Decision support for utilization review using the HELP hospital information system

The {dollar}{lcub}{bsol}rm{bsol}underline A{rcub}{dollar}utomated {dollar}{lcub}{bsol}rm{bsol}underline S{rcub}{dollar}upport {dollar}{lcub}{bsol}rm{bsol}underline S{rcub}{dollar}ystem for {dollar}{lcub}{bsol}rm{bsol}underline U{rcub}{dollar}tilization {dollar}{lcub}{bsol}rm{bsol}underline R{rcub}{dollar}eview (ASSURE) was developed at LDS Hospital in Salt Lake City, Utah, where the electronic patient record is in an advanced state in the HELP computerized hospital information system. The development, verification, and validation of ASSURE, a HELP application, are presented.; ASSURE focuses the efforts of Utilization Managers on inappropriate patients by the automatic, daily, concurrent screening of inpatient days of care. Inappropriate patients are those who can be treated more cost-effectively in an outpatient facility or who are unnecessarily delayed in their care. Utilization Managers intervene with physicians when necessary to advocate more efficient care.; ASSURE's expert system implements the Appropriateness Evaluation Protocol (AEP) Day of Care criteria for adult patients in acute care. The AEP is a validated, diagnosis-independent criteria set in the public domain. Satisfaction of any criterion indicates appropriateness; satisfaction of none indicates presumptive inappropriateness. When the expert system finds a criterion satisfied, an explanatory message is formulated from the patient's data. A user interface supports interaction with expert system findings, charting of other data, worklist management, and reports.; In ASSURE's development and verification phase, the expert system and a Utilization Manager agreed on satisfaction of single criteria for 92% of 560 randomly sampled current inpatients. Analysis using Cohen's kappa showed statistically significant agreement ({dollar}{bsol}alpha{dollar} = 0.05) for all 20 AEP Day of Care criteria implemented. The overall kappa of 0.84 showed statistically significant agreement, P {dollar}<{dollar} 0.0001.; In ASSURE's validation phase, expert system decisions on the appropriateness of days of care were compared to the independent judgement of two Utilization Managers in a random sample of 168 current inpatients. Agreement was analyzed using prevalence-adjusted bias-adjusted kappa (PABAK). In Manual reviews, the expert system and the Utilization Managers agreed on 80.4% of 148 patients, for PABAK = 0.608. In Computer-Assisted reviews, the expert system and the Utilization Managers agreed on 85.8% of 148 patients, for PABAK = 0.716.; Research data show that ASSURE is twice as efficient as random sampling at detection of inappropriate days of care, enabling cost-effective daily review of all patients on three nursing divisions. Experimental data support a projected annual health care cost reduction of approximately {dollar}600,000.