Literature Evaluation Table: level vs strength of evidence

The table asks for two ratings per article and they are not the same rating, which is the distinction this whole assignment turns on. **Level of evidence is a property of the design**: where the study sits in the hierarchy, from systematic reviews and meta-analyses of randomised trials at the top, through single randomised trials, non-randomised controlled studies, cohort and case-control designs, down to descriptive and qualitative work and expert opinion. **Strength is a property of this particular study's execution**: sample size and power, how participants were selected, attrition, blinding, whether the outcome measured is the one that matters, and whether the findings are consistent with the wider literature. A badly conducted randomised trial is a high level and a low strength. Filling both columns with the same judgement is the commonest way to complete this table wrongly while completing it fully.

The eight articles are constrained twice over and both constraints eliminate most of what a search returns. They must be current within the last five years, and they must relate closely to **the PICOT statement you developed earlier in this course**. That second constraint is the one that bites, because it means an excellent, recent, well-designed paper on an adjacent question does not qualify. Go back to your own PICOT, pull out its population, intervention, comparison and outcome as separate search terms, and require each candidate article to touch at least the population and the intervention. Articles chosen because they were easy to find will show as a set that does not converge on a single question. The set is also read as a set: eight articles that each touch a different corner of the topic look like a search that was never narrowed, whatever their individual quality.

Read the sentence about systematic reviews carefully, because it is doing something specific. The permitted article types are listed as quantitative research, descriptive analyses, longitudinal studies and meta-analyses — and then systematic reviews are mentioned separately, as something that **may be used to provide background information for the purpose or problem** rather than as one of the eight. The safe reading is that your eight should be primary studies and any systematic review sits in your background section. If you want to count one among the eight, check with your instructor rather than assuming, because a table of eight that turns out to contain three reviews of the same trials is thinner evidence than it appears. Where the wording is genuinely ambiguous, saying in your submission how you interpreted it costs one sentence and removes the risk entirely.

A practical point about the search that saves the most time: filter by date and by design before you read anything. Databases will let you restrict to the last five years and to specific publication types, and doing that first means every article you open is already eligible. Record the level of evidence as you go, in the table, at the moment you identify the design — going back through eight articles at the end to work out what each one was is the slowest possible way to complete this, and it is where inconsistencies between rows creep in. Screening on design first also means you notice early if your PICOT has almost no randomised evidence behind it, which is itself worth knowing before you build a capstone project on it.

Finally, note what this assignment is for. It is not a literature review and it is not asking for synthesis prose; it is a structured appraisal that feeds a proposed capstone project. That means the value of the table lies in whether someone reading it could tell how much confidence your project's premise deserves. Eight rows all rated highly and uncritically communicate nothing. A table that says plainly which two articles carry the argument and which three are weak but included for coverage is more useful, and it is also more honest about what the evidence base actually supports. That candour is not a weakness in the submission; it is the thing the appraisal exists to produce, and a marker can tell the difference between eight rows read carefully and eight rows filled in.