Any successful project starts with a well-defined research question.
1. Identify Research Question
Every successful research project starts with a well-defined research question. The PICOS method is one way to structure a research question. PICOS stands for Population, Intervention, Comparison, Outcomes, and Study Design. By specifying each of these components, a very specific and researchable question can be developed.
2. Create your Research Team
A systematic review requires a multifaceted research team to carry out. Multiple reviewers of articles reduce biases and increase reliability of the results. In addition, your team should have expertise in the following areas: Domain Knowledge of the subject (e.g., epidemiology or education), Research Methods of the studies being reviewed (as well as of the systematic review method itself), appropriate Statistics (to understand the results of the studies being reviewed), and Information Retrieval.
3. Create (and register) a Protocol
Your protocol describes the steps you will use to conduct your review. PRISMA provides guidelines for what a protocol should include.
Registering your protocol will improve transparency/reproducibility as well as alerting other researchers of your intentions, so efforts are not duplicated. PROSPERO, BioMedCentral Protocols, and Open Science Framework are places.
The protocol should be completed before you start, to ensure a level of objectivity of the review. The protocol should include:
Create a comprehensive search strategy, using multiple databases. Include synonyms and stemmed terms. The advanced search features of databases will allow you to construct a precise, transparent search strategy. Ask a librarian for assistance on how to access and utilize these search features.
Determine which studies are relevant to your research question. Utilize your inclusion/exclusion criteria. Each study should be independently reviewed by different team members, resolving disagreement, to ensure reliability.
Determine potential biases of studies, based on study design and reporting. These can include selection bias, attrition bias, detection bias, and reporting bias.
Extract data from all relevant studies. There are several software programs that assist in organizing and visualizing your data.
Writing up your results! Your methodology should be clearly documented so others can evaluate the validity of your work, and updates can easily be generated. Your results can include recommendations for practice and policy and identification of gaps in research.