Study methods were approved by the institutional review board at Edward Via College of Osteopathic Medicine. A retrospective chart review was used to gather data. This method is well-established in free clinic literature as being a cost- and time-efficient means of reporting outcomes and comparing them to existing professional standards [8-13]. The use of hemoglobin A_1c as a diabetes quality-of-care metric is endorsed by the US government and the American Diabetes Association (ADA) [23-25]. Hemoglobin A_1c is an indicator of 3-month average blood glucose as well as a predictor of diabetes-related morbidity and microvascular complications [26-28]. The ADA endorses a hemoglobin A_1c threshold of 7.0 as being appropriate for most adults [23].

All patients’ medical charts at the SLFMC between January 1, 2018, and January 1, 2021, were screened for a diagnosis of T2DM. Inclusion criteria were as follows: (1) a diagnosis of diabetes, prediabetes, hyperglycemia, or metabolic syndrome; and (2) two or more different hemoglobin A_1c levels documented, with at least 6 months between the first and last recorded hemoglobin A_1c levels. Patients with a documented pregnancy within the study timeframe were excluded owing to the potential confounding effect of gestational diabetes. Patients with diabetes were stratified into two groups: controlled (baseline hemoglobin A_1c<7.0) and uncontrolled (baseline hemoglobin A_1c≥7.0) [12,13]. Stratification was essential because the two groups have two different therapeutic goals: the goal when treating controlled diabetes is to maintain hemoglobin A_1c levels under 7.0, while the goal when treating uncontrolled diabetes is to reduce hemoglobin A_1c levels to <7.0 [23].

All available descriptive data and hemoglobin A_1c readings—including dates—for individual patients were transcribed from a hard copy medical record into Microsoft Excel. Descriptive data included age, gender, race/ethnicity, and BMI. Smoking status and the presence of concomitant diagnosis with essential hypertension (ICD-10-CM I10) were also recorded. Characteristics of the SLFMC sample were compared to data from Spartanburg County to explore what kinds of patients the SLFMC attracts [29,30]. Baseline hemoglobin A_1c values were defined as each patient’s first recorded hemoglobin A_1c value within the study window. To facilitate comparison, subsequent hemoglobin A_1c values of each patient were rounded to either 6 months, 1 year, or 2 years from baseline. The baseline hemoglobin A_1c value was also used to sort individuals into 1 of 2 groups on the basis of whether they had controlled (hemoglobin A_1c<7.0) diabetes or uncontrolled (hemoglobin A_1c≥7.0) diabetes relative to a best-practice hemoglobin A_1c treatment goal of 7.0, as recommended by the ADA [23]. Finally, each patient’s most recent hemoglobin A_1c values within the study timeframe were recorded. The most recent hemoglobin A_1c value on file was necessary for comparing treatment efficacy because not all patients received treatment for the same duration—some of the participants enrolled at the SLFMC in the middle of the study’s timeframe, and some patients also left the clinic outright or aged out into Medicare.

The primary outcome measured per individual was the net change in hemoglobin A_1c (△A_1c) from baseline to the most recent visit on record. At the group level, successful treatment was defined as maintenance of hemoglobin A_1c in patients with controlled diabetes and reduction in hemoglobin A_1c in those with uncontrolled diabetes. The primary outcome measured per group was median hemoglobin A_1c. Finally, the success of the SLFMC diabetes treatment program was assessed by stratifying the clinic’s A_1c outcomes into three tiers of <7.0, 7.0-8.0, and >8.0. Proportions of patients within each range were compared to the hemoglobin A_1c data from various populations in the recent literature.

A Wilcoxon matched-pairs signed rank test was performed for the median group’s hemoglobin A_1c outcomes (controlled vs uncontrolled diabetes). The hemoglobin A_1c distribution was expected to show a skew, so the median value was chosen over the mean value to dampen the influence of outliers. A Bonferroni-Dunn post-test was applied to the resultant P values to standardize them to a single significance level (a) per group. Between-group medians, P values of <.05 were considered significant. Graphs were generated using GraphPad Prism (version 9; GraphPad Software, Inc).

In analyzing the follow-up behaviors of the SLFMC’s patients, we noted no significant difference in follow-up rates between patients with controlled and those with uncontrolled diabetes. However, there were some differences between the follow-up rates of different ethnic groups at the SLFMC. Prior research shows that Black and Hispanic patients tend to have higher baseline hemoglobin A_1c levels [34-37] and poorer self-management of diabetes, are less likely to achieve good glycemic control with treatment [38], and are more likely to experience diabetes-related complications resulting in worse diabetes outcomes and higher rates of mortality than non-Hispanic White patients [39]. Hispanic patients may specifically face obstacles associated with cultural differences between patients and providers, placing them at an increased risk for underuse of services, poor-quality care, and worse outcomes compared with non-Hispanic Whites [40]. The literature suggests that language barriers have a negative influence on health behaviors and outcomes among Latino patients [41]. To our knowledge, no unifying theory explains these known discrepancies in hemoglobin A_1c values among ethnic groups.

