The ‘real’ on REaL data collection: A case study

In its simplest, apolitical and non-ideological state, race is the grouping of people based on shared physical traits, while ethnicity classifies people based on shared ancestry, language and cultural traditions. Both are social constructs with no biological basis. However, failing to self-report these demographics, or allowing others to report one’s race/ethnicity based on bias, has far-reaching implications for the quality of care, including mortality rates. Thus, it is vital for stakeholders in healthcare organizations to educate and accurately collect race and ethnicity data. This article highlights the journey Montefiore Medical Group took to identify these challenges, develop interventions and meet operational success.

Data collection on a national scale

The U.S. Census Bureau, a division within the U.S. Department of Commerce, is tasked with attaining, analyzing and reporting quality data about Americans. Mandated by Article I, Section 2 of the Constitution, every 10 years, the U.S. Census Bureau is responsible for collecting an array of demographic information for all residents of the nation. This data is aggregated and utilized diversely to address national mandates and social obligations, such as apportioning seats in the U.S. House of Representatives, assigning school districts, providing services for the indigent, addressing deteriorating roads and assessing racial disparities in health.

Category labels

The Census Bureau collects self-reported race data as outlined by the Office of Management and Budget (OMB). The available categories represent an accepted definition of race in the country. These categories include: White, Black or African American, American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander. The Census Bureau is authorized to add an “Other” category; this option enables respondents to identify a choice if the other categories do not align with their racial identity. With respect to selecting an option for ethnicity, the categories are far more narrow, essentially directing respondents to identify as Hispanic or non-Hispanic. With Diversity Index calculations for redistricting purposes, the Census found that the chance of two people chosen at random will be from different racial or ethnic groups increased from 2010 (54.9%) to 2020 (61.1%).¹ 

Many Americans can no longer be classified or quantified within the proverbial checkbox — and they refuse to do so. According to the Census Bureau, “our research has found that over time, there have been a growing number of people who do not identify with any of the official OMB race categories.”² This sentiment typically results in patients not providing the requested information, as opposed to selecting “Other,” which may be deemed as offensive and exclusive. As Khunti, et al, note: “An important first step in improving the use of ethnicity categorization is to ensure labels are sufficiently granular as to capture important heterogeneity, and that they are employed systematically to decrease confusion … and permit easier data collection and pooling.”³

Healthcare disparities and mitigation

Society and the global economy are gradually returning to normal after facing one of the worst modern public health crises in the form of the COVID-19 pandemic. According to the World Health Organization, more than 770 million people reported confirmed COVID cases, with a death toll approaching 7 million.⁴ Initially believed to be immune to the disease, the African American community was ravaged by the virus.⁵ CDC data showed that “African Americans have been disproportionately affected by the virus at much higher levels than all other races in the United States.”⁶ This grim reality highlights the biological and system health inequalities affecting minority groups. Literature review points to ethnic minorities being “at a higher risk for infectious diseases like tuberculosis, HIV and hepatitis B and C, or higher premature death rates from heart disease and stroke.”⁷ Accurately recording a patient’s race and ethnicity helps reveal healthcare inequities in care and track health trends and results. Collecting and sharing this data helps advocate for change, raise awareness, and research conditions that disproportionately impact minority groups.

Massachusetts General Hospital developed a patient navigator program “after finding a significant gap in colorectal cancer rates between Latino and white populations.”⁸ Methodist Le Bonheur Healthcare in Memphis implemented a program to help African Americans transition from hospital to home after learning this group experienced high readmission rates. This program positively contributed to a reduction in readmissions and a decrease in average healthcare costs. These scenarios offer empirical evidence for the benefits of collecting race and ethnicity for patients.

Regulations

Although various entities and regulatory agencies have enacted legislation mandating collection of race and ethnicity data, application of the process has been inadequate. Per Khunti, et al.: “The completeness of ethnicity data within healthcare and routine databases has been poor historically … only 27% of patients in Clinical Practice Research Datalink had ethnicity recorded … accuracy is often low, with 20% to 35% error in coding of major ethnic minority groups in NHS hospital records when compared to self-reported ethnicity.”⁹ Due to such collection rates, regulations have been instituted to enhance data collection.

The Patient Protection and Affordable Care Act (PPACA) of 2010 enacted Section 4302 the following year to strengthen federal data collection by requiring all health surveys sponsored by the Department of Health & Human Services (HHS) to collect race and ethnicity.

