Four studies investigated Salmonella in chicken meat but one of them (Somda et al., 2016) assessed this pathogen in different chicken products (grilled chicken, fumed chicken, flamed chicken and chicken prepared around fire). We entered Somda et al 2026 as four entries (just treated as if it were different studies/articles) while conducting the meta-analysis. What is your opinion/advice on this approach? See snipped part of the forest plot for clarity (attached as file). Thank you for your expertise and time.
I think you can dig in methods using hierarchical models where your first random effect is at the trial level and your second random effect is at the study / paper level.
You can look for a good review of the topic in the following website done by Mathias Harrer and colleagues:
Chapter 10 “Multilevel” Meta-Analysis | Doing Meta-Analysis in R (bookdown.org)
I am no expert neither as experienced as you are but I don't think the comparison is right and even the forest plot shows the bias very clearly.
Uncooked meat would definitely have more pathogens and cooked meet will have low pathogens. That is the entire base of the debate which says to cook meat properly before eating to avoid diseases
You cannot compare the biased data similar as you cannot compare raw apple and overripe oranges for their nutrient level.
The problem I see is that the "population" under study (meat) is different, and therefore I find it difficult to compare these analyses. In my opinion, and without being an expert in this specific area, the studies of chicken meat should go together and without mixing other meats. What happens for example if hypothetically the prevalence of Salmonella in other meats is much higher or lower? or if some other biological parameter changes?
- Badges
- Science method