Hello all,
For a meta-analysis, we need to calculate effect sizes for our different outcome measures, in order to perform the analysis.
We aim to compare surgical cohort studies, there are no control groups available. Results are all presented in incidence rates (eg. % of patients in cohort undergoing secondary intervention). We struggle to find the correct way to write these incidence rates as odds ratio or relative risk in order to calculate effect sizes.
For our statistics we are using Stata software.
Does anyone know a way to perform these calculations?
Hello Iris,
I hope I'm not being too simplistic, but you could use:
1. Proportion (p) of cases that meet the target category. In his text, Jacob Cohen (1988), Statistical power analysis for the behavioral sciences (2nd ed.), defined an ES for differences in proportions, "h" as .2 = "small" ES, .5 = "medium" ES, and .8 = "large" ES. However, these thresholds might not be suitable for your application.
2. Odds of a sample meeting the target category: odds = p / (1 - p), where p is the proportion of cases. The key for interpretation would be determining what difference in odds (whether crude or adjusted) is noteworthy for your research question.
3. If you absolutely, positively have to have an OR (odds ratio), then: (a) select one study or extant data point as the baseline or referent; (b) create an OR for each of the remaining study results (as in option 2 above); then (c) record OR = odds for the study / odds for baseline-referent study.
Good luck with your work.
You need the effect size of an incident rate
see
https://scholar.google.co.uk/scholar?q=effect+size+incidence+rate&hl=en&as_sdt=0&as_vis=1&oi=scholart
https://www.tandfonline.com/doi/abs/10.1080/03610926.2015.1134575?src=recsys&journalCode=lsta20
https://campbellcollaboration.org/escalc/html/EffectSizeCalculator-OR-main.php
https://campbellcollaboration.org/escalc/html/EffectSizeCalculator-OR5.php
Article A Simple Method for Converting An Odds Ratio to Effect Size ...
https://www.meta-analysis.com/downloads/Meta-analysis%20Converting%20among%20effect%20sizes.pdf
Thank you David Morse and Loopinder Sood both for your answers,
This way we can continue with our analysis
https://training.cochrane.org/handbook/current/chapter-10
section 10.8 Meta-analysis of counts and rates
Short and clear
Good luck!
Just forgot https://training.cochrane.org/handbook/current/chapter-06
Chapter 6: Choosing effect measures and computing estimates of effect
6.7 Count and rate data
You can use psychometrica website.
It provides tools to calculate many types of effect sizes.
Also, it is a valid and credible tool and I have seen many scholarly articles that cite it.
URL: https://www.psychometrica.de/effect_size.html
- Badges
- Science topic
- Similar topics
- Psychology