Risk adjustment for measuring health outcomes: an application in VA long-term care

UMMS Affiliation

Department of Quantitative Health Sciences

Publication Date


Document Type



Aged; Diagnosis-Related Groups; Evaluation Studies as Topic; Frail Elderly; Humans; Insurance, Health; Long-Term Care; Models, Organizational; Nursing Homes; *Outcome Assessment (Health Care); Quality of Health Care; Risk Adjustment; United States; *United States Department of Veterans Affairs


Biostatistics | Epidemiology | Health Services Research


An empirically derived risk adjustment model is useful in distinguishing among facilities in their quality of care. We used Veterans Affairs (VA) administrative databases to develop and validate a risk adjustment model to predict decline in functional status, an important outcome measure in long-term care, among patients residing in VA long-term care facilities. This model was used to compare facilities on adjusted and unadjusted rates of decline. Predictors of decline included age, time between assessments, baseline functional status, terminal illness, pressure ulcers, pulmonary disease, cancer, arthritis, congestive heart failure, substance-related disorders, and various neurologic disorders. The model performed well in the development and validation databases (c statistics, 0.70 and 0.68, respectively). Risk-adjusted rates and rankings of facilities differed from unadjusted ratings. We conclude that judgments of facility performance depend on whether risk-adjusted or unadjusted decline rates are used. Valid risk adjustment models are therefore necessary when comparing facilities on outcomes.


Am J Med Qual. 2001 Jul-Aug;16(4):118-27. Link to article on publisher's site

Journal/Book/Conference Title

American journal of medical quality : the official journal of the American College of Medical Quality

PubMed ID


Related Resources

Link to Article in PubMed