Dear users of Statalist.
I am conducting a research project where I examine time to return to work in patients with ischaemic heart disease and with non-ischaemic heart disease.
However, I am having trouble calculating the overall (and group differentiated) cumulative incidence accounting for death and retirement as competing risks.
There are three patients who never returned to work but did not die or retire.
My question is: wouldn't the cumulative incidence never reach 100% in both overall and group differentiated cumulative incidence since three people will never reach the event of either returning to work, dying, or retiring?
I used the following code to calculate the overall cumulative incidence for return to work:
And the following code to calculate the cumulative incidence in ischaemic heart disease patients and non-ischaemic heart disease patients:
I am conducting a research project where I examine time to return to work in patients with ischaemic heart disease and with non-ischaemic heart disease.
However, I am having trouble calculating the overall (and group differentiated) cumulative incidence accounting for death and retirement as competing risks.
There are three patients who never returned to work but did not die or retire.
My question is: wouldn't the cumulative incidence never reach 100% in both overall and group differentiated cumulative incidence since three people will never reach the event of either returning to work, dying, or retiring?
Code:
* Example generated by -dataex-. For more info, type help dataex clear input float(event_RTW count genesis) 2 1 2 2 1 1 1 1 1 2 1 1 1 1 1 2 1 1 1 1 1 2 3 2 2 4 2 2 4 2 2 5 2 1 7 1 1 10 2 2 10 1 1 12 2 2 20 1 1 24 1 3 30 1 2 30 1 1 30 1 2 34 1 3 34 2 2 45 1 3 50 2 1 56 1 1 59 2 1 64 2 1 66 2 1 89 2 1 100 1 0 260 1 0 260 1 0 260 2 end label values event_RTW event_lbl label def event_lbl 0 "Never returned to work", modify label def event_lbl 1 "Returned to work", modify label def event_lbl 2 "Dead", modify label def event_lbl 3 "Retirement pension", modify label values genesis genesis_lbl label def genesis_lbl 1 "Ischaemic heart disease", modify label def genesis_lbl 2 "Non-ischaemic heart disease", modify
I used the following code to calculate the overall cumulative incidence for return to work:
Code:
stset count, fail(event_RTW=1) stcompet cif=ci hi=hi low=lo, compet1(2) compet2(3) gen ci_case=cif if event==1 gen lo_case=lo if event==1 gen hi_case=hi if event==1 sort _t stcomlist if event==1, compet1(2) compet2(3)
And the following code to calculate the cumulative incidence in ischaemic heart disease patients and non-ischaemic heart disease patients:
Code:
drop ci_case lo_case hi_case cif hi low // CIF for ischaemic heart disease patients stcompet cif=ci hi=hi low=lo, compet1(2) compet2(3) by(genesis) gen ci_case=cif if event==1 & genesis==1 gen lo_case=lo if event==1 & genesis==1 gen hi_case=hi if event==1 & genesis==1 sort _t stcomlist if event==1 & genesis==1, compet1(2) compet2(3) // CIF for non-ischaemic heart disease patients drop ci_case lo_case hi_case gen ci_case=cif if event==1 & genesis==2 gen lo_case=lo if event==1 & genesis==2 gen hi_case=hi if event==1 & genesis==2 sort _t stcomlist if event==1 & genesis==2, compet1(2) compet2(3)
