I am Masters student and working on Stata 14.2 obtained by my University.
I am attempting to calculate the measure the magnitude of antibiotic usage prior to a new incidence of a repeated illness. I ought to calculate two variables namely, the direct and indirect effects of using prescribed antibiotic on a set of illnesses.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input str23 ID str9 Illness float(YearofContracting DosageOfAntibiotic ClassofIllness) "Patient 1" "Illness 1" 2010 37 1 "Patient 1" "Illness 1" 2008 100 1 "Patient 1" "Illness 1" 2011 49 1 "Patient 1" "Illness 2" 2001 15 1 "Patient 1" "Illness 2" 2012 60 1 "Patient 1" "Illness 3" 2006 26 2 "Patient 1" "Illness 4" 2018 80 2 "Patient 2" "Illness 1" 2013 89 1 "Patient 2" "Illness 1" 1999 15 1 "Patient 2" "Illness 2" 2016 90 1 "Patient 3" "Illness 1" 2001 18 1 "Patient 3" "Illness 4" 1987 10 3 "Patient 4" "Illness 7" 2002 30 3 "Patient 5" "Illness 8" 2004 60 4 end
I would like to explain with an example:
Direct effect
As per the Patient1 below, Illness 1 struck in year 2008, 20010 and 2011, hence, each successive occurrence of the same illness and its prescribed dosage has some impact on the patient's next occurrence of the same illness. By some logic, we believe that there is an impact of time gap signified by a constant multiple (let's call it gamma).
For Time lapse and constant multiple relationship, please refer the table below
Time lapse | constant scale multiple (=gamma) |
1 | .1 |
2 | .05 |
3 | .007 |
4 | .001 |
5 and above | .0003 |
Incident1 of Illness 1: Direct effect (in year 2008)= 0
Incident2 of Illness 1: Direct effect (in 2010) = (corresponding gamma for time lapse between 2010-2008 )* DosageofAntibiotic (in 2008) * Count of same illness in year 2008
Incident3 of Illness 1: Direct effect (in 2011) = (corresponding gamma for time lapse between 2011 and 2008)* DosageofAntibiotic (in 2008)*Count of same illness in year 2008 + (corresponding gamma for time lapse between 2011-2010)* DosageofAntibiotic (in 2010)* Count of same illness in year 2010
Essentially:
[ATTACH=CONFIG]temp_16563_1577542093155_488[/ATTACH]
Indirect effect
For patient1, impact of illness 1 in the "year n" is dependent upon the prior incidents of illness 2 (because they are in same "Class of illness" - a variable in my data) that is incidents of illness 2 in year X (where X<n).
Time lapse | constant scale multiple (=alpha) |
1 | .08 |
2 | .021 |
3 | .0078 |
4 | .0004 |
5 and above | .000065 |
Incident2 of Illness 1: Direct effect (in 2010) = (corresponding gamma for time lapse between 2010-2008 )* DosageofAntibiotic (in 2008) * Count of same illness in year 2008
Array
This is my first post in this community and I would appreciate all the help.
Regards,
Ram.
0 Response to Generate sum of preceding values conditional on dates, id, and illness type
Post a Comment