The impact of tailored intervention services on charges and mortality for adult super-utilizers

Interventions designed to improve care and reduce costs for patients with the highest rates of hospital utilization (super-utilizers) continue to proliferate, despite conflicting evidence of cost savings.

We evaluated a practice transformation intervention that implemented team-based care and risk-stratification to match specific primary care resources based on need. This included an intensive outpatient clinic for super-utilizers. We used multivariate regression and a difference-in-differences approach to compare changes in mortality, utilization, and charges between the intervention group and a historical control. Sensitivity analyses tested the robustness of findings and revealed the inherent challenges associated with quasi-experimental designs.

Observed charges for the intervention group were significantly lower than expected charges as derived by the trend of the historical control (p<0.04) resulting in total charge avoidance of approximately $26 million. While inpatient admissions were significantly higher (p<0.01), charges associated with total inpatient (p=0.01), intensive-care unit (p<0.05, not robust to sensitivity analyses), and surgery (p<0.01) were significantly lower than expected in the intervention group. One year mortality was significantly less in the intervention group (12.6% vs 11.5%, p<0.01).

The use of tailored services, including a dedicated intensive outpatient clinic, for super-utilizers within a larger primary care practice transformation reduced mortality and provided significant savings, even while total hospitalizations increased. These savings were achieved through a reduction in the intensity of inpatient services. The unexpected finding of a reduction in ICU charges deserves further exploration.

These findings suggest that intensity of inpatient service, and not merely volume of services, should be considered a focus for future intervention design and evaluated as an outcome.

Level III (Quasi-Experimental Design).