stpp calculates the Pohar-Perme non-parametric estimate of marginal relative survival, which under assumptions can be interpreted as marginal net survival. The estimate is a step function and changes at each event time. This is a different implementation to strs and stnet where the time-scale is split into a number of intervals. stns is another command that can be used to give an estimate of marginal relative survival. The motivation for developing stpp was to directly standardize the estimates using both traditional standardization and by using individual weights and to extend to estimation of other related measures. The other measures include crude probabilities and the ability to incorporate a second expected rate file, which can be used to estimate measures defined by Sasieni and Brentnall and non-parametric estimates of reference adjusted measures.

The single example currently just replicates the help file, but with some added description.


Basic examples

  • Using stpp
  • All cause and crude probabilities
  • Reference adjusted all cause and crude probabilities

Version History

Version 1.2.6

  • Use of the by option without the list option gave an unhelpful error. Now Fixed.

Version 1.2.5

  • Add minexpsurv and dropexpected options

Version 1.2.4

  • Add better return matrices when using by option

Version 1.2.3 - 2021-10-14

  • fixed mata error when having a time in list() option less than first event

Version 1.2.2 - 2021-08-16

  • removed output from error checks
  • better check for no events

Version 1.2.1 - 2021-08-15

  • updated variance estimation for crude probabilities of death
  • added levels option
  • Use of if statement returned unnecessary values
  • Improved error checks

Version 1.2.0 - 2021-06-08

  • fixed links for example code in help file

Version 1.1.9 - 2021-06-04

  • graph options added by Mark Rutherford

Version 1.1.8 - 2021-02-10

  • stpp gave an unuseful mata error with delayed entry. This has been fixed.

Version 1.1.7 - 2020-12-22

  • When listing results on the screen using the list() option, a missing value was given if the time was greater than the largest event time, but less than the largest censoring time. This has been fixed.

Version 1.1.6 - 2020-12-13

  • Fixed mata error when list() option had time greater than in data
  • some minor speed gains
  • help file examples now produce nice graphs (previously not sorted by _t)

Version 1.1.5 - 2020-10-19

  • fixed bug introduced in 1.1.4: ederer2 option now works again.

Version 1.1.4 - 2020-10-06

  • check for problems with merging lifetable
  • Big speed up in writing results back to Stata
  • Change default for transformation to survival to product integration. (added “fh” option for Fleming-Harrington estimates)

Version 1.1.2 - 2020-07-14

  • changed weights for references adjusted measures
  • added error if try to use ederer2 option with using2()
Professor of Biostatistics