BACKGROUND/RATIONALE:
The VA has funded studies that have examined the structure, process, and outcomes of both VA and community nursing homes. To date, there have been no published studies that have examined validated quality of care outcomes for veterans in community nursing homes that are being paid for through the VA's per-diem program. This study seeks to describe the utilization and quality of care for veterans diagnosed with stroke whose care in community nursing homes from 1999 to 2002 was paid for by the VA.
OBJECTIVE(S):
The objectives of this study are: 1. Define the characteristics of community nursing homes where veterans receive their care versus community nursing homes that do not treat veterans. 2. Compare Centers for Medicare & Medicaid Services (CMS) Minimum Data Set (MDS) nursing home facility-level quality indicators for residents diagnosed with stroke. 3. Examine the patient-level variation in therapy utilization for veterans who had stroke and are being treated in community nursing homes using the CMS MDS. 4. Determine the patient-level outcome variation for residents who had strokes using CMS MDS.
METHODS:
Retrospective cohort study data was gathered from the Online Survey Certification and Reporting (OSCAR) system and the CMS MDS. Quality measures examined include: meeting CMS nurse staffing levels, restraint use, new catheter use, licensing survey deficiencies, actual harm citations, development of new pressure sores, pain, delirium, provision of rehabilitation restorative care, and receipt of specialized speech, occupational, physical and respiratory therapy. Statistical methods: Logistic regression/negative binomial regression is used to model the relationship between presence of VA per-diem residents and quality of nursing home care (objective 1). Ordered probit is used to model the relationship between presence VA per-diem residents and post-acute quality of nursing home care (objective 2). The RAND Health Insurance 2-part model is used to model the relationship between being a VA per-diem resident and rehabilitation therapy utilization (objective 3). Logistic regression is used to model the relationship between being a VA per-diem resident and post-acute outcomes for pain, delirium, and pressure sores (objective 4).
FINDINGS/RESULTS:
OSCAR quality of care measures in nursing homes that serve per-diem veterans were poorer than facilities that do not serve veterans. However, among VA per-diem facilities, those with higher proportions of VA residents were more likely to do better in the quality of care measures. MDS post-acute quality indicators show nursing homes serving veterans with stroke in general have poorer scores than non-VA per-diem facilities. At admission, VA per-diem stroke veterans were less likely to receive speech, occupational, physical, and respiratory therapy than non-VA stroke residents. VA per-diem residents were also less likely to receive rehabilitation restorative care for bed mobility, transfer, dressing and grooming, eating or swallowing, and communication. There were no significant differences at the resident level between VA per-diem veterans with stroke and non-VA stroke residents for 3 post-acute outcome measures: pain, delirium, and pressure sores.
IMPACT:
This is the first study to compare quality of care indicators in VA per-diem CNHs and non-VA per-diem facilities. Initial results suggest that there may be differences in quality indicators and measures by per-diem status. Additional research is needed to understand why these differences exist (e.g., further volume- and/or geographic-stratified analyses may explain the disparities observed). Future research is also needed to understand why VA per-diem residents, even when age and other risk factors are controlled, appear to receive fewer therapy minutes and restorative techniques when compared to non-VA residents in community nursing homes. Finally, VA per-diem stroke veterans experienced similar post-acute outcomes compared to non-VA per-diem stroke residents. This finding may suggest that VA oversight and monitoring of care at different local sites is effective in ensuring that VA per-diem residents receive comparable quality of care to that of non-VA residents. Conversely, it may be other outcome measures are more responsive. Further research is needed to determine whether the observed effect is truly an effect of VA oversight or whether it is a result of selection bias or data quality and/or data collection issue. Currently, we are continuing our analysis on selection bias. Our initial findings may help to improve the care quality provided by VA per-diem CNHs to thousands of our nation’s veterans.
PUBLICATIONS:
Journal Articles
- Laberge A, Weech-Maldonado R, Johnson CE, Jia H, Dewald L. Outsourcing veterans for long-term care: comparison of community and state veterans' nursing homes. Journal of Health and Human Services Administration. 2008; 30(4): 441-67.
- Buchanan RJ, Johnson C, Wang S, Cowper DC, Kim MS, Reker D. Analyses of male residents in community nursing facilities: comparisons of Veterans Health Administration residents to other residents. Journal of Rehabilitation Research and Development. 2004; 41(6A): 847-60.
