- In recognition of patient matching as a persistent obstacle to seamless information exchange between organizations, the Sequoia Project on Nov. 10 released a proposed framework for patient identity management. Sequoia is collaborating with the Care Connectivity Consortium (CCC) on the HIT standards effort, which will develop minimal acceptable cross-organizational patient matching rules, suggested matching traits, a framework for improvement and a maturity model intended to serve as a roadmap for future growth.
The draft framework document acknowledges a current “blind spot” for many healthcare organizations: the inability to accurately match patients across organizations.
“Patient identity issues are more daunting when they cross organizational lines. Such issues often involve six or more organizations (the two health information organizations, their two vendors, and often an intermediary such as a health information organization and their vendor). In such an environment, even mundane items such as scheduling cross-organizational working sessions often introduce days and weeks of delay in resolving each issue due to lack of availability of key personnel. In essence, health data sharing introduces dependencies upon these independent organizations, and intertwines the workflows of the organizations, where no single organization has direct control over the other,” the framework states.
The result is that patient matching processes across organizations are inconsistent and yield match rates as low as 10 to 30 percent, according to the paper.
Part of the framework analysis follows a case study of a live production pilot using CCC Shared Services between Intermountain Healthcare, based in Salt Lake City, and local exchange partners including the Utah Health Information Network. The project started with a baseline test of 10,000 patients known to have been treated by both Intermountain and one of its exchange partners. The surprising result was a match rate of only 10 percent. “The sample data were fraught with quality issues,” the report states. However, through an iterative process of data cleaning and normalization; algorithmic refinement and operational improvement; and application of pre-worked and reused correlations and identified best practices, a 95 percent match rate was eventually achieved.
Among the case study’s lessons learned:
Fragile identities. A category of patients repeatedly failed to match correctly and identified as “fragile identities.” In most cases, the problem was due to “thin” demographics, such as just a first and middle initial being used instead of full first and middle names, a missing address or an address using non-standard abbreviations.
Well-behaved group. Another category of patient identities exhibited the opposite behavior — their identities seemed to almost always match correctly. Analysis showed they had a full, correctly spelled name (including middle name and any special characters), complete current address and phone number, and historical name and address information. The traits of this group are being used to inform best practices for other patient groups.
Knowledge reuse. Although manual work on a patient’s identity is expensive and slow, it yields very valuable information. “Once a patient’s records have been manually analyzed and remediated, that information can and should be leveraged in the future to prevent repeated manual rework on the same patient,” the report states.
Patient involvement in identity management. The case study identified two methods to bring patients into the cycle of maintaining their identities: (1) ask patients at the point of care to help update their records and (2) a portal or self-service application could help patients understand their identity completeness.
Separately, the Sequoia framework proposes a maturity model for patient matching deployments across organizational boundaries. The model establishes five levels, from 0 to 4, starting with ad hoc processes and outcomes with little or no management oversight to the ultimate phase indicating innovation, ongoing optimization and senior management’s active involvement.
“We hope this leads to national-scale improvements in our ability to accurately exchange patient information while honoring patient privacy preferences, and ultimately providing better care and outcomes to those patients,” the document states.
Finally, the framework solicits feedback on a list of “minimal acceptable cross-organizational patient matching principles.”
“Once published in final form, we believe this list of principles will serve to create a component of a Level 1 adoption model that organizations can target, test against as appropriate, and declare conformance to,” the report says.
Comments on the framework are due by Jan. 22, 2016. Sequoia plans to publish the final version subsequently some time in 2016. All comments should be submitted to email@example.com. Comments will be reconciled by interested parties during public working sessions.