Susan Shenkin is Associate Editor for Systematic Reviews at Age and Ageing journal. She has recently published Systematic reviews: guidance relevant for studies of older people. She tweets at @SusanShenkin
Systematic reviews and meta-analyses are increasingly common. Our recent article in Age and Ageing journal aims to provide guidance for people conducting systematic reviews relevant to the healthcare of older people. It’s essential that systematic reviews are performed by a team which includes the required technical and clinical expertise: if you’re planning to do a review, ask for advice and support early. We hope that highlighting these issues will also help people reading systematic reviews to determine whether the results will influence their clinical practice. Here is a summary of ‘good practice points’.
- Decide on a focused question that you wish to answer, using the PICOS approach:
- Population – consider which ages, genders, settings (e.g. hospital-based, care home-based or community-based) and disease you wish to study
- Intervention or exposure – the precise type of intervention or exposure
- Comparators – what will you compare the intervention to? (if applicable)
- Outcomes – the outcome of interest
- Study design – e.g. randomised controlled trials only, or whether observational studies such as cohort studies or case-control studies will also be included
- Ensure this question has not already been covered by a systematic review before (or, if covered, there are still important questions that remain unanswered and/or an update is needed, suggesting a new review may be useful)
- Work with an experienced team, with expertise in literature searching, performing reviews, and statistical analysis (if meta-analysis included)
- Use a template to help you devise a protocol such as PRISMA-P and register/publish it if possible
- Consider your eligibility criteria in relation to your review question. This will be easier if you used the PICOS approach above
- If restricting by age, what will you do if other ages are included in the study?
- Will you contact authors for additional information?
- Which languages? Will you obtain translations for articles in other languages?
- Will you only include published papers, or also abstracts, or book chapters, theses etc?
- Devise the search criteria and which electronic databases you will use to identify studies
- If you are aware of some key studies that would be included in this systematic review, test your search criteria to check these papers are identified
- Seek expert help from a librarian or information scientist to help you to refine the search criteria and to choose which databases to search
- Outline all the information that will be extracted from included studies. Designing the data extraction tool in advance will help you focus on which information should be collected
- Which outcome measures will you report? How will you deal with missing outcomes, death of participants, discrepancies from published protocols?
- Consider potential confounders or effect modifiers
- Pilot the data extraction tool in advance of starting the review
- Decide how you will assess for risk of bias and any other features that touch on the quality of the evidence
- Decide whether a meta-analysis will be performed, or whether there will be a narrative summary of included studies. Plan your analysis and analytical decision rules in advance
- Register your review protocol (e.g. https://www.crd.york.ac.uk/PROSPERO/; http://www.researchregistry.com/).
- After the review is performed, write up using the PRISMA checklist and flow diagram as a guide.
Read the full Age and Ageing article: Systematic reviews: guidance relevant for studies of older people