Related Experiment Videos
Does diagnostic information contribute to predicting functional decline in long-term care?
1Center for Health Quality, Outcomes and Economic Research, Bedford VAMC, Massachusetts 01730, USA. akrosen@bu.edu
Medical Care
|June 8, 2000
Summary
Adding diagnosis codes to administrative data modestly improves prediction of functional decline in long-term care residents. This offers a cost-effective way to obtain data for risk adjustment.
Area of Science:
- Gerontology
- Health Services Research
- Medical Informatics
Background:
- Risk-adjusted outcomes are increasingly important in long-term care.
- Administrative databases are becoming more prevalent in long-term care settings.
- Predicting functional decline is crucial for resident care.
Purpose of the Study:
- To evaluate the predictive value of ICD-9-CM diagnosis codes from administrative data for functional decline in long-term care.
- To compare the performance of a base regression model with one enhanced by diagnostic codes.
Main Methods:
- Retrospective analysis of 15,693 long-term care residents in VA facilities.
- Functional decline defined as an increase of ≥2 in Activities of Daily Living (ADL) summary score.
- Comparison of a base model versus a model augmented with ICD-9-CM codes, with performance validated in an independent cohort.
Main Results:
- The enhanced model with ICD-9-CM codes significantly improved data fit (chi2 = 179, P <0.001).
- The full model showed better prediction (R2=0.06 vs 0.05) and discrimination (c-statistic=0.70 vs 0.68) compared to the base model.
- Validated performance metrics confirmed the modest but significant improvement offered by diagnostic codes.
Conclusions:
- Incorporating specific diagnostic variables from administrative data modestly enhances the prediction of functional decline in long-term care.
- Administrative data with diagnostic codes offer a potentially cost-effective alternative to chart abstraction for risk adjustment.