Session 9 - Literature Searches and Systematic Literature Reviews PDF

Session 9: 8th Nov Created @September 24, 2024 3:23 PM Tags Literature Searches and Systematic Literature Reviews: Lead Teachers: Sarah Rowe To be covered: Principles and Processes of Systematic Reviews....

Session 9: 8th Nov Created @September 24, 2024 3:23 PM Tags Literature Searches and Systematic Literature Reviews: Lead Teachers: Sarah Rowe To be covered: Principles and Processes of Systematic Reviews. Reading list: PSBS0002: Core Principles of Mental Health Research | University College London (talis.com) Class Prep: Additional: Read: A 24-Step Guide on How to Design, Conduct, and Successfully Publish a Systematic Review and Meta-Analysis in Medical Research Webpage: Maudsley Debates Read: A Systematic Review and Meta-Analysis of Randomised Controlled Trials of Peer Support for People with Severe Mental Illness Webpage: LibGuides at UCL Read: 5 Key Things to Know About Meta-Analysis Read: Guidance Notes for Registering a Systematic Review Protocol with PROSPERO Webpage: PROSPERO Website: The Centre for Evidence-Based Medicine Website: The Cochrane Collaboration Website Session 9: 8th Nov 1  Webpage: University of York: Centre for Reviews and Dissemination Key: How to do a Systematic Review Planning the Review Define the Research Question Clearly specify the research question, including the population, interventions, comparisons, outcomes, and study designs of interest (PICOS framework). Develop a Protocol Create a detailed protocol outlining the methods for the review, including: Inclusion and exclusion criteria Search strategy Data extraction process Quality assessment approach Analysis plan Register the protocol with a database like PROSPERO to enhance transparency. Conducting the Review Literature Search Search multiple electronic databases (e.g., PubMed, PsycINFO, Cochrane Library) Use a comprehensive search strategy with relevant keywords and MeSH terms Include grey literature sources Document the search process thoroughly Study Selection Screen titles and abstracts independently by at least two reviewers Assess full texts of potentially eligible studies Resolve disagreements through discussion or involvement of a third reviewer Document reasons for exclusion Data Extraction Session 9: 8th Nov 2  Use a standardized form to extract relevant data from included studies Extract information on study characteristics, participants, interventions, outcomes, and results Have two reviewers independently extract data to ensure accuracy Quality Assessment Assess the methodological quality of included studies using appropriate tools (e.g., Cochrane Risk of Bias tool for RCTs) Evaluate factors such as selection bias, performance bias, detection bias, and reporting bias Analysing and Reporting Results Data Synthesis Conduct a narrative synthesis of the findings If appropriate, perform meta-analyses to pool quantitative results Assess heterogeneity between studies Consider subgroup analyses or meta-regression to explore sources of heterogeneity Interpret Findings Summarize the main results Discuss the strength of evidence and limitations of the review Consider implications for practice and future research Reporting Follow the PRISMA guidelines for reporting systematic reviews Clearly describe the methods, results, and conclusions Include a flow diagram of the study selection process Present results in tables and figures where appropriate Additional Considerations Be aware of potential biases, such as publication bias Session 9: 8th Nov 3  Consider involving stakeholders, including patients and clinicians, in the review process Update the review periodically to incorporate new evidence By following these steps, researchers can conduct a comprehensive and rigorous systematic review in mental health, synthesizing the available evidence to inform clinical practice and policy decisions. Key: PRISMA The PRISMA 2020 statement is an updated guideline for reporting systematic reviews, designed to improve transparency and completeness in systematic review reporting. Here are the key points: Purpose and Scope PRISMA 2020 replaces the 2009 statement, reflecting advances in systematic review methodology. It is primarily designed for systematic reviews evaluating health interventions but is applicable to other types of reviews as well. Key Components 27-item checklist detailing reporting recommendations Expanded checklist providing more detailed guidance for each item Abstract checklist for concise reporting in review abstracts Revised flow diagrams for original and updated reviews Main Updates New guidance on reporting search strategies, study selection, and data analysis methods Modified structure and presentation to facilitate implementation Expanded options for reporting different types of searches (e.g., databases, registers, grey literature) Intended Benefits Session 9: 8th Nov 4  Improved transparency in reporting why the review was conducted, what methods were used, and what was found Enhanced ability for readers to assess the conduct and findings of systematic reviews Facilitation of evidence-based decision-making in healthcare Implementation Authors are encouraged to use the explanation and elaboration document alongside the checklist The guideline is applicable to various types of systematic reviews, including those with and without meta-analyses PRISMA 2020 represents a significant update to systematic review reporting standards, aiming to improve the quality and usefulness of published reviews across various fields of research. Key: Pre-Lecture Literature Review: A summary and synthesis of previously published academic work on a research question, usually including: - Details of primary studies through a literature search on the topic - Synthesis of the findings of these studies - Appraisal of the quality of the studies - Commentary on the current state