Don't Let Perfect Be the Enemy of Good: A Proof of Concept for a Custom National Data Repository of Quality Measures for Free and Charitable Clinics

Purpose: Free and charitable clinics (FCCs), nonprofits that utilize volunteer licensed health care professionals to provide health services at no cost or a small fee to low-income uninsured patients who are disproportionately from underrepresented communities, have been part of the safety net for over a century. Approximately 1400 known FCCs serve two million patients annually. Despite their longevity and sizable number, evidence regarding the quality of care in FCCs is lacking. We report new evidence generated by a national initiative, the Roadmap to Health Equity. Established in 2017, this consortium is co-led by two national organizations serving FCCs and an academic institution. It has involved more than 150 FCC stakeholders with the shared goal of improving the quality of care and reducing inequities. The centerpiece is a custom national data repository of 15 validated clinical quality measures and patient-level characteristics. Methods: Fifty FCCs pilot tested the data repository. Clinics submitted patient-level data on two blood pressure (BP) measures and at least one additional measure. Descriptive statistics were stratified by sex, race, ethnicity, and language. Results: In 2021, 33 pilot FCCs from 21 states reported data across 13 of the 15 clinical measures, representing 34,359 unique patients. For example, on average, 60% of patients had controlled BP, but Black patients had lower rates of BP control than Hispanic and White patients (55.9% vs. 62.1% and 63.0%, respectively). Conclusion: Our findings demonstrate a proof of concept. By standardizing quality measures alongside patient characteristics, clinics can become aware of racial/ethnic inequalities in health outcomes. This information can motivate clinics to investigate the causes and implement solutions. In an environment where outcome data from FCCs are scarce, the new national data repository lays the foundation for routine stratified reporting of a range of quality outcomes for an important safety net for the uninsured.


Introduction
Free and charitable clinics (''FCCs''), volunteer-based nonprofits that provide a range of health care services at no cost or for a small fee to low-income uninsured patients who are disproportionately members of underrepresented minority groups, have been part of our nation's safety net for over a century. 1 The *1400 known FCCs serve more than two million patients each year. 2 Despite their longevity and sizeable number of patients, evidence about the quality of care in FCCs is deficient.
Data on the quality of care in FCCs beyond singleclinic narratives are scarce. Statewide-level data exist only in North Carolina where, since 2011, the North Carolina Association of Free and Charitable Clinics (NCAFCC) report on select patient outcomes. 18 As NCAFCC data represent the largest effort to aggregate standardized quality outcomes, their results best approximate the level of quality of care in FCCs. In their 2020 report based on 65 FCCs, the NCAFCC found that 30.9% of diabetic patients had HbA1c levels above 9%. This is comparable to 34.3% of patients at Health Resources and Services Administration (HRSA) federally funded health centers in North Carolina. 19 For hypertensive patients, 55.3% had controlled blood pressure (BP) ( < 140/90) 18 versus 58.9% at federally qualified health centers (FQHCs). 19 Limited data on the quality of care among the vast majority of FCCs across the country creates numerous obstacles. For FCCs, the lack of quality data inhibits clinics' abilities to (1) benchmark against their peers, thereby improving care; (2) secure resources from funders that increasingly expect evidence showing outcomes; (3) form partnerships with hospitals and health systems to extend the scope of services; and (4) attract volunteer licensed providers. 20 For other stakeholders, this lack of data means policymakers, other safety net providers, private philanthropy, the media, and the general population have an incomplete picture of the quality of care. The absence of quality outcome data helps perpetuate myths and misconceptions about FCCs because ''free'' care is often presumed to be of low quality. 21 An explanation for the lack of data on the quality of care in FCCs is the lack of a national infrastructure to systematically collect standardized evidence-based data that can indicate the level of quality of care that FCCs provide to their patients (akin to the Uniform Data System [UDS] used by HRSA for health centers; see data. HRSA.gov). To address this shortcoming, a consortium led by two national organizations serving FCCs (Americares, National Association of FCCs) and an academic institution (Loyola University Chicago) joined FCC leaders and state associations representing FCCs in 2017 to create a national initiative called Quality of Care in FCCs: Roadmap to Health Equity: (hereinafter ''Roadmap'').
To date, more than 150 FCC stakeholders have worked together to improve quality of care and reduce inequities in FCCs. The centerpiece is a custom national data repository of 15 externally validated clinical quality measures, which are a mechanism to assess processes, experiences, and/or outcomes of patient care, such as BP control, combined with patient-level characteristics, including race, ethnicity, and language. In this study, we report our progress, share early results, and discuss the next steps.
Roadmap was launched in preparation for a National Institute on Minority Health and Health Disparitiessponsored conference. At the 2018 conference, key FCC stakeholders connected to address the lack of national-level quality data for FCCs. There, Roadmap stakeholders reached consensus on a ''starter set'' containing 15 clinical quality measures, sociodemographic factors, and clinic-level characteristics (Table 1) and agreed to a process for identifying a vendor that would house the quality data. Following the conference, Roadmap participants continued to work through committees aptly named the ''How'' Committee, which explored how the data would be collected; the ''What'' Committee, which examined what data ought to be reported into the repository; and the ''Why'' Committee, which articulated the value proposition.
A steering committee, comprising nine leaders, the majority of whom are from underrepresented communities, guides the initiative. This original structure ( Fig. 1) is intact 5 years later, although the activities subsumed under the committees have changed as the initiative has evolved. Figure 2 highlights the key accomplishments.

