Systematic Reviews and if included meta-analysis are in depth reviews of a specific clinical question. The systematic review contains the narrative portion and the meta-analysis contains the data and graphs portion. Quality systematic reviews follow a strict set of criteria for creation.
There are a number of organizations that can help you through the process of choosing your topic, registering your review, and writing your review. The first one we will start with here is PRISMA or Preferred Reporting Items for Systematic Reviews and Meta-Analyses. PRISMA provides a statement for their guidelines. We will discuss many of the PRISMA guidelines throughout this guide. You should also begin by printing out the PRISMA checklist. This will help guide you and your team through the process.
Before starting a systematic review, consider the following:
If you have the resources and time to devote to a systematic review, then the following topics will provide assistance with this process.
Clinical questions for systematic reviews need to be specific enough to be answerable, but not so specific that you exclude too many studies. Questions should be focused on an area of clinical importance that may not be well reported or where a clear consensus has not been reached in the literature. A systematic review and meta-analysis may help to clear up debate on certain topics. To start on the clinical question, use the PICO criteria.
Using the PICO format, first identify the Problem,Patient or Population. Next, look at the Intervention or what treatment or diagnosis will be used. List any comparisons, such as placebo, specific therapy or procedure. Additionally, there may be no comparison used. Lastly, list the desired outcome. For example:
Using the PICO example above, clinical questions could be: For women undergoing emergent cesarean section will the addition of morphine result in higher rates of respiratory depression? Note that the question is neither too broad, nor too narrow: Will women ages 35-45 experience more respiratory depression (RR<10) with higher doses of IV morphine (2mg, 6mg, 10mg) during cesarean section? Being too specific will make it difficult to find enough studies to combine into a review.
Once the clinical question and PICO format have been developed, the next step is to look at inclusion and exclusion criteria for eligible studies. Carrying on the above example, will you include or exclude studies which look at variants of morphine? What will be the definition you use for respiratory distress? Will you include studies using IV, SubQ and oral morphine or only one or two of these? What type of study are you looking for? Only Randomized controlled trials? All clinical trials? Basic criteria will help with the literature search as well as when the review of articles. Criteria may change during the review as you come up against articles which you are unsure about. These can be discussed in your team and added to the criteria list.
Once you completed the processes above, the next step is to work on the primary outcomes. The process defines the expectation of the main outcome of the research. Continuing with the example above, a primary outcome may simply be rates of respiratory depression. Secondary outcomes are also explored in this step. A secondary outcomes from the example above might be mortality from respiratory depression or morbidities that may result from respiratory depression. A primary outcome is always needed and secondary outcomes may of interest but are not necessary.
It is highly encouraged that you both register your systematic review and search for similar reviews in PROSPERO. PROSPERO is the international database of prospectively registered systematic reviews in health care. Searching PROSPERO for similar reviews allows you to see if another group is currently working on the same review, necessitating a change in topic. Registering your topic also helps in letting others know your topic is being addressed. Your protocol for the PROSPERO record should include the following:
Literature Searching is the next big step in your systematic review. A big part of a systematic review is a through and systematic review of the literature. This is where your local medical librarians can be of assistance. Librarians can work as part of the systematic review team as a kind of project manager to conduct the literature search (including grey or unpublished sources), organize citations in bibliographic citation managers and article review software, help to narrow down the initial number of articles to review, create a PRISMA flow diagram for publication, and write up the methodology of the search for publication. Librarians can also conduct basic editing on the document.
When searching the literature it is suggested that at least three databases are searched. This creates overlap in your articles, but also ensures a higher rate of coverage on pertinent articles. Database access varies by library, so you may need to check to see what databases your library can access. Suggested databases include:
Once searches are completed, the entire search strategy and resulting citations should be saved in a file along with the total number of records retrieved, the date of the search, and the database searched. Any limits or focus elements used should be noted.
Once the literature search has been conducted the articles titles and abstracts must first be searched for articles relevant to your question. A librarian can assist in first removing any duplicates and many articles which fall under the exclusion criteria. A librarian can also assist with putting the citations into an online database for review, such as Rayyan or Coevidence. These online software tools allow reviewers to vet articles separately and then come together for a consensus. Articles can also be labeled with inclusion or exclusion data.
At least two content experts should look through all articles with a third content expert as a tie breaker. After the initial articles are screened and agreed on, the reviewers should obtain full-text of the remaining articles for a more in depth review. A secondary review can then be conducted to narrow the list down even further. This should leave the reviewers with a final list of articles which fit the study criteria.
Your process should follow the PRISMA flow diagram below. You should also use a copy of this flow diagram to document your search for publication or presentation.
Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7): e1000097. doi:10.1371/journal.pmed1000097.
Data extraction can be one of the tougher and more time consuming parts of the systematic review. All content experts on the systematic review team should decide how they would like to collect data, with written checklists or more likely on an electronic spreadsheet. The use of Microsoft Excel or Google Spreadsheets can make the process much easier. The content experts also need to decide what data should be collected. This will depend heavily on your question and both primary and secondary outcomes. You will need at least enough data to cover these areas as well as information for a basic demographic profile of participants.
The evaluation of study quality can be done at the same time as data extraction. This is simply taking each study and deciding on the quality of the methodology used by the study. There are a few ranking systems which can be used:
Assessing risk of bias may also be included in your systematic review. This is a portion required by Cochrane Reviews, but not included in all reviews. However, bias in individual studies will effect the overall outcome of the systematic review. General types of bias to look for:
Once all studies have been mined for data it is time to decide which studies had enough data to be included in a meta-analysis(if one is being done). Generally, a statistician should be used for this portion of the review to assist the content experts in choosing and carrying out the best statistical tests for the data given. On the UAMS campus, you can contact
the Translational Research Institute for biostats assistance. If a content expert can also perform the statistical portion of the review the following parts should be considered:
You now have a set of articles with the data abstracted, your data analysis and you are ready to write the narrative portion of your systematic review. A systematic review is a type of research study, so will be written up similar to any research study with a few additional headings:
Writing a great abstract is key to having a good final research product for presentation or publication. As a summary of your research and findings, your abstract should be limited to 100-250 words. Since it can be difficult to condense your research these few words, the tips below will help with this process.
The title of your project is the first thing that people will see. Creating a good title for your presentation/publication/poster will capture the attention of visitors. The following tips will help you as you develop your title:
Review articles should have anywhere from 2-5 authors. Students, residents and fellows should keep in mind that they will always need to have a practicing clinician as an author on their article.
Authorship can be one of the trickiest parts of writing an article and should be dealt with up front. The International Committee of Medical Journal Editors (ICMJE) has set guidelines for what authors are expected to contribute and what qualifies someone as an author. These are the suggested guidelines to follow for the majority of clinical journals. The criteria for authorship are as follows:
Anyone not meeting all four criteria should be listed in the acknowledgements section of your paper.
Along with meeting author criteria, authors must decide amongst themselves who will be credited with the first author slot and the order of authors. Traditionally the lead or corresponding author, who will handle organization, final edits, and submission, or the one who does the most writing will be the first author. Mentors, senior faculty, and project PI's are generally placed as the last author.
© 2017 University of Arkansas for Medical Sciences | Little Rock, AR