Detecting score streak disruptions

[Maximilian Mayrhofer]

Hi everyone,

I have a dataset in long formate showing patients with their date of surgery and pain scores taken at postoperative visits (in no strict intervals). Then I got a variable (by the help of this forum) for pain remission which is == 1 if patients report a pain score of <=3 for any timeframe of >= 91.3125 days during their postoperative followup. Now I would like to know which patients are able to sustain that pain remission for the whole time of their follow up and which have a rebound to pain >3.

For that I tried creating variables which show the beginning and the end of that timeframe of pain remission and next I would try to combine them with the date of the last pain score reported (end of follow up) and then try to find out if there is anything >3 in between their timeframe of remission and the end of their followup. That is the code I started writing to find out the beginning and the end of that timeframe:

by record_number (time_surgerytopainscore), sort: ///
gen rem_start = time_surgerytopainscore[_N-1] if time_surgerytopainscore[_N] - time_surgerytopainscore[_N-1] >= 91.3125 & pain_remission == 1 & low_pain_score == 1 & score_pain_ <= 3

by record_number (time_surgerytopainscore), sort: ///
gen rem_stop = time_surgerytopainscore[_N] if time_surgerytopainscore[_N] - time_surgerytopainscore[_N-1] >= 91.3125 & pain_remission == 1 & low_pain_score == 1 & score_pain_ <= 3

I'm sure there is an easier way to do this, so I would appreciate any help!


Here is the post for the original detection of pain remission: https://www.statalist.org/forums/for...ions-in-a-loop

Example data:

* Example generated by -dataex-. To install: ssc install dataex
clear
input int record_number str5 no_painscore byte score_pain_ float time_surgerytopainscore byte low_pain_score int run_num byte long_run float(pain_remission rem_start rem_stop)
113 "1"   3    1 1  1 0 1   9 106
113 "2"   2    2 1  1 0 1   9 106
113 "3"   2    3 1  1 0 1   9 106
113 "4"   2    4 1  1 0 1   9 106
113 "5"   3    5 1  1 0 1   9 106
113 "6"   4    6 0  2 0 1   .   .
113 "7"   4    7 0  2 0 1   .   .
113 "8"   1    8 1  3 1 1   9 106
113 "9"   2    9 1  3 1 1   9 106
113 "10"  3  106 1  3 1 1   9 106
834 "1"   7    1 0  1 0 1   .   .
834 "2"   5    2 0  1 0 1   .   .
834 "3"   7    3 0  1 0 1   .   .
834 "4"   6    4 0  1 0 1   .   .
834 "5"   6    5 0  1 0 1   .   .
834 "6"   6    6 0  1 0 1   .   .
834 "7"   5    7 0  1 0 1   .   .
834 "8"   4    8 0  1 0 1   .   .
834 "9"   6   17 0  1 0 1   .   .
834 "10"  4   62 0  1 0 1   .   .
834 "11"  0   83 1  2 1 1  83 216
834 "12"  1  216 1  2 1 1  83 216
920 "1"   7    1 0  1 1 0   .   .
920 "2"   6    2 0  1 1 0   .   .
920 "3"   6 1270 0  1 1 0   .   .
911 "1"   4    1 0  1 0 0   .   .
911 "2"   0   27 1  2 0 0   .   .
911 "3"   1   69 1  2 0 0   .   .
221 "1"   7    1 0  1 0 1   .   .
221 "2"   8    6 0  1 0 1   .   .
221 "3"   3   34 1  2 1 1  34 174
221 "4"   3  174 1  2 1 1  34 174
102 "1"   2    1 1  1 1 1 482 649
102 "2"   0   77 1  1 1 1 482 649
102 "3"   0  146 1  1 1 1 482 649
102 "4"   0  181 1  1 1 1 482 649
102 "5"   4  228 0  2 0 1   .   .
102 "6"   0  229 1  3 1 1 482 649
102 "7"   0  230 1  3 1 1 482 649
102 "8"   1  231 1  3 1 1 482 649
102 "9"   2  232 1  3 1 1 482 649
102 "10"  0  233 1  3 1 1 482 649
102 "11"  0  234 1  3 1 1 482 649
102 "12"  0  235 1  3 1 1 482 649
102 "13"  0  236 1  3 1 1 482 649
102 "14"  0  286 1  3 1 1 482 649
102 "15"  1  391 1  3 1 1 482 649
102 "16"  0  482 1  3 1 1 482 649
102 "17"  6  649 0  4 0 1   .   .
123 "1"   7    1 0  1 0 1   .   .
123 "3"   8    3 0  1 0 1   .   .
123 "4"   0    4 1  2 1 1 367 598
123 "5"   2    5 1  2 1 1 367 598
123 "6"   0    6 1  2 1 1 367 598
123 "7"   0    7 1  2 1 1 367 598
123 "8"   0    8 1  2 1 1 367 598
123 "9"   0    9 1  2 1 1 367 598
123 "10"  0   10 1  2 1 1 367 598
123 "11"  1   67 1  2 1 1 367 598
123 "13"  0  134 1  2 1 1 367 598
123 "12"  1  158 1  2 1 1 367 598
123 "2"   7  367 0  3 0 1   .   .
123 "14"  1  598 1  4 0 1 367 598
end

[Clyde Schechter]

If you want the dates of start and end of each remission, that's fine. But you don't need them to identify which patients had a sustained remission. For that, all you need is:

by record_number (run_num time_surgerytopainscore), sort: ///
    gen byte had_sustained_remission = long_run[_N] == 1 & score_pain_[_N] <= 3

[Maximilian Mayrhofer]

Works perfectly, thank you once again!