Protocol and registration
This systematic review was conducted following the methodological standards of Cochrane Handbook. [9] We prospectively registered the protocol in Open Science Platform. The full text of the protocol is available in an open access registry and as an online as Supplemental file 1. We reported the review using the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) standards. [10] The full checklist is available as Supplemental file 2.
Eligibility criteria
We included published randomized controlled trials that recruited women with HELLP syndrome, confirmed by objective testing. We included studies comparing corticosteroids versus placebo or no treatment. The primary outcome measure was maternal death. Other outcomes included acute pulmonary edema; acute renal failure; dialysis, liver morbidity (hematoma, ruptured liver, and failure), need for platelet transfusion, and perinatal death.
Information sources
A comprehensive literature search was initially conducted on September 20, 2023. We did not impose language or other restrictions on any of the searches. We searched bibliographic databases (Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE/PubMed) and citation indexes (Web of Science and Scopus). We included the terms (HELLP Syndrome) AND (corticosteroids or glucocorticoids or Dexamethasone or Betamethasone or Prednisolone). The detailed exact strategy adapted for each database is provided in Supplemental file 3 and is available as an open access registry document. We peer-reviewed the search strategy and further tested it with a set of known relevant, ‘gold standard’, reports. We also searched clinical trial registries (ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry Platform) to identify ongoing trials. We finally searched reference lists and explored the cited-by logs of identified studies and previously published reviews.
Study selection
All reports identified in the databases were imported to Bibtex library using Jabref version 5. After removing duplicates, two authors independently screened all titles and abstracts for eligibility. We retrieved and assessed the full text of all reports that potentially met our eligibility criteria during screening. Two authors (AFK, HBA, MAE, RHA) independently assessed each full-text article. Disagreements regarding trial eligibility was resolved by consensus and finally resolved by a third author (AFN).
Data collection process
For eligible studies, we extracted the data in duplicates using an offline electronic form. We resolved discrepancies through discussion. Extracted data were transcribed to a spreadsheet and checked for accuracy. We contacted authors of the original reports, if needed, to provide details regarding unclear or missing data.
Data items
Extracted data included study design, sample size, description of included participants, description of the intervention, outcomes, trial registration, and funding sources, and country.
Study risk of bias assessment
Two authors (AFK, HBA, MAE, RHA) independently used the Risk of Bias 2 (RoB 2) tool to assess the risk of bias of study results contributing information to each of the outcomes specified for inclusion in the Summary of Findings table.
We assessed the following risk of bias domains as outlined in Cochrane Handbook for Systematic Reviews of Interventions: 1) risk of bias arising from the randomization process; 2) risk of bias due to deviations from the intended interventions (effect of assignment to intervention); 3) risk of bias due to missing outcome data; 4) risk of bias in measurement of the outcome; and 5) risk of bias in selection of the reported result. Each domain was judged as being at “low risk of bias”, “some concerns”, or “high risk of bias”. Trials with “low risk of bias” in all domains were classified as being at overall “low risk of bias”. RCTs with one domain judged to be at “some concerns”, but no domain judged to be at “high risk of bias”, were classified as being at overall “some concerns” of risk of bias. RCTs were classified as being at overall “high risk of bias” if at least one domain was judged as being at “high risk of bias”. However, if a trial was judged to be at “some concerns” due to risk of bias for multiple domains, it was judged as being at overall “high risk of bias” if the assessors judged that the multiple concerns amounted to a serious risk of bias. In case of discrepancies among their judgments and inability to reach consensus, we consulted the senior author (AFN) to reach a final decision.
Effect measures
For dichotomous data, we presented results as summary risk ratio (RR) with 95% confidence intervals (CI). None of the outcomes of interest were meta-analyzed as a continuous variable. The unit of analysis was the individual participant. We used a complete case approach for analysis. Data related to participants reported as not compliant was analyzed on an intention-to-treat basis.
Synthesis methods
Fixed-effect meta-analysis was performed to combine data of trials that are judged to be sufficiently similar in terms of intervention, populations, and methods. We planned to investigate substantial statistical heterogeneity, defined as I² statistic ≥ 50% or P < 0.1.
We performed the planned subgroup analysis by gestational age at enrollment (ante- vs postpartum) and by type of corticosteroids. We assessed subgroup differences by interaction tests. Results of the subgroup analyses were reported by mentioning the Chi² statistic and P value, and the interaction test I² value.
Sensitivity analysis was performed to explore robustness of pooled estimate using outcome data from trials with a low risk of bias.
Synthesis was performed using RStudio 2023.06.1 Build 524 (MacOS, Apple Silicon version), R 4.3.1 (2023-06-16) [11] and R package meta version 6.5. [12]
Reporting bias assessment
We explored whether the study was included in a trial registry and whether a protocol was available. We planned to examine funnel plots to assess the potential for publication bias if we found 10 or more studies reporting on a particular outcome.
Certainty assessment
We used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach to create the Summary of Findings table. [13] Briefly, GRADE uses study limitations, consistency of effect, imprecision, indirectness, and publication bias to assess the certainty of evidence for each outcome. A summary of the intervention effect and a measure of certainty was produced using the GRADE Profiler Guideline Development Tool (GRADEpro GDT) software [14] for the prespecified important outcomes: maternal death, pulmonary edema, renal failure, dialysis, liver morbidity, need for platelet transfusion, and perinatal death. One author (A.N.) conducted GRADE assessments and the decisions on downgrading. This was discussed for final approval by all authors.
Patient and public involvement
We consulted with patients by asking a group to review and comment on an early draft of the manuscript.