3064 — Maximizing Consent and Minimizing Bias: What is the Best Way to Request Permission to Access Patient Medical Records?
Partin MR (Minneapolis VA Medical Center and University of Minnesota Department of Medicine) , Halek K
(Minneapolis VA Medical Center), Noorbaloochi S
(Minneapolis VA Medical Center and University of Minnesota Department of Medicine), Grill J
(Minneapolis VA Medical Center), Burgess D
(Minneapolis VA Medical Center and University of Minnesota Department of Medicine), Fisher D
(Durham VA Medical Center and Duke University Department of Medicine), vanRyn M
(University of Minnesota Department of Family Medicine and Community Health), Vernon S
(Univeristy of Texas at Houston School of Public Health)
To assess the effect of two approaches to requesting permission to access patient medical records on response bias and the proportion of patients with complete medical records data available for validating self-reported screening behavior.
The study sample consisted of 890 veterans age 50-75 receiving primary care at the Minneapolis VA Medical Center who were mailed a questionnaire to assess colorectal cancer screening behaviors. For half of the sample, authorization to access medical records was included as part of the survey questionnaire (group 1); for the other half, authorization was requested in a separate mailing sent out to those returning a completed questionnaire (group 2). Analyses compared response rates, response bias and the proportion and characteristics of patients providing access to complete medical records data, by authorization group.
Survey response rates did not vary significantly by authorization group (76% for group 1 and 78% for group 2, p=.45). The proportion of patients providing access to complete medical records data was 41% in group 1 and 26% in group 2 (p <.0001). In group 2, patients with complete medical records information were significantly more likely to be female, Caucasian, married, age 65-75, to have a family history of colorectal cancer, and to report being screened for colorectal cancer. In group 1, patients with complete medical records information were significantly more likely to be married but otherwise did not vary significantly from those without complete medical records information.
Requesting medical records as part of a self-administered questionnaire did not significantly reduce survey response rates relative to requesting through a separate communication, was less likely to bias the sample available for validating self-reported health behaviors, and produced a higher number of cases with complete medical records information.
These findings provide practical guidance on data collection methods that will benefit future survey research studies requiring authorization of access to medical records for self-report validation, treatment ascertainment, or outcomes documentation.