We are comparing the incidence rate of two age groups in a cohort study; those under and over 50 years of age. As patients age, they will pass from one group to the next, so their count for some person-years in the first group, and than the second. Is there a way to adjust for the time since they are in the cohort?
In this particular case, we are dealing with HIV patients, so as patients become infected, they will be included in the cohort, and during their follow-up they might have some sort of event due to the infection. What I'm wondering is, if a patient has an event at 52, and counts as 0.5 person-years for the >50, is there a way to adjust for time depending on when this person was infected at 58 or at 30 (so, in essence, to adjust for time since infection)?
If the comparison is incidence of event (superinfection say) in persons wth HIV > 50 yrs age with incidence of event in persons with HIV < 50 years of age why not just count the person years of exposure (HIV) from start of HIV and compare no.. of cases of eventin persons>50 / py of all HIV cases >50 with no. of cases of event in persons
Ia the patient excluded from cohort If the had an opportunistic infection or can that patient be reincluded in the cohort after the recovery of infection?
If the patient is not reincluded, It easier to understand the research as the study of two independent cohort. One with patient ≤50 whose follow-up time ends at 50 Years old. Other with > 52 with no previous infection, with no precise end of follow-up time (death). There would be patients considered member of the two cohort, tipically those who started in the first cohort and that had no infection ou other events, and that has been censured at 50 Years. The follow-up time in the first cohort count from hiv infection to 50 Years. And in the Second counts from hiv infection to event or to the end of follow-up time as define in research.
If the patient would be allowed to be reincluded the same rules and procedure should be use in both cohort.
The difference would be the interpretation of the event previous of 50 Years in the cohort of > 50. It should be analysed as potential risk factor for development of event after 50 Years.
James Leigh thanks! That's what I was planning on doing, calculating the incidence rates and then the ratio between the two age groups. What I worry is that does patients that have an event at >50 will probably have been in the cohort for much longer, but will not weight as much in person years of exposure (in my example, a 52 year-old in treatment since he is 30 will only add 2 person years to the >50 group, but he might be at more risk since he was infected a while back, while a 52 year-old infected at 48 will also add 2 person years to the >50 group, but might be at less risk - that's what I was wondering on how to adjust for).
Marcelo G Land thanks!
The patients will not be excluded. The event that we are looking at is viral failure, so if a patient presents viral failure, that will count as an event, but since the patient is still in follow-up, he will not be excluded.
The conceptual dynamic cohort is based on those diagnosed with HIV. The person years are person years of exposure (being HIV +).
so the 52 years old event case diagnosed HIV at 30 and event at 52 would have contributed 22 py exposure, while the 52 year old diagnosed at age 48 would have conrtibuted 4 py exposure. This adjusts for time in cohort.
James Leigh
Well, the idea was to have the 52 year old diagnosed at HIV to count as 20 person-years for the 50 group, but talking to my collegues, it seems this is not done, to have an individual count in two different groups of follow-up.
Marcelo G Land
the idea was to take as zero time the date of cohort uptake, when the patients started being followed.
if you take this approach, alternative methods would be to restict study to comparable exposure durations or to use regression modelling with length of exposure as a covariate.
- Badges
- Science topic
- Similar topics
- Psychology