Referral and ascertainment bias in patients with synchronous and metachronous endometrial malignancy.

Abstract
The purpose of this study was to evaluate the frequency in patients with endometrial cancer of other malignancies and the influence of referral and ascertainment biases on these associations.
Analysis of 1,028 local and referred patients who had a hysterectomy for endometrial cancer was based on residence at the time of diagnosis.
Altogether, 208 patients had a history of another malignancy, most frequently breast, colon, and ovary. At the time of surgery for endometrial cancer, the prevalence of lymphoma and breast and ovarian cancers was greater than expected although the higher prevalence of lymphoma was limited to referred patients. During follow-up after hysterectomy, the incidence of lung cancer was lower than expected, whereas the incidence of lymphoma was higher. Breast, colorectal, and bladder cancers were more common than expected although this finding was limited to local patients.
We concluded that results of epidemiologic studies from tertiary care centers may be misleading if they do not account for referral and ascertainment biases.