Driving Change With Data For Clinicians And Policymakers

Health data is beginning to play a greater role in clinical and policy decision-making for clinicians and policymakers to decide how to organize and prioritize healthcare delivery, how to distribute resources and funding, and how to measure quality. Hospitals, clinics, payers, and government agencies are generating and using an unprecedented amount and type of health data to make a variety of clinical, operational, and strategic decisions, such as whether a patient receives cancer screening, how many nurses should be on duty today, or what reimbursement rate will pay providers to deliver the highest quality of care.

Data-Driven Healthcare Transformation

Health data is becoming a lot more than just numbers. As medicine moves further into the digital age, the information in your electronic health record is no longer just limited to your lab values and vital signs. Numbers representing your blood pressure and glucose levels are only part of the story; your address, the nearest bus stop, and even your insurance provider can be tracked and shared with your health care providers.

Why It Matters

When data such as clinical, demographic, and administrative information is accurate, up-to-date, and useful, it can be a very powerful tool to improve patient outcomes. Doctors and clinicians can identify which patients are most at risk; stratify their patients for the most appropriate treatment based on clinical and other factors; and assess the effect of their interventions. Accurate and useful data is especially important in the treatment of patients with chronic medical conditions. A recent cohort study identified the effect of social factors on biological risk and on the chances of a patient with diabetes receiving the eye care that he or she needs to avoid advanced diabetic retinopathy (DR). The researchers utilized data from the Outcomes Consortium, a diabetes outcomes network that includes over 37,000 patients at 11 medical centers across the U.S.

Precision Beyond the Biological

Studies of disparities in eye health can inform several policy and programmatic decisions. They can aid in the design of programs and interventions to reduce disparities; inform allocation of community resources to needy populations or communities; and inform policies and regulations affecting access to health care for general and vulnerable populations. A recently completed study found that for both Black and White patients, the odds of receiving indicated eye health visits were substantially lower for those living in rural locations than for those living in urban locations. Specifically, rural-dwelling Black patients were 88% less likely to receive indicated eye health visits than their urban counterparts, and rural-dwelling White patients were 25% less likely. These findings identify geography as a major “smoking gun” barrier to health care access for patients for whom clinical indicators suggest that they need eye health services.

As we evaluate the clinical effectiveness of data, we are often surprised to discover deeper flaws in the healthcare system. Do patients and families fully understand the options available to them when they sign informed consent? Are there enough specialists in all geographic locations to meet demand? Are doctors and clinicians choosing to practice in underserved communities due to financial pressure to see more patients per hour or week? As we provide data to the market, we are being forced to respond to growing demands for tangible improvement, and are well-positioned to have payer and regulatory stakeholders join the conversation about what can happen and what is needed to make it happen.

The Impact of Social Determinants

Collecting and utilizing data on the social determinants of health (SDOH), nonmedical determinants such as housing, income, and language that affect health, has immense value for health care providers. By linking SDOH data to clinical data, care teams can connect their patients with resources in the community, develop outreach strategies that target the precise audience they are trying to reach, and design strategies for the prevention of costly complications.

Data from several thousand electronic medical records reveal that Black patients with newly diagnosed diabetic retinopathy are less likely to receive follow-up eye care than their White counterparts. Furthermore, when compared to patients without diabetic retinopathy, the odds of follow-up care for newly diagnosed diabetic retinopathy in Black patients with pre-existing diabetic retinopathy are actually lower. The Health Systems project provides a wealth of high-quality data, exposing in detail the extent of the problem as well as the specific areas that need to be addressed. Most importantly, the project uncovers the “inverse care law” – that those who need the most care receive the least – and health systems now have the opportunity to address this disparity through targeted follow-up strategies and interventions for high-risk patients. Clinicians will be able to understand why a high-risk Black patient does or does not return for a dilated fundus exam and design individualized follow-up programs to bring much-needed treatment to patients who require it most.

