The Impact of Utilizing Shared Decision-Making AI Software in Patients with T2D

Results from using GlucosePATH in Medicaid patients reveal the benefits of clinical decision-making AI support software for type 2 diabetes patients in reducing HbA1c, reducing the risk of disease complications, and improving access to care. 

Type 2 diabetes mellitus (T2DM) is a growing global health crisis, and prevalence has also dramatically increased within the United States. According to the Centers for Disease Control, over 10% of the population have T2DM, and 33% have prediabetes. Unfortunately, there have also been other negative implications with this rise, including disease-related complications and healthcare costs. A large population that is impacted by this is patients covered by Medicaid. In 2019, Medicaid paid $177 million on diabetes services alone. In addition, a recent study focused on this population within Kentucky, where there has been an exponential increase in those with diabetes from 6.5% in 2000 to 12.9% in 2017. Due to these reasons, there has been an increased effort among many different sectors to develop tools to help reduce disease development and progression. One primary tool developed is clinical decision support systems (CDSS). These programs utilize software that allows for collaboration between clinicians and patients to help customize treatment regimens that assess for desired health outcomes, individual patient demographics, and barriers to care, including cost. These studies on these programs have shown them to be beneficial in reaching HbA1c goals faster and improving access to antidiabetic medications. GlucosePATH is a CDSS that uses software to accomplish these goals and is utilized in the before mentioned study to show evidence of the importance and need for implementing these programs in real-world clinical practice to help combat the growing issues related to diabetes care. 

The study out of Kentucky focused on quality improvement and used both longitudinal (Using CDSS) and control group (No use of CDSS) comparisons of outcomes as a mixed-method model. In addition, it utilized online surveys from physicians and pharmacists and clinical measures to determine its efficacy. The data was collected from September 2018 to June 2020 and focused on uncontrolled patients with diabetes (HbA1c > 8%) covered through Kentucky Medicaid and seen at an extensive urban healthcare system. The study split the participating subject into four groups, those who were placed on the whole medication regimen provided by CDSS, those who partially use the CDSS recommendations, those whose providers put them on an alternative medication outside of the recommendation, and those who had no changes to their current regimen. A Statistical Package for Social Sciences (SPSS) 26 was used to assess the data. Additionally, to compare the results of the study group and the control group’s initial HbA1c, final HbA1c, and mean difference of HbA1c, bootstrapping t-tests were used, and to determine statical significance, alpha was set at 0.05.  

This comparison showed that the initial HbA1c of the study group was 10.11 (95% CI: 9.82-10.41), and the control group was 9.91 (95% CI: 9.62-10.2). The final HbA1C for the study group was 8.37 (95% CI, 7.98-8.76), the control group was 8.66 (95% CI, 8.3-9.01). The mean difference in HbA1c in the study group was 1.74 or 15.3%, whereas the mean difference in HbA1c in the control group was 1.26 or 11.4%. Initially, the bootstrapping t-test for the final HbA1C and mean difference did not demonstrate statistical significance (p-value= 0.29 and p-value=0.11, respectively). However, since the original t-test did not combine the full and partial groups, an additional bootstrapping t-test was performed combining the two, since they did utilize CDSS recommendations, and then were then compared to a control group made up of those in the alternative and those who had no change in their current therapy group. After correcting for this bias, the new outcome for the final HbA1c showed a p-value of 0.04 (95% CI 0.06-1.14), and the mean difference in HbA1c resulted in a p-value of 0.02 95% CI 0.17-1.44).  

Overall, this quality improvement project demonstrated efficacy and a clear need for integrating CDSS combined with team-based care. The surveys reveal that physicians and pharmacists enjoyed the tool and gave patients a more significant role in their care. In addition, these programs allow for more options to the wide variety of pharmacotherapy options for those with diabetes, which results in fewer cost-related barriers and adherence. It also was able to help achieve goal levels for HbA1c at a higher and faster rate than those who did not have access to the CDSS program. While there is still a need for more extensive studies that look at the use of CDSS combined with team-based care in other patient populations, these results should serve as evidence for the importance of moving towards shared decision-making software to reduce barriers to care, lower healthcare costs, and improve health outcomes for those with T2DM. 

Practice Pearls: 

• Innovative artificial intelligence (AI) tools such as GlucosePATH optimize care for patients with T2DM. 
• GlucosePATH software uses an algorithm that would focus on A1C control, body weight, adherence, cost, and side effects. Thus, it improves the outcomes for patients with T2DM. 

• Decision support software like GlucosePATH provides aid to the prescriber in developing treatment plans, thus, helping patients get to their goals sooner, in the most effective way. 
• For more information on GlucosePath Software, email publisher@diabetesincontrol.com 

References for “The Impact of Utilizing Shared Decision-Making AI Software in Patients with T2D”:
Zhang, M. Svec, R. Tracy, G. Ozanich, Clinical Decision Support Systems with Team-based Care on Type 2 Diabetes Improvement for Medicaid Patients: A Quality Improvement Project, International Journal of Medical Informatics (2021),

Ozanich, G. et al. “A STUDY ON TYPE 2 DIABETES MELLITUS FOR PATIENTS AMONG MEDICAID BENEFICIARIES IN KENTUCKY”.Kentucky Progress Report. Accessed February 11, 2022.  

Alexa Rodriguez, PharmD Candidate 2022, University of South Florida, Taneja College of Pharmacy  

See more about GlucosePATH shared decision-making AI software for diabetes.