Computerized order entry systems have the potential to improve the quality of care in nursing homes, but few previous studies have evaluated their feasibility or impact. We developed evidence-based computerized order entry algorithms for 5 common nursing home problems, and tested their acceptance, use, and preliminary impact on quality indicators in 2 VA nursing homes.
Although providers were uniformly enthusiastic about the computerized order entry algorithms, they used them infrequently. The only computerized order entry algorithm used frequently was for falls, which may have occurred for several reasons. First, falls occur more frequently than the other conditions. Second, higher staff interest in fall prevention due to regulatory and legal implications may have driven use. Finally, post-fall evaluation generally occurs during regular working hours when the providers are in the facility and can easily access the computerized order entry algorithms. In contrast, orders for acute conditions such as fever or pneumonia are often given over the telephone if they occur when a provider is not present. Osteoporosis, the only target condition without an acute event to trigger its use, was rarely used despite large numbers of patients with a history of fracture in the homes. The statistically significant decline in several of the osteoporosis quality indicators therefore likely reflects the lower proportion of patients with recognized osteoporosis in the post-implementation group. It is not clear whether providers with more experience in using computerized order entry systems would be more or less likely to use the algorithms.
When computerized order entry algorithms were used frequently, as in the case of falls, we saw a trend toward clinically meaningful improvements in quality indicators. While larger, randomized controlled studies are necessary to determine whether the change in quality indicators will translate into improved processes and outcomes for residents, this result is important because prior studies of fall-prevention quality improvement programs have not been widely successful.34, 41-44
The lack of success of previous quality improvement efforts may, in part relate to the need for coordination among multiple disciplines for multi-factorial fall risk reduction. Our computerized order entry algorithms provided a means for various groups to communicate efficiently, for example through automatic alerts sent to the pharmacist when the provider accessed the fall algorithm or through optional consults to physical and occupational therapy and nursing. Given the Joint Commission, State regulatory agency, and CMS's focus on fall prevention, we believe that the fall prevention computerized order entry algorithm merits further study as a means of post-fall assessment, and risk reduction for high risk residents.
The minimal impact on indicators for the acute and chronic conditions suggests that additional interventions might be necessary to prompt clinician use. For example, osteoporosis alerts could automatically send providers information about patients with a prior fracture history or who take chronic corticosteroids. Because providers are frequently not present in the home when an acute illness occurs, training nurses to use the computerized order entry algorithms may increase their use and facilitate team communication. However, nurses were not allowed to place orders in the computer in one of the study homes, and scope of practice limitations in other facilities would require that providers still be contacted for verbal approval. Algorithm-based standing orders for fever or falls evaluation that are automatically initiated by nursing staff is another option, although this eliminates the customization to the specific resident's presentation and goals that is valued by nursing home providers.
Provider turnover is often cited as a barrier to clinical practice guideline use in nursing homes, and was indeed observed in our study. Interestingly, the impact of provider turnover differed in the 2 study facilities. In the academic center, new providers were trained to use the computerized order entry algorithms by the medical director (who was part of the study team) during orientation, and they welcomed the computerized order entry algorithms as a valuable tool to help them “learn the ropes”. They became more frequent users than the previous providers who were accustomed to the old system. Thus, staff turnover may actually represent an opportunity to introduce new technology such as computerized order entry algorithms, particularly if the nursing home medical director and other leaders are committed and supportive. In the second facility, provider turnover exceeded 100% (including 3 medical directors) during the study period, and the facility was frequently understaffed. In this challenging environment, computerized order entry algorithm use was minimal despite repeated reminders and visits. If computerized order entry algorithms are employed by a facility, we suggest incorporating training in their use to the routine employee orientation, and actively engaging the facility leadership in promoting their use.
Our study adds to the emerging literature on the use of information technology in improving nursing home care. A recent randomized study of computer decision support to reduce adverse drug events in nursing home residents found no significant benefit.45
The authors hypothesize that limited scope of the medications covered, a high proportion of inappropriate provider alerts, and targeting of only the medical team members may have explained the lack of effect. Our study suggests that it is feasible to use computer order entry algorithms to enhance communication about geriatric syndromes with other team members (e.g., the pharmacist, therapist, and nursing staff in a patient with falls). While allowing providers to access order entry algorithms voluntarily avoids the problem of burdensome inappropriate reminders, it also may result in low use and therefore limited impact. Combining reminders to providers (e.g., reminder to screen/treat for osteoporosis triggered by an admission for hip fracture) with the appropriate computerized order entry page may allow the advantages of both strategies to be realized.
There are several limitations to this pilot study. First, it is small and the observed trends to improvement in falls quality indicators may be due to chance alone. Our pre-post study design does not allow us to make causal inferences. However, we sought to examine the feasibility and acceptance of these computerized order entry algorithms. Second, our study was conducted at VA nursing homes, which limits our generalizabitity. Notably, VA nursing homes have greater on-site availability of providers and an electronic order medical record system which may not be available in community nursing homes.
In summary, in this pilot test we successfully developed and implemented computerized order entry algorithms for common geriatric problems in 2 VA nursing homes. Despite high staff turnover, trends toward improvements in quality indicators were measured for the most frequently used computerized order entry algorithms (falls). This technology offers promise for improving clinical practice guideline use in nursing homes, but further modifications to adapt to the community nursing home system, and to prompt use for chronic conditions are needed. Formal evaluation of computerized order entry algorithms in nursing homes requires a randomized trial.