DISABILITY STATUS IN EHRs

Megan Morris recounted how when she went to visit her uncle, who had a developmental disability, in the hospital after he suffered a serious fall, she found his hands tied to either side of his bed, with no access to a nurse call button. When she asked the nurse about this, the nurse told her that because he had a developmental disability, he was a danger to himself and others and did not have communication skills, so he did not need access to the call button. Morris walked the nurse back to her uncle’s room and began asking him yes/no questions—including a question about a recent presidential debate; he could answer yes with a thumbs-up with his right hand and no with a thumbs-up with his left hand, and he answered correctly each time. “That team had made assumptions about David because of what was written in his medical chart, that he had a developmental disability,” she said. “I believe one of the first steps to begin to address these disparities and these challenges that he experienced is through consistent documentation of disability status.”

Morris explained that diagnostic codes are based on a medical model and are used to inform billing and medical treatment. A clinician documents them, and they are located throughout a patient’s chart. In contrast, disability status is based on a social model, and there are two main purposes of documenting disability status in the EHR. The first is to inform the provision of accessible health care as the Americans with Disabilities Act and Affordable Care Act mandate. “If you do not know who has a disability, you cannot provide them with accommodations,” said Morris. The second purpose of disability status documentation in the EHR is to identify and address disparities. Disability status needs to be self-reported, and it should appear in a prominent location in a patient’s EHR. One way to elicit information about a person’s disability is to ask a series of questions (Figure 8-1).

FIGURE 8-1

Disability status questions. SOURCE: Morris presentation, April 5, 2024.

There are many reasons diagnostic codes are insufficient for identifying and addressing disparities, starting with the inconsistency with which clinicians document them. For example, a patient could have a diagnostic code related to stroke with hemiparesis in their EHR, but if the clinician at a follow-up visit does not use that code, it could mean the clinician decided not to use that code anymore, or it could be that the person has recovered from their stroke and does not have hemiparesis anymore. “Since disability status can change over time,” said Morris, “we need consistent and regular documentation of disability status.”

In addition, she explained, diagnostic codes do not provide information about accommodations, so if someone has a cerebral palsy diagnosis, the associated code does not say anything about that individual’s specific limitations. Without knowing if the individual has difficulty with cognition, mobility, vision, or hearing, it is difficult to have necessary accommodations ready when the individual has a medical appointment.

There are two types of disability bias in the EHR. The first is stigmatizing language, such as wheelchair-bound or retarded, and the second is language suggesting biases and stereotypes, such as lazy or noncompliant (Figure 8-2). Morris noted that research shows that when health care team members read biased language in the medical chart, it affects the medical care they provide and their decision making (Casau and Beach, 2022).

FIGURE 8-2

Stigmatizing language and disability bias in clinical notes. SOURCE: Morris presentation, April 5, 2024.

Morris and her colleagues have been conducting studies over the years to document disability status in the EHR. Patients, she said, support these efforts and do not object when asked about their disability status. One finding from her research points to the importance of tying disability status to the legal requirement for providing accommodations to give health care teams a reason to document this information. However, health care systems are having trouble implementing disability status in their EHRs. The lack of standardized tools to collect disability status in the EHR is one impediment. Another is a lack of federal, state, and local policies that require documenting disability status.

There are still significant biases around documenting disability status. One common statement she hears from clinicians is that documentation is great, but not for those faking a disability. Morris recalled how one director of a primary care clinic told her they could not document disability because everyone would claim low back pain and disability. “We need to think about addressing those biases and that lack of education,” said Morris.

The situation is not completely dire, as there have been positive advances in the policy and research areas. For example, in July 2022, the Office of the National Coordinator for Health Information Technology released the third edition of its interoperability standards that contain elements representing disability status. In 2023, the Joint Commission released a new health equity certificate that requires disability status documentation, and in 2024, the Centers for Medicare & Medicaid Services (CMS) released the CMS Enhancing Oncology Model that requires disability status documentation. Morris noted that while the Health Resources and Services Administration requires documentation of race, ethnicity, sexual orientation, and gender identity, it does not mandate collecting disability status. “This is hampering research and advancement in this area,” said Morris.

Recently, the National Institutes of Health awarded funding to Morris and colleagues to develop and evaluate workflows for consistently documenting disability status in the EHR and then use the data to inform provision of accommodations. Morris and her collaborators are working with Epic to create a standardized approach for documenting disability status. On a final note, Morris said the Disability Equity Collaborative, which includes members from health systems, providers, insurers, and patients, issued an implementation guide to help health systems integrate disability status collection into the EHR and workflow processes.

