Patients need to know that despite all the potentially highly sensitive personal information being collected in an electronic health record, their data will be contextually protected from inappropriate disclosure, says a paper prepared for the Office of National Coordinator for Health Information Technology.

The paper, dated Sept. 29 but only recently posted on the ONC website, acknowledges difficulties with what it terms data segmentation, but says that absent the ability to block the uniform distribution of patient data, some patients would react by hiding information from doctors.

Data segmentation nonetheless faces several technical hurdles, notes the paper, which was primarily written by Melissa Goldstein of George Washington University and Alison Rein of AcademyHealth, a Washington, D.C.-based professional society for health services researchers and health policy analysts.

Among them is the fact that clinicians often resist filling out structural data record forms and simply rely on the free text field to enter all information about a patient encounter. Structural data, of course, is easier to selectively segment. Were a free text field to reference a condition deemed by a patient as being too sensitive to share across practice boundaries, the entirety of the text field would have to be withheld from exchange, the paper says.

Even segmenting structural data isn't necessarily easy, the authors add. Data indicating HIV status, for example, could be located in numerous fields, from white blood cell counts to the presence of related diagnoses such as Kaposi's sarcoma in a problem list.

Also, because there is no standard code set used in EHRs, segmentation across organizations is more complex than it would be were there a single code.

"Some have described the current situation as similar to using multiple-gauge railroad tracks within a single rail system and have expressed a desire for policy makers to choose one particular gauge," the authors say.

A one-size-fits-all approach to data segmentation could fall short of patient objectives, the paper says. But an individualized approach to segmentation, even if technically feasible and desirable, could be impractical or difficult to achieve logistically, the paper adds.

"Many technology vendors, for example, have expressed a preference for the specification and definition of what constitutes sensitive information so that they can develop one solution that meets the needs of multiple clients."

For more:
- download the paper, "Data Segmentation In Electronic Health Information Exchange: Policy Considerations and Analysis." (.pdf)

Related Articles:
Study says EHRs don't improve outpatient care 
HHS gearing up for second round of 'meaningful use' objectives 
White House council calls for health IT universal data exchange