JLepidata
commented
5 months ago Received here. I will try to find a source Bw Jens
Hent BlueMail til Android
Den 23. jun. 2024 22.00, fra 22.00, Jamie @.***> skrev:
There is a problem when the last row of a stratified survival table has 1 person at risk. The estimate of hazard ratio uses r in the denominator, where r is the number at risk at any time. when r=1 in the final time period, the denominator of that row's contribution to the variance of the hazard ratio includes the term r r (r-1), which is
- Even though the numerator at this point is also 0, the calculation 0/0 returns nan.
The code for hazard ratio estimate was taken from classic analysis, but I don't have a reference. While I'm inclined to just not use this final term (0/0) as it is non-informative, I'd like to be able to back this up.
We need a proper source for validation of the estimated hazard ratio for stratified survival analysis.
-- Reply to this email directly or view it on GitHub: https://github.com/EpiData-dk/analysis/issues/95 You are receiving this because you were assigned.
Message ID: @.***>
There is a problem when the last row of a stratified survival table has 1 person at risk. The estimate of hazard ratio uses r in the denominator, where r is the number at risk at any time. when r=1 in the final time period, the denominator of that row's contribution to the variance of the hazard ratio includes the term r r (r-1), which is 0. Even though the numerator at this point is also 0, the calculation 0/0 returns nan.
The code for hazard ratio estimate was taken from classic analysis, but I don't have a reference. While I'm inclined to just not use this final term (0/0) as it is non-informative, I'd like to be able to back this up.
We need a proper source for validation of the estimated hazard ratio for stratified survival analysis.