Purpose: To determine whether interviewed non-Hodgkin's lymphoma (NHL) patients differed from non-interviewed NHL patients and to revise the population-based risk estimate for NHL associated with HIV infection in homosexual men in the San Francisco Bay Area. Differences between study participants and non-participants are a concern in epidemiologic studies. Recruitment techniques can be modified when nonparticipation is a result of subject refusal. However, the solutions for nonparticipation due to death are less clear, especially when rapid case ascertainment (RCA) is used to identify newly diagnosed patients.
Methods: A large population-based case-control study was conducted in the San Francisco Bay Area between 1988 and 1995 to investigate risk factors for NHL. Nearly 1600 patients were interviewed and included in this study. A total of 1047 NHL patients were not interviewed because they had died prior to contact (602 patients) or we had received their RCA information after the study had ended (445 patients). We compared available demographic, clinical, and histologic profiles of the interviewed and non-interviewed NHL patients.
Results: Compared with the interviewed group, the non-interviewed group was younger, had more single HIV-infected homosexual men and a higher incidence of high-grade and "not-otherwise-specified" lymphoma. When categorized by sex and sexual preference, median age at diagnosis of NHL was the same for the interviewed and non-interviewed homosexual men. Combining interviewed and non-interviewed patients, the estimated risk for NHL associated with HIV infection in homosexual men [odds ratio (OR) = 50] is substantially higher than we had estimated in our earlier publications (OR = 20).
Conclusions: High-grade lymphomas associated with poor survival in homosexual men are likely to be under-represented in case-control studies of NHL.