of the evidence on the relevant topic Systematic Review: Reports a comprehensive and reproducible search. Exhaustive steps to trace all relevant evidence Session 9: 8th Nov 5  Developed as a means of synthesising randomised controlled trial evidence Now systematic review not just of RCTs → all types of data, incl. recently qualitative Has a methods section which allows full replication of review methods Meta-Analysis Statistical pooling of results from different, but similar results Often results in much great statistical power than original studies Modern meta-analyses → assess both the pooled result and its significance, and the extent of variation between the studies Central to clinical guideline → e.g. NICE on healthcare interventions that should be available through UK Notes: Literature Review Narrative vs systematic review: Narrative review = more subjective → no inclusion/exclusion criteria Possible bias → search for data that supports argument Session 9: 8th Nov 6  Systematic review = include a quality review → not really done in a narrative review (more of a summary of studies findings) Advantages of Systematic Reviews and Meta-Analyses Comprehensive data search and systematic extraction avoids: Subjectivity Significant evidence being missed When data can be aggregated, pooling evidence makes overall finding clearer (meta- analysis) Clear positive finding from substantial and good quality meta-analysis can be seen as good support for a treatment Limitations of Systematic Reviews Studies included may vary greatly in time, culture, type of intervention: Publication bias: E.g. Cochrane review of crisis Negative findings less often intervention for severely ill: published Publication dates range from Key stakeholders have an interest 1964 to 2010; in not publishing Experimental conditions range Around 50% trials currently not from weekly visits to twice published daily intensive support or Initiatives to increase publication: crisis house; trial registries, requirements to Control conditions range from publish within a year asylum to community Data invisibility even higher for Negative conclusions may mean: other types of study Treatment doesn’t work or Restricted search strategies, especially regarding language. Evidence doesn’t show an effect due to small overall numbers or Positive findings more likely poor quality of studies to be published in English. Session 9: 8th Nov 7  Quickly out of date trials of complex mental health interventions often Analyses that are wrong or even rated low quality fraudulent, data manipulation Important effects may not be Trial registration & making data measured e.g. companionship from available – part of Open Science peer support Designing Systematic Review 4. Specify sources to be searched and 1. Define a review question – as precise language and date ranges e.g.: as possible. a. Bibliographical databases 2. Consider PRISMA guidelines for (Pubmed, PsychInfo, Web of reporting and consider talking to a Knowledge etc.) librarian/information technologist. b. Grey literature (e.g. theses, a. Preferred reporting method for government reports, charity systematic review/meta-analysis reports) 3. Define inclusion and exclusion criteria c. Research registries, searching for selecting studies for the review, reference lists, contact with considering PICOS: authors a. Population: which types of 5. Design search strategy research participant will be included? 6. Decide on process for screening papers, usually: i. e.g. age or diagnostic groups a. Titles and abstracts first b. Intervention (where relevant): which types of experimental i. 1,000s of papers often intervention will be included? retrieved by search, many of limited relevance c. Comparator: which types of control group? b. Then full papers for potentially relevant studies d. Outcomes: which types of outcome? i. Does it meet inclusion criteria? Session 9: 8th Nov 8  e. Study design: which types of c. Multiple raters if possible – second study design? ratings at each stage for at least a proportion a. randomised controlled trials only? Any quality or size 7. Specify what data/variables will be criteria? extracted and write data extraction forms a. Describing the paper, where it’s done, how many people used, study design, etc. b. Related to information needed to answer review question i. e.g. effect sizes, themes → findings 8. Decide how to store and process data a. e.g. Endnote or other bibliographical software, specialist systematic review software 9. Decide how to synthesise studies: a. Is meta-analysis possible? i. consider similarity of studies, quality of statistical reporting? b. How will you summarise papers found if not? c. Will a quality rating tool be used? 10. Register protocol – PROSPERO: International prospective registry of systematic reviews Session 9: 8th Nov 9  📎 Example: PICOS criteria for a review of crisis house care Objective: To investigate the clinical effectiveness of crisis houses in managing relapses among people with psychosis. The following will be included: Population: Studies with participants (1) aged over 18 (2) with a diagnosis of psychosis, or with mixed populations and separate reporting of psychosis outcomes. Interventions: Studies of short term residential care in a non- hospital site with staff available 24 hours a day and a maximum stay under 2 months. Comparators: Studies with any control condition. Outcomes: Studies reporting as outcomes symptoms, quality of life, social functioning or any other validated clinical measure Study design: Randomised controlled trials. Key Sources for Systematic Reviews Centre for Reviews and Dissemination, York – Prospero registration Session 9: 8th Nov 10  Cochrane database of systematic reviews – largest source of reviews in healthcare, standard procedures for all aspects of review. Campbell collaboration – reviews relevant to public policy Centre for Evidence