I'm conducting a systematic review reporting dropout rates and demographic characteristics in social prescribing.
I've searched for other reviews that have reported on average dropout rates and many appear to conduct a meta-analysis comparing interventions to controls and then report the pooled estimate for attrition.
My early scoping review has found most studies to be before-after/pre-post studies so this is unlikely to be possible.
What is the best statistical way for me to report average dropout rates and the heterogeneity across the studies?
Thanks.
If I understand you correctly what I would do is report the drop out rates in the before and after study - that is how many participants recruited and assessed at baseline are followed up (n/N). In this respect, it's not really any different to an RCT. My guess from your question is that the studies you are looking at do not report this and so it is entirely unclear how many people who were 'prescribed' a particular course of action actually undertook it. This is clearly a huge source of bias and a major limitation of research. It's something to comment on in your review but if the information is not there you can't analyse it! Sometimes one of the most important results of a review is the identification of what important things have NOT been studied.
Considering your question, the studies that particularly indicate the number of participants enrolled initially in the study and the number of participants that have completed the study would be the primary inclusion criteria in your review.
Additionally, if the studies have denoted the sample size calculation method, the reason for the observed drop-outs, and the power of the study after removing the dropouts would be an additional benefit for your review question.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude