Figure/graph of bootstraped cost-effectiveness/cost-utility

[Hanna Gyllensten]

Hi,

This has been a week of frustration with regards to statistical analyses, and happiness after several very good suggestions from list members on how to solve coding issues. Hopefully someone in this forum has suggestions for how to get on also with this problem that I have got stuck on:

I am conducting a cost-utility analysis using cost data collected from healthcare administrative registers and EQ-5D questionnaires collected from participants in a clinical trial of two alternative treatments. Thus, I have conducted multiple imputation (mi impute chained) to handle missing responses to the questionnaire, but the cost data is expected to be complete.

After minor adjustments to my data I have managed to use the code provided by Faria and colleagues (A Guide to Handling Missing Data in Cost-Effectiveness Analysis Conducted Within Randomised Controlled Trials. from 2014), step 5 in their description (titled "Analysis of multiple imputed datasets (post ice or mi impute chained)"), to identify ICERs and the probability of the intervention being cost-effective at different thresholds. I have so far used their described solution using "sureg", but have got stuck on trying to apply the alternative bootstrap solution due to differences in names of variables (possibly a result of them using ice and my use of mi impute?).

However, I would also like to create one of those graphs with QALYs on the x-axis and costs on the y-axis, and a "cloud" of bootstrapped comparisons. Does anyone here have suggestions on how to get there, or how to get bootstrapped results for a variable to go into a graph? Any suggestions on how to get forward would be very welcome!

Kind regards,
Hanna

[Carlo Lazzaro]

Hanna:
Henry Glick and co.'s textbook may have some do.files about that issue: https://global.oup.com/academic/prod...?lang=en&cc=it

Otherwise, you may want to try:

scatter Incremental_Cost Incremental-QALYs

For those engaged in other research fields:
EQ-5D questionnaire (and related Visual Analogic Scale - VAS) compose a set of instruments conceived for eliciting health-related quality of life (that health economist name utility. It ranges, in general, from 0, indicating death of a health state perceived as worse than death, to 1, meaning perfect healh) (https://www.euroqol.org/)
QALYs stays for Quality-Adjusted Life Years (that is, expected or registered survival adjusted for utility) (https://global.oup.com/academic/prod...-9780199665877).

[Weiwen Ng]

Glick actually posted the .do files on his website at UPenn, and I have used them in a non-MI setting. They are bsceaprogs.do nad bsceagraphs.do. You'll need to run both, and I think you need to run the first file first. The site may look a bit old, but I used these two programs last year in Stata 14.

http://www.uphs.upenn.edu/dgimhsr/stat-cicer.htm

[Hanna Gyllensten]

Thanks a lot, both of you, this appears to be very useful. I made an attempt to find the book, but it was apparently not available at my university, I'll just have to order it, but I start with the webpage. Kind regards, Hanna

[Carlo Lazzaro]

Hanna:
please note that the link I pointed you out to refers to the 2nd edition of the textbook.
I've bought the 1st one some years ago and the reading (as well as the Stata files included in the supplementary materials) were very interesting.
As an aside, I usually prefer preenting the cost-effectveness accepatbility curve vs the cost-effectiveness plane, as per my expereince any audience with a limited smattering of health economics gets the meaning of the first one, whereas some tricky situations (such as the straddling limits of the 95 confidence interval of the incremental cost-effectivenessa ratio) are difficult to convey via the second one.