PSet 8 Question 1C clarification

PSet 8 Question 1C clarification

by Melanie Molina -

Hi.

Myself and several of my classmates are confused by question 1C: 

"C.  Although these data are not provided in the paper, assume that in both randomization groups, the 5-year survival of those whose tumor size was < 1 cm at the time of diagnosis was longer than those whose tumor was ≥ 2 cm at the time of diagnosis.  

Could this difference in survival by tumor size at diagnosis be at least partly due to lead time bias?  Explain. [2]"

When the question asks about "difference in survival," is it asking about difference in survival between < 1cm and ≥ 2 cm tumors? Or is it asking about "difference in survival" BETWEEN intervention and control groups, assuming the intervention group was more often diagnosed with <1cm tumors because they received specific instruction on BSE versus the control group, which presumably presented with ≥2cm tumors more often because they weren't instructed on BSE? Or were these "differences in survival" both present equally(?) in the intervention and control groups because we are assuming this "in both randomization groups"?

Thanks.

Re: PSet 8 Question 1C clarification

by Nicole Rodriguez -
Hi Melanie,

The question is asking about the difference in survival for people exhibiting the different tumor types - not the difference in survival between the intervention and control groups. You do not need to consider group assignment to answer this question, just focus on the impact of the distinct tumor classifications on the possibility of lead time bias.

Hope this helps clarify!

Best,
Nicole

Re: PSet 8 Question 1C clarification

by Thomas Newman -

Dear Melanie,

Thanks for asking!  

The difference in survival we are talking about is between those whose tumors were small (< 1 cm) and larger (>= 2 cm).  You can ignore randomization groups for this part of the problem.

Also, a clarification suggested during office hours was to reword the question to ask if it could be due to "lead time" rather than to "lead time bias."

Let me know if this is still not clear.

Best wishes,

Tom