The primary outcome evaluated in this study, A1C value, did not show a statistically significant difference between intervention and comparison groups. While the lack of improvement in A1C value is paradoxical to what was anticipated at the onset of our study, similar data was seen in other studies published in literature. Specifically, the change in A1C value did not significantly improve if baseline A1C was approximately near 8% (10.2 mmol/L) [13]. Within our own resident clinic, the baseline A1C value for intervention and comparison groups was 7.9% (10.0 mmol/L) and 7.8% (9.8 mmol/L), respectively. Per Lancet’s meta-analysis of quality improvement studies evaluating A1C, this lack of improvement was also demonstrated in settings where the QI intervention involved clinician reminders and auditing, similar to the interventions used in our study.13 Similar findings were observed in previous studies examining the effect of quality improvement interventions on resident clinics alone. For example, in a study investigating foot examinations performed in resident clinic, the HbA1c value increased from 7.9% (10.0 mmol/L) to 8.1% (10.3 mmol/L) over the span of the QI intervention [14]. Additionally, another study performed at an internal medicine resident clinic also demonstrated a lack of decrease in A1C, irrespective of intervention or comparison group [15].

There may be several underlying reasons for the lack of improvement in A1C values. For one, the residents were unaware that A1C value was the primary outcome in our study. This was implemented to avoid bias by steering the residents away from focusing solely on the primary outcome. If our study was developed with a singular focus on A1C improvement, the results may have demonstrated improved A1C values. Another reason specific to our study is that the quality improvement interventions that were implemented in our resident clinic did not directly involve activities that decrease A1C. The interventions incorporated were primarily targeted at examinations or laboratory tests that should be performed by the clinician in the clinic, rather than patient interventions. We hypothesized that the preventative examinations may still indirectly affect A1c value but our study did not demonstrate that correlation. A possible cause for this lack of improvement is time utilization during a visit. For example, a resident who is focused on performing a foot examination may not have spent the necessary amount of time counseling on diet or medication regimen, thus mitigating the beneficial effect of a foot examination. With limited time during a clinic visit, the residents’ ability to impact both glycemic control and provide appropriate preventative care may be decreased. These results demonstrate the importance of such comprehensive diabetes care in patients with diabetes and the definite need in this patient population for parallel diabetes diet education, pharmacist education and intensive lifestyle changes [16, 17].

While foot examination and A1C test did not show a statistically significant improvement with the quality improvement intervention, each secondary outcome demonstrated an absolute increase in the percentage of patients who received those tests 1 year after the interventions were implemented. Compared to the national data as shown earlier in Fig. 2, the adherence rates in the clinic are still lagging behind, however there were significant improvements from the pilot study data. Specifically, the national adherence rate of foot and eye examinations are 68 and 62% respectively. In our clinic’s intervention groups, these two adherence rates improved from 48 to 61% in foot examinations performed and 42 to 47% in eye examinations performed. Similarly, with A1C testing, the intervention group improved from 57 to 65%, similar to the national rate of 68%. While these QI interventions have not completely eliminated the gap between our clinic and the national averages, the significant improvements in these rates indicate the QI interventions as potential solutions to the low adherence rates. Considering the given trend, we hope that there is further improvement in adherence rates with continued use of QI interventions.

Comparing with present literature, there were limited studies evaluating the secondary outcomes from our study. One previous project evaluating foot examinations showed similar improvements in the number of foot examinations performed post-intervention.14 In another research article, there was significant increase in A1C and LDL testing obtained in intervention versus the comparison groups [15]. While there were few studies reporting on these secondary outcomes, we were unable to identify any projects that showed a lack of improvement with a quality improvement intervention.

We also compared the difference in the secondary outcomes between the intervention groups. For foot examinations, there was improvement in all intervention groups compared to baseline. This may reflect the ease of performing a foot exam versus the other preventative examinations. Whereas eye examinations by ophthalmologists or laboratory tests obtained outside of clinic visit depend partly on the patients, the foot exam can be performed directly in the clinic. Thus, this may have contributed to the effectiveness of the quality improvement study on obtaining more foot examinations in patients. With regards to the eye examinations, three teams (Purple, Red, and Green) had improvements whereas two teams (Yellow and Blue) did not show improvement. Both Yellow and Blue teams identified pre-visit labs as an emphasis of their quality improvement study. Given that eye examinations/referrals are more likely to be performed post-visit, their emphasis on pre-visit testing may have lowered the effect of their QI interventions on eye examination adherence rates.

