2.1. Protocol and registration

This paper presents a protocol for a systematic review and meta-analysis. The synthesis follows the guidelines outlined in the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 and is registered with PROSPERO under registration number CRD42023464826. The review will be based on existing scholarly literature and does not require ethical approval. The results will be presented in subsequent publications.

2.2. Inclusion criteria

The selection criteria for this research follow the PICOS structure (population, intervention/exposure, comparator, outcomes, study design) [9]. Studies fulfilling these criteria will be encompassed, with no limitations pertaining to language or the year of publication.

2.3. Exclusion criteria

Investigations focusing on populations affected by ischemic stroke, spontaneous intracerebral hemorrhage, chronic subdural hematoma, or other types of non-traumatic cerebral hemorrhages will be excluded.

Research papers that lack comprehensive clinical data, even after diligent efforts to reach out to the respective authors, will not be included.

Exclusion will apply to correspondence, conference summaries, editorial pieces, case studies, review articles, and any studies lacking clinical data.

Accessible full-text scholarly works that cannot be located through thorough search efforts will not be considered for inclusion.

2.4. Study selection and search strategy

An extensive literature search will be performed across multiple databases, including PubMed, Embase, Web of Science, The Cochrane Library, CNKI (China National Knowledge Infrastructure), VIP (WeiPu), SinoMed, and Wanfang, covering all available records from the inception of each database to the current date. The search strategy has been carefully designed with input from a professional librarian and information specialist to ensure comprehensiveness and precision. It is informed by existing literature and incorporates Medical Subject Headings (MeSH) terms, keywords, and synonyms relevant to the research topic.

Search terms will include variations of “traumatic brain injury,” such as “traumatic cerebral contusions,” “TBI,” “head injury,” “cerebral blunt trauma,” and “diffuse axonal injury,” combined with terms like “adverse prognosis,” “unfavorable outcome,” “risk factors,” “predictors,” “impact factors,” “determinants,” and “predictive indicators.” To maximize retrieval, Boolean operators (AND/OR) and database-specific controlled vocabulary will be applied. For instance, MeSH terms will be used in PubMed, and EMTREE terms in Embase, supplemented by title and abstract keywords tailored to the syntax of each database.

To ensure thoroughness, the reference lists of included studies will be screened for additional articles, and gray literature will be identified via OpenGrey. Additionally, clinical trial registries, including ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP), will be searched to identify ongoing or unpublished randomized controlled trials (RCTs). Authors of ongoing studies will be contacted directly to gather the most recent findings. These steps aim to reduce the risk of publication bias and to capture a comprehensive scope of relevant research.

2.5. Search terms for PubMed

#1: ((traumatic brain injury[MeSH Terms]) OR (traumatic brain injuries[Title/Abstract])) OR (TBI[Title/Abstract])) OR (brain injuries, traumatic(Title/Abstract])) OR (head injury[Title/Abstract])) OR (cerebral contusion(Title/Abstract])) OR (brain contusion[Title/Abstract])) OR (closed head injury[Title/Abstract]) OR (diffuse axonal injury[Title/Abstract]))

#2: ((((prognosis[MeSH Terms]) OR (outcome[Title/Abstract])) OR (mortality[Title/Abstract])) OR (recovery[Title/Abstract])) OR (functional outcome[Title/Abstract])) OR (neurological recovery[Title/Abstract])) OR (cognitive outcome[Title/Abstract])) OR (disability[Title/Abstract])) OR (quality of life[Title/Abstract])

#3: ((((risk factors[MeSH Terms]) OR (predictors[Title/Abstract])) OR (determinants[Title/Abstract])) OR (factors[Title/Abstract])) OR (clinical predictors[Title/Abstract])) OR (biomarkers[Title/Abstract])) OR (severity score[Title/Abstract])

#4: #1 AND #2 AND #3

2.6. Literatures collection and organization

All identified literature will be imported into the NoteExpress reference management tool. After duplicates are removed, Zhaoyin Su and Yao Liu, as the first and second reviewers, will conduct article selection based on the previously established criteria. The process will begin with an initial screening of titles and abstracts to assess relevance. Subsequently, both reviewers will independently examine the full texts of all potentially relevant studies. The article selection process is illustrated in Fig 1. In case of discrepancies between the two reviewers, a third expert will be consulted to reach a consensus.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart of literatures screening.

