Agreement between internet-based self- and proxy-reported health care resource utilization and administrative health care claims

Objectives: Although Internet-based surveys are becoming more common, little is known about agreement between administrative claims data and Internet-based survey self- and proxy-reported health care resource utilization (HCRU) data. This analysis evaluated the level of agreement between self- and proxy-reported HCRU data, as recorded through an Internet-based survey, and administrative claims-based HCRU data. Methods: The Child and Household Influenza-Illness and Employee Function study collected self- and proxy-reported HCRU data monthly between November 2007 and May 2008. Data included the occurrence and number of visits to hospitals, emergency departments, urgent care centers, and outpatient offices for a respondent's and his or her household members' care. Administrative claims data from the MarketScan® Databases were assessed during the same time and evaluated relative to survey-based metrics. Only data for individuals with employer-sponsored health care coverage linkable to claims were included. The Kappa (κ) statistic was used to evaluate visit concordance, and the intraclass correlation coefficient was used to describe frequency consistency. Results: Agreement for presence of a health care visit and the number of visits were similar for self- and proxy-reported HCRU data. There was moderate to substantial agreement related to health care visit occurrence between survey-based and claims-based HCRU data for inpatient, emergency department, and office visits (κ: 0.47-0.77). There was less agreement on health care visit frequencies, with intraclass correlation coefficient values ranging from 0.14 to 0.71. Conclusions: This study's agreement values suggest that Internet-based surveys are an effective method to collect self- and proxy-reported HCRU data. These results should increase confidence in the use of the Internet for evaluating disease burden.