Development of Clinical Nutrition Staffing Model

Abstract

Study objectives were to (1) determine which patient and organizational variables affect time spent by clinical nutrition staff in performing direct patient care activities; (2) incorporate these variables into a model for predicting staffing requirements for the provision of clinical nutrition services at VA medical centers (VAMCs); and (3) validate the model. Approximately 230 registered dietitians and dietetic technicians in 27 randomly selected VAMCs completed logs of time spent and direct patient care services provided over a two-week period, resulting in time data on 1289 inpatient episodes and 1074 outpatient visits. Automatic Interaction Detection analysis software was used to identify the variables with the most significant main and interaction effects on time requirements. Model validation was performed by comparing model predictions with actual FY 1996 staffing at 80 VAMCs. The resultant inpatient model includes 18 patient groups categorized by nutrition status, bedsection, and length of stay. These groups account for 32% of the variance in time requirements. The outpatient model consists of six patient groups categorized by education services (yes/no), interventions (yes/no), visit type (recurring or non-recurring), and medical center type. These groups account for 25% of the variance in time requirements. The model accurately predicted (within 95% confidence interval) 1996 staffing for 69 of the 80 validation sites (69%, predictive validity). All sites except one (99%) accepted the model as valid (face validity). Results demonstrate that a simple, valid model with modest data requirements can be developed which accounts for a significant portion of the variability in time spent providing clinical nutrition services. The model can be used for determining staffing requirements and monitoring productivity.