o******6 发帖数: 538 | 1 ☆─────────────────────────────────────☆
bigwing2 (titi) 于 (Thu May 15 23:29:00 2008) 提到:
4 treatment: 1(control),2,3,4
5 timepoint: baseline, Month 1,2,3,4(repeated measure)
Already substract baseline value from the value of Month 1,2,3,4.and can
find mean change from basline for control and each treatment at each follow-
up month.
Now, want to find the difference of mean change from baseline between
treatment and control at Month 1, 2,3,4 respectively
i.e. want to find:
what's the differe | q****k 发帖数: 1023 | 2 lsmeans treatment*time | slice=time?
I like the SLICE statement.
follow-
【在 o******6 的大作中提到】 : ☆─────────────────────────────────────☆ : bigwing2 (titi) 于 (Thu May 15 23:29:00 2008) 提到: : 4 treatment: 1(control),2,3,4 : 5 timepoint: baseline, Month 1,2,3,4(repeated measure) : Already substract baseline value from the value of Month 1,2,3,4.and can : find mean change from basline for control and each treatment at each follow- : up month. : Now, want to find the difference of mean change from baseline between : treatment and control at Month 1, 2,3,4 respectively : i.e. want to find:
| c**d 发帖数: 104 | 3 when you look for mean change over time, you first have to check interaction between month and treatment. if significant interaction,
treat month as a class variable (you can create a new time variable ) to find mean change over time.
Then you can get what you want by using lsmeans trt*monthclass/
By the way, model can be complicated to check linear and quadratic trend with random effects
follow-
【在 o******6 的大作中提到】 : ☆─────────────────────────────────────☆ : bigwing2 (titi) 于 (Thu May 15 23:29:00 2008) 提到: : 4 treatment: 1(control),2,3,4 : 5 timepoint: baseline, Month 1,2,3,4(repeated measure) : Already substract baseline value from the value of Month 1,2,3,4.and can : find mean change from basline for control and each treatment at each follow- : up month. : Now, want to find the difference of mean change from baseline between : treatment and control at Month 1, 2,3,4 respectively : i.e. want to find:
|
|