Patient-Centered Outcomes Measurement: Does It Require Information From Patients?

Area of Science:

• Health Informatics
• Patient-Reported Outcomes
• Medical Data Analysis

Background:

• Collecting patient outcome data is costly and prone to bias.
• Electronic health records (EHR) and insurance claims offer potential alternative data sources.

Purpose of the Study:

• To determine if patient-important outcomes can be accurately extracted from EHR or insurance claims data.
• To assess the feasibility of using secondary data sources for outcome measurement.

Main Methods:

• Compared patient survey data with EHR audits and claims data for 21 pre-identified outcomes.
• Utilized telephone surveys from 321 patients undergoing advanced imaging for back or abdominal pain.
• Assessed data agreement using kappa scores for chart audits and algorithmic extraction from claims.

Main Results:

• Few outcomes met a minimal agreement threshold (kappa ≥ 0.2) when comparing survey data to EHR audits (2 for back pain, 3 for abdominal pain).
• No outcomes showed adequate agreement with algorithmically extracted EHR/claims data.
• Limited agreement was found between patient surveys and EHR audits for outcomes where claims data was initially considered primary.

Conclusions:

• EHR and claims data are generally inadequate for capturing patient-important outcomes.
• Patient-reported outcomes via surveys are superior, except for specific instances where patient recall may be less accurate.