Public awareness about risk factors could pose problems for case-control studies: the example of sunbed use and...

In a large case-control study we found no association between sunbed use and melanoma risk, but indications for potential recall and recruitment biases made the interpretation of the results difficult. Associations with skin phototype (adj OR for skin type I vs. IV: (2.6, 95% CI 1.5-4.8)), hair colour (adj OR red/blond vs. brown/black 2.0 (95% CI 1.4-2.8)) and number of naevi on both arms (OR>10 vs. 10 3.13 (95% CI: 2.47; 3.97)) were comparable to previous studies, but negative associations were found between sun exposure and melanoma risk (adj. OR 0.87 (95% CI: 0.65-1.18)) and in cases between sun exposure and naevus count. These observations led us to speculate that cases may have underreported their sun exposure and, most likely, their sunbed exposure. High percentages of sunbed use among controls indicated possible recruitment bias: eligible controls who were sunbed users were probably more likely to accept the invitation to participate than non-users, possibly due to a feeling of 'guilt' or 'worry' about their habits. Such selective participation may have strongly influenced the risk estimates of sunbed use in our study.