Purpose: Estimation of nutritional intake from photographs of served meals has been proposed. We examined the inter-observer agreement and validity of this method for assessment of the daily diet.
Methods: Twenty-five family members (13 men and 12 women, mean age +/- SD: 47.3 +/- 5.6 years) of students in a dietetic course performed both meal photography and traditional weighed dietary records for four days. For the photo method, two observers independently identified foods and estimated their portion sizes from the photographs, converting them into energy and nutrient intake with a food composition table. The inter-observer agreement with the photo method was assessed in terms of inter-observer ratios, correlations, and coefficients of variation (CV), and the method was validated using the weighed dietary records as the reference. Its running cost was also calculated.
Results: The inter-observer ratios for estimation of mean daily nutritional intake ranged from 0.89 (magnesium) to 1.14 (retinol) with a median of 1.03. The correlation coefficients between observers varied from 0.65 (saturated fatty acids [SFA]) to 0.92 (vitamin C) on a daily basis (median, 0.79), and from 0.65 (SFA) to 0.96 (vitamin C) on an individual basis (median, 0.78). The CVs ranged from 7.9% (energy) to 23.8% (carotenc) (median, 13.3%), and from 5.2% (energy and magnesium) to 17.8% (carotene) (median, 8.8%) for daily and individual intake, respectively. Regarding validity, the ratios between methods (the photo method/weighed dietary records) ranged from 0.96 (potassium and SFA) to 1.11 (retinal and salt) with a median of 1.00. Correlations between the two methods were distributed from 0.40 (salt) to 0.82 (vitamin C and retinol) on a daily basis (median, 0.67), and from 0.47 (salt) to 0.90 (vitamin C) on an individual basis (median, 0.74). The CVs ranged from 10.5% (energy) to 39.6% (carotene) (median, 16.9%), and from 6.1% (protein) to 20.6% (carotene) (median, 11.2%), respectively. It was calculated to cost 105 Japanese yen per meal when using a lens-attached film for 25 shots.
Conclusions: Inter-observer agreement and validity of the photo method, though varying with the nutrient, were generally acceptable. Although some modifications are needed for diners-out, the method appears to be useful as a dietary recording tool.