Comparison of the Accuracy of Several Systems for Hospital Quality Screening andAssessment

摘要

The purpose of the study was to measure the sensitivity, specificity, andpredictive values of three different medical-record based screening systems for quality of care problems experienced by patient with coronary artery disease hospitalized for myocardial infarction (MI), percutaneous transluminal coronary angioplasty (PTCA) or coronary artery bypass graft (CABG) surgery. The three screening systems were: (1) an early prototye of the Uniform Clinical Data Set (UCDS); (2) Johns Hopkins hopsital-wide (JHH) screens; and (3) Harvard Medical Practice Study (HMPS) screening criteria. For the cardiac patients at Johns Hopkins Hopsital, the verion of UCDS tested was more sensitive and less specific for detecting quality problems than the other two screening methods, but was much less efficient per quality problem detected. There are differences in sensitivity and specificity depending on whether patients were admitted for MI, angioplasty, or bypass surgery, with UCDS performing significantly better in MI patients than the other two systems. Positive predictive values were similar for all three methods treated and were only about 25% in this population with a prevalence of 18% quality problems. All systems could be streamlined for these populations of patients to include only the highest yield screens with little loss of sensitivity.