The effects of a low-cost intervention program on hospital costs.

OBJECTIVE

To assess the impact of a low-cost education and feedback intervention designed to change physicians' utilization behavior on general medicine services.

DESIGN

Prospective, nonequivalent control group study of 1,432 admissions on four general medicine services over 12 months. Two services were randomly selected to receive the intervention. The other two served as controls. Admissions alternated between control and intervention services each day. Results were casemix-adjusted using diagnosis-related groups (DRGs). Three internists blinded to patient study group assignment assessed quality of care using a structured implicit instrument.

SETTING

Four general medicine services at a university hospital.

INTERVENTIONS

A brief orientation, a pamphlet of cost strategies and common charges, detailed interim bills, and information about projected length of stay and usual hospital reimbursement for each patient.

PATIENTS/PARTICIPANTS: Each service was staffed by a full-time internal medicine faculty member, one third-year and two first-year internal medicine houseofficers, three medical students, and a clinical pharmacist. Physicians were assigned to services for one-month periods by a physician unaware of the study design. To prevent crossover, houseofficers assigned to a service returned to the same service for all subsequent general medical inpatient assignments.

MEASUREMENTS AND MAIN RESULTS

Geometric mean length of stay was 0.44 days (7.8%) shorter for the intervention services than for the control services (p less than 0.01), and geometric mean charges were $341 (7.1%) less (p less than 0.01). Effects persisted despite using a more precise cost estimate or casemix adjustment. Intervention houseofficers demonstrated superior cost-related attitudes but no difference in knowledge of charges. Audits of quality of care detected no significant difference between groups.

CONCLUSION

This low-intensity intervention reduced length of stay and charges, even under the cost-constrained context of the prospective payment system.