Expanding Diabetes Specialty Care in a Community Primary Care Practice Center

Parkland Health is the safety-net health system of Dallas County, TX, serving the indigent and underserved individuals of the county. The Diabetes Specialty Clinic is located on the main campus of Parkland Health adjacent to the main hospital. Primary care is provided to residents at community-oriented primary care clinics (COPCs), which are located in underserved areas throughout Dallas County.

Traditionally, to receive diabetes specialty care, residents would have had to travel to the main campus of Parkland Health. This project was undertaken at one of the COPCs, located at Hatcher Station in southeastern Dallas.

The Hatcher Station COPC (HSC) is staffed by 10 physicians, five advanced practice providers, a nutritionist, a social worker, a clinical pharmacist, and other staff members, including nurses, medical assistants, and clerical assistants. Of the patients seen at this center, the largest racial/ethnic group is Hispanic (67%), followed by non-Hispanic Black (24%); 61% of patients at the clinic are female. The most common health care funding source for these patients is Medicaid (30%), followed by charity care (17%) and Medicare (12%). The clinic handles roughly 50,000 patient visits annually.

The Diabetes Specialty Clinic staff consists of two endocrinologists, seven advanced practice providers (nurse practitioners and physician assistants), two certified diabetes care and education specialists (CDCESs), a clinical pharmacist, a clinical social worker, nurses, and clerical staff. To reduce barriers to accessing diabetes specialty care, a diabetes satellite clinic at the HSC (DHSC) was set up with one advanced practice provider, one nurse, and one clerical staff member to see patients at the HSC on 1 day every other week. On occasion, a CDCES would also see patients on the same day, if available.

One of the main barriers that Parkland Health patients face is access to care, particularly specialty care. The creation of the DHSC had already reduced this barrier somewhat for patients in need of diabetes specialty care. However, with the onset of the coronavirus disease 2019 (COVID-19) pandemic, health care institutions had to quickly assimilate to providing both telehealth and face-to-face visits, and Parkland Health was no exception.

Thus, the first aim of this project was to expand diabetes specialty care using telehealth and face-to-face encounters to improve patient access. Specifically, the goal was to increase the number of people with diabetes requiring specialty diabetes care to be seen at the DHSC, either through telehealth or in-person visits or both.

The second aim was to show that HSC primary care patients referred to the DHSC (designated the HSC + DHSC group) had superior glycemic control as evidenced by lower A1C levels compared with those who were treated for diabetes only by the usual primary care service (designated the HSC-only group). Of note, the DHSC also accepted non-HSC patients who were directed to the clinic after being discharged from the hospital or who chose to come to the satellite clinic based on its location.

Success in meeting this second aim would be shown if HSC + DHSC patients had a greater improvement in A1C than HSC-only patients in the baseline period, with maintenance of this difference or further improvement in the final study period.

Baseline data were extracted from Parkland Health’s electronic health record system. Baseline data were collected from 1 April 2019 through 31 March 2020. This was the period immediately before the onset of the COVID-19 pandemic. The structured data elements continued to be populated in a similar manner during the subsequent 12 months, which made up the intervention phase of the study.

During the baseline period, the HSC saw 7,084 unique patients with diabetes in 8,507 encounters (an average of 1.2 visits per patient), with 88% of the visits conducted in person. The DHSC saw 88 unique patients (55 who were referred from the HSC and 33 who were referred from other sources such as at hospital discharge) in 137 encounters (an average of 1.5 visits per patient), with 95% of the visits conducted in person. Patients were referred to the DHSC for diabetes specialty care based on established criteria, which identified patients with an A1C >9% despite being on insulin, recurrent hypoglycemia while on insulin, multiple comorbidities complicating diabetes management, type 1 diabetes, or diabetes during pregnancy. Supplementary Figures S1 and S2 provide a summary of the data for aim 1.

The initial data analysis examining A1C change over a 12-month period showed that the mean baseline A1C of the HSC patients with diabetes was 7.8 ± 2.1%, and their mean final A1C was 7.6 ± 2.0%. Thus, the mean difference in A1C was −0.2 ± 1.3% (P >0.05).

