Citation

Richmond TS, Kauder D, Hinkle J, Shults J. Am. J. Crit. Care 2003; 12(3): 197-205.

Affiliation

School of Nursing (TSR) and School of Medicine, University of Pennsylvania, Philadelphia, Pa., USA.

Copyright

(Copyright © 2003, American Association of Critical-Care Nurses)

DOI

unavailable

PMID

12751393

Abstract

BACKGROUND: Improving outcomes after serious injury is important to patients, patients' families, and healthcare providers. Identifying early risk factors for long-term disability after injury will help critical care providers recognize patients at risk. OBJECTIVES: To identify early predictors of long-term disability after injury and to ascertain if age, level of disability before injury, posttraumatic psychological distress, and social network factors during hospitalization and recovery significantly contribute to long-term disability after injury. METHODS: A prospective, correlational design was used. Injury-specific information on 63 patients with serious, non-central nervous system injury was obtained from medical records; all other data were obtained from interviews (3 per patient) during a 2 1/2-year period. A model was developed to test the theoretical propositions of the disabling process. Predictors of long-term disability were evaluated using path analysis in the context of structural equation modeling. RESULTS: Injuries were predominately due to motor vehicle crashes (37%) or violent assaults (21%). Mean Injury Severity Score was 13.46, and mean length of stay was 12 days. With structural equation modeling, 36% of the variance in long-term disability was explained by predictors present at the time of injury (age, disability before injury), during hospitalization (psychological distress), or soon after discharge (psychological distress, short-term disability after injury). CONCLUSIONS: Disability after injury is due partly to an interplay between physical and psychological factors that can be identified soon after injury. By identifying these early predictors, patients at risk for suboptimal outcomes can be detected.