top of page

Six Ways Data Can be Used to Address Healthcare Inequities

Inequities in healthcare are nothing new. Decades of research have shown that race makes a difference in how patients are diagnosed, treated, and receive follow-up care. One of the most significant ways that long-term inequities can be addressed is by accurately collecting and using data.

Data are essential to understanding where and why disparities exist, how they can be mitigated, and where resources need to be deployed. There is no better example of data's impact than how it was utilized during the pandemic. It helped answer questions like how COVID was spreading in communities, what was the best allocation of resources and capacity and how to determine the amount of equipment, such as ventilators, that would be necessary to treat ICU patients. Healthcare organizations also realized the value of sharing information and working together to find solutions.

The volume of big data is projected to increase faster in health care over the next seven years than in any other field, and it has already increased by 47% in the previous seven years. Medical imaging, health information exchanges and health records, labs and genome testing, wearable fitness devices, pharma research, medical research, search engines, and smartphones provide actionable insights. But how can the medical field disseminate all of this information to not only use it to solve complex challenges but also to provide the highest quality of care for all their patients? BCT Partners, a company dedicated to accelerating equity, shows how the accuracy of REaL data — (attributes of race, ethnicity, and language) tied to individual data records can address healthcare disparities.

1. Proper data collection

The beneficial aspects of utilizing data can begin once the information is accurate, so it must meet the following standards:

Accuracy: Self-identified, correctly recorded with consistent categorization

Completeness: Is REaL data captured across all services?

Uniqueness: Are individual patients represented only once?

Timeliness: Are data updated regularly?

Consistency: Are data internally consistent, and does it reflect the patient population served?

2. Provide Staff Training and Support

To stratify, characterize, and assess REaL data, organizations must first develop a collection plan and then train staff. Support staff must understand the value of consistently interacting with patients to collect information to serve them better. They also must be able to address concerns the patient may have about why the data is necessary and how it is secured. For example, they need to articulate why there are questions about race, ethnicity, and language and how that helps to identify inequities and the steps to address gaps in patient care.

3. Collect Missing Data

Segmenting rates of missing data by region is the first step in identifying areas for improvement. As with other patient-provided information, incomplete REaL data will likely vary by collection mode (e.g., in person, mail, patient portal). At this point, organizations can begin an outreach project with patients to collect missing data points. Of course, organizations are unlikely to get to a 100% completion rate; however, most agree that a threshold of 95% is achievable.

4. Assess the accuracy of data

Best practice for REaL data requires self-identification by the patient or patient proxy, with the implication that the patient's choices will provide the most accurate records. As with any data item, organizations need quality assurance to ensure that categories indicated in the data record accurately match patient choices. Some methods of achieving this include validation sampling, where patients are randomly selected to be interviewed to ensure their selections match the recorded information. Another way to ensure the accuracy of reports is to ask patients how well they understand the questionnaire and if steps can be taken to make it easier to use. Finally, they should observe staff to ensure consistency of requests for information and their following of protocol.

5. Mitigate bias

When including health equity data sets in analytics, there is the potential to use that data as a source of bias or unintended discrimination. Some have suggested that a concept of "fairness through unawareness" should be implemented to avoid bias, which means avoiding collecting specific data points like race and ethnicity to prevent discrimination. However, that is not an option for healthcare because inequities already exist and will continue until we can better understand how to develop solutions to address them. So, to avoid the possibility of discrimination while still collecting granular data, organizations should adopt validated bias identification and mitigation strategies. This means that designers, practitioners, and organizational leaders should consider legally protected groups or subgroups that might be vulnerable to unfair outcomes and use bias-aware algorithms during the design phase and development lifecycle of an AI system.

In addition, healthcare demographic data, when collected at an aggregate level, is less useful for health equity purposes. For example, the industry standards for race and ethnicity defined in 1997 by the Office of Management and Budget (OMB) need to be more granular for the racially diverse world we live in today. Comprehensive data needs to be disaggregated by race/ethnicity to address inequities. Updated and increased racial/ethnic data standardization at both a state and federal level would aid in interpreting information.

6. Analyzing the data

Once patient information is as accurate as possible, organizations need advanced tools and talent to analyze the data to discover meaningful insights and mitigate bias, as described above. Due to the high volume and variety of data, machine learning (ML) and artificial intelligence (AI) are essential. Artificial intelligence can help improve the accuracy of clinicians' information so they can better prioritize their time, empowering them to focus on patient care. AI can also be integrated into diagnostic imaging analysis so that clinicians can identify conditions more quickly and intervene earlier. And ML tools can provide various treatment alternatives and individualized treatments that improve the overall efficiency of hospitals and healthcare systems.


Healthcare organizations embracing the accurate collection, use, and secure data sharing will ensure much better patient outcomes. They will also increase productivity and save money and resources by embracing technology to gain insights from all the information at their disposal. And most importantly, they can decrease healthcare inequities that have existed for far too long.

To learn more about how BCT Partners uses data analytics to address inequities, click here.

To read more BCT blogs, click here.


bottom of page