Based Medicine, Oxford Beyond Systematic Reviews As well as classical systematic reviews, which are time-consuming and do not always fit a research question, other methods have been developing, all also based on following a pre- specified systematic method. Some examples: Scoping review: a broad review aimed at mapping out what the literature in a particular area looks like → to identify knowledge gaps, more broad and less rigorous than a systematic review e.g. Using technology to deliver mental health services to children and youth: a scoping review Qualitative reviews: methods now developed for synthesis of qualitative literature e.g. Patients' experiences of assessment and detention under mental health legislation: qualitative meta-synthesis Umbrella review: a systematic review of review, good for getting a handle on a large body of literature on a question quickly → a review of reviews, to give it a high level review → more commonly conducted when competing interventions for a condition e.g. Preventive interventions for psychosis: umbrella review Notes: Developing a Search Strategy First Principles The research question informs the search strategy: Choice of databases Choice of search terms Search limits (language, time period, study type) Be over-inclusive/broad rather than under-inclusive in your search. Be clear what you have done Session 9: 8th Nov 11  Someone else should be able to repeat your search Prisma Guidelines Prisma provides internationally recognised standards for conducting and reporting systematic reviews http://www.prisma- statement.org/statement.htm The Prisma Checklist is a helpful prompt re decisions you need to make for your systematic review, including: Research question Study eligibility criteria (PICOS) Information sources (which databases) Search terms Study selection (screening/decision- making process) Data extraction process Risk of bias/quality assessment Session 9: 8th Nov 12  Reviewers will be expected to present a chart like this, so need to keep track of numbers and stages of search throughout Information Sources UCL library services provide access to a wide range of bibliographic databases Bibliographic databases are the best way to search for peer-reviewed journal articles → Use at least 3-6 It’s good practice to search several databases You can search several databases simultaneously on the UCL system (e.g. via Ovid) Useful Databases for UCL Explore “Grey Literature” MH Research There may be other relevant UCL Explore Medline / Pubmed offers search all literature not retrieved through (biomedicine and UCL library searches of electronic databases health) resources e.g.: Web of Science Guidance PhD theses (broad range of available online Session 9: 8th Nov 13  bioscience) https://www.ucl.ac.uk/library/electronic- Conference proceedings resources/about- Embase Government or expert bodies’ explore/#whatisexplore (biomedicine and guidelines pharmaceutical) Good starting Surveys/reports from point for finding PsycINFO campaigning organisations out about a topic (psychology) Some databases available: Not ideal for CINAHL (nursing OpenGrey, EThOS etc. systematic and allied health) Additional search strategies may be reviewing Cochrane (searches not required to find this “grey (including replicable outside literature” databases of UCL) systematic reviews (DARE) and RCTs (CENTRAL)) Defining Search Terms Identify the main concepts within your research question Identify all key terms within each concept Use inverted commas to search for whole phrases only (e.g. “supported employment”) Use truncation (* or $) to include various spellings of the word (e.g. behavi* will capture both behaviour or behavior) Combine terms for each concept using “booleian operators” E.g. “and” / “or” Subject Terms (MeSH) and Text Words Subject term searches are best used in combination with free-text searches Some problems: Many databases use standardised Subject terms differ across subject terms as part of a controlled databases (can’t use MeSH terms) vocabulary Session 9: 8th Nov 14  Assigned by indexers Subject terms not available for all search terms Used to identify articles that relate to the same construct, but use different Incorrect indexing due to poor descriptors: description of methods, populations or interventions in E.g. self-harm, deliberate self- article harm, self injurious behaviour, self injury, parasuicide Indexing is relatively new: Some older articles will not have been Can be useful in finding articles that assigned subject terms use alternative, unusual, or different terminology Exp. = indicator of MeSH term → all words associated with suicide Search Limits Setting limits can make your searching more precise, e.g.: Publication dates Language of paper Type of study (e.g. systematic reviews only) Search limits must have a clear rationale → justify why limits chosen Use “Limits” options within bibliographic databases Save Your Searches! Session 9: 8th Nov 15  During a period of activity, your “search history” will be saved and you can retrieve searches. To retrieve searches after this: Use “my account” to create a personal account and save your search And/Or Export references to Refman or Endnote + record search terms used and date of search Practical 💡 You want to review the effectiveness of psychological interventions for adolescents that self-harm. Use the PICOS framework to develop a search strategy. Inclusion criteria for this search are: Population: 12-18 years, engaging in self-harm behaviour Intervention: CBT, Psychodynamic therapy, Problem-solving therapy Comparators: Treatment as Usual (TAU) or wait-list control Outcomes: Self-harm (frequency, intensity etc), inpatient bed use, social functioning Study design: You decide Session 9: 8th Nov 16  Keep log of results for Prisma diagram → export all papers pulled from data bases → then can get rid of duplications Session 9: 8th Nov 17