Developing Electronic Health Record (EHR) Strategies Related to Health Center Patients' Social Determinants of Health

Rachel Gold, PhD, MPH; Erika Cottrell, PhD, MPP; Arwen Bunce, MA; Mary Middendorf, BS; Celine Hollombe, MPH; Stuart Cowburn, MPH; Peter Mahr, MD; Gerardo Melgar, MD


J Am Board Fam Med. 2017;30(4):428-447. 

In This Article


Which SDH Measures?

Our stakeholders asked that the SDH tools include all the patient-reported IOM-recommended domains, made minor adaptations to the wording on some of these domains, and added a few questions (Table 1 and Table 3). For example, the IOM's single question on financial resource strain asks, "How hard is it for you to pay for the very basics like food, housing, heating, medical care, and medications?" (not hard at all, somewhat hard, very hard). Because CHCs treat low-income patients, many of whom are likely to screen positive for financial hardship, the CHC stakeholders wanted to augment this broad question with more granular questions about specific areas of strain (eg, food, utilities, transportation). The hope was that this granularity would identify the specific areas in which assistance was needed. The stakeholders also preferred to not use the IOM-recommended screening tool for intimate partner violence; they considered its questions too sensitive for general SDH screening. They opted for a broader question about exposure to violence that was taken from KP's SDH questionnaire. They also opted to add 2 questions on social isolation from KP's questionnaire (eg, "How often do you feel lonely or isolated from those around you?"; "Do you have someone you could call if you needed help?"), along with the IOM-recommended questions on social isolation. They also added a question on preferred learning style (eg, reading, listening, viewing pictures).

Collecting SDH Data

Stakeholder feedback, and our understanding that CHC workflows vary, indicated the need to enable SDH data collection by different care team members. Because EHR security measures limit which staff can access aspects of the EHR (for example, front desk staff often cannot access the problem list), we created several options for SDH data entry:

  • SDH "documentation flowsheets" were accessible to front desk staff at check-in, rooming staff, or community health workers (Figure 1).

  • Article versions of the SDH questions, in English or Spanish, that can be printed out and handed to the patient to complete at check-in or rooming, were provided on OCHIN's member wiki site. These data would have to be hand-entered by CHC staff into 1 of the EHR flowsheets described above.

  • A questionnaire on the patient portal allowed patients who had an online portal account to be emailed and asked to enter the data online before a visit. The EHR's panel management tool can identify patients with pending visits and enable bulk secure messages to these patients. Within the portal, patients can choose navigational instructions in Spanish, but the screening questions are available only in English.

Figure 1.

Social determinants of health flowsheet in EPIC.

We discussed various considerations during this process:

  • Making an electronic tablet available in the clinics' waiting rooms or examination rooms, on which patients could complete their SDH screening. Two of the pilot CHCs decided it would be too complex to manage, for example, identifying who would be the tablet's "keeper," where it would be stored, and how to identify which patients should use it.

  • Creating a setting on the computer in the examination room where patients could sign up for a patient portal account then complete the SDH data through the portal immediately. In the end, this proved unfeasible because the patient must be sent the questionnaire after they sign up for the portal, necessitating an impractical multistep workflow.

  • Clinicians did not want to collect SDH data themselves, preferring to transfer that responsibility to another team member. Two of the pilot sites opted to use the Article forms for data collection, then have a staff person enter the data into the EHR. This approach creates potential workflow barriers to use of the SDH tools, because until the responses are manually transferred into the chart, the data will not be available to care team members to act on during the encounter.

  • All options for reminding the team to conduct SDH screening were considered inadequate. Clinics said that best practice advisories (also known as alerts) are largely ignored. They preferred health maintenance advisories (HMAs), which are closely integrated into clinic workflows. However, HMAs must be standardized across all clinics using a shared EHR; because a universal HMA was not possible, HMAs were not a feasible option.

  • Similar to other screening questionnaires administered in clinical settings, clinics asked that the patient-facing data collection form not include a "refuse to answer" option. The staff-entered methods did include this option.

Reviewing Data on Patients' SDH Needs

SDH data might be collected via multiple routes, and certain SDH data are already collected regularly by most CHCs. Thus, there was a need for an EHR-based summary that contains all of a patient's SDH data. We created an SDH data summary that is automatically populated with data from any of the SDH data entry options and from SDH-related data elsewhere in the EHR. The SDH Summary also shows any SDH-related International Classification of Diseases, Tenth Revisions (ICD-10), codes from the patient's problem list and any past SDH referrals if they were associated with an SDH-related ICD-10 code (see more on this in "Tracking Past Referrals," below). "Positive screens" for SDH needs are visually highlighted. The algorithm used to identify positive screens is shown in Table 4. This summary could be accessed in 2 ways:

  • An SDH Summary tab can be accessed in an open Office Visit or Patient Outreach encounter. The most recent SDH data for the patient is displayed, and the date(s) of data collection and referral are shown (Figure 2).

  • A view in the EHR's Synopsis window can be accessed in a closed chart or open encounter displays a patient's SDH questionnaire responses over time, both as text and graphically (Figure 3).

Figure 2.

Social determinants of health summary tab.

Figure 3.

Social determinants of health summary in Synopsis.