THE EHR AND WHAT IT DOES NOT TELL US

When a member of someone’s care team opens their EHR, said Carolyn Petersen, they can see the individual’s personal and family histories, test results, and any diagnoses. There will be treatment history, some health outcomes and patient-related outcome measures, maybe some information on social determinants of health, and perhaps some person-generated health data, such as sleep patterns and other information an individual maintains for themselves and through an agreement with their care team, though the latter is not standard. The care team can update personal and family histories; review diagnoses and previous care; check test results and patient-reported outcome measures; order tests, medications, and durable medical goods; make referrals; and schedule appointments and consultations, both internal and external.

Petersen said when people get into their EHR through their patient portal, they may see inaccurate or incomplete information, which she said can be concerning at a minimum and even enraging and frustrating. Though individuals may attempt to correct misinformation in their EHR, many EHR systems do not allow that. Regarding what gets missed in the EHR, Petersen said the EHR does not capture the effects of disability or illness on daily life. The EHR also does not capture the effects of any changes in a person’s health and ability to function in all the environments and roles of which they are a part.

Petersen presented a case study involving fragrance sensitivity, an invisible disability that affects some 20 to 25 percent of people, said Petersen (de Groot, 2020). Today, over 2,000 fragrances occur in various consumer products, and a fragrance can include 10 to over 100 chemicals. Some chemicals help the fragrance linger in the air so the fragrance can persist. She said it is hard to know the chemical names of a fragrance’s constituents, making them difficult to study. There are many symptoms of fragrance susceptibility, including respiratory distress, skin rashes, headaches, other neural symptoms, and sometimes nausea. Skin sensitivity tests can detect many allergens, but not all.

Regarding what the care team can do with the EHR, they can document discussions about fragrance-related issues, order tests, annotate recommendations for over-the-counter medications, prescribe medications, and create a referral to a dermatologist, allergist, or other specialists. From an individual’s perspective, the patient is aware of all the challenges they encounter in trying to manage those symptoms, which can change from day to day, and they are aware of the limitations to their lives when they cannot adapt their environment or roles.

What is missing from the EHR are the reduced social interactions given the need to avoid public transportation with assigned seating because of the possibility someone wearing too much perfume will sit next to them. These things cause individuals with a sensitivity to have few career options and opportunities. They may, for example, have to work remotely or be ineligible for company-provided health care insurance, and their income may suffer, reducing their access to supportive services such as home food delivery. Finally, there are symptom-specific risks. Antihistamines, for example, may increase a person’s experience with hazards; light sensitivity may make people susceptible to falls given their use of sunglasses; and nausea can increase the risk of falls that result from dizziness from skipping meals.

Petersen concluded her remarks with four key points:

1.

EHRs document health information but not the effects of impairments on individuals’ lives, thereby limiting care teams’ ability to recognize needs.

2.

People have varying degrees of health literacy and digital skills, and they may not document health and disability issues and their effect on function in the terminology of medical professionals and agencies.

3.

Health conditions and disabilities are dynamic, with variable and changing effects on functions not captured in the EHR.

4.

Determination of function and disability is not a function of technology but a process between an individual and their care team; technology may be a facilitator, not a solution.

SOCIAL DRIVERS OF HEALTH IN THE MEDICAL RECORD

Prerana Laddha said there is substantial focus on equitable care in the EHR resulting from its ability to collect accurate data on race, ethnicity, sexual orientation, gender identity, and social drivers. The EHR also enables compiling these data on a population level to understand where disparities in the health system are, and it can provide interventions in a provider’s workflow to promote equity. She noted that over the last decade, health systems have become increasingly interested in documenting social determinants of health and addressing them for their patient population.

Laddha said her company includes validated clinical assessments to document this information in its EHR and makes the information available as a social determinants of health wheel in the patient’s chart that a provider accesses in the normal course of their workflow. “Having that type of data front and center not only helps this provider make the right decision, but it is also constantly promoting equity as they are going through their patients in their busy schedule,” said Laddha. There is value, too, in showing these data over time, said Laddha. For example, a clinician who sees their patient has food insecurity and connects them with Meals on Wheels can see if that connection helps the patient improve their health. She added that a reliable method for documenting social drivers and disabilities is to have the patient answer directed questions through their patient-facing portal outside of the clinical setting. Sometimes, doing this in the privacy of their own homes makes it easier for patients to answer questions sensitive to them.

Using social data promptly within workflows is something today’s EHRs can do. For example, said Laddha, a scheduler can see a patient has transportation risks and contact the patient before their appointment to ask if they need a rideshare service to get them to the office. Interoperability standards, she added, can enable health systems and organizations to exchange these data so as someone moves between health systems and organizations, their data can move with them.

Laddha said EHR vendors such as her company will soon incorporate artificial intelligence (AI) tools in the EHR to extract information from clinical notes. “Clinical notes is our primary focus because we have seen statistics that over 50 percent of this social driver data still lives in clinical notes, so it is a good place for us to start extracting that information using AI,” said Laddha.

