Linet MS, Taggart T, Severson RK, Cerham JR, et al. (2006)
This study was part of a comprehensive investigation of possible causes of non-Hodgkin lymphoma (NHL). The authors carried out a multicentre case-control study of NHL in 4 regions covered by the National Cancer Institute's population-based surveillance. Cases had a primary diagnosis of NHL at ages 20-74 between July 1988 and June 2000. All consecutive cases were included in Iowa and Seattle, and all African-American and a random sample of Caucasian cases in Detroit and Los Angeles. Controls were selected from persons aged 20-74 residing in one of the four regions at initial contact and who had no previous diagnosis of NHL or HIV infection. Controls < 65 were identified using random digit dialing, and those 65 or older from Medicare files.
Information was obtained from a self-administered questionnaire and a computer-assisted questionnaire obtained in a home visit. Included were questions on use of cordless phones, car phones, and hand-held cell phones.
Of 905 presumed eligible cases, 204 were excluded because they had died, could not be found, or moved out of the region before contact, or because their physicians refused permission to contact them. Of the 701 remaining cases, 99 declined to participate and 51 were excluded because of serious illness, impairment, or failure to respond to multiple contacts. Thus, 551 cases (79% of those the authors attempted to recruit) participated. Using similar criteria, 462 eligible controls participated. This was 55% of the contacted persons.
Compared to those who had never used cell phones, risks were not increased among individuals whose lifetime use was less than 10 times (odds ratio (OR) = 0.9, 10-100 (OR = 1.0), or more than 100 times (OR = 0.9). Among regular users compared to those who had never used a cell phone, risks were not associated with minutes per week, duration, cumulative lifetime or year of first use. NHL was non-significantly higher in men who used cell phones for more than 8 years, but the number of cases was very small (7).
Limitations of the study were low participation rates, possible selection or recall bias, and the small numbers of long-term or heavy users.