search

audrey-b / BUGSnet

BUGSnet: a new comprehensive package for network meta-analysis
Other
23 stars 9 forks source link

Support for direct analysis of log hazard ratios and SEs? #61

Open RichPollock opened 4 years ago

RichPollock commented 4 years ago

Is it currently possible to use BUGSnet with data that is already in the form of hazard or log hazard ratios and standard errors? This is pretty straightforward in gemtc by specifying a cloglog link and binomial distribution for the model, "NA" for the reference treatment outcomes, and the LHR and SE on the comparator, but I can't quite figure out how to convince BUGSnet to run the analysis.

audrey-b commented 4 years ago

Hello Rich,

At the moment, BUGSnet does not have the capability to analyse this kind of data. It's definitely on our list of future improvements but I cannot say when it will be done. We are happy to accept contributions if you want to implement it. It's a fairly simple analysis in terms of the BUGScode but the bulk of the work is with prepping the data and whatnot.

Best,

Audrey

RichPollock commented 4 years ago

Thank you, Audrey! I'll have a think about it... I guess the input format would ideally be similar to gemtc using NA values for the reference treatment and the contrast measure and variance for the comparator? I'll pull the code down and take peek. Presumably a fork followed by a pull request would be the best way to go about it?

Alternatively, I'd be very happy to provide some sponsorship for the project if that would allow you/the team to free up some capacity to implement it? If you think that could work, feel free to drop me a quick note at pollock@covalence-research.com

Many thanks and best wishes, Rich