Disability and accommodation needs are a part of what the EHR can capture within various workflows, including during scheduling, registration, and clinical encounters. “Just like with social drivers, having that ability to document the disability status and accommodation needs before my upcoming visit [can help] the clinic or the hospital be better prepared to accommodate me when I get there,” said Laddha. She noted that she and her colleagues are working with Morris’s team to standardize disability data collection to improve interoperability and visualize the information for providers.

As mentioned in an earlier presentation, providers are sometimes hesitant to document social information because they do not know how to help the individual. To address this, Laddha’s company’s EHR has a resource directory available. When a patient screens positive for any of the social needs in the directory, the system automatically selects some community resources and prompts the provider. She explained the automation is based on several factors, including the patient’s demographics, their insurance coverage, veteran status, and location. This does not take additional time from the office visit, nor does it increase the need for documentation or to search through a list of services. The provider can text or mail these resources to the patient and communicate bidirectionally with the community providers to let them know a referral is coming.

The ability of EHRs to capture Z codes can help justify needed interventions. A patient who presents with chronic conditions exacerbated by homelessness, and whose EHR contains a Z code denoting that, can help the clinician justify the extra interventions and services the patient needs. It can also help with reimbursement, a further encouragement to document this information, said Laddha.

As she mentioned earlier, EHRs can provide information at the population level. This can enable a health system to see overall screening rates and how many people within a population are screening positive. Geographic information can also pinpoint problem areas where housing or transportation are major social drivers, enabling health systems to target those areas with strategic initiatives, such as establishing a food pantry or lobbying for an additional bus line. EHRs can help health systems analyze the effects of social drivers on health outcomes. An analysis of outcomes and social drivers might show, for example, that patients with diabetes with adverse outcomes are affected more by a lack of transportation than by an elevated A1C level.

STANDARDS AND OPPORTUNITIES

Community health centers, said Julia Skapik, arose as an outgrowth of the civil rights movement to address a lack of culturally competent health care in health care access deserts. There are five essential elements to a community health center. They are in high-need areas and provide comprehensive health and wraparound services, including enabling services or social care services. They are open to all residents regardless of insurance or the ability to pay, with a sliding scale fee based on income, and they are nonprofits governed by community boards to ensure responsiveness to local needs. Finally, they follow performance and accountability requirements regarding their administrative, clinical, and financial operations.

Today, said Skapik, the nation’s 1,487 community health centers serve about 9 percent of the U.S. population, or 31.5 million people, and over 14,000 sites. They disproportionately serve underserved communities, and the majority of health center patients come from minoritized populations. They also serve a large proportion of people who are unhoused or who are uninsured. At her health center, over 40 percent of its patients are best served in a language other than English.

Skapik, who serves as a part-time primary care physician at a community health center, commonly interacts with people with a disability. In that role, her primary goal is to help these individuals achieve and maintain their goals and functionality. However, her community health center’s information technology infrastructure has a limited focus on assessing and improving functional status, understanding a patient’s story, and supporting their goals. Skapik said,

There is a duality of being a health care provider in this space because on the one hand, I genuinely want to help meet my patients’ needs, and on the other hand, what I see as the activities around disability are administrative, burdensome, confusing, and frustrating.

She added that some of this frustration stems from working to document social drivers and disabilities for patients and not being reimbursed for that work.

Theoretically, the EHR has the information clinicians need, and in fact, there is many times more information in a patient’s EHR than anyone will ever look at or use. What the EHR does not adequately support—and this, she said, might be a generous categorization—is functional status and disability status and the workflow around that. She also commented that medicine still treats health data as little fragments of something at one moment in time tied to a specific encounter. “We do not think about these things as episodic, so it is difficult to understand a patient’s story by looking at those fragments of data,” said Skapik. In her opinion, the time is right to use AI, health IT standards, and fast processing to unlock the information in the petabytes of data in an EHR.

Skapik noted that while the Office of the National Coordinator for Health Information Technology required health organizations to have access to Fast Healthcare Interoperability Resources application programming interfaces, community health centers are often last on the implementation priority list. In fact, too many community health centers do not have access to this interface. One problem with the current standards is that capturing disability status is limited to finding a Logical Observation Identifiers Names and Code that denotes an individual’s disability status. What is needed, said Skapik, are sound data models built with the input of patients with lived experience and subject matter experts who understand the germane science and research.

Skapik said she dreams of the day when a dashboard tracks over time a person’s functional status and sends her alerts when there are changes in an individual’s functional status. This requires identifying the data providers will record over time and setting thresholds for notifying the clinician when functional status has changed. For cognitive status, there may not be an easy way to identify changes over time in the EHR. She mentioned the Pacio Project as a successful partnership that aims to create formal standards for postacute, home, and functional status improvement and build use cases before building these standards.

EHRs, said Skapik, contain data that could help identify where disability and the social determinants of health intersect. Those health care organizations that are not doing this analysis are not tracking health disparities in their organization, adding,

If we are not setting up dashboards and support for analytics at the point of care to look at the intersection of all of these different domains, we are going to fail to see that there are some really big signals and big opportunities to address those.

Skapik mentioned the Gravity Project, which aims to accelerate the adoption of nationally recognized standards to advance identifying and acting on social determinants of health. She also briefly discussed the validated PRAPARE tool, a national standardized patient risk assessment protocol built into the EHR. PRAPARE is designed to engage patients in assessing—and importantly—addressing social determinants of health. She noted that the focus on actions to address social determinants of health is important for ameliorating the “moral hazard” people experience when asked about social determinants of health without having an intervention to deal with them. Regarding the Z codes that health care organizations use to capture social determinants of health, Skapik said they do not contain enough information to understand what a patient is experiencing.

Skapik offered suggestions for improving how the EHR can support disability. There is a concept called the care plan that aims to link these pieces of information with related information and track them over time to generate a complete picture of what is going on with a patient. The data should come from both the patient and everyone involved in the care ecosystem, including caregivers the patient authorizes to contribute data. Federal EHR regulations support this concept, she said, though one challenge is convincing the care team there is value in documenting a patient’s goals and what the health care system is doing to meet those goals.

Skapik also listed opportunities to support disability in the EHR. These included:

• Standardizing disability templates and data elements;
• Enabling electronic submission of forms for disability determinations, using the model of electronic prior authorizations;
• Better supporting care teams by providing regular evaluation and documentation of a patient’s functional status, cognitive and behavioral health status, social determinants of health, and health-related social needs; and
• Integrating patient-generated health data via apps and allowing them to track their own status.

Q&A WITH THE PANELISTS

Laura Jantos, from RecastHealth, asked how bidirectional communication with community services is happening and what systems community providers have that allow them to receive this information in a structured format if they do not have EHRs? Laddha replied that providing community resources has been an area her company has been working on in terms of improving the software and workflows. Her company’s EHR, for example, integrates with FindHelp and Unite Us, both of which evaluate social care investments. Community-benefit organizations, said Laddha, use this software to receive electronic referrals and accept or decline them. She noted her company is working with the Gravity Project to standardize the interfaces so this software can be adopted broadly.

Maggie Downey, a former medical social worker, said she is excited about health care reforming how it addresses the social determinants of health, but she struggles with how a community resource directory, even with bidirectional communication, is better or different than a social work model, which has not meaningfully addressed the social determinants either. Laddha said what her company has seen over the last few years is that connecting patients or assessing patients for social needs is happening across different settings. Previously, she said, it was care managers, social workers, or community health workers who focused on this work, but it is now happening in hospital settings. Therefore, providing a quick and easy tool that can be automated and help workflows is what the resource directory aims to accomplish. Skapik added that developers could use the Fast Healthcare Interoperability Resources standard to create a smartphone app that social care organizations can use without needing an EHR to enter the information they want displayed and that consumers could access directly and enter information.

Amy J. Houtrow commented that if EHRs are going to document disability status and providers are biased against people with disabilities, it is surprising that patients with disabilities favor having that information in their EHR. She also noted the importance of acting on disability status and providing accommodations, yet the disability status questions Morris listed do not provide the information needed to address accommodations. Given this, she wondered how the field can get to a place that identifies what people need and provide it and not have them face discriminatory practices in health care.

Morris replied that people with disabilities have told her it is key to ask questions about disability and accommodations early, before a clinical appointment. The challenge with the questions Houtrow raised is that they must serve two purposes—identifying patients who require accommodations and tracking health inequities—that are often at odds with each other. Regarding the bias people with disabilities face, she said if someone is in a wheelchair, they will experience biases whether they are asked about their disability or not. However, acquiring that information is a first step toward providing equitable care and getting health care teams to think more explicitly about their biases.

Skapik said asking people in their own words is undervalued in health care. The advent of AI and natural language processing creates the ability to take large groups of similar disparate concepts and group them in a meaningful way, “but that is not worth anything if we do not display that information to the care team and let them understand what supports they must figure out if they are appropriately accommodating a condition,” she said.

Petersen, speaking as someone who has had a disability long enough to remember when employment forms said people with physical defects need not apply, said that to realize the potential of improving care for people with disabilities, people with disabilities, their partners, and patient advocates must continue pushing for the positive uses of these data and tools and not expect it to organically occur on its own.