Format

Send to

Choose Destination
J Expo Sci Environ Epidemiol. 2015 Jan;25(1):72-9. doi: 10.1038/jes.2014.76. Epub 2014 Nov 5.

Validation of self-reported start year of mobile phone use in a Swedish case-control study on radiofrequency fields and acoustic neuroma risk.

Author information

1
Unit of Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden.
2
Unit of Biostatistics, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden.
3
Department of Clinical Neuroscience, Karolinska Hospital, Stockholm, Sweden.

Abstract

The possible effect of radiofrequency exposure from mobile phones on tumor risk has been studied since the late 1990s. Yet, empirical information about recall of the start of mobile phone use among adult cases and controls has never been reported. Limited knowledge about recall errors hampers interpretations of the epidemiological evidence. We used network operator data to validate the self-reported start year of mobile phone use in a case-control study of mobile phone use and acoustic neuroma risk. The answers of 96 (29%) cases and 111 (22%) controls could be included in the validation. The larger proportion of cases reflects a more complete and detailed reporting of subscription history. Misclassification was substantial, with large random errors, small systematic errors, and no significant differences between cases and controls. The average difference between self-reported and operator start year was -0.62 (95% confidence interval: -1.42, 0.17) years for cases and -0.71 (-1.50, 0.07) years for controls, standard deviations were 3.92 and 4.17 years, respectively. Agreement between self-reported and operator-recorded data categorized into short, intermediate and long-term use was moderate (kappa statistic: 0.42). Should an association exist, dilution of risk estimates and distortion of exposure-response patterns for time since first mobile phone use could result from the large random errors in self-reported start year. Retrospective collection of operator data likely leads to a selection of "good reporters", with a higher proportion of cases. Thus, differential recall cannot be entirely excluded.

PMID:
25352163
DOI:
10.1038/jes.2014.76
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Nature Publishing Group
Loading ...
Support Center