Hi so I am doing a systematic review to examine the effect of a surgical produce A vs surgical procedure B on patients by using patient reported outcome scores.
There is only 1 RCT and the rest are case series for surgical procedure A or surgical procedure B with a mean no of patients of 50.
Please understand that most case series done by surgeons are usually retrospective and done within their unit. By aim is to highlight what we know so far and the future, where i will be doing large data registry work.
I understand I will use the GRADE approach to check for bias.
Let's say you have 4 studies, AA and aa for procedure A; BB, and bb for procedure B. I think combining mean from AA and aa together is ok. However, comparing AA+aa to BB+bb is another story. The quality of the comparison between AA+aa and BB+bb is at most equal to a retrospective observational study. When the baseline characteristics between arms are not similar, bias occurs. I would combine and compare the case series as an overview and to generate a hypothesis. I do not think meta-analysis would yield anything useful. This is just what I think. I'm waiting for the experts too.
Due to the nature of the case series, I would not recommend any further analysis, as there is no control group or any comparison. And so, no you can't do a pooled analysis on the result of these case series, as there are too many various differences between these cases that might generate further bias. I would simply do a simple comparison using tables and elaborate it on the discussion section.
I think meta-analysis requires PICO elements. If they are not present in the included studies then do only systematic review without a MA.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis