Observational Tools Counter Weight Loss

Abstract

NEW YORK — A nursing facility resident's medical record says she eats 75% of her meals with 5 minutes or more of feeding assistance 100% of the time, but she still loses weight unintentionally. How can that be? According to Sandra Simmons, Ph.D., who spoke at the New York Medical Directors Association recent education meeting, “medical record documentation overestimates care quality, feeding assistance,” supplements offered, and the oral intake of meals. Dr. Simmons, associate professor of medicine at the Vanderbilt University School of Medicine division of general internal medicine and public health, noted residents' unintentional weight loss has adverse clinical outcomes such as increase in pressure ulcers, hospitalizations, and mortality rates, and offered key care process measures of food and fluid consumption: ▸ Accuracy of medical record documentation (a resident who eats less than 50% of meals requires assistance); ▸ Adequacy of feeding assistance (spend at least 15–20 minutes assisting residents at risk of weight loss); ▸ Quality of assistance (give residents physical assistance in addition to verbal cueing and social stimulation). To improve facility feeding programs, Dr. Simmons conducted direct observational studies at 40 California facilities. Comparing resident charts and researcher observations of food consumption, she found that residents received assistance 40% of the time, versus the 100% charted. She also found residents were offered supplements once daily, rather than three times daily as listed in their charts. They also ate 20% less than documented. Additionally, observation indicated each resident received less than 10 minutes of feeding assistance, and little or no verbal cueing or social interaction. During snack time, foods and fluids were offered less than once a day and supplements ordered by physicians were rarely given. In another study, Dr. Simmons introduced feeding assistance programs for 134 residents with low oral intake. Such assistance required 35 minutes per resident per meal to significantly increase meal intake. Staff spent 13 minutes per resident, offering snacks between meals, versus less than 1 minute for the average nursing facility. “While it may be impractical in the typical nursing home to spend as much time with residents as feeding assistants did during the program, grouping three to four residents during meals” or grouping eight during snacks has the same effect, Dr. Simmons said. Results were striking: 90% of residents improved their food/fluid intake. To determine optimal staffing for such programs, Dr. Simmons used expert consensus panels, computerized simulations, and a validation study. The findings show: ▸ Staffing levels are consistently a significant predictor of quality; ▸ 5–7:1 ratio support for nurse aides; ▸ 5:1 ratio is necessary to consistently provide care to residents in need; ▸ Eighty percent of physically dependent residents got 5 minutes or more of feeding assistance per meal, versus 55% with traditional care; ▸ Staffing below the 5–7:1 level may require targeting residents who are most in need and/or using non-nursing staff for some daily mealtime tasks. Direct observational studies are critical to Continuous Quality Improvement (CQI) and to CMS' sponsorship of studies evaluating such programs. CMS funding of pilot studies of paid feeding assistants to cost-effectively augment nursing staff's efforts have shown that results with feeding assistants were at least comparable to results with nursing assistants. Dr. Simmons also presented results of a 4-month CQI pilot study in one skilled nursing facility targeting high-risk residents who ate less than 50% of most meals and had a history of weight loss. Results were dramatic (see graphic). According to Dr. Simmons, plans for improving care should be based on realistic assumptions about reimbursement and staff shortages. A pillar of change is “Train the Trainer” programs, in which supervisory-level RNs and dieticians train feeding assistants and explore other staffing models.