The current pandemic has highlighted enduring health inequities that are contributing to higher COVID-19 infection and survival rates among certain populations. Lower-income individuals, people of color, and those without ready-access to healthcare, healthy foods, and adequate medication have disproportionately suffered severe health complications and death from COVID-19.
The health crisis also spurred an economic downturn that drove unemployment rates as high as 14.7% in April 2020 and created a desperate situation for millions who lost both their incomes and health insurance. With the increased number of potentially vulnerable individuals, healthcare leaders committed to the health and wellness of their patient populations must take measures to identify those at risk for poorer health outcomes and/or patients who could benefit from additional health resources.
Social determinants of health (SDoH) can provide a wealth of information about non-clinical factors impacting a patient’s overall well-being. By looking at environmental and physical influences, access to medical care, and social factors, providers can glean a wide variety of insights into which patients might be at higher risk of COVID-19 infection, such as those living in multi-generational homes or working in conditions where social distancing is difficult.
Unfortunately, the actual identification of a patient’s SDoH is challenging because many of these details are essentially trapped as unstructured text within clinical notes, telehealth transcripts, patient portal messages, or secure email exchanges. EHRs and other clinical systems were typically designed to support billing processes and often have too many structured fields that commonly overwhelm clinicians. Even then, most EHR’s lack the structured fields to store vital information such as a recent job loss or crowded living conditions or risky behavior that increases contamination risk. A careful clinician may document such observations into a patient’s chart, but the details are usually captured within free text fields that are not easy to search. As a result, clinicians are often unaware of key SDoH information until after a patient’s health has been negatively affected.
To efficiently identify important SDoH red flags and proactively address patient and population health needs, healthcare organizations can leverage technologies such as AI-based natural language processing (NLP). Tools such as NLP allow healthcare providers to take large volumes of unstructured data from clinical records and automate the identification and extraction of key concepts. This eliminates manual and time-consuming chart reviews to find critical information about patients and populations, including specific SDoH details. When combined with structured data, organizations have a more complete, 360-degree-view that provides direction for proactive outreach.
A complete view of a patient that includes SDoH details provides care coordinators with better insight into the health of individual patients, allowing them to focus their outreach efforts on higher risk patients. For example, we now know that individuals with certain underlying medical conditions, including diabetes, obesity, and heart disease, are at increased risk for severe illness from COVID-19.
If an individual with one of these underlying medical conditions also works in a crowded meat processing facility, a care coordinator can engage the patient and make sure they are compliant with recommended treatments, including taking medications as prescribed or getting routine screenings. For patients that require additional services, such as financial assistance with medication or transportation to the doctor’s office, care coordinators can take appropriate action to ensure the best possible health outcomes.
Managing patient populations
Another way healthcare leaders can work to reduce health inequities is by analyzing SDoH details across their patient population. With a more enriched picture of the patient population that includes SDoH details, an organization can customize its population health programs to address unmet patient needs.
Consider, for example, how SDoH details can provide deeper insight into a population’s risk for diabetes, which along with prediabetes, affects more than 100 million Americans and accounts for approximately 25% of health care dollars spent in the U.S., according to a study in Diabetes Care. Diabetes risk is closely associated with social and economic factors, and is more common among non-white populations, with black, Hispanic, and Native American populations experiencing the disease at much higher rates than whites.
Though structured data can detail risk factors tied to weight, race and age, it is likely to miss additional risk factors that could be noted in free text within physicians’ notes, such as limited access to proper medications and healthy foods, barriers to physical activity, high stress levels and social isolation. Additional free text notes might mention complaints of excessive thirst or hunger, frequent urination, fatigue, or blurred vision. Further, information pertaining to laboratory values such as hemoglobin A1c and blood glucose levels are also markers that may appear in free text lab reports and be missed when relying solely on structured data.
Using NLP to identify critical information from unstructured notes, an organization can more accurately and efficiently assess the need for additional resources to educate patients about diabetes risk factors or to offer additional nutritional counseling services, or, for partnerships with payers to reduce out-of-pocket drug costs for medications or to provide affordable gym memberships.
The pandemic is taking a toll on all of us – but some populations continue to suffer poorer health outcomes than others. By automating the discovery of critical SDoH detail within unstructured patient data, healthcare organizations can glean deeper insight into the needs of their patients, take proactive measures to reduce disparities, and improve overall patient health.