Dear Statalist,

I am using Stata 14.2 on Mac. I have created a propensity score IPT weighted sample of breast cancer patients and am comparing survival (months) between those who did and did not receive radiotherapy.

The sample size is 7,708, 6,908 of which received radiotherapy while the remaining 802 did not.

A Kaplan Meier curve has been creating using the risktable option.

I am confused because the numbers at risk at time 0 are not the same as stated above, there now seems to more patients than exist in my sample (ie 6,918 in the radiotherapy group, 824 in the no radiotherapy group). Is this something to do with the p weighting?

Your insights would be greatly appreciated.

Code:
logistic radiotherapy covariate1 covariate2 covariate3          /* Propensity score model */
predict ps                                                                               /* Predict propensity scores */
propwt radiotherapy ps, ipt                                                    /* Calculate IPT weights */

stset death_censor_date [pweight=ipt_wt], failure(death==0) origin(diagnosis_date) scale(30.4)
sts graph, by(radiotherapy) risktable
Array ​​​​​​​