Using Item Response Theory to Summarize Health Care Quality Measures

• Activity Number: 376
• Type: Invited
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): John L. Adams*+ and Marc N. Elliott
• Companies: Kaiser Permanente and RAND Corporation
• Keywords: Quality of Care ; Item Response Theory ; Health Outcomes

Abstract

As the interest in measuring the quality of care delivered to users of health services has grown there has been a corresponding proliferation of quality measures. One prominent example is the HEDIS (c) measures. Process measures (e.g. did a diabetic get their blood sugar checked) are typically binary pass/fail measures. Intermediate outcome measures (e.g. was the patient's blood pressure below a threshold) are also frequently reported as binary measures. This increase in the size of the measure sets has driven a need for summaries of patient quality of care. In this talk I will explore the potential for using item response theory (IRT) to estimate a latent quality of care score from a collection of binary measures. An example using the HEDIS measures from the Medicare diabetic population will be presented. Rather than just averaging measures together IRT has the potential to use measures with high or low pass rates to more fully explore the range of quality of care.