Differences in dietary supplement use and secular and seasonal trends assessed using three different instruments in the EPIC-Norfolk population study

Objective: Supplement use has increased over time and measurements of supplement use are dependent on instruments chosen. Therefore, we investigated three different questionnaires to measure supplement use and whether these results were associated with age, year of recruitment (secular trend, 1993-1998), and seasonal trends. Design: The questionnaires were self-administered within a median time interval of 54 days and included a Health and Lifestyle Questionnaire (1-year recall), a Food Frequency Questionnaire (1-year recall), and a 7-day diet diary. Setting: Men and women, aged 40-79 years from the general population living in Norfolk (East Anglia, UK), recruited between 1993 and 1998 in the European Prospective Investigation into Cancer Study (EPIC-Norfolk). Results: The prevalence of supplement use estimated with different instruments ranged from 31.7% to 39.0% for men and 45.0% to 54.3% for women. Agreement was substantial with kappa-statistics between 0.72 and 0.81. Participants' age (men only) and recruitment year were independently associated with supplement use; season showed inconsistent results. Conclusions: The diary provides a good agreement as measured by the kappa-statistic, compared to more long-term measures of supplement use classification. The secular and seasonal trends in supplement use and type of assessment instrument need to be taken into account in studies on health and supplement use.