While you see both these with respect to cancer screening, I wanted to know the difference between the two to compare between interventions and survival.
Consider two groups A and B.
Group A: Receives treatment with A and then is followed until death.
Group B: Receives treatment with A. After several weeks receives treatment with B. And then followed until death.
Note: This is not a prospective randomized trial.
Now, an early death prevents a patient from getting B. Patients who died
quickly had less time available to get transplants and will be counted as group A. Is this (Time A to B) causing an immortal time bias or lead time bias?
Hi, Kushal,
To me, the scenario you described is called allocation bias or misclassification bias.
I would also like to follow this question.
Victor
Dear Kushal,
Lead time bias occurs when you evaluate the effect of a screening programme and the outcome measure is survival time. Consider two persons with same etiological disease and same progression with respect to time and same severity and same time of death, but one (A) is in the population with screening programme and the other (B) is in the population without screening programme. There is potential for A to be detected earlier (during DPCP - Detectable pre-clinical phase) than B (after clinical signs and symptoms). Thus you happen to overestimate the survival for A just because A happened to be in the population with screening and therefore detected early not because A survived longer than B.
Immortal time bias occurs when there is gap between the two timings; one, time of start of follow-up, and two, time of administration of drug distinguishing the two groups.
The situation you have described has the potential for immortal time bias.
THUS, THERE IS CONCEPTUAL SIMILARITY IN LEAD TIME BIAS AND IMMORTAL BIAS, HOWEVER, THE TWO HAVE DIFFERENCE IN THE CIRCUMSTANCES IN WHICH THEY OCCUR, ALSO IN THE WAY TWO ARE DEALT WITH.
About the difference between the two biases: Lead-time bias is due to an intervention (e.g. screening) that moves diagnosis to an earlier point in time. Immortal time bias is due to defining exposure/treatment in a way that the outcomes cannot occur during a period of time.
But: Why are you so focused on getting the right label on your bias? There are infinite ways of producing biased results and there is only a limited number of established labels for biases, many of which are not very well defined. Therefore it seems better to me to put your effort into clearly describing what is the actual bias you are dealing with, instead of trying to reduce your description to an ambiguous label.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package