AI identifies reasons for statin nonuse in diabetics at a single center

Important points:

• AI model identifies reasons for non-use of statins in diabetic patients.
• The most common reasons include statin hesitation, guideline inconsistency, and clinical inertia.

A deep learning model utilized unstructured electronic health record data to identify specific patient-, physician-, and system-level reasons for statin nonuse in people with diabetes, the researchers reported.

According to a study published in , the reasons for not using statins vary by age, race/ethnicity, insurance, and type of diabetes. American Heart Association Journal.

“Identifying reasons for statin abstinence can guide targeted interventions to improve guideline statin use. To understand it, we need to analyze unstructured EHR data at scale.” Ashish Saraju, M.D., a resident of the Department of Cardiovascular Medicine and the Cardiovascular Institute at Stanford University and the Department of Cardiovascular Medicine at the Cleveland Clinic Foundation, colleagues wrote. “Artificial intelligence approaches to natural language processing (NLP) can help analyze large amounts of unstructured data like this.”

To test the hypothesis, Sarraju et al. recruited 33,461 diabetics from the Stanford Health Care Alliance (mean age 59 years; 49% female, 36% Caucasian).

Accuracy of clinical BERT

The algorithm’s performance was evaluated against expert clinician reviews and compared with other NLP approaches.

Sarraju et al successfully identified statin abstinence in the area under the receiver operating characteristic curve for the Clinical BERT of 0.99 and successfully identified the reasons for statin abstinence in patients, clinicians, and systems with an AUC of 0.9 I observed what you did.

Researchers reported that Clinical BERT showed good agreement with expert clinician reviews and was superior to other NLP approaches.

Overall, 47% of the cohort did not receive a statin prescription and 16% were using statins despite having no documented statin prescription.

Researchers found that statin hesitation (19%), non-guideline-consistent practices (19%), and clinical inertia (18%) were more common reasons for statin nonuse than side effects or contraindications (12%). reported to be common.

Specific reasons for not using statins

Research shows that reasons for non-use vary by age, race/ethnicity, insurance, and type of diabetes.

Patients aged 75 years and older are more likely than younger patients to experience statin-related side effects and contraindications.P. < .05).

Hispanic individuals were most likely to experience non-guideline-consistent practice compared to most other reasons for statin abstinence (P. < .05), whereas black patients were most likely to experience clinical inertia as the reason (P. < .05), according to research.

Patients with Medicaid insurance were more likely to experience non-guideline-consistent practice compared with other reasons for non-use (P. < .05).

Furthermore, patients with type 1 diabetes were more likely to experience non-guideline-consistent exercise compared to other reasons for statin abstinence (P. < .05), according to research.

“Statins definitively reduce cardiovascular events in diabetes, but despite guideline recommendations, statin utilization in practice remains inadequate and represents an important and well-recognized target for population intervention. “Our findings clearly demonstrate a gap in structured statin prescribing in patients with diabetes. Nearly half of the diabetic cohort Disparities were found among younger, female, and black people in not receiving statin prescriptions according to guidelines.These results add to previous literature demonstrating gaps and disparities in statin prescribing in diabetes, leading to major It underscores the need to understand the reasons for non-use in order to ultimately improve guideline adherence in diseases that represent a public health burden.”