https://doi.org/10.1371/journal.pone.0319146.g001

2.7. Data extraction and management

Two other independent reviewers will employ preformatted Excel templates to extract a set of predetermined data. The extracted information will include study characteristics (such as the primary authors, journal, publication year, study design, blinding methods, and inclusion criteria), participant characteristics (e.g., diagnostic criteria for traumatic CC, age range, and gender ratio), exposure factors (e.g., biochemical marker values), outcome measures (e.g., mortality and GCS scores), as well as follow-up duration. Where quantitative data are not directly provided in the text or additional materials, Engauge Digitizer 11.1 will be used to extract data from graphical representations. For studies with incomplete data, the reviewers will reach out to the corresponding author via email to request the missing information, offering to acknowledge their assistance. If no response is received within two weeks of the initial inquiry, further attempts to contact the study’s researcher will be halted, and the study will be excluded from this systematic review and meta-analysis. Effect estimates for the outcomes of interest will be extracted for statistical analysis, with a preference for adjusted data that controls for confounding variables. To reduce bias and errors in the data extraction process, two researchers will independently extract data from the selected literature. Following extraction, a cross-verification process will take place, with any discrepancies resolved through discussion. If an agreement cannot be reached, a third reviewer will be involved to facilitate a resolution.

2.8. Data synthesis and statistical analysis

Statistical analyses and graphical representations will be executed utilizing STATA 16.0. One researcher will be responsible for entering the post-treatment mean values and standard deviations into the software, while a second researcher will cross-check these entries for correctness. The assessment of data heterogeneity will be performed via the chi-square and I^2 tests. In the event of non-significant heterogeneity (P ≥  0.10 or I^2 ≤  50%), the fixed-effect model will be chosen for the meta-analysis. On the other hand, if significant heterogeneity is detected (I^2 > 50% or P < 0.10), the random-effects model will be applied. To examine the impact of baseline attributes on the variability of primary outcomes, subgroup analyses will be carried out focusing on the following patient-related factors: (1) Glasgow Coma Scale scores of the patients, and (2) cerebral perfusion pressure indices. Additionally, subgroup analyses will be extended to different study design categories, complemented by meta-regression analyses, to explore potential origins of heterogeneity. We will strive to precisely identify and address heterogeneity exceeding 50%. In cases where heterogeneity is excessively high, a qualitative synthesis will be conducted. To ascertain the stability of our findings, a sensitivity analysis will be performed, removing studies one at a time to assess the impact on the overall results. A significance level of p < 0.05 will be denoted as statistically significant. Following this analytical process, a detailed examination and comparison of the outcomes will be undertaken to gauge the consistency of the study results. This rigorous evaluation is intended to confirm the dependability and veracity of our findings.

2.9. Assessment of risk of bias

A duo of investigators will independently apply the Cochrane Collaboration’s Risk of Bias tool [10] to gauge the potential bias in the randomized controlled trials (RCTs) incorporated into this review. The categorization of studies into high, low, or unclear risk of bias will be determined by examining various domains of bias, including: 1) selection bias, encompassing random sequence generation and allocation concealment; 2) performance bias, concerning the blinding of participants and outcome evaluators; 3) attrition bias, associated with incomplete outcome data; 4) reporting bias, pertaining to the selective reporting of results; and 5) other conceivable sources of bias. For non-randomized controlled trials, the methodological quality will be evaluated using the Newcastle-Ottawa Scale [11], which consists of three main domains: selection of participants, comparability of study groups, and assessment of outcomes. Each study will be assigned a score out of a possible nine stars, with scores ranging from 0 to 5 stars indicating low quality and 6 to 9 stars indicating high quality. Any conflicts in the evaluation process will be resolved through dialogue between the two assessors or, if required, by seeking the input of a third reviewer. The outcomes of these assessments will be compiled into a summary table. In cases where bias is deemed substantial, we may need to reconsider the literature gathered or potentially replace certain studies to enhance the synthesis of data.

The examination of reporting bias will be undertaken as necessary to corroborate the validity of the study findings. Should the review include more than ten studies, the assessment of reporting biases will involve the use of Egger’s test and a visual examination of the funnel plot, which plots the estimated effect size (on the horizontal axis) against its standard error (on the vertical axis). An asymmetrical funnel plot, particularly an inverted one, may suggest the presence of publication bias.