Perspectives in Primary Care: The Foundational Urgent Importance of a Shared Primary Care Data Model

During my office visits, I too often mumble, "There has to be a better way" while wondering why my electronic health record system seems decades behind my smartphone.

Fixing this problem starts with re-design of our data models as Drs Green and Klinkman astutely note. It will require active involvement of family medicine and other primary care disciplines.

Larry Green and Michael Klinkman should be applauded for sketching out their vision and for inviting others to join them in this creative process.

I like much of their model. My comments focus on suggestions for refining it.

1) I really like the inclusion of patient defined goals. This helps ensure that care revolves around patient autonomy and choice. While establishment of goals should be a collaborative process between the patient clinician, ultimately no goal should appear unless it is endorsed by the patient. Concerns by the treating clinician that are not endorsed as a goal by the patient can placed in other places, e.g. a risk factor, problem or concern. I might even feature patient goals at the center. Goals, explicit or not, not only represent patients' reasons for obtaining health care but also provide a unifying focus for not only the patient's primary team, but their broader health team which might include specialists, navigators, and community programs.

2) Functional status does not belong under goals any more than risk factors do. Functional status reflects another dimension of the person. When a patient indicates that their goal is improve their functional status then improving functional status becomes a goal.

3) The IOM issued a report this year, Capturing Social & Behavioral Domains & Measures in Electronic Health Records. These domains warrant consideration for inclusion under "person." Calling out these key health determinants will help ensure they receive the attention. Examples from these domains include collection and standardization of patient race, ethnicity, language, educational level, SES of area of residence, and health literacy, and key behavioral determinants including physical activity, smoking, and alcohol use.

4) In the future, many person-level factors will be derivable using "Big Data" gleaned from electronic health record data. These profiles could be added and continuous updated under "person."

5) Consideration might be given to a sixth, major category- resources. Its inclusion focuses on patient's assets (not just financial), strengths and other enablers of care such as insurance. This section could be expanded to include patient-level and community level resources that can be potentially brought to be bear when achieving patient goals. Note: these data do not necessarily need to appear on the screen. In many instances, relevant might exist behind the screen. For example, if the patient's goal is to lose weight. This goal might trigger local resources. Information would include location, language, out-of- pocket costs, and ideally information about quality, e.g. success rates, use ratings. The same would apply to specialists. If a patient's goal is to improve function and the patient desires knee replacement surgery and rehab to achieve this goal, then a list of knee surgeons with the corresponding out-of-pocket costs (ideally bundled with hospital and rehab costs) along with complication rates and patient reports of experience of care would appear to guide patient decision making. Factors such as insurance ad language etc would trigger relevant resources minimizing the burden in matching patient need to resources.

6) Insurance (type, deductibles, co-payments etc.) should be called out. As more of us enroll in high deductible insurance plans and cost of health care climbs, the cost of any test, referral, medication or procedure will become increasingly important to patient decision-making. Many health plans currently allow patients to determine and compare their out-of-pocket payments depending the specialist, laboratory, or hospital. This information needs to be integrated in the EHR at the point-of-care so that the clinician and patient can determine the cost at the time the patient is making a decision. Health care is one of the few sectors where we learn of our costs after the fact. This needs to change.

7) I realize this perspective focuses largely on conceptual issues. However, the issue of expanded transmission of structured data is needed. Examples include dates, diagnoses, and actions from ED visits, hospitals and specialists in addition to any tests ordered. Common structured data will decrease clinician burden and improve the validity of population level reports and tracking and improve quality reporting.

8) Once consensus on core elements are agreed upon, then data sources (office EHR, patient portal, patient self monitoring systems e.g. blood pressure, glucose, physical activity etc), hospital, laboratory, ED, specialists, community programs e.g. DPP, and insurance plans can be integrated.

9) While the issue of user-interface is beyond the scope of Drs Green and Kinkman's paper, user input combined with the science of cognition and human factors needs to be considered to re-design patient and clinician interfaces that efficiently engage patients and clinicians in addressing patients goals.

Competing interests: None declared