Search

Predictive models have been used to identify factors that may prolong hospital length of stay (LOS). However, because predictors of LOS are collinear, the proportion of variance associated with each factor in a multivariate stepwise regression model may not reflect its mathematical contribution in explaining LOS. In an attempt to model factor contribution to LOS more realistically, we evaluated a clinically based clustered model. This model uses classes of candidate predictors, that is, patient attributes, adverse events, treatment modality, and health provider identity. Clusters of variables are permitted to enter into the model in a theoretically based predetermined sequence, so that the additional contribution of each cluster of factors can be assessed while the contribution of preceding factors is preserved. The clustered model was tested and compared with a free stepwise multivariate analysis in a cohort of patients undergoing prostatectomy for benign prostatic hypertrophy. We found that both models explained a similar proportion of the variance in LOS (56%-57%). However, some important differences were evident. Prostate size, associated with 12% of the variance in the clustered model, was not an independent predictor in the free model. A higher proportion of variance was associated with process variables, such as treatment modality in the free model. We conclude that use of a clustered model may facilitate more realistic assessment of the relative contribution of factors to LOS.

Objectives: To evaluate whether routine laboratory data can improve the ability to compare risk-adjusted outcomes of different medical wards, and to detect 'outlier' wards with significantly better or worse outcome., Methods: Patient data were taken from the Combined Patient Database Systematic Management and Research Tool, a database created by merging different computerized sources at a tertiary care hospital. All patients admitted to internal wards with the diagnosis of pneumonia during the years 1991-1995 were included (n = 2734). The outcome variable was mortality 30 days post-admission. We used three comorbidity measures based on ICD-9-CM codes as possible predictors of mortality: secondary diagnoses; the Health Care Financing Administration severity index; and the Charlson comorbidity index. Models were created using logistic regression. To each model, laboratory data gathered in the first 48 hours after admission were added. To identify 'outlier' services we determined whether the patients' ward was an independent predictor of mortality. The area under the receiver operator curve (ROC) of the models was used for comparisons., Results: The area under the ROC was 0.65-0.72 for the models based on age and comorbid diagnoses. The addition of laboratory data improved markedly the discriminatory ability of each of the models, as reflected by an increase in the area under the ROC to 0.83-0.84. An 'outlier' ward with a higher risk-adjusted mortality rate was identified only by the models that included laboratory data., Conclusion: Basic, automated, routinely gathered laboratory data added significantly to the discriminatory power of risk models based on administrative data with abstracted diagnoses. Addition of laboratory data improved the ability to identify providers with possible exceptional quality of care.