Dear researchers:
I am doing a meta-analysis to assess the impact of an intervention on mortality or death. Some of the included studies have reported the risk ratio, while others are survival analyses that merely indicate the number of patients who survived or died over a predicted length of time, such as five years. As you are aware, the figures supplied are estimates, but I need to find a way to compute the number of dead and living patients, but I have no idea what the real follow-up duration is. For one, two, or more years. Please inform me as to the time period I should use. Mean or median follow-up duration, or...
Thank you in advance.
Best regards
Hello Hossein Motahari-Nezhad, May I just clarify, how many studies are there, and what is the time variation of follow up?
Dear Deborah J Hilton
There are 14 studies that reported the odds ratio and seven studies are survival analyses. Actually, I want to combine odds ratios with the results of those survival analyses to conduct my meta-analysis.
Those seven survival analyses:
Five articles reported 5-year survival analysis and the rests 3-year survival analysis. One study explained that "The median follow-up for survivors was 65 months (minimum 36 months)." Another study explains that "The median clinical follow-up was 36 months (interquartile range, 21–59 months)".
Also in another study, It was mentioned that "This study included patients with follow-up data that spanned at least 3 months". In another article, "Mean follow-up for the entire patient population was 50.1 (± 1.1) months", and in another study, "The minimum follow-up period was 15 months". The rests did not provide any information about the timing. Therefore, I'm looking for a way to calculate the odds ratio, variance, and standard error to run the meta-analysis.
Thank you
Hossein
It would be difficult to merge these 2 types of data because of different reporting format. The survival analysis theoretically allows you to estimate a proportion of survival after a specific period (using the reported models => however, this would be an estimation from a model and not the observed data). One of the strength of survival analysis is to account for different period of observation for the included cases (with censoring).
So for me it would be difficult to merge these 2 types of studies vs doing 2 subgroups (1 for survival analysis model and 1 for others). This would be a good opportunity for you and your group to report this dual format of reporting and insisting on future study standardization.
Good luck!
Dear Sébastien Buczinski,
Thank you very much for your feedback.
Best regards
Hossein Motahari-Nezhad
Sébastien Buczinski May I ask a follow-up question? In inputting data studies from survival analysis, which do we input as the event? the Deaths or number of survivals?
Thank you
- Badges
- Science topic
- Similar topics
- Engineering
- Computer Engineering
- Computer