1. Results Providers (n=20) were more willing to share patient data when: Sharing with universities vs. private firms and non-government organizations No effects of benefits Providers were more willing to share: Survey responses over anonymized paper and electronic records Aggregate statistics over individual-level anonymized records Electronic records over paper records (perhaps easier to share) No effects of familiarity Conclusions Providers appear to place a high value on patient data remaining confidential. They are willing to share when objective risk of re-identification is low (data is aggregate, or data related to survey responses rather than health record), and when perceived risk of re- identification is low (sharing with a university vs. private firm). Providers are more willing to share electronic data, suggesting that ease may be a factor in sharing data. Acknowledgements This work is funded by: Background A growing problem facing researchers and health care professionals is getting access to linked and anonymized individual- level patient data sets, for purposes of research, accountability, and improved care. Providers, however, hold the reigns of patient data, and may be reluctant to share. The current research examines when providers are comfortable sharing patient data for purposes of secondary use. Methods Physicians responded to six unique vignettes. We varied two factors in each: Organization: sharing with public vs. private organization Benefits: direct patient benefit vs. no direct patient benefit We surveyed providers’ willingness to share different types of data: Method of collection -Health record audit -Survey response Level of identification -Individual-level -Aggregate Health record -Paper record -Electronic record Familiarity with sharing practices Caring is sharing: Provider perceptions of secondary use of anonymized health information MEASURING AND IMPROVING THE PERFORMANCE OF PRIMARY HEALTH CARE IN CANADA Ester Moher, William Hogg, Sharon Johnston, Mehdi Ammi, and Khaled El Emam Contact: emoher@ehealthinformation.ca Figure 1. Vignette Figure 4. Method of collection and health record type Figure 2. Response options for each vignette Institution was either a public (academic research center) or private (insurance or pharmaceutical company) Benefits were either direct to the patient, or no direct benefit to the patient * p <.01 Figure 5. Level of identification Withholding all data Sharing all data * p <.01 * p <.01 ns, p =.14 Figure 6. Organization type Withholding all data Sharing all data * p <.01 * p <.01 More willing to share survey data than record data More willing to share aggregate data than individual-level data Figure 3. Demographics Gender61.1% Male, 38.9% Female AgeMean = 46 years Years in practiceMean = 16.9 years Role33.3% residents, 66.7% physician Record type 61.1% Electronic only, 27.8% mixed paper and electronic For each item, we summed the number of times each participant checked either Yes or No, and converted the total into a proportion of Yes’, which was used as the DV across conditions.