What Does “Levels of Evidence” Mean in Evidence-Based Practice?
In evidence-based practice (EBP), we talk a lot about “Levels of Evidence.” You will see levels of evidence (LOE) ratings on critically appraised topics or synopses of original research, or in the methodology section or evidence table of a clinical practice guideline or systematic review. But what does an LOE rating tell you?
The last two blog posts focused on pre-appraised evidence and on the quality of evidence pyramid – information that is related to levels of evidence or an evidence hierarchy. This post will focus on level of evidence scales – what they are and how to interpret them.
What is a Levels of Evidence Scale and What Does it Tell You?
Anytime we consult the literature to answer a clinical question, we must evaluate the quality of the study. Once we critically appraise a study, the quality of the evidence is graded according to an established scale or hierarchy.
Level of evidence scales are rating scales. They provide a rating of the strength of the evidence, ranging from a summary to a manufacturer’s recommendation. The higher the evidence is in the hierarchy, the stronger the evidence. How the “strength” of the evidence is judged varies among the rating scales. Strength may be identified according to the type of study design and may also take into account how rigorous the research methodology was in relation to the type of study design and/or the results of the study.
Randomized controlled trials (RCTs) are the strongest research designs for treatment or intervention studies because they exert the most control over the methods. More control means there is less of a chance of systematic or random error and therefore, the results are considered more trustworthy. However, not all RCTs are well-conducted nor always reported in the literature with the level of detail needed to critique the quality of the study; therefore, it can be difficult to decide whether the results can be believed and used in clinical practice.
Because summaries contain evidence from the lower levels of the hierarchy, including RCTs and evidence syntheses, they are considered the highest form of evidence accessible to all clinicians. (See my post on the 6S pyramid for more explanation.)
The strength of the evidence is usually designated with numbers (e.g., Level 1,2,3 or Level I, II, III) or qualifiers of study quality such as “high” or “low.” Some LOE scales use letter grades (e.g., Level A, B, C); these scales can be confusing when paired with grades for practice recommendations — I’ll talk about practice grades next time.
By the way, just because I say a study is at a “lower level” on a hierarchy of evidence scale, doesn’t mean that it was a poorly done study – be clear about that! It just means that there are other study designs that are stronger because they afford more control for the researcher to reduce or eliminate bias. So we have less “confidence” in the results of lower level studies because we know there is more chance for error.
Also, know that there are poorly done RCTs – and they would be judged lower on the quality scale than a well-done observational study! Expert opinion or expert consensus evidence is at the lowest level of evidence for most hierarchies, but personally, I’d rely on these evidence sources over a “higher level” study design about which I had many questions related to methodology.
Which Levels of Evidence Scale Should You Use?
There is NO universal LOE scale that everyone uses so there are lots of different LOE scales out there.
You might be asking why there is more than one levels of evidence hierarchy. That’s a good question, but the bottom line is that different clinical questions require different levels of evidence (see the table below). Professional organizations frequently use their own scales (e.g., American Heart Association, American College of Chest Physicians); publishing companies and evidence-based practice textbooks have developed their own scales, too (e.g., Elsevier, UpToDate). Melynk & Fineout-Overholt (2015) developed a widely used LOE scale for questions about therapy based on the literature.
In some scales, Level 1 evidence is the strongest evidence and in others, Level 5 is the strongest. Some LOE scales are based only on the research design and others provide levels identified by the judgment of the quality of the study itself.
Originally, levels of evidence systems were designed to assess the quality of intervention (treatment) studies only, but there are now levels of evidence scales for studies questioning etiology, diagnosis, prognosis, and harm. Just realize that a high level of evidence for a treatment study (e.g., a summary or systematic review of RCTs) may not be the highest level of evidence for a study of prognosis (Glasziou, Vandenbroucke, & Chalmers, 2004).
(Sources: Fineout-Overholt & Johnson, 2005; Howick et al., 2011; Winona State University, 2017)
There are some LOE scales that are widely known and used by many researchers and scholars: Oxford Centre for Evidence-Based Medicine, US Preventive Services Task Force, Joanna Briggs Institute, The Johns Hopkins Nursing Evidence-based Practice Rating Scale, AACN’s Levels of Evidence, and the LOE scale developed by Melynk and Fineout-Overholt (2015) are some examples.
The important point for you to understand right now is that there is no commonly or universally used LOE scale! That means that you cannot assume that a Level 1 is the same “strength” in all scales. So the caution here is to understand which LOE scale is being used for the article or source you are reading. This information is usually found in the introduction or the methods section or identified in an evidence table.
If you have to determine the level of evidence for studies you are reading for a school paper or project, the first thing to do is see if the instructor has prescribed a specific LOE scale to use in the assignment directions. I would assign a specific LOE scale for my students to use so that they were all “speaking” the same language. So check first before searching!
If the instructor wants you to choose which scale to use, then you have a lot of choices. You want to use the scale that makes sense with the clinical question you are asking. Are you writing an evidence paper about medical or nursing interventions? Then your LOE scale needs to identify pre-appraised evidence of summaries or syntheses as the highest level of evidence. If your paper is about the cost of a specific intervention, then I’d be looking for synthesis evidence of economic evaluations. If you are writing a paper about patient experiences, I’d be looking for a meta-synthesis.
Because nursing research uses a variety of research designs, I’d suggest using a scale that includes descriptive and qualitative studies in the evidence hierarchy. The Joanna Briggs Institute Levels of Evidence, The Johns Hopkins Nursing Evidence-based Practice Rating Scale, AACN’s Levels of Evidence, or the LOE scale by Melynk and Fineout-Overholt (2015) are good choices to rate your research studies.
As the Howick et al. (2011) noted: “no evidence ranking system or decision tool can be used without a healthy dose of judgment and thought.” Keep that statement in mind as you explore and use these LOE scales.
Fineout-Overholt, E., & Johnston, L. (2005). Teaching EBP: Asking searchable, answerable clinical questions. Worldviews on Evidence-Based Nursing, 2(3), 157-160.
Glasziou, P., Vandenbroucke, J., & Chalmers, I. (2004). Assessing the quality of research. British Medical Journal, 328(7430), 39-41.
Howick, J., Chalmers, I., Glasziou, P., Greenhalgh, T., Heneghan, C., Liberati, A., … Thornton, H. (2011). The 2011 Oxford CEBM levels of evidence: Introductory document. Oxford Centre for Evidence-Based Medicine. http://www.cebm.net/wp-content/uploads/2014/06/CEBM-Levels-of-Evidence-Introduction-2.1.pdf
Melnyk, B. M., & Fineout-Overholt, E. (2015). Evidence-based practice in nursing and healthcare: A guide to best practice (3rd ed.). Philadelphia, PA: Wolters Kluwer Health.
Winona State University Library. (2017). Levels of evidence. http://libguides.winona.edu/ebptoolkit