For laboratory tests such as lipid panel, A1C testing, and microalbumin/creatinine ratio, Red and Purple teams consistently had better improvements than the remaining teams. Both of these teams implemented interventions (as listed in Table 2) that involved obtaining overdue labwork right after the clinic visit. This was unique to these two teams as the other three intervention teams did not consider this in their intervention plans. Given this, there may be increased effectiveness in adhering to ADA guidelines, specifically with regards to laboratory testing, if the testing is performed right after a clinic visit. However, further studies are necessary to evaluate this intervention further.

In our study, Purple team had the largest improvements in all secondary outcomes with the exception of microalbumin/creatinine ratio (2nd highest improvement in this outcome). In retrospect, we evaluated the different interventions implemented by the resident teams to potentially identify the reason for such significant improvement in the Purple team versus the remaining teams. One particular unique intervention by the Purple team consisted of a protected one-half day block for each resident where they identify their patients who are overdue for the required examinations and subsequently call the patients to schedule appointments for these tests. This intervention may have been beneficial because residents were given just one task for the half-day, allowing them to better focus on identifying patients who are due for these examinations. The focused half-day may also have helped the residents better understand the ADA guidelines and made them more likely to perform these measures at their patients’ clinic visits. Additionally, personally speaking to the patients over the phone may indirectly have decreased the no-show rate and increased the compliance rate due to this increased communication. Given the significant improvements seen with the Purple team, expansion of this intervention for the remaining resident teams will be necessary to identify if it is a truly beneficial intervention for resident clinics. The significant time commitment associated with this intervention also necessitates further studies to evaluate whether ancillary staff can perform this intervention with similar improvements in outcomes.

The consistent improvement in adherence rates amongst comparison groups during this study was an interesting observation. One reason may be the proximity of the comparison groups to the intervention groups. Since the residents in the program work so closely together, it may be that strategies from the intervention groups were discussed with those from comparison groups and possibly implemented by individual residents. Another cause may be that comparison groups also were able to attend the discussion sessions every 3 months that evaluated the progress of QI interventions. During this time, these groups may have discussed strategies to improve their adherence rates but did not write down an official intervention strategy.

Regardless, the significant improvement in adherence rates in several secondary outcomes amongst intervention groups, especially Purple and Red teams, demonstrates that certain quality improvement interventions in resident clinic can be beneficial in better adhering to the ADA guidelines. The implementation of a quality improvement intervention not only allows for better preventative care in patients with diabetes in the resident clinics but also helps residents understand how to implement quality improvement into daily practice beyond residency and in their own clinics and hospital settings.

There are several limitations present in our research study. As mentioned prior, there is a large no-show rate for patient visits at the resident clinic [9]. The no show rate limits opportunities for the residents to provide the preventative care that is expected by the ADA guidelines and may lower the adherence rates compared to the national averages. Additionally, the comparison groups were in the same hospital location as intervention groups, which may have influenced sharing of intervention strategies amongst residents and may limit the pure random allocation of these groups. Furthermore, the comparison groups were not randomly assigned but were made of teams that opted not to design a QI intervention plan. This can be considered a limitation since teams that did not design a plan may consist of residents who are less motivated to engage in improving their quality measures for patients with diabetes.

Another limitation is the underreporting of tests performed. Residents may have performed foot examinations but did not report it in the EMR due to lack of time or disruption in the workflow. This could also inaccurately lower the adherence rates in the clinics. Similar phenomenon may have occurred with the laboratory tests, which could have been deferred if the patient had instead received the laboratory tests at an outside facility. If these facilities were not associated with Beaumont, those laboratory tests are not recorded in the EMR system, thus falsely lowering the adherence rates.

We also did not collect attendance data on the educational sessions prior to the study period, thus we were unable to evaluate the correlation between attendance at these sessions and the outcome measures. This can be considered a limitation since teams that had more members attend the sessions could have been more motivated to engage in their team’s quality improvement plan and have better outcomes.

Additionally, we did not address all of the variables listed in the ADA guidelines such as blood pressure and vaccinations. Thus, the effect of the QI interventions on these variables is unknown and needs to be addressed in future studies. The blood pressure was not measured in this study because of the wide variability present between different visits. The influenza vaccination was not measured because this study was started during the middle of flu vaccination season so we did not feel that our baseline and post year 1 results would be an accurate reflection of vaccination rates.

Beaumont Hospital – Royal Oak is an academic medical center nearby the large metropolitan city of Detroit. Our outpatient clinic has patients from both rural and urban areas of Southeast Michigan. Thus, the results of our study can be expected at other resident clinics in academic medical centers around the country. Additionally, the primary and secondary outcomes of our study were investigated from the national guidelines set by the ADA so we believe that this study can be replicated in other resident clinics and interventions can be implemented in different hospitals to a similar effect.