Meaningful Use Certification Criteria also required “the recoding of demographic information in such a way that enables the user to electronically record, modify, and retrieve patient demographic data including preferred language, sex, race, and ethnicity.”¹⁰

The Joint Commission’s standards for accreditation or certification of healthcare organizations also requires hospitals to collect race and ethnicity; however, as of Jan. 1, 2023, “new and revised requirements to reduce health care disparities will apply to organizations in the Joint Commission’s ambulatory health care.”¹¹ This will require all outpatient physician practices to develop workflows and technology to collect and report this information.

Other regulations and policies requiring collection of this data include, but are not limited to, HIPAA, Initiative to Eliminate Racial and Ethnic Disparities in Health, the OMB and the HHS standards for Culturally and Linguistically Appropriate Services (CLAS).

Case study

Montefiore Medical Group (MMG) is an ambulatory network of Federally Qualified Health Centers (FQHCs) and Article 28 and non-Article 28 physician practices within the Montefiore Medical Center. Totaling 20 locations throughout the Bronx and Westchester County (NY), MMG provides primary care, specialty, behavioral health and urgent care services to more than 270,000 patients annually. Accredited by the National Committee on Quality Assurance (NCQA), MMG is a Level III Patient Centered Medical Home (PCMH). This designation represents MMG’s model of delivering high-quality, cost-effective, patient-centered care, leveraging a team-based approach to coordinate care across the health system. Under the direction of OMB, the NCQA requires all PCMH-accredited organizations to collect and report race and ethnicity data of patients served.

As a measure of focus, MMG partnered with the Community Health Care Association of New York State (CHCANYS) to enhance the current workflow in collecting race and ethnicity for our patient base.

Assembling the team

Thoughtfully assembling a project team is key to successful planning and implementation for all quality improvement (QI) initiatives. MMG’s University Avenue Family Practice (UAFP), an FQHC, was selected as the pilot site based on their current race and ethnicity collection performance and bandwidth for participation in a formal Plan-Do-Study-Act (PDSA) process. The health center’s Operations Manager, Medical Director and a Patient Services Representative (PSR) were selected to join the project team. Frontline representation via the PSR was foundational to the pilot study. Additional team members assembled included the Senior Director of Operations, Director of Strategic Operations, Director of Revenue Cycle, Director of Informatics, and QI Coaches. The interdisciplinary team represented diversity of discipline, expertise and individual races and ethnicities.

Grasping the situation and planning the PDSA

Before commencing the formal PDSA cycle, the team focused on grasping the current situation. According to baseline data from January 2022 through May 2023, UAFP averaged 73% successful collection of race and ethnicity for the health center’s unique patient volume.

From the perspective of visit-based weekly data (i.e., the percent of patients with race and ethnicity categories collected who had a visit during the reporting period), 73% of patients had race and ethnicity categories complete at the start of the PDSA (see Figure 1). Visit-based data was selected to guide this project for various reasons:

1. To provide the team with a better locus of control, by including patients with completed visit encounters during the period of intervention implementation
2. To decrease the denominator to allow for greater perceptible data change weekly. 

The team created a fishbone diagram to identify possible barriers to collecting this information (see Figure 2). Through team brainstorming sessions and discussion with colleagues, the decision was made to focus initial interventions on addressing:

1. Patient hesitancy to disclose sensitive information
2. The limited understanding of the importance of collecting race and ethnicity data (for patients and associates).

For the first PDSA cycle, the team selected the following initial interventions to test:

1. Education for staff, including scripting guidance, to highlight the importance of race, ethnicity and language (REaL) data and to equip front desk staff with the tools to encourage hesitant patients
2. A visual (English and Spanish versions) for patients to circle and/or point to their respective race/ethnicities to eliminate the need to disclose sensitive information aloud (see Figure 3)
3. A patient education visual (English and Spanish versions) visible in the waiting room area summarizing the importance of obtaining race and ethnicity information (see Figure 4)
4. Daily huddle reminders to front desk staff to prioritize race/ethnicity data collection
5. An automated warning within the EHR alerting front desk staff when values were incomplete.

In addition to organizing how to put these interventions into action, the project team also identified process measures to track their use and effectiveness. To gather this information, front desk staff were asked specific questions about how they used the interventions during a set time. For example, “Last week, for what percent of patients did you utilize scripting when obtaining race and ethnicity information?” By measuring how much each intervention was used, the project team could analyze their individual impact on performance.

