Usage
  • 12 views
  • 1 download

Developing computerized adaptive tests to improve the efficiency of patientreported outcome assessment: Clinically feasible procedures

  • Author(s) / Creator(s)
  • Presentation at 13th Annual Glenrose Rehabilitation Hospital Spotlight on Research Breakfast. Background Patient-reported outcome measures (PRO) are considered valuable sources of information. A clinically useful PRO will balance efficiency (quick administration), while maintaining precision (capturing individual variation) in measuring various health outcomes (e.g., physical function, quality-of-life). Computerized adaptive testing (CAT) can achieve these aims. In CAT, patients receive a unique set of items from a large item bank targeted towards their own health status. The individualized test produces a reliable measurement with far fewer items than traditional questionnaires. Objective To demonstrate procedures to develop a computerized-adaptive PRO. Method We used the items in the ‘Lower Extremity Functional Scale’, the Medical Outcomes Study Short Form-36 ‘Role-Physical’, and ‘Physical-Function’ subscales to create a CAT for physical functioning. We analyzed an existing dataset of responses (n=1,429) to the scales, collected from workers with lower extremity impairment. First, we calibrated the items on the same metric, using Item Response Theory. Then, we used computer simulations to design the CAT, and to evaluate the measurement precision of CATs of varying lengths (20 items, 12 items, 8 items, 5 items). Result Comparing the CAT to the three questionnaires (34 items total), we found that a CAT of 8 items in length was sufficient to maintain 95% of measurement precision. If greater precision is desired (99%), 20 items would be sufficient. We developed a CAT for physical function that is fully functional, and ready for implementation into clinical practice. Conclusion CATs are an efficient method of PRO measurement. We demonstrate how practitioners can develop and implement CATs using freely available online platforms.

  • Date created
    2017-11-22
  • Subjects / Keywords
  • Type of Item
    Conference Poster
  • DOI
    https://doi.org/10.7939/R3J679985
  • License
    Public Domain Mark 1.0