Implementing Month-Specific Treatment Duration Carryover in Panel Data

[Sergio Serrano]

Hi everyone,

I'm working with a panel dataset where each row corresponds to a monthly observation for a subject. One of the variables, months_on_treatment, indicates the month when a treatment was prescribed and shows the duration of the treatment in months. The column months_fu refers to the current observation in months.

Here's my current problem: I need to carry over the treatment duration data only for the months where the treatment is actually supposed to be active, post-prescription. However, I am struggling to apply this only to the relevant months for each subject after the initial prescription observation.

Could someone help me devise a Stata code or suggest a method to correctly propagate the treatment duration data across the relevant months?


This is an example of how my data looks:

id	treat_duration	months_fu
1	5	0
1	.	1
1	.	2
1	.	3
1	.	4
1	.	5
1	.	6
1	.	7
1	.	8
1	.	9
1	4	10
1	.	11
1	.	12
1	.	13
1	.	14
1	,	15
1	.	16
1	.	17
1	.	18
1	.	19
1	.	20
1	3	21
1	.	22
1	.	23
1	.	24
1	.	25

What I need is to generate the "on_treatment" variable:

id	treat_duration	months_fu	on_treatment
1	5	0	yes
1	.	1	yes
1	.	2	yes
1	.	3	yes
1	.	4	yes
1	.	5	no
1	.	6	no
1	.	7	no
1	.	8	no
1	.	9	no
1	4	10	yes
1	.	11	yes
1	.	12	yes
1	.	13	yes
1	.	14	no
1	,	15	no
1	.	16	no
1	.	17	no
1	.	18	no
1	.	19	no
1	.	20	no
1	3	21	yes
1	.	22	yes
1	.	23	yes
1	.	24	no
1	.	25	no

Thank you in advance for your help!

[Benno Schoenberger]

Hi Sergio,

the following code should work for your snippet of data and I hope it also will for your whole dataset:

clear
input int (id     treat_duration     months_fu)
1     5     0
1     .     1
1     .     2
1     .     3
1     .     4
1     .     5
1     .     6
1     .     7
1     .     8
1     .     9
1     4     10
1     .     11
1     .     12
1     .     13
1     .     14
1     ,     15
1     .     16
1     .     17
1     .     18
1     .     19
1     .     20
1     3     21
1     .     22
1     .     23
1     .     24
1     .     25
end


gen on_treatment = treat_duration + months_fu -1
bysort id (months_fu): replace on_treatment = on_treatment[_n-1] if !missing(on_treatment[_n-1]) & on_treatment[_n-1]>=months_fu
recode on_treatment (.=0)(else=1)
label define yn 1 "yes" 0 "no"
label values on_treatment yn

All the best,
Benno

[Sergio Serrano]

It works perfectly, very elegant!
I was stuck with this. Thank you so much, Benno.