For technical reasons, it was not feasible to show problem list data or referrals in the Synopsis version of the SDH Summary. Thus, each summary had information that the other lacked; that is, 1 had past referral information but only the most recent SDH data for a given patient; the other did not have past referrals but did present patients' SDH history, rather than just their most recent SDH data.

Identifying Referral Options

The pilot CHCs already had lists of SDH-related local resources in binders or on shared drives. These were not updated systematically, but rather only when someone on the team received new information and thought to update the list. The options for how CHC teams could do this systematically, using EHR-based tools, are shown in Table 2. All of them would be accessed via a hyperlink on the SDH Summary.

The preference list option was selected for several reasons. Creating linkages to an external agency's website was cost-prohibitive and required organizational contracts; thus, the study clinics might learn to rely on something that would incur costs after the study. Furthermore, some searches on these websites yielded results that were not specific to a location but rather gave statewide or nationwide data. The wiki options were rejected because users would have to leave the EHR system to access them, and the study sites were concerned about how to ensure that these documents were updated. The preference lists, however, used the same EHR function that the CHCs used for other referrals; involved discrete data fields, creating trackable data; and built on the CHC teams' local knowledge. One concern about the preference lists was that they must be updated manually. However, the study CHCs currently designate a staff member to update other preference lists (eg, for ordering laboratory tests), and the same person could be responsible for updating the SDH lists.

We helped the study clinics create "starter" preference lists for the SDH areas they prioritized (Figure 4). The resources listed in each were populated with data from each clinic's current method for keeping such information, then augmented by Web searches and reviewed by staff. The lists include names and contact information of relevant services and agencies, and include information such as "women and children only" and hours of operation, when available.

Figure 4.

Social determinants of health preference lists.

Ordering Referrals

The SDH referrals preference lists can be used to make internal referrals (eg, to a community health worker), have clinic staff facilitate external referrals (eg, calling an agency to schedule an appointment for the patient), or share agency information with patients at the encounter or in the after-visit summary so patients can follow up on their own. To make these easier to use, we created a new referral priority option of "no follow-up needed," which, if selected, informed CHC staff that they were not required to follow up on SDH referrals as they would for others. We also created a new referral type—"community referral, nonmedical"—so that SDH referrals would be excluded from related care quality measures. Another consideration here is that only certain care team members are authorized to make referrals of any kind; thus, support staff may need to be trained and authorized to use these tools.

Tracking Past Referrals

As described above, the SDH Summary accessed through the Summary tab (Figure 3) is automatically populated with information on past SDH-related referrals in order to enable CHC teams to track them. Referrals are shown in the SDH Summary if they are tied to a relevant ICD-10 code and/or if the SDH referral preference list was used. Presented data include the date of referral, contact information about the community resource, status of the referral, and who ordered it. Care team members authorized to edit referrals can manually update the referral status.

Lessons Learned

Lessons learned here may inform future efforts to build EHR tools for collecting and acting on SDH data. Because these lessons come from a pilot study conducted in 3 CHCs, we present them for consideration, not as a set of directions for SDH data tool development.

Considerations for which SDH questions to include. Consider striking a balance between standardized SDH data collection (ie, aligned with the IOM-recommended measures) and the need to adapt to meet local needs, especially given that SDH data collection may become required for EHR certification and Uniform Data System reporting.

Considerations for designing SDH data collection tools. Patients may decline to answer SDH questions. Consider having SDH tools include a "patient refused to answer" option. Consider the advisability of including a "decline to answer" option on patient-facing data collection tools, which might make it too easy for patients to decline. Also, ensure that EHR-based SDH data tools do not require duplicate entry of SDH data collected elsewhere in workflows.

Patients with a positive SDH screening result may not want assistance in addressing the identified need. Consider creating EHR-based SDH data tools that include response options to indicate this preference, or to otherwise note that help was offered and declined.

Considerations for designing SDH data summary tools. Carefully consider which SDH data sources should populate the SDH data summary and how to manage potentially conflicting data.

Considerations for designing SDH referral tracking tools. Monitoring the outcomes of past SDH-related referrals is challenging, and often requires outreach calls to patients. Consider whether this ability is desired.

ICD-10 codes related to SDH needs enable the tracking of such needs, but they may add to the complexity of the problem list. Consider creating an SDH "box" within the problem list.

Considerations for maintaining up-to-date SDH referral tools. SDH referral tools rely on updated lists of local resources. Consider whether established processes for maintaining other referral lists can be applied to SDH tools. Consider partnering with organizations that maintain such lists.

Considerations for SDH-related workflows. EHR-based SDH data tools need to accommodate diverse staffing structures, resources, and workflows. Consider ensuring that the appropriate care team members are authorized to access all aspects of the tools.

To avoid overwhelming clinic staff and care teams with SDH-related work, consider limiting SDH screening to a subset of patients and ensuring that EHR-based SDH data tools enable targeting this subset. Consider creating an alert to identify overdue patients. To avoid overwhelming care teams, consider designing the EHR tools so that SDH-related referrals can be marked "no follow-up needed."

Consider using electronic tablets[66–68] to enable SDH screening at registration or upon rooming, with workflows for using and tracking them. Clinics will need wireless Internet to enable tablets to transmit SDH data to the EHR.

To use patient portals for SDH data collection, consider developing workflows for helping patients create portal accounts at registration then enter their SDH data through the portal on the spot. Tablets may be useful here as well.