PDSA Cycle #1: Implementation and findings

Implementation of interventions began on June 13, 2023, and was recorded within an annotated run chart (see Figure 1). Throughout the initial round of intervention testing, process measures were collected from the front desk staff and incorporated into the analysis, or “Study,” portion of the PDSA.

Despite the team’s expectations, PSRs used scripting and visuals less than anticipated, with only 13.75% using them during the first PDSA cycle. However, the percentage of patients with complete race and ethnicity information continued to increase on a weekly basis, barring the 1-percentage point decrease mid-July.

Additionally, the team collected a balancing measure to gauge if the focus on this initiative was inadvertently disrupting other processes. When asked if other daily tasks were delayed or incomplete due to the time it took to collect race and ethnicity data, 100% of PSRs responded “no.”

The team used the above findings to address the “Study” and “Act” portions of the PDSA. While data collection was undoubtedly improving, initial interventions weren’t used much. This led the team to conclude that educating the front desk staff on the importance of collecting race and ethnicity really made a significant impact. Key to the initial success of this initiative was explicitly elevating the importance of race and ethnicity data and assigning the same urgency for collection for insurance verification or copayment collection (other key components of patient registration during point of service). Further, the balancing measure showed that PSRs found this task manageable, even with their other daily responsibilities.

The interdisciplinary team opted to continue the use of scripting and the visual without change and focus on collecting additional data to inform how to best modify interventions to meet the needs of patients. The Operations Manager also performed routine chart audits for “missed opportunities” and reviewed results with the front desk staff to further understand the underlying reasons for incomplete information.

Additional PDSA cycles and sustaining results

During subsequent weeks, the site team continued interventions and collecting process and balancing measures. An automated survey was also distributed to patients via text message providing them with the opportunity to self-select race and ethnicity. Analyzing survey results and talking with the project team highlighted ways to modify existing interventions to increase their effectiveness. For example, UAFP has a large Hispanic/Latino/Spanish population that increasingly expressed feeling unrepresented in race categories. This finding aligns with other case studies noting patients of Hispanic origin may not fully understand race and ethnicity definitions and therefore choose not to report.¹² In response, the order of race and ethnicity were switched for the patient visual to “soften” introduction to the concepts. Discussion is ongoing regarding methods of incorporating education into existing scripting.

By Aug. 21, 2023, two months after initiating the interventions, the team decided to continue interventions with the previously mentioned changes and monitor the data for several weeks to determine if improvements were sustained. During this period, the team observed performance measures of more than 90% for four consecutive weeks. As a result, a control plan was designed to hardwire changes and identify a reaction plan should performance diminish. This plan included:

1. The identification of process owners and outlining of individual responsibilities in response to performance changes
2. Highlighting the importance of this workflow in the New Associate Orientation packet (documents provided to all new hires)
3. Identifying a system for ongoing monitoring via chart audits
4. A specific reaction plan if performance decreases below the identified threshold of 90%.

Challenges

Sustaining performance for the long term remains a challenge for the team and will require further attention. As performance drivers naturally shift due to the success of particular interventions, ongoing assessment of these drivers and identification of strategic interventions will be necessary. Additionally, it will be crucial for the team to continue to reassess balancing measures to ensure the front desk staff continues to satisfy this requirement.

Likely, the biggest challenge is using REaL data accurately to tackle and reduce health disparities among MMG’s patients. To achieve this goal, additional focus is needed to:

1. Ensure the collection of accurate and specific race and ethnicity data, including granular categories
2. Continue to design targeted interventions to ensure patients feel comfortable disclosing this information and are informed about how to self-identify
3. Leverage race and ethnicity data to improve clinical outcomes for patients.

Successes

The rapid initial success of this QI initiative set the tone for the project, energizing and empowering the project team and frontline associates. Identifying early on that education for front desk staff was a major driver of performance improvement was a powerful starting point. This finding motivated the team to start the project at the pilot site and illuminated a key intervention to test when expanding to other practices. Additionally, increasing staff comfort with REaL data may encourage patient comfort in providing this information.¹³ Using multiple PDSA cycles and a concerted effort to analyze quantitative and qualitative data led to the ongoing improvement of selected interventions to better meet the needs of patients. Scripting, visuals, handouts, etc. were edited and improved in real time to respond to increasing understanding of the hesitancy patients have in divulging this information.

Finally, engaging in structured QI methodology with a multidisciplinary team provided a ripe environment for learning. The QI concepts and tools used during this pilot study can and should be translated to other QI projects and initiatives throughout the medical group. Engaging both leadership and the frontline in this important work encourages team members to view all processes through a QI lens, paving the way for improved care for MMG’s patients.

