I am conducting a qualitative synthesis and a couple of papers relate to the same study (so using the same sample of participants but generally with a discrete focus and not necessarily reporting the same original data).
I know that in traditional effectiveness reviews, especially where meta-analyses are conducted, it is important to avoid double-counting.
However I'm less sure how this principle applies to qualitative syntheses.
I would be grateful for any advice from people experienced in qualitative syntheses on how this should be reported and considered in my results.
For quantitative analysis, studies can be categorized by outcomes and be analyzed separately.
For example, 10 studies have examined blood pressure as a continuous variable, 11 studies have examined hypertension as a dichotomous variable, 2 studies have examined both outcomes.
Then in the meta-analysis, there'll be 12 studies examining blood pressure, and 13 of them examining hypertension.
Not sure what outcomes or categories you're interested in, but double counting may not be a major issue when using this strategy.
In both meta-analysis and meta-synthesis you should consider the 'Study' as the 'Unit of Currency' in the review not the references or reports.
So you refer to the study as Smith 20172-8 which means the references (reports) number 2 to 8 belong to one study.
While analysing or synthesising however you do not count duplicate data or information from difference reports of the same study.
You only consider the new information/data from each report. The information/data that are unique and have not been reported in the other papers of same study.
Also it is not just to avoid over-counting but also under-counting and relevancy check. Please see
Box 1. Reasons for assessment and concatenation of all references of one study in this paper: Article Study-based registers of randomized controlled trials: Start...
For outcomes where double counting may be an issue, I would suggest an analysis of all studies followed by a subgroup analysis. The subgroup analysis would then have any duplicate data removed with reasons why. Then any difference between the intial analysis and subgroup analysis could be accounted for by duplication bias.
If you do the analysis only with one of several 'duplicate' papers, epecially without a thorough explanation, this could then be interpreted as bias from the systematic review that you are performing.
Hope that helps!
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation