Fruit and vegetable intake of US adults: comparing intake by mode of survey administration

Modes for collecting dietary data vary across studies and include in-person/interviewer-administered surveys, mail, and telephone surveys. Few studies use mixed modes to assess dietary intakes. Using data from the 2007 Health Information National Trends Survey, we compared fruit and vegetable intake of adults measured through dual modes (mail and random-digit dial [RDD] telephone), and discussed potential factors that could account for mode differences. The Health Information National Trends Survey data were collected through mailed (n=3,582), and RDD (n=4,092) surveys from December 2007 to May 2008. Data were weighted and analyzed in SUDAAN. Unadjusted mean fruit and vegetable intake was 0.48 servings higher (P<0.001) among mail (mean 5.40) vs RDD (mean 4.09) participants. In a multivariate model that controlled for other predictors, the odds of consuming ≥5 servings of fruit and vegetables per day was 83% higher among mail respondents compared to RDD (odds ratio 1.83, 95% confidence interval 1.62 to 2.07). Other predictors of fruit and vegetable intake were sex, education, participation in physical activity, self-rated health, and knowledge of the fruit and vegetable recommendation. Methodologic issues may account for modal differences in fruit and vegetable intake. Different measures (cups, servings) were used to assess fruit and vegetable intake in both modes, details about portion sizes were provided on the mail mode vs RDD, and closed-ended responses were provided on the mail vs open-ended responses for RDD. We cannot recommend one mode over the other nor attribute mode differences to real differences in reported fruit and vegetable intake between participants from both modes. Future research that uses dual modes needs to use identical methods of dietary assessment to minimize these potential sources of error. Further research is needed to validate the use of dual modes to assess dietary intake and inform research practice.