When examining only patients with a baseline A1C >9%, 1,678 of these patients (24% of all HSC patients with diabetes) were seen by HSC only and not referred to the DHSC. These patients had a baseline A1C of 11.0 ± 1.6% and a final A1C of 10.0 ± 2.1%. The mean difference in A1C for these patients was −1.0 ± 2.0% (P <0.001). This difference was a reduction of 3.6% from baseline, and 1,426 of these patients had an A1C >9% at the end of the baseline period. Supplementary Figure S1 provides more details on the A1C distribution of these patients and a summary of the aim 2 data for the HSC.

For the 55 patients in the HSC + DHSC group, the mean baseline A1C was 9.4 ± 2.1, and the final A1C was 8.9 ± 2.2%. Thus, the mean difference in A1C was −0.5 ± 2.1% (P >0.05). In this group, 25 patients had a baseline A1C >9%. Among those 25 patients, the mean baseline A1C was 11.2 ± 1.4%, and the final A1C was 9.9 ± 2.3%. The mean difference in A1C for these 25 patients was −1.3 ± 2.3% (P = 0.005), which represented a 9.1% decline in A1C, with five patients ending the initial data period with an A1C <9%. Supplementary Figure S3 provides more detail on the A1C distributions of patients in the HSC + DHSC group and a summary of the aim 2 data for these patients.

The time frame from initiation of the QI initiative to its completion was 12 months (1 April 2020 through 31 May 2021).

The lead physician of the Diabetes Specialty Clinic served as the project leader, as she was able to review charts and ascertain which patients could be triaged to be seen via telehealth and could assess the time frame within which these patients needed to be seen. Additional QI team members included the clerical staff who called patients and made the appointments, clinical staff who called patients and reminded them of their appointments, and providers (endocrinologists, nurse practitioners, and physician associates) who saw the patients.

Although the DHSC existed before the start of this intervention, the in-person clinical sessions were increased from every-other-week clinics with one provider to weekly clinics with one to two providers. In addition, four telehealth-only clinics were held per month during the intervention period.

The most important change in our care delivery process was to leverage the use of telehealth visits to continue delivery of diabetes care. The ability to change how patients are seen greatly improved our ability to provide access to care in this underserved area of Dallas.

As the intervention period continued, more telehealth-only clinics were added based on demand. We advertised to the HSC providers that we were still seeing patients and had increased availability to do so, including urgent appointments. More in-person clinic sessions were also added during periods when COVID-19 surges were not occurring.

By reviewing charts more expeditiously, we were able to schedule patients for visits sooner. By reviewing no-show data for 3 months before the start of the intervention (December 2019 to March 2020), we were able to reschedule previously established patients who had been lost to follow-up and get them reconnected to care.

DHSC visits included titration and intensification of medications as needed, in addition to new medication starts when needed. Patients also had more individualized diabetes education provided by CDCESs.

During the 12-month intervention phase (1 April 2020 through 31 March 2021), the HSC saw 8,749 unique patients with diabetes in 9,530 encounters (an average of 1.1 visits per patient), with 43% of the visits conducted in person. The HSC saw a 23% increase in the number of unique patients seen during the intervention period and continued to have the same average number of visits. Additionally, there was a 50% reduction in the number of patients seen in person (because of the COVID-19 pandemic).

The DHSC saw 314 unique patients (85 patients referred from the HSC and 229 patients referred from other sources) in 413 encounters. This volume represented a 356% increase in unique patients seen compared with the baseline 12-month period. Of note, 17% of visits were conducted in person (a 78% reduction), with the majority of visits occurring via telehealth. Supplementary Figures S1 and S2 summarize these data.

During the intervention period, the baseline A1C of the HSC patients with diabetes was 8.0 ± 2.2%, and the final A1C was 7.8 ± 2.1%. Thus, the mean difference in A1C was −0.2 ± 1.3% (P <0.001) and was not clinically significant.

