Tag only one observation per value per variable per ID

[Cathrine Christiansen]

How can I count the number of episodes per patID? And tag only one of the observations, per episode, per patID?
I would like to have something like
patID 76779 episodes:2, tot_dose in episode 1: 50, tot_dose in episode 2: 50.
patID 76812 episodes: 1, ttot_dose in episode 1: 250
patID 77072 episodes: 2, tot_dose in episode 1: 100, tot_dose in episode 2: 200

* Example generated by -dataex-. To install: ssc install dataex
clear
input long patID float(date med1 dose gap episode tot_dose)
76779 19058 1  50    . 1  50
76779 21626 1  50 2568 2  50
76796 19109 1  50    . 1  50
76804 21509 1  50    . 1  50
76806 20612 1  50    . 1 100
76806 20685 1  50   73 1 100
76812 19108 1  50    . 1 250
76812 19197 1  50   89 1 250
76812 19319 1  50  122 1 250
76812 19423 1  50  104 1 250
76812 19696 1  50  273 1 250
76837 19709 1  50    . 1  50
76911 20403 1  50    . 1 200
76911 20458 1  50   55 1 200
76911 20511 1  50   53 1 200
76911 20581 1  50   70 1 200
77072 20485 1  50    . 1 100
77072 20544 1  50   59 1 100
77072 21565 1  50 1021 2 200
77072 21595 1  50   30 2 200
77072 21862 1  50  267 2 200
77072 21875 1  50   13 2 200
77103 19355 1  50    . 1 100
77103 19425 1  50   70 1 100
77103 20381 1  50  956 2 200
77103 20441 1  50   60 2 200
77103 20597 1  50  156 2 200
77103 20780 1  50  183 2 200
77516 19086 1  60    . 1 120
77516 19107 1  60   21 1 120
77675 20170 1  50    . 1 100
77675 20251 1  50   81 1 100
77745 21265 1  50    . 1  50
77745 21803 1  50  538 2  50
77906 19149 1  50    . 1 100
77906 19286 1  50  137 1 100
77930 21346 1  50    . 1 250
77930 21428 1  50   82 1 250
77930 21503 1  50   75 1 250
77930 21566 1  50   63 1 250
77930 21632 1  50   66 1 250
77939 19976 1  50    . 1  50
78173 20801 1  60    . 1  60
78624 19607 1  60    . 1 110
78624 19767 1  50  160 1 110
78624 21909 1  50 2142 2  50
78663 20045 1  50    . 1  50
78663 21256 1  50 1211 2 100
78663 21365 1  50  109 2 100
end
format %td date

[Øyvind Snilsberg]

maybe,

bys patID (episode): gen episodes = episode[_N] 
bys patID (episode): gen tot_dose_ep1 = tot_dose[1] 
bys patID (episode): gen tot_dose_ep2 = tot_dose[_N] if episode[1]!=episode[_N]