Kaplan-Meier plot - How to edit the numbers at risk

[Tiago Pereira]

Consider this example:

use https://www.stata-press.com/data/r18/drug2, clear
stset, noshow
sts graph, failure by(drug) risktable

I have compelling reasons to adjust the number of participants at risk, but I aim to automate this process.

Do you know of any method to, for instance, triple the numbers at risk without the need for manual adjustments through the graph editor?

[Andrew Musau]

Go back to the stset and specify frequency weights.

use https://www.stata-press.com/data/r18/drug2, clear
gen weight=3
stset studytime [fw=weight], failure(died)
sts graph, failure by(drug) risktable

[Tiago Pereira]

Andrew, as always, brilliant solutions! Thank you so much.

Are you aware if there is a workaround non-integer weights? Some of my analyses involve reducing the numbers at risk, say, by 10 fold.

[Andrew Musau]

For graphing purposes, specifying pweights seems to work.

use https://www.stata-press.com/data/r18/drug2, clear
gen weight=.1
stset studytime [pw=weight], failure(died)
sts graph, failure by(drug) risktable

[Tiago Pereira]

Thank you so much again, Andrew. Indeed, I tried the option aw, but fw does the job. A million thanks!

[David Fisher]

On a related note, the KMunicate study (https://github.com/tpmorris/kmunicat...ster/README.md) elicited views from a range of stakeholders on how best to present Kaplan-Meier-style data. The Stata implementation of the suggested solution is not automated (to my knowledge) but does suggest a coding approach for non-standard KM plots.

[Tiago Pereira]

Thank you, David! KM plots are always a challenge. I did not know that initiative, but it does look remarkably necessary.