Setting Use of dietary supplements has increased over recent decades in the UK. Although dietary supplement use has been described in a number of populations we know little about patterns of supplement use in older people.

Objective The objectives of this study were to describe dietary supplement use in the Hertfordshire Cohort Study (HCS), to determine patterns of supplement use, and to investigate the extent to which supplement user groups differ in terms of their characteristics, diets and morbidity.

Methods 3217 HCS participants, aged 59 to 73, were interviewed. Diet over the preceding 3 months was assessed by FFQ; compliance with “healthy” eating recommendations was defined using individual scores for a “prudent” dietary pattern, identified using principal components analysis. Details of the brand and dose of all dietary supplements taken in the preceding 3 months were recorded. Individual supplements were allocated to one of 10 groups based on their nutrient composition, and cluster analysis was used to define groups of supplement users. The demographic and lifestyle characteristics, and morbidity of the supplement user groups was examined.

Results 45.4% of men and 57.5% of women reported taking at least one dietary supplement in the previous 3-month period; the most commonly taken type of supplement was oils (43% of total supplement products taken), particularly fish liver oil. There were five distinct clusters of supplement users; these were common to men and women. They were labelled according to the principal supplement taken; oils, glucosamine, single vitamins, vitamins and minerals, and herbal products. For men, there were differences between the groups for BMI (p=0.030), “prudent” diet score (p=0.002), Hospital Anxiety and Depression score (p=0.013) and social class (p=0.012). For women, there were differences between the groups for age (p=0.030), “prudent” diet score (p=0.014) and social class (p=0.005). There was no significant difference between the groups for Hospital Anxiety and Depression score among the women. With the exception of a difference in diagnosis of diabetes among the women (p=0.021), there were no differences in morbidity between the supplement groups in either men or women.

Conclusion There were distinct patterns of supplement use in the HCS. Supplement user groups differed in their characteristics, but there were few differences in morbidity.