Question
The effect of sexually transmitted human papillomavirus (HPV) infection on the risk of cervical cancer was investigated in a study. A gynecological screening program in Sydney, Australia, identified 1,347 women free from cervical cancer who had abnormal epithelial cells suggestive of HPV infection. These exposed women were asked to return repeatedly for follow-up examinations (Pap smears) to identify cases of cervical cancer during an 8-year period following the initial examination. However, 311 women did not return for any followup visits despite reminders. The remaining 1,036 women, returning for one or more follow-up visits, contributed a total of 4,222 woman-years among the exposed. A total of 36 cases were identified in the exposed, all at the end of the 8 years. NOTE: You are only given information about the exposed people in this cohort. Additional information about the unexposed is not necessary to do the problem. Assume the unexposed numbers are unaffected by any bias.
a. Among the exposed, how many woman-years were lost to follow-up during this study?
b. Women could be more likely to return for follow-up visits if they developed minor bleeding from a cervical cancer. If so, in what direction would the IDR be biased by the loss of women to follow-up?
c. Assume that the 36 identified cases were all the cases that occurred in the 1,347 exposed women during the 8 years. By how much could the loss to follow-up of exposed women, according to B (above), have biased the incidence density ratio?
