contrast of joint (dx,bar{n}*) interventions defined by same dx but different bar{n}*

[romainkp]

Example: UACR/TI 90days.

dx with x=7.5
bar{n}* = continous 1, 101010..., 1001001...

In this example, we are interested in an MSM for 3 rules.

survMSM however ignores the fact that each intervention is indexed by a particular monitoring intervention and defines model terms solely based on the various levels of treatment interventions. In this case, there is only one dx so survMSM gives:

MSM.IPAW$MSM.fit
$coef
Periods.0to0_new.d7.5 Periods.1to1_new.d7.5 Periods.2to2_new.d7.5 Periods.3to3_new.d7.5
-5.362056 -3.953137 -4.067354 -3.152000
Periods.4to4_new.d7.5 Periods.5to5_new.d7.5 Periods.6to6_new.d7.5 Periods.7to7_new.d7.5
-6.936838 -3.870630 -4.138810 -2.931822
Periods.8to8_new.d7.5 Periods.9to17_new.d7.5
-5.218133 -5.028533

$linkfun
[1] "logit_linkinv"

$fitfunname
[1] "h2o.glm"

I'll email template code that describes the problem in more details