Sample
In fall 2019, the Roadmap coleads solicited FCCs through a request for application process to pilot test the data repository created by VisionTree, a company with a cloud-based patient-centered outcomes platform. Of 57 applications received, the coleads selected 50 pilot FCCs. The clinics received $1000 to participate in the pilot phase, which continues through June 2023.

Materials
In 2021, we asked clinics to submit their 2020 deidentified patient-level data using measure-specific data.csv files on two BP measures and at least one additional clinical quality measure into the new repository ( Table 1). The BP measures assess the process and the outcome of patient care. Specifically, the BP screening measure indicates the percentage of adult patients who were screened for high BP and have a recommended follow-up plan documented if BP is elevated. In addition, the BP outcome measure indicates the percentage of hypertensive patients who have controlled BP (i.e., systolic BP is < 140 mmHg and diastolic BP is < 90 mmHg). A user guide, data dictionary, and training videos provide instructions for submitting the data. Clinics can report the population for a given measure or a random sample of 70 patients using approved methods that mimic the health center UDS methodology. 22 Statistical analysis Using SAS 9.4, we performed descriptive statistics for the clinic characteristics and reported clinical quality outcomes, stratified by sex, race, ethnicity, and language, where possible. We created a single race/ethnicity  measure using separate questions on race and ethnicity. We coded patients as having a ''potential language barrier'' if responses indicated patients needed an interpreter, spoke a language other than English at home, or spoke English less than ''Very Well.'' The Institutional Review Board of Loyola University Chicago approved the study.

Results
In 2021, 33 pilot FCCs (66%) from 21 states submitted data across 13 of 15 clinical measures, representing 34,359 unique patients. Table 2 describes the organizational characteristics of the participating FCCs. Pilot clinics submitting data are located in all four regions of the United States, but more than half are from the South. In a logistic regression model, we detected no statistically significant difference by region between pilot clinics that submitted clinical quality data (n = 33) and pilot clinics that did not submit data (n = 17). On average, participating clinics have operated for 20 years as of 2020, but the number of years open ranged from 1 year (2019) to 90 years (1930). Half have budgets that are $1 million or greater, but about one-quarter report annual cash operating budgets of < $300,000. More than half of participating clinics are open at least 5 days per week.
More specifically, clinics report being open to serve patients, on an average of 117 h per month. Participating clinics include both volunteer-run enterprises and heavily staffed operations, with a mean of 15 fulltime employees and more than 8300 annual volunteer hours. On average, the annual number of total unduplicated patients served is 3322, of which 540 (roughly 15%) are new patients. These patients receive, on average, 5802 medical visits, 980 dental visits, and 948 mental health/behavioral health visits. Table 1 shows the mean percentage of patients who met each measure, stratified by race/ethnicity and language, if available. For example, among patients with a diagnosis of hypertension, for the BP control performance measure, on average, across 22 reporting clinics, 60.0% of patients had controlled BP, but Black patients had lower rates of BP control than their Hispanic and White counterparts (55.9% vs. 62.1% and 63.0%, respectively). Turning to the diabetes control performance measure, an optional measure, among nine reporting clinics representing 1367 patients with diabetes, 34.4% of patients, on average, had poor control of their diabetes (defined as > 9.0% HgA1c). Latinx patients had slightly higher rates of poor diabetes control than White patients (23.5% vs. 21.1%).
Among patients of unknown race, however, 36.9% had poor control of diabetes. When we stratify results by potential language barrier, diabetic patients with a potential language barrier have much higher rates of poor control of diabetes compared with patients without a known language barrier (34.7% vs. 25.5%).

