Meta-analysis

#1
Hello, I need help with meta-analysis. I want meta-analyze data from studies regarding histological features of a series of disease (A,B,C,D..). Fo this, do I need to create a table for every features for every disease?The aim of the work is to obtain the most frequent histological characteristics of every disease. Collected data are collected from studies in which control groups are different. For this, in the table for the meta-analysis, who are controls?Please,help me!
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Meta-analyses work best for randomized studies of a treatment. Are these studies randomized studies or observational in nature? Also, you need to better define what you mean by control groups. Are these comparator treatments?

More information is needed!
 
#3
Meta-analyses work best for randomized studies of a treatment. Are these studies randomized studies or observational in nature? Also, you need to better define what you mean by control groups. Are these comparator treatments?

More information is needed!
Hi,
The studies describe histological features in biopsies of patients with different pathology (A,B,C,D...) comparing them with biopsies of patients with another disease. No treatment. I collected the frequence of histological features for each pathology in each studies (ex.
Pathology A
Study 1 Three patients of 5 have necrosis, three of 5 inflammation

Study 2 2/3 patients with necrosis,1/3 inflmmation

Pathology B
Study 1 4/7 necrosis , 1/10 inflammation

Study 2 3/4 necrosis,4/8 inflammation

I want obtain the more frequent features for each pathology.For this I want obtain a forest plot for every features for every pathology.
Is it possible?For forest plot ,I need for controls!My controls would be the controls used in the studies..right?Can I use these controls though they are different?
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
Are controls Pathology B? Control part is still not clear to me. Also, you are saying pathology B, so is pathology B made up of different pathologies?
 
#5
No,controls aren't pathology B.
I have studies of 8 different pathologies.
10 studies of pathology A
8 pathology B
5 pathology C
6 pathology D
20 pathology E
7 pathology F
3 pathology G
But each study uses controls that are not patients with pathology A,B,C,D,E,F,G but others.
I want obtain forest plot for each pathology (A,B,C,D,E,F,G) for each histological features.But forest plot, for definition, is based on comparative studies with the same control (ex untreated),right?
In my case,is it possibile create a forest plot ,for example for necrosis for pathology A with studies in which control groups are different between them?
 

hlsmith

Less is more. Stay pure. Stay poor.
#6
This sound kind of like a hot mess. You will really need to use evaluation criteria to review studies (e.g.,Jade scores or Ottawa Scores).

There is something called network meta-analyses. In them given you have A vs B and B vs C, you can then attempt to infer the differences between A vs C. I think you need to post some links to a couple of these studies so that we have an idea of what types of studies you are referring to. Also if the patient samples are different enough you will need to address heterogeneity.
 
#7
For example, I attach 2 links for studies of pathology A "dermatomyositis" and two links for studies of pathology B "toxic myopathies". Tell me if it's possibile infer the differences between them considering that I want 1)obtain the main features (more frequent) of dermatomyositis from the 2 studies and obtain the main features (more frequent) of toxic myopathies of the other two and then 2)compare features of dermatomyositis with toxic myopathies , in particular know if there are freatures that are more frequent in one over the other.
Pathology A:
https://doi.org/10.1136/jnnp.73.4.420
https://doi.org/10.3346/jkms.2009.24.6.1015
Pathology B
https://doi.org/10.1016/j.ajpath.2012.08.033
https://doi.org/10.1371/journal.pone.0162760