While there are numerous benefits to incorporating SDOH into clinical practice, the costs and complexities of SDOH data collection to support improved health for individuals and populations should not be underestimated. There is time required in the clinical visit to collect data and ensure that it is accurately integrated into the EHR. There is also the challenge of having a workforce trained to collect and use SDOH data with cultural humility. In the end, SDOH data collected that is not used is useless to health system leaders.

The Rise of Digital Intelligence

As technology advances and more data becomes available for analysis, there are increasing potential benefits and risks. While models that provide predictive information or clinical decision support can surface timely alerts and enable the development of a personalized treatment plan for each patient, effective governance of these resources is paramount to preventing bias in the underlying data from being exacerbated. As researchers and clinicians develop algorithms to diagnose and monitor a variety of eye diseases, the majority of the training data must be comprised of commercially insured, urban patients—individuals far removed from those on Medicaid from rural areas, whose neglect of eye disease has been highlighted in recent research as the greatest risk.

Ultimately, physicians must remain in control of technology used to care for patients, having a firm grasp of the capabilities and limitations of each tool to deploy them responsibly and maximize the technology’s potential to become a valuable asset in delivering high-quality patient-centered care.

Who It Affects

Patients: The Quest for Equitable Access

Data about patients can matter deeply. It can help ensure that patients receive the correct diagnosis at the right time, receive the clinical and non-clinical care and services in the most appropriate setting, and receive health information, education, and services in their preferred language and in culturally competent ways. For vulnerable patients, who often face significant challenges to receiving high-quality care, having the right data can be life-changing. This is the case for patients with limited English proficiency, patients living in rural communities far from specialists, the homeless, and others.

New research from the Harvard T.H. Chan School of Public Health finds that patient experiences are affected by both insurance status and ethnicity. Hispanic White patients had 15% lower odds of eye-care visits than non-Hispanic White patients in the same area. Patients with Medicare or Medicaid had lower odds of eye care than those with commercial insurance. Data-driven policy change could be lifesaving for many of these patients.

Patients are taking on the risk of collecting and using the data as well. While there are legitimate privacy concerns around the sharing of social determinants of health data (such as information about a patient’s unstable housing or food insecurity), there are also concerns around “algorithmic redlining.” Insurers could use all this data to make predictions about future costs, and then use those predictions to deny patients access to pricey treatments they may need. Patients and providers want more transparency around this process.

Clinicians and Care teams

Clinicians and care teams know what works well and what doesn’t for their patients, and they are interested to know what others are doing, what has been tested, and what works best.

As the clinical workforce faces increasing pressure, having a data-driven workflow can enable clinicians to make higher-quality decisions, reducing unnecessary uncertainty in patient care. Having a point-of-care risk score for common ocular conditions can inform patient selection at triage, and additionally, having a follow-up dashboard can highlight those patients who are overdue for follow-up care, including those in rural areas, as highlighted in this study. This allows clinics to contact those patients who are most at risk of having advanced vision impairment before they come into the clinic. Using zip code and other demographic data, providers can use clinical data to identify those most likely to fall through the cracks.

While improved integration of data systems can help to identify patient needs more accurately at the time of exam, poor integration can do the opposite by adding steps, decreasing accuracy, increasing alert fatigue, and decreasing time for the doctor-patient interaction that brought both to the exam. High-tech must be paired with high-touch to identify the need, and then allow the clinician to provide the care to meet that need. High-quality training, system design, and clinician input in the design of these tools are critical to this end.

Healthsystems and Payers: The Infrastructure of Value

Building Information-Infrastructure Capabilities: Providers must develop capabilities to gather, store, and utilize data (including shared information across providers) and information-technology systems such as interoperable electronic health records systems and sufficient numbers of trained data analysts and scientists.

Outcomes-based and Value-based Reimbursement

As payments for services shift from a fee-for-service model, insurance companies will rely on a set of common metrics to determine which providers should be paid for delivering high-quality services.

