Background: The assessment of the impact of data quality issues, such as omitting to answer questions on a food frequency questionnaire (FFQ), is important in all study populations, including those in early old age. Assumptions about the limited nature of diets of older participants may influence the treatment and interpretation of their dietary data.
Subjects/methods: The Boyd Orr cohort is a long-term study based on 4999 UK men and women whose families took part in a survey of diet and health during 1937-1939. In 1997-1998, all 3182 traced, surviving study members, then aged 60 years and over, were sent a health and lifestyle questionnaire, including a 113-item FFQ, primarily to examine relationships between childhood and adult fruit, vegetable and antioxidant intakes. In-depth interviews were conducted with a purposively sampled subset of 31 respondents.
Results: Of the 1475 subjects who returned the questionnaire, 11% (n=161) had missing data on their FFQ. Those who omitted answers to more than 10 questions (n=127; 8.6%) were more likely to be aged over 70, to be female, but no more likely to report being overweight than those with 10 or fewer missing answers. Follow-up by telephone or post to reassess missing FFQ data was successful for 102 of the subjects with more than 10 omitted answers. Mean intakes of energy, fruit and vegetables, and selected nutrients were significantly increased after reassessment. The use of 'cross-check' questions to weigh fruit and vegetable intake (n=1383) showed potentially systematic errors in the reporting of these foods, vitamin C and carotene. Analysis of interview data among a subset of participants partially challenged stereotypical views of the diets of older people with, for example, increased freedom in food choice associated with life transitions.
Conclusions: Food frequency questionnaires for those in early old age, as for others, need to meet competing demands of being comprehensive for those with varied diets, while not being so onerous that they deter completion. Reviewing questionnaires with participants remains important in this group, as omitting to answer questions on the FFQ does not necessarily equate to non-consumption. Qualitative interviews may aid in the interpretation of the quantitative data obtained.