Accuracy and reliability of direct observations of home-packed lunches in elementary schools by trained nutrition students.

Increased attention has been directed toward the school food environment because children consume important contributions toward their daily food intake while at school. In Canada, most elementary school students bring a lunch to school and there are minimal data on the composition and consumption of these lunches. Dietary assessment of home-packed lunches is challenging compared with assessment of standardized school meals due to greater diversity of items, nonstandard portions, and opaque containers. We assessed accuracy and reliability of a food observation method whereby upper-year nutrition students (n=15) were trained to assess packed lunch contents and intake in elementary schools. Accuracy and reliability was assessed during 2010-2011 in three observational phases: sample lunches, volunteer-consumed lunches, and elementary school students' lunches (n=32). Observers accurately identified 96% and 95% of items in the sample and volunteer lunches, respectively. Similarly, they accurately reported portion sizes for 86% and 94% of the items in the sample and volunteer lunches, thus showing improvements in successive phases. Interobserver reliability for amount consumed, by portion size and macronutrient content, ranged from 0.79 to 0.88 in the volunteer-consumed lunches and 0.78 to 0.86 in the students' lunches, with a majority ≥0.80. It is noteworthy that the analyses for the amount consumed were conducted as absolute amounts with no allowances for discrepancies, which differs from other interobserver reliability assessments where as much as 25% discrepancy is considered agreement. Observers with prior nutrition knowledge assessed packed lunch contents and intake accurately and reliably by direct observation in an elementary school setting.