Discussion
Our findings demonstrate a proof of concept and suggest that the national quality data repository is highly promising in three respects. First, with standardized, evidence-based quality outcomes alongside patient characteristics, clinics can become aware of inequalities in health outcomes based on race, ethnicity, and language. While stratified performance data, by themselves, do not reduce inequities, they can motivate clinics to investigate causes and implement solutions. Second, clinics will be able to compare their individual clinic's performance to other FCCs as well as to related peers, such as health centers. Finally, the new data repository can help clinics meet funders' expectations concerning outcomes, supporting sustainability. Anecdotal evidence from pilot clinics that are involved in the Roadmap initiative suggests that the COVID-19 pandemic may have hindered clinics' participation (n = 33 of 50), underscoring the importance of making the data submission process feasible for low-resource settings. Efforts are ongoing to enhance the training materials and simplify reporting. In our preparations for the second wave of data collection, which is currently underway, we have highlighted these improvements in our messaging to pilot clinics, and remain hopeful that we can reengage the pilot clinics that did not submit 2020 data.
Better training materials can also help to prevent data irregularities. For example, we observed that some clinics had misinterpreted the BP screening measure, mistakenly extracting hypertensive patients. In addition, some clinics appeared to report the most recent visit rather than all visits for the documentation of current medications measure. Spotting these issues during the pilot phase offers an opportunity to resolve them before scaling up, ensuring more accurate reporting.
Committed to consensus decision-making, Roadmap has been responsive to FCC leaders' suggestions to improve data accuracy and completeness. For example, clinic leaders have explained that many patients have difficulty answering ethnicity and race with two questions, the federal standard and format we adopted. 23 In particular, Latinx patients often reportedly do not identify as another race. Consequently, we have observed high levels of ''Unknown'' race in our 2020 sample, which is a current limitation that constricts our ability to detect racial/ethnic inequalities in the quality of care from the inaugural dataset. As a remedy, starting in the 2021 reporting year, we now give clinics two options to report race and ethnicity: using two questions, or a single question that treats Hispanic origin as a race. We anticipate that this change will result in more complete information about race and ethnicity.
Along with data collection, Roadmap workgroups are bringing resources to clinics to enhance equity and quality improvement efforts. For instance, through a partnership with the Institute for Healthcare Improvement (IHI), four pilot clinics received $30K scholarships, enabling them to receive coaching and peer support through the IHI's Pursuing Equity initiative.
A formal mixed-methods evaluation of pilot clinics is underway to more systematically identify the barriers and facilitators to participating in Roadmap. Feedback from focus groups and surveys will inform our efforts to update our starter set, modify the process for data collection, identify needed resources, refine our messaging about the value of Roadmap, learn about how clinics might use the results to advance their equity and quality improvement activities, determine the optimal pace for scaling up, and decide a cost-sharing strategy to assure long-term sustainability of the national data repository. With initial funding from NIMHD and subsequent external funding from corporate donors as well as from Americares, one of the project coleads, we have the means to add up to 100 pilot clinics over the next year.
Our 4-year pilot phase, begun in late 2019 and continuing through June 2023, enables us to collect quality measures from FCCs across three time periods (2020, 2021, 2022), giving us ample time to identify and address concerns and secure resources for continued growth. This pilot study has several limitations. Similar to the UDS, the data were self-reported. In addition, there are insufficient data to draw definitive conclusions about the quality of care in FCCs, particularly within racial/ethnic groups where the sample sizes can be quite small. Nevertheless, these are the first known nationallevel quality data and extend the evidence base beyond individual clinics and North Carolina.

Conclusion
In an environment where outcome data from FCCs are scarce and, if available, often reported in the aggregate, the new national data repository lays the foundation for routine stratified reporting of a range of quality outcomes for an important safety net for the uninsured. This foundation is imperfect, but improved thanks to pilot clinics, and could ultimately help drive increased evidence-based decision-making and better outcomes in FCCs.