Timeline for Possible non-linearity pspline Cox model
Current License: CC BY-SA 4.0
3 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 18, 2021 at 16:02 | comment | added | Frank Harrell | psplines work well so don't switch just for that reason. | |
| Oct 18, 2021 at 14:45 | comment | added | psoares |
Hi, thank you for your reply! Just to give a bit more background info. delay is the time from symptoms until diagnosis, so I don't expect this variable to have linear behaviour, but I got confused with the output. We used the pspline especially to avoid specifying the location of the knots, but maybe it's best to use a cubic spline and define knots at the percentiles, or maybe at the beginning? Two, three months of delay is quite large theoretically though the median is a bit higher.
|
|
| Oct 18, 2021 at 13:14 | history | answered | Frank Harrell | CC BY-SA 4.0 |