Payers are looking at data from this study of 17 health systems to understand why certain populations may have long-term costs that are higher than others. For example, the cost of blindness and disability due to vision loss will likely far exceed any savings related to staffing efficiency unless all diabetic patients in a health system receive a dilated fundus exam to prevent vision loss. Data from this study can help payers move from a pay-for-volume model to a pay-for-value model.

Policymakers and Regulators: Setting the Standard

Data can be particularly valuable in helping to make policy and programmatic decisions about how to assign workers, design incentives, and set health priorities. For example, the finding that individuals living in rural areas have 44% to 88% lower odds of having their essential eye care visits can be used to justify increased telehealth infrastructure and incentives for ophthalmologists and optometrists to practice in rural areas.

Regulators, policymakers, and health administrators will need to consider how the distributional effects of algorithms that predict need influence the allocation of resources. In addition to the above worries, such tools could systematically discriminate against the very populations who are most underserved and in need of health care services.

What Changes

A Roadmap for Action

Moving from a focus on data collection to outcomes is not a function of willpower or good intentions. It requires a basic shift in how an organization is structured and how it conducts business on a day-to-day basis.

1. Integration of Actionable Data

Clinicians and other health professionals need access to relevant data at the point of care, in a format that is no more burdensome to process. For social determinants of health, EHRs should integrate social needs data and patients’ preferred languages so that clinicians are prompted to make relevant referrals during the care delivery process. For example, if a patient lives in a rural zip code with few eye doctors, after screening for vision loss, the clinician should be prompted to schedule a telehealth consultation or mobile screening unit appointment.

2. Redesigning Payment Models

Payment models and incentives need to be revamped to encourage the use of data to improve the quality of care. Payment for care coordination, telehealth, and community linkages must be reimbursed by payers, in addition to payment for face-to-face time with patients. Current reimbursement models are not meeting the needs of rural and minority populations. By investing in the infrastructure of access, successful interventions can become more fiscally sustainable and address health disparities. For this study, the researchers invested in transportation vouchers and community health workers who do bridge work specifically for Black patients with pre-existing conditions.

3. Investing in Interoperability

The need for interoperability and data standards has never been more pressing. A person with diabetes who attends their primary care practice for routine monitoring and then, a few weeks later, attends an ophthalmology outpatient clinic may receive a less-than-optimal experience of continuity of care. The lack of interoperability between systems may produce poor linkage of care, completely invisible to both providers. Improving continuity of care requires an integrated, unified digital view of the person’s experience of health and care services across the entire end-to-end journey.

4. Establishing Ethical Governance

As predictive tools and algorithms are implemented into clinical workflow, there must also be a framework of governance and transparency around their use. This means performing audits for bias, annually reporting performance by demographic for all modules, and ensuring meaningful input from patients and clinicians alike as tools are implemented in practice. The ultimate measure of success will be to strive toward odds ratios close to 1.0, signifying true equity in healthcare delivery and a “level playing field” in the diagnosis and triage of eye disease relative to current levels of knowledge in ophthalmology, as cited in the literature.

What will it take to truly use data to achieve equity and opportunity in urban communities? The answer to this critical question is now, and that time is running out.

It’s time for healthcare to go from data-rich to data-driven. Our latest study shines a harsh light on the inequities built into the current system for treating blinding diseases. Large and often unpredictable disparities emerged in outcomes for patients with the same disease, treated by the same doctors with the same technology, depending on their race, whether they had Medicaid or commercial insurance, and even whether they lived in a city or a rural town.

By utilizing more advanced data and analytics incorporating SDOH, we can move beyond simply identifying disparities in health and begin to address them. We must first ask if there are disparities; then, where and why they occur; and finally, how we can intervene to address them. The data is the map. The policy is the vehicle. The end destination is a healthcare system in which a patient’s zip code or race no longer determines the health of the life they will lead or the quality of care they receive.

References

1. https://pubmed.ncbi.nlm.nih.gov/39264618/
2. https://pubmed.ncbi.nlm.nih.gov/40669406/