My query is a bit complicated, and possibly more to do with logic than statistics.
I have a large population (300,000) and a common condition (10,000 cases)
Everyone is a potential control, but cases that are also controls are censored as controls when they become cases. Three controls are chosen for each case. No-one is used twice as a control. No case is its own control.
The event I am interested in occurs before the condition is diagnosed. An event in a control is only counted once. An event in a case may be counted twice if the case is also a control.
Is this analysis valid?
ZJ Nugent, PhD
Health Outcomes Analyst
Epidemiology & Cancer Registry
CancerCare Manitoba
675 McDermot Ave
Winnipeg R3E 0V9
CANADA
|