Recent research provides evidence that care coordination improves outcomes such as hospitalization, rehospitalization, emergency room use, and reduces cost. One program using nurse care coordination as an intervention to improve care for community dwelling older adults'was Aging in Place (AIP) at the University of Missouri's Sinclair School of Nursing. AIP's service provider for the enhanced nurse care coordination intervention was Senior Care, a home health agency and Home and Community Based service provider. After the AIP study ended Senior Care continued to exist as a home health care (HHC) agency. Taking advantage of the unique data from the AIP study, and the HHC agency we propose to perform a secondary analysis to (1) identify care coordination interventions, intervention activities, and intervention doses used by AIP nurse care coordinators using enhanced nurse coordination, and HHC nurses providing traditional HHC , (2) describe any differences in interventions used by AIP nurse care coordinators, and HHC nurses, and finally (3) to identify whether outcomes of hospitalization, rehospitalization, emergency room use, and cost differ for older adults in the AIP and HHC groups. The original AIP analysis did not link specific nurse care coordination interventions with outcomes, identify the specific activities associated with interventions, explore intervention dose, or compare outcomes to traditional HHC. Data will compare two groups of older people who received services, those with enhanced nurse care coordination in the AIP group from the years 2000-2002 and those who received traditional HHC without enhanced nurse care coordination from the years 2003-2005. Natural Language Processing (NLP) methods will be used to identify interventions and activities nurse care coordinators used when coordinating care for older adults. NLP transforms human-generated text into machine-readable (structured) information. We propose to use a unique iterative approach to NLP in which the available activities are extracted from existing electronic medical records from Senior Care, while new activities are identified by hand from records in which not enough information was found in the first pass. After each pass, a team of experts will check the extracted activities to assure that they are relevant to care coordination. Intervention profiles that are representative for AIP and HHC will be developed by determining what types of activities occur most frequently for each intervention category and target, and which ones are most representative of the category and target. Dose profiles of interventions will be developed, and finally, outcomes of hospitalization, rehospitalization, emergency room use, and cost will be compared between the AIP and HHC groups.
This study informs research and clinical practice by identifying actual care coordination intervention activities and dose, and developing intervention profiles for future use in research and practice. The Natural Language Processing method will add to our knowledge of how to successful use electronic medical record text data.
| Popejoy, Lori L; Galambos, Colleen; Stetzer, Frank et al. (2015) Comparing Aging in Place to Home Health Care: Impact of Nurse Care Coordination On Utilization and Costs. Nurs Econ 33:306-13 |
| Popejoy, Lori L; Khalilia, Mohammed A; Popescu, Mihail et al. (2015) Quantifying care coordination using natural language processing and domain-specific ontology. J Am Med Inform Assoc 22:e93-103 |
