Research suggests that efforts are needed to help identify patients with uncontrolled type 2 diabetes who have a low probability of A1C goal attainment. “The development of a prediction model could serve as a means of identifying patient subgroups that require additional resources or alternative approaches to managing their disease, with the goal of improving rates of A1C goal attainment,” explains Kevin M. Pantalone, DO, ECNU, FACE. “This is important for both healthcare organizations and physicians because having a model that identifies these patients may improve the likelihood of meeting quality metrics, improving patient outcomes, and ensuring maximal reimbursement for care delivery.”
Dr. Pantalone and colleagues had a study published in Diabetes Care that assessed patient characteristics and treatment factors associated with uncontrolled type 2 diabetes (A1C >9%) and the probability of A1C goal attainment. Of the 6,973 patients with uncontrolled type 2 diabetes in the study, only 1,653—fewer than 25%—achieved an A1C <8%. “Only a minority of patients with an A1C >9% achieved an A1C <8% at 1 year,” says Dr. Pantalone. “Simply put, a subset of patients is predicted to have a low probability of A1C goal attainment in the current real-world patient care environment. It is important to identify these patients so that alternative treatment approaches may be pursued.”
Several predictors were indicative of a higher probability of A1C goal attainment, including older age, white or non-Hispanic race and ethnicity, Medicare health insurance, lower baseline A1C, and a higher frequency of endocrinology or primary care visits. Other predictors of a higher likelihood of attaining A1C goals were use of DPP-4 inhibitors, thiazolidinediones, metformin, and GLP-1 receptor agonists and use of fewer classes of diabetes medications. Insulin use and longer time in the type 2 diabetes database—both of which were presumed as likely surrogates for duration of disease—were identified as factors associated with lower probability of A1C attainment.
Using the study data, the researchers then developed a prediction nomogram that identified 17 variables to help predict A1C goal attainment based on patient and treatment characteristics (Figure). Available online at http://riskcalc.org:3838/Type2DiabetesA1CGoalAttainment/, this model is one of the first of its kind to integrate a wide variety of clinical and nonclinical patient-specific factors into a single predictive tool rather than focusing on outcomes influenced by clinical interventions.
“When applied to patients with uncontrolled type 2 diabetes, this prediction model may help with current population health initiatives,” Dr. Pantalone says. “It can also help clinicians identify patients who may need additional help or alternative care approaches to obtain better control of their A1C. Our nomogram was created to predict the probability of goal attainment based on patient and treatment characteristics. Instead of having busy clinicians use the nomogram while seeing patients, we see the ultimate goal as having the model applied to electronic health record (EHR) systems and reviewed behind the scenes. The EHR can flag patients who are predicted to have a low-probability of A1C goal attainment.”
“Patients with an A1C >9% are often among the most challenging to manage, and many will have numerous characteristics and barriers to care that impact our ability to help them improve their A1C levels,” says Dr. Pantalone. “There is an urgent need for more precise ways of identifying patients with an A1C >9% who will require more effort or alternative approaches to reach their target A1C. Identifying these patients is only the first step. Once identified, we must then implement and evaluate the effectiveness of alternative care models. Ultimately, we cannot treat all patients with an A1C >9% the same, especially given the limited resources of our current healthcare environment and the COVID-19 pandemic.”
In the future, Dr. Pantalone would like to see the study team’s prediction tool and other models implemented in routine clinical care. “We need to leverage the power of our EHR systems,” he says. “Currently, EHRs function more like data repositories and they are not being utilized to their full potential. Once prediction models are incorporated into EHR systems, we will be able to more easily identify patients with a low probability of A1C goal attainment. In turn, this will allow us to evaluate the effectiveness of a variety of treatment strategies to improve their A1C goal attainment.”