During the intervention year, the HSC-only group included 2,215 patients with an A1C >9%. These patients represent 25.3% of all of the HSC patients with diabetes. The initial A1C in this group was 11.1 ± 1.6%. At the end of the intervention year, 1,973 patients had an A1C >9%, and their mean A1C was 10.1 ± 2.1%, which represented an improvement of 1.0 ± 2% from baseline (P <0.001). Overall, the percentage of patients with an A1C >9% fell to 22.5%, which represents a 2.8% decline from the start of the period. This glycemic improvement during the intervention year was similar to that in the prior 12 months, during which the HSC-only patients had an A1C decline of 0.2%, and the number of those with an A1C >9% decreased from 1,678 to 1,426, representing a change of 3.6%. Supplementary Figure S1 summarizes these data.

During the intervention year, the 85 HSC + DHSC patients had an initial A1C of 10.1 ± 2.4% and a final A1C of 9.4 ± 2.0%. Thus, the change in A1C was −0.7 ± 2.3% (P >0.05). Fifty-seven of these HSC + DHSC patients (64.8%) had a baseline A1C >9%. These patients had an initial A1C of 11.3 ± 1.9% and a final A1C of 9.9 ± 2.0%. Thus, the change in A1C was −1.4 ± 2.1% (P <0.001). Overall, the proportion of patients in this group with an A1C >9% fell to 57.9%, which represents a 6.9% decline from the start of the period. Supplementary Figure S3 summarizes these data.

Comparing the baseline and intervention periods, the HSC + DHSC group had a similar overall decrease in A1C over each period (−0.5 ± 2.1 and −0.7 ± 2.3%, respectively), as did patients with a baseline A1C >9% (−1.3 ± 2.3 and −1.4 ± 2.1%, respectively). In the baseline period, there was a 9.1% reduction from baseline, whereas in the intervention period, there was a 7.1% reduction from baseline.

The first aim of this project was to expand the number of patients seen in the DHSC between the baseline and intervention periods. This aim was accomplished with a 356% increase in unique patients seen. The increase represented more specialty diabetes access to both HSC patients and other patients with diabetes within the Dallas County service area. Much of this increase was driven by the use of telehealth.

The second aim, to maintain or further improve glycemic control using an A1C comparison between the HSC-only and the HSC + DHSC groups, was also accomplished. The overall reduction in A1C in all HSC patients of ∼0.2% was maintained between the baseline and intervention periods, and the HSC + DHSC group also maintained an ∼0.5–0.7% A1C reduction over both periods. This is particularly notable in patients who began each period with A1C levels >9%.

Glycemic control improvements for patients in both the HSC-only and the HSC + DHSC groups are remarkable given that they were accomplished through a combination of in-person and telehealth appointments. Reallocating Diabetes Specialty Clinic staff to a satellite location once weekly to bring about an A1C reduction of 0.5 and 0.7% (intervention year A1C reductions in the HSC + DHSC and DHSC, respectively) may seem to be poor use of resources. However, it is important to note that ∼27% of the patients who came to the DHSC in the intervention period were also being followed in the HSC (compared with 62% in the baseline period). So, during the intervention period, the DHSC was being used as both a local referral center for HSC patients and also as an access point for patients within the wider Parkland Health community who lived closer to the DHSC and either had primary care services at another location or no primary care services at all.

In looking at more recent preliminary data from 2021 and 2022 (not reported), there has now been a shift toward more in-person visits (similar to the baseline study period) versus more telehealth visits (as in the intervention period), as the nature of the COVID-19 pandemic has changed. This shift underscores the importance of continued allocation of these resources to the DHSC, both to handle the increase in in-person visits and to continue serving as an access point for non-HSC patients with diabetes.

Having shown that the DHSC was able to improve glycemic control using telehealth or in-person visits or a combination of the two, we are now planning to expand COPC-based satellite specialty clinics. Specifically, we are planning to strategically place more satellite diabetes specialty clinics at some of the larger COPCs to serve patients who live in that service area and do not want to travel to the main Parkland Health campus for specialty services.

The key to our success in this QI process was to secure buy-in from our entire clinic, from the clerical and nursing staff to the providers, allowing us the flexibility to provide patient care in a number of ways to expand overall access to diabetes specialty care throughout our service area.

No potential conflicts of interest relevant to this article were reported.

As the sole author, U.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.