Background:

Recent research shows that inpatient continuity has been declining. Little is known about the association between continuity during hospitalization and important outcomes such as cost and hospital readmission.

Methods:

We conducted a retrospective data analysis of inpatient care provided exclusively by hospitalists from the Northwestern University Feinberg School of Medicine. Quantile regression models were used to predict the median increase in cost of hospitalization as a function of two measures of continuity: the number of hospitalists writing progress notes and the Usual Provider of Continuity (UPC) index. Covariates included age, sex, race, payor, case mix, Diagnosis Related Group (DRG) weight, Charleson comorbidity score, admission source, discharge disposition, admission weekday, and length of stay (LOS). Because of multicollinearity among variables, models were constructed with LOS and DRG weight treated alternatively as continuous and categorical variables (i.e. quartiles). We report results from models with optimal fit according to the Akaike information criterion and c–statistic values. In a similar fashion, we used logistic regression models to predict the odds of readmission as a function of measures of continuity.

Results:

The dataset included 11,234 hospitalizations. The average hospitalization lasted 3.5 days, had notes written by 1.9 hospitalists, had a UPC index of 0.76, cost $7,385 and resulted in readmission with a 22% likelihood. Across all models, lower continuity was significantly associated with higher costs, regardless of the continuity metric used or whether LOS and DRG weight variables were continuous or categorical. We found inconsistent results for models evaluating the association between continuity and readmission rates.

Conclusions:

Lower hospitalist continuity is associated with higher hospital costs. Further research may clarify whether patients undergo additional testing and/or management changes as they transition from one hospitalist to another.

Table 1Predicted Changes in Cost and Readmission

Outcome Variable Predictor Variable* Predicted Change
Odds of readmission Number of hospitalists (increase of 1) -3.7% (CI: [-10.0%, 3.0%])
UPC (decrease of 10%) -3.0% (CI: [-0.5%, -5.6%])  
Cost of hospitalization Number of hospitalists (increase of 1) 7.2% (p value < 0.0001)
UPC (decrease of 10%) 12.5% (p value < 0.0001)  
*Increases in the number of hospitalists and decreases in UPC result in lower continuity.