Comparison of visual estimation methods for regular and modified textures: real-time vs digital imaging.

A variety of methods are available for assessing diet; however, many are impractical for large research studies in an institutional environment. Technology, specifically digital imaging, can make diet estimations more feasible for research. Our goal was to compare a digital imaging method of estimating regular and modified-texture main plate food waste with traditional on-site visual estimations, in a continuing and long-term care setting using a meal-tray delivery service. Food waste was estimated for participants on regular (n=36) and modified-texture (n=42) diets. A tracking system to ensure collection and digital imaging of all main meal plates was developed. Four observers used a modified Comstock method to assess food waste for vegetables, starches, and main courses on 551 main meal plates. Intermodal, inter-rater, and intra-rater reliability were calculated using intraclass correlation for absolute agreement. Intermodal reliability was based on one rater's assessments. The digital imaging method results were in high agreement with the real-time visual method for both regular and modified-texture food (intraclass correlation=0.90 and 0.88, respectively). Agreements between observers for regular diets were higher than those for modified-texture food (range=0.91 to 0.94; 0.82 to 0.91, respectively). Intra-rater agreements were very high for both regular and modified-texture food (range=0.93 to 0.99; 0.91 to 0.98). The digital imaging method is a reliable alternative to estimating regular and modified-texture food waste for main meal plates when compared with real-time visual estimation. Color, shape, reheating, mixing, and use of sauces made modified-texture food waste slightly more difficult to estimate, regardless of estimation method.