Inherent flaws in a method of estimating meal intake commonly used in long-term-care facilities

Objective To evaluate the accuracy of a method used to estimate nursing home residents' meal consumption, where the meal tray is assessed as a whole and assigned a value of 0%, 25%, 50%, 75%, or 100% consumed, under both routine and controlled conditions. Subjects/setting This study was conducted with certified nursing assistants at a 180-bed long-term-care facility in Miami, Fla. Methods Study 1 evaluated the method under routine conditions by comparing nursing assistants' estimates to actual meal consumption of 42 residents over 109 meals. A second study evaluated the method in a controlled setting where nursing assistants were free of disincentives and distractions that might contribute to inaccurate reporting. In a crossover design, nursing assistants estimated consumption on 4 manipulated trays under conditions of both immediate and delayed reporting. Statistical analysis An intraclass correlation coefficient and percent agreement were used to compare nursing assistants' estimates to weighed meal intakes. Results Under routine conditions, the intraclass correlation coefficient between nursing assistants' estimates and the actual resident meal consumption was weak at 0.464 (95% confidence interval=0.146 to 0.664). The correct estimate was recorded 44% of the time. In the controlled setting, the nursing assistants' estimates for percent consumed agreed with weighed intakes 44% and 38% of the time with immediate and delayed recording, respectively. Applications/conclusions This 1-step method of estimating meal consumption with an overall percentage is not sufficiently accurate to identify residents who are eating less than 75% of most meals.