When I use the atlas to do the incidence rate analysis like the picture, I am a little confused about how to compute the Time At Risk(years). I know the formula of the incidence rate is equal to the case num/time at risk(years) just like in thebookofohdsi, but I am a little not sure how-to compute ‘person-time at risk contributed by persons in cohort’, time at risk(years), thank you very much
Each person that is included in the IR calculation (those people who did not have the prior outcome) contributes ‘time at risk’ (TAR) to the IR calculation.
Time at risk starts at a certain offset from the Target Cohort’s cohort_start date (by default, the Time at Risk starts at cohort_start date, but you can change the analysis settings to have a ‘fixed offset’ from the cohort start/end date). So, the point here is that you have to understand how your Target Cohort’s start and end dates are defined: by default, the cohort’s end date will be set to the enclosed observation period’s end date, and the IR analysis will use the cohort start-end dates as the Time At Risk.
A patient’s Time at Risk is calculated as starting at the TAR start date, and ending at the earliest of 1) the TAR end date (and this is not a case) or the occurrence of the outcome in your Outcome Cohort (and this is a case).
Once you have all the TAR that was contributed by all patients, you can, as you described, calculate the IR as the cases / Total TAR.
Note, if the outcome event happens at the date of the TAR end date, the outcome event takes priority, so you will get all the TAR for the person, and the person will appear as a case in the result. If the outcome occurs after the TAR, the only the time during TAR is contributed to the IR analysis, and this person is not a case. A patient is only a ‘case’ if the outcome appears during the time at risk.
Last thing: while we use ‘days’ in the underlying calculations, once we find the total number of days, we convert it to years for purposes of calculating the IR in ‘person years’.