Black and Hispanic patients at the SLFMC defied the data cited above, with higher use of resources and better health outcomes than non-Hispanic White patients at the clinic (Table 3). Hispanic patients had a mean ΔA_1c of –0.6 while enrolled at the SLFMC. Black patients had a remarkable 61% glycemic control rate by the end of the study, which was the highest of any other race. SLFMC provides free care, including many free prescription medications, and we believe that ease of access to care is a primary driver of the trend-reversing outcomes achieved by the SLFMC’s Black and Hispanic patients with diabetes. Generally, Black and Hispanic populations are subject to disproportionately low health care access [42,43]. Furthermore, patient trust in providers is a direct predictor of subjective treatment effect [44,45]. The authors offer the following original thoughts on care for minorities within the free clinic setting: we suggest that patients are more likely to trust practitioners who provide their services free of charge. It is likely that patients feel more comfortable and less intimidated in the setting of charitable care compared to payment-based models. We also suggest that the ratio of immigrants in the Hispanic population is higher than that in other ethnicities [46]. We think that immigrants are more likely to use free health services out of appreciation for some of the things that American culture takes for granted. The aforementioned factors probably influenced the stronger improvements in hemoglobin A_1c values observed in Black and Hispanic patients than in Non-Hispanic White patients (Table 3).

Free clinic populations universally show hemoglobin A_1c averages at least a full point above nationwide averages for adults with diabetes (Table 4). The NHANES, which was chosen as the national comparison data set, generates annual samples with the intent of representing the American population as a whole. Most of the patients included in the NHANES sample have health insurance coverage; hence, NHANES suitably represents the glycemic control of the insured population. The SLFMC data are encouraging and suggest that free clinics may be able to approach the quality of care provided by traditional models of outpatient practices.

The glycemic control data gained in this study add to the growing body of knowledge characterizing the health of patients who utilize free clinic services. At baseline, The SLFMC population demonstrated the highest rate of controlled diabetes (38%) and the lowest mean hemoglobin A_1c value (8.3) observed in free clinic outcome literature, as well as the smallest mean improvement in hemoglobin A_1c values (–0.3). These findings are likely attributable to the relatively good health of the clinic’s population at the start the study. The COVID-19 pandemic occurred within the study timeline, and it probably influenced the modest mean improvement in hemoglobin A_1c value of –0.3. Altogether, this study presents the current best-case scenario of free clinic glycemic control at baseline.

Quantifying free clinic quality of care is a matter of importance to individual free clinics for the purpose of fundraising, but this aim is complicated by many factors. One such factor is time. The SLFMC has used paper charts for decades and has not completely transitioned to electronic medical records (EMRs). Many other free clinics are likely in the same situation because EMRs are costly to adopt. Sifting through paper charts takes more time than searching through a well-kept EMR. Another factor is available labor. Volunteer providers may not always be able to spare the time it takes to write up and publish outcomes. Despite obstacles of time and cost, the NAFC recently approved quality reporting standards, which it expects each member clinic to abide by. This presents a valuable professional growth opportunity for medical students and other postgraduate medical professionals. We recommend that free clinics take further initiative and publish outcomes when possible.

The retrospective cross-sectional design is limited in scope and cannot explain relationships between variables. Study data were extracted largely from paper charts and were subject to potential inaccuracies. The clinic is currently transitioning from a paper chart system to an EMR, and some charts were unavailable for evaluation. The SLFMC yielded a population of 140 patients with diabetes with a unique demographic profile that may not generalize well to other clinics. Furthermore, this study did not consider the specific diabetes treatment modalities provided to each patient (pharmaceuticals, nutrition counseling, home glucose monitoring, etc) which were excluded from the study because treatment varies widely across patients and providers. One final limitation is that not all patients had the same number of follow-up hemoglobin A_1c tests—such variations were possibly due to, for example, duration of enrollment (including loss to follow-up) and missed appointments.

This report highlights a free clinic diabetes treatment program that provides significant benefits to patients with uncontrolled T2DM. The SLFMC was successful in treating minority patients, with uncontrolled Black and Hispanic patients showing a mean 1% improvement in hemoglobin A_1c values. Rates of glycemic control at the SLFMC were superior to those observed at other free clinics and within 10% below two recently reported nationwide averages. This study presents one of the largest sample sizes yet observed in the free clinic hemoglobin A_1c literature. Limitations include loss to follow-up and incomplete patient records. Meta-analysis of comparable outcomes data is the intuitive next step in the journey toward a better understanding of the quality of care provided at free clinics.