Step 3: Select full text of articles that you will review and abstract information
In our case, we have identified a systematic review \cite{baker2016interventions} and a study from the paper \cite{cooper2016effectiveness}. You can download the Cooper Paper from the following link to get started:
In real life, you will need to abstract data from more than one paper to complete an evidence based review. Also, in this case, we have just used the Baker meta analysis and systematic review to collect the Cooper paper, you will need to identify your paper based on the search you have conducted.
Step 4: Assess using GRADE
We will now present a step by step use of the GRADEpro tool to conduct an evidence appraisal and portfolio development. Follow these steps to conduct a GRADE based assessment. We will use the Cooper paper (see above link) to conduct the assessment.
1. First log in to the GRADEpro website (
link to GRADEpro website) and click "Log in". If you have not logged in before, you will need to create a sign up.
2. When you click on "Log in" and input your username and password, the programme may take a while to start up. This is normal.
3. If this is the first time you are working with the problem, create a new project. Otherwise, if you have already entered data before, use the project you created before. For this project, we will create a new project.
4. In the "Create new project" box, type the name of the new project. We will call it "Elder Abuse Project". Set the type of the project to "GRADE Evidence Profile" from the drop down list, and click on "Create Project" to get started.
5. Next, click "Add management question" box. In our case we are working on a "management question" because we are interested to find out what interventions will work best to prevent elderly abuse. If your project is one where you want to test the effectiveness of a diagnostic or screening test, use "Add diagnostic question" box.
6. In the next box, you will need to type the interventions and comparisons, the setting where the study results should be applied and optionally the bibliography and the name of the question author or the authoring team. Here, we will enter "Elder abuse prevention programmes" in place of "intervention" and "Treatment as usual" in place of "comparison in the top two boxes. Then we enter "Carer Abusive Behaviour" in the outcome box. We will also enter "Residential care settings" in the box titled "Setting". We can leave the boxes "Bibliography" and "Question author(s)" empty. You can optionally fill in these boxes.
If you want, you can add more management or diagnostic questions. Here, we are interested in "Strategies for Relatives" (START programme), and this is the only one, so we did not add any more question to the mix.
7. Then we click on the "Training of caregivers vs Treatment As Usual ..." box in the resulting window from the above step to get started. In this window we see several boxes starting at the top with "Quality Assessment" and "Summary of Findings" and ending with "Add outcome". We will click on "Add outcome" to get started. A point to remember is that, GRADE works on the basis of outcomes. So, we use outcomes as the unit of analysis.
8. As you click on the "Add outcome" box, it expands the screen where you can add the details.
In order to do this, you must read the "Methods" and the "Results" section of the paper you choose to work with. The methods and the results section will provide you will nearly all information that you may need in order to fill in the specific boxes. We will use these two sections in the Cooper et.al. (2016) paper to show the details.
You will note that not all papers will report the outcomes right away so you will need to dig for the outcomes through the paper. Also, you will note that for many research papers, the authors do not place the primary and secondary outcomes in proper order and it is not uncommon to find several outcomes are reported in the same research. When we talk about GRADE approach, our focus in on outcomes and we assume that one outcome can be achieved in a number of different studies. Therefore, it is the outcome that we are interested to work with. So,
- You can have one outcome and one study per outcome (simplest of the choices). Report one study in details (the approach we will take in this example)
- You can have one outcome that is covered by a number of different studies. There are two ways to approach this situation:
- The simplest approach is to use one study per outcome and report individually as many studies as you want, each time abstracting information from that one study
- A more complicated approach is to use that one outcome and multiple studies. In these situations, results are abstracted from multiple studies. There are two ways of doing this: first, you can conduct a systematic review of the studies put together without pooling the results from the study statistically. This is a simpler approach and will work in most cases.
- You can pool the results of two or more studies together using an approach of meta analysis. When you conduct a meta analysis, you will need to ensure that the study results can indeed be pooled or that the studies are sufficiently homogeneous or at least they fail the test of heterogeneity.
In this case, we will conduct a GRADE exercise assuming that there is one outcome and that there is one study on the basis of which we will evaluate that one outcome. This is for practice purposes but this is rather rare in real life. Anyway, let's continue to fill in the specific boxes:
So, let's review for the Cooper study, and we see that here is a section on "Outcomes":