O'Donnell JC; Association for Health Services Research. Meeting.

Abstr Book Assoc Health Serv Res Meet. 1999; 16: 119.

National Center for Health Promotion, VAMC, Durham, NC 27705, USA.

RESEARCH OBJECTIVE: To identify determinants of length of stay (LOS) for patients with traumatic brain injury (TBI) cared for in Department of Veterans Affairs (VA) rehabilitation units. STUDY DESIGN: Prospectively collected data on clinical, demographic and rehabilitation unit characteristics were extracted from the VA database of the Uniform Data System for Medical Rehabilitation (VA-UDSMR). The sample included 628 first-admission TBI patients discharged from 47 hospital-based rehabilitation units in fiscal years 1993-1996. A regression analysis was conducted in which variables measuring clinical, demographic and rehabilitation unit characteristics were entered hierarchically and R-squared change was computed. PRINCIPAL FINDINGS: The model explained 39 percent of the variance in logged LOS (model R2=.386, p<.001). The majority of the explained variance was accounted for by the clinical variables (R2 change =.339, p<.001). However, the demographic and rehabilitation unit variables also attained joint significance in the model (R2 change =.030 and .017, respectively, p=.023). Longer LOS was associated with closed traumatic injury (p=.048), referral from external acute hospitals (p<.001), severity of comorbidity (p=.012), interrupted rehabilitation (p=.011), and care in the Northeast region (p=.018). LOS was negatively associated with functional status at admission and age (p<.001), although the relationships were nonlinear and interacting. Significant inter-unit variation in risk-adjusted LOS was detected. CONCLUSIONS: The results of this study indicate that LOS for TBI in VA rehabilitation has determinants similar to those driving LOS in nonfederal rehabilitation units. To a large extent, LOS was contingent upon the severity of functional impairment at admission and age. However, determinants of LOS also included factors mutable to policy intervention such as referral mechanisms and unit characteristics. IMPLICATIONS FOR POLICY, DELIVERY OR PRACTICE: Although clinical characteristics explain the majority of the variance in LOS, demographic and unit characteristics also maintain predictive salience in multivariate modeling. Strategies to enhance the efficiency of care for TBI will need to consider these domains simultaneously. VA planners should capitalize on the significant variation in adjusted LOS system-wide. As the largest unified health system in the U.S., the VA has a unique opportunity to seek out and disseminate highly efficient rehabilitation practices.

Publication Types:

• Meeting Abstracts

Keywords:

• Activities of Daily Living
• Brain Injuries
• Comorbidity
• Hospital Units
• Hospitals
• Humans
• Patient Discharge
• Referral and Consultation
• Regression Analysis
• United States
• Veterans
• rehabilitation
• hsrmtgs

Other ID:

• HTX/20602702

UI: 102194391

From Meeting Abstracts