This bias occurs when survivors of a highly lethal disease are more likely to enter a study than other cases.
Example:
let's suppose we study the role of age as a potential risk factor for
viral haemorrhagic fever (VHF), and that the study includes only those
who are still alive at the time of the study. If older age is associated
with VHF death, this will decrease the proportion of cases over a
certain age in the study, and consequently underestimate the odds ratio. This is illustrated in the following table:
Age group |
VHF survivors |
VHF deaths |
Total cases |
Controls |
ORs |
ORtt |
< 50 years |
60 |
80 |
140 |
50 |
1.5 |
2.3 |
≥ 50 years |
40 |
20 |
60 |
50 |
ref. |
ref. |
Total |
100 |
100 |
200 |
100 |
|
|
ORs = OR with all surviving cases |
ORtt = OR with all cases |