA using a random-effects model within a frequen- available, focusing on villous branching abnormalities tist framework. We will calculate pooled ORs combining and maternal decidual arteriopathy. the estimates reported in each study using random-effects Safety outcomes will be the number of patients (either Mantel-Haenszel method. mothers, fetuses or newborns) experiencing any adverse For safety outcomes, no quantitative synthesis will event out of the total number of treated patients. Adverse be performed, and the proportions of each reported events will be identified based on specific authors’ defi- adverse event will be described at study level. Similarly, all nitions, and will be classified using the Medical Dictio- reported maternal adverse events will be described, and nary for Regulatory Activities (MedDRA) classification, no quantitative analysis will be performed. according to preferred terms and system organ class For placental lesions, data from included studies classification.19 Similarly, maternal adverse events (such will be narratively synthetised according to the type of as pre-eclampsia, placental abruption, etc) will be the described lesions (lesions related to maternal vascular

4 Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467 Open access underperfusion versus fetal vascular underperfusion), Patient and public involvement BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from and no quantitative analysis will be performed. The project for this systematic review and NetMA was conceived within the frame of the CiaoLapo Onlus—a Unit of analysis issues charity for support to stillbirth and perinatal loss in All analysis will be conducted per trial arm, rather than at response to an unmet need suggested by healthcare individual patient level. professionals and patients.

Dealing with missing data Ethics and dissemination When there are missing data, we will attempt to contact There is no primary data collection involved in this study, original study authors to obtain the relevant missing thus research ethics approval is not required. information. Important numerical data will be carefully Results will be disseminated by release of findings in evaluated. If missing data cannot be obtained, the study a peer-reviewed scientific journal, and by abstracts and will be excluded from the related analysis. speeches at international meetings and congresses.

Assessment of heterogeneity Contributors AB, NL and GC designed the study, drafted the PROSPERO protocol We will evaluate the clinical heterogeneity by considering and the manuscript. LA, AM, CR and AV contributed to study design and provided the variability in participants’ features among studies critical revisions on the manuscript. As this is a protocol paper, the research has not yet been conducted, no data have been acquired or interpreted. and in study characteristics (study design, intervention, follow-up). Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. We will evaluate statistical heterogeneity across studies using the I-squared and Cochran’s Q tests, and publica- Competing interests None declared. tion bias using plots of SE against effect estimate (bias is Patient consent for publication Not required. likely to cause asymmetry in such plots), or using formal Provenance and peer review Not commissioned; externally peer reviewed. tests such as Egger one or similar. Open access This is an open access article distributed in accordance with the If high levels of heterogeneity exist (I-squared ≥50% or Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which p<0.1), we will try to explain the source of heterogeneity permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is by conducting subgroup or sensitivity analysis. properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://​creativecommons.org/​ ​licenses/by-​ ​nc/4.​ ​0/. Subgroup and sensitivity analyses Subgroup analysis will be conducted for timing of inter- vention (within vs after the first trimester of pregnancy). References 1. Groom KM, David AL. The role of aspirin, heparin, and other Additional subgroup analysis will be performed, if appro- http://bmjopen.bmj.com/ interventions in the prevention and treatment of fetal growth priate, according to the clinical characteristics of patients restriction. Am J Obstet Gynecol 2018;218:S829–40. in included studies. 2. Gordijn SJ, Beune IM, Thilaganathan B, et al. Consensus definition of fetal growth restriction: a Delphi procedure. Ultrasound Obstet We will conduct a sensitivity analysis including only clin- Gynecol 2016;48:333–9. ical trials versus only observational studies. If possible, a 3. McCowan LME, Thompson JMD, Cronin RS, et al. Going to sleep second sensitivity analysis will be performed including in the supine position is a modifiable risk factor for late pregnancy stillbirth; Findings from the New Zealand multicentre stillbirth case- only high-quality clinical trials. control study. PLoS One 2017;12:e0179396. 4. Gardosi J, Kady SM, McGeown P, et al. Classification of stillbirth by relevant condition at death (ReCoDe): population based cohort study. Meta-biases BMJ 2005;331:1113–7. on October 2, 2021 by guest. Protected copyright. To determine whether reporting bias is present in 5. Brosens JJ, Pijnenborg R, Brosens IA. The myometrial junctional zone spiral arteries in normal and abnormal pregnancies: a review of included clinical trials, we will evaluate whether the the literature. Am J Obstet Gynecol 2002;187:1416–23. protocol of the clinical trial was published before recruit- 6. Levine RJ, Lam C, Qian C, et al. Soluble endoglin and other ment of study patients. Specifically, for studies published circulating antiangiogenic factors in preeclampsia. N Engl J Med 2006;355:992–1005. after July 2005, we will screen the Clinical Trial Register 7. Lyall F, Robson SC, Bulmer JN. Spiral artery remodeling and at ClinicalT​ rials.gov​ . We will evaluate whether selective trophoblast invasion in preeclampsia and fetal growth restriction: relationship to clinical outcome. Hypertension 2013;62:1046–54. reporting of outcomes is present (outcome reporting 8. Sakamoto M, Osato K, Kubo M, et al. Early-onset fetal growth bias). The potential for reporting bias will be evaluated restriction treated with the long-acting phosphodiesterase-5 inhibitor tadalafil: a case report. J Med Case Rep 2016;10:317. using funnel plots (if ≥10 studies are present). 9. Lausman A, Kingdom J. Maternal Fetal Medicine Committee. Intrauterine growth restriction: screening, diagnosis, and management. J Obstet Gynaecol Can 2013;35:741–8. Confidence in cumulative estimate 10. "Institute of Obstetricians and Gynaecologists, Royal College of The quality of evidence will be judged using the Grading Physicians I, ‘Directorate of Clinical Strategy and Health Service of Recommendations Assessment, Development and Eval- Executive Programmes’. Fetal growth restriction - recognition, diagnosis & management. 2017. uation working group scale, considering the domains of 11. ‘Royal College of Obstetricians and Gynaecologists’. The risk of bias, consistency, directness, precision and publica- Investigation and Management of the Small–for–Gestational–Age Fetus. 2014. tion bias. Quality will be adjudicated as high, moderate, 12. Vayssière C, Sentilhes L, Ego A, et al. Fetal growth restriction and 20 low or very low. intra-uterine growth restriction: guidelines for clinical practice from

Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467 5 Open access

the French College of Gynaecologists and Obstetricians. Eur J 18. Khong TY, Mooney EE, Ariel I, et al. Sampling and Definitions BMJ Open: first published as 10.1136/bmjopen-2019-029467 on 26 July 2019. Downloaded from Obstet Gynecol Reprod Biol 2015;193:10–18. of Placental Lesions: Amsterdam Placental Workshop Group 13. ACOG. ACOG Committee Opinion No. 743: Low-Dose Aspirin Use Consensus Statement. Arch Pathol Lab Med 2016;140:698–713. During Pregnancy. Obstet Gynecol 2018;132:e44–52. 19. MedDRA classification. https://www.​meddra.​org/​how-​to-​use/​basics/​ 14. Rodger MA, Gris JC, de Vries JIP, et al. Low-molecular-weight hierarchy heparin and recurrent placenta-mediated pregnancy complications: a 20. Higgins JPT, Altman DG, Gotzsche PC, et al. The Cochrane meta-analysis of individual patient data from randomised controlled Collaboration's tool for assessing risk of bias in randomised trials. trials. Lancet 2016;388:2629–41. BMJ 2011;343:d5928. 21. Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale 15. Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for (NOS) for assessing the quality of nonrandomized studies in meta- systematic review and meta-analysis protocols (PRISMA-P) 2015: analyses. Ottawa Hosp Res Inst 2013. elaboration and explanation. BMJ 2015;349:g7647. 22. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of 16. Nardozza LM, Caetano AC, Zamarian AC, et al. Fetal multiple treatments: combining direct and indirect evidence. BMJ growth restriction: current knowledge. Arch Gynecol Obstet 2005;331:897–900. 2017;295:1061–77. 23. Salanti G. Indirect and mixed-treatment comparison, network, or 17. Lawn JE, Blencowe H, Pattinson R, et al. Stillbirths: multiple-treatments meta-analysis: many names, many benefits, Where? When? Why? How to make the data count? Lancet many concerns for the next generation evidence synthesis tool. Res 2011;377:1448–63. Synth Methods 2012;3:80–97. http://bmjopen.bmj.com/ on October 2, 2021 by guest. Protected copyright.

6 Bettiol A, et al. BMJ Open 2019;9:e029467. doi:10.1136/bmjopen-2019-029467