Expanding to other MMG practices

Team leads initiated expansion of the pilot following UAFPs:

1. Performance improvement with a peak performance of 96%
2. Identification of effective and feasible interventions
3. Performance stabilization
4. Creation of a control plan, team leads initiated the expansion of the pilot to three additional practices.

Similarly, these sites were selected based on their current performance and bandwidth for participation in a structured QI project. While there were many lessons to share from the initial pilot, the approach to expansion was not to scale identical interventions to the three additional practices. Given performance drivers differ across the medical group, the assumption can’t be made that the same interventions will work equally well elsewhere. Past performance is not an indicator of future results.

Teams from the newly participating sites, pilot site representatives and project leads gathered in a collaborative kick-off meeting to discuss individual PDSAs for each practice, requiring the data-driven identification of performance drivers before identifying interventions to test. This expansion is in progress; it is anticipated the shared learnings and resources from the pilot site will contribute to efficient and sustainable race and ethnicity data collection among the new pilot practices.

Conclusion

Complete and accurate data collection is a key component in identifying healthcare disparities within a patient population. Healthcare employees, specifically registration/front desk personnel, must be equipped with the appropriate knowledge, training and resources to solicit this information from the patients within their organization. Selecting “Unknown” as an option for race and/or ethnicity exacerbates the care gap and perpetuates adverse health outcomes for marginalized groups. Collection of this demographic information should be considered a key intervention in addressing these well-documented inequities.

Notes:

1. Jensen E, Jones N, Rabe M, Pratt B, Medina L, Orozco K, Spell L. “The Chance That Two People Chosen at Random Are of Different Race or Ethnicity Groups Has Increased Since 2010.” U.S. Census Bureau. Aug. 12, 2021. Available from: https://bit.ly/48tMrQb.
2. U.S. Census Bureau. “Research to Improve Data on Race and Ethnicity.” Available from: https://bit.ly/3NCbWqf.
3. Khunti K, Routen A, Banerjee A, Pareek M. “The need for improved collection and coding of ethnicity in health research.” J Public Health (Oxf). 2021 Jun 7;43(2):e270-e272. doi: 10.1093/pubmed/fdaa198. PMID: 33283239.
4. World Health Organization. “WHO Coronavirus (COVID-19) Dashboard. Available from: https://covid19.who.int.
5. Mock B. “Why You Should Stop Joking That Black People Are Immune to Coronavirus.” Bloomberg. March 14, 2020. Available from: https://bit.ly/3tAQLxY.
6. Reed DD. “Racial Disparities in Healthcare: How COVID-19 Ravaged One of the Wealthiest African American Counties in the United States.” Soc Work Public Health. 2021 Feb 17;36(2):118-127. doi: 10.1080/19371918.2020.1868371. Epub 2020 Dec 28. PMID: 33371822.
7. Drewniak D, Krones T, Wild V. “Do attitudes and behavior of health care professionals exacerbate health care disparities among immigrant and ethnic minority groups? An integrative literature review.” Int J Nurs Stud. 2017 May;70:89-98. doi: 10.1016/j.ijnurstu.2017.02.015. Epub 2017 Feb 14. PMID: 28236689.
8. AHA. “Reducing Health Care Disparities: Collection and Use of Race, Ethnicity and Language Data.” Aug. 20, 2013. Available from: https://bit.ly/4arPcDf.
9. Khunti, et al.
10. IFDHE. “Why Collect Race, Ethnicity and Primary Language.” Available from: https://bit.ly/3TzOzBI.
11. The Joint Commission. “R3 Report Issue 36: New Requirements to Reduce Health Care Disparities.” June 20, 2022. Available from: https://bit.ly/41uaagM.
12. Lee WC, Veeranki SP, Serag H, Eschbach K, Smith KD. “Improving the Collection of Race, Ethnicity, and Language Data to Reduce Healthcare Disparities: A Case Study from an Academic Medical Center.” Perspect Health Inf Manag. 2016 Oct 1;13(Fall):1g. PMID: 27843424; PMCID: PMC5075235.
13. Wynia M, Hasnain-Wynia R, Hotze TD, Ivey SL. “Collecting and using race, ethnicity and language data in ambulatory settings: A white paper with recommendations from the Commission to End Health Care Disparities.” AMA. 2011. Available from: https://bit.ly/3GQeb5x.