Cotinine as a measure of smoking in observational studies of multiple sclerosis.

Studies using cotinine levels to define smokers have generally failed to detect an association between smoking and multiple sclerosis (MS). Using a Swedish population-based case-control study, we show that associations in relation to MS risk and progression differ considerably depending on how smoking is measured. The risk of conversion into secondary progressive disease was increased among smokers when self-reported smoking history, but not presumed cotinine levels, was used to define smokers. Defining smoking by cotinine levels without distinguishing between different sources of nicotine may lead to severely biased estimates of the association between smoking and both MS risk and progression.