AI identifies reasons for statin nonuse in patients with diabetes at a single center

Key takeaways:

• An AI model identified reasons for statin nonuse among patients with diabetes.
• The most common reasons included statin hesitancy, guideline-discordant practice 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 among patients with diabetes, researchers reported.

Reasons for statin nonuse varied across age, race/ethnicity, insurance and diabetes type, according to the study published in the Journal of the American Heart Association.

“Identifying reasons for statin nonuse can guide targeted interventions to improve guideline-directed statin use. … To understand statin nonuse in diabetes across a health system, it is therefore necessary to analyze unstructured EHR data at scale,” Ashish Sarraju, MD, resident in the division of cardiovascular medicine and cardiovascular institute at Stanford University and the department of cardiovascular medicine at Cleveland Clinic Foundation, and colleagues wrote. “Artificial intelligence approaches to natural language processing (NLP) may help analyze such large-scale unstructured data.”

To test their hypothesis, Sarraju and colleagues developed a deep learning NLP algorithm using Clinical Bidirectional Encoder Representations from Transformers (BERT) to identify statin nonuse and actionable reasons for nonuse among 33,461 patients with diabetes at Stanford Health Care Alliance (mean age, 59 years; 49% women; 36% white).

Accuracy of Clinical BERT

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

Sarraju and colleagues observed Clinical BERT successfully identify statin nonuse with an area under receiver operating characteristic curve of 0.99 and patient, clinician and system reasons for statin nonuse with an AUC of 0.9.

The researchers reported that Clinical BERT demonstrated good concordance with expert clinician review and outperformed other NLP approaches.

Overall, 47% of the cohort had no statin prescriptions and 16% were using a statin despite no documented statin prescriptions.

Researchers reported that statin hesitancy (19%), guideline-discordant practice (19%) and clinical inertia (18%) were more common compared with side effects and/or contraindications (12%) as the reason for statin nonuse.

Specific reasons for statin nonuse

Reasons for nonuse varied by age, race/ethnicity, insurance and diabetes type, according to the study.

Patients older than 75 years were more likely to experience statin-associated side effects and/or contraindications than statin hesitancy, clinical inertia or discordant practice compared with younger patients (P < .05).

Hispanic individuals were most likely to experience guideline-discordant practice compared with most other reasons for statin nonuse (P < .05), whereas Black patients were most likely to experience clinical inertia as their reason (P < .05), according to the study.

Patients with Medicaid insurance were more likely to experience guideline-discordant practice compared with the other reasons for nonuse (P < .05).

Moreover, patients with type 1 diabetes were more likely to experience guideline-discordant practice compared with the other reasons for statin nonuse (P < .05), according to the study.

“Although statins conclusively reduce cardiovascular events in diabetes, real-world statin utilization remains poor despite guideline recommendations, representing an important and well-recognized target for population interventions,” the researchers wrote. “Our findings clearly demonstrated gaps in structured statin prescriptions in patients with diabetes. Nearly half of our cohort with diabetes lacked guideline-directed statin prescriptions, with disparities observed in younger, female and Black individuals. These results add to prior literature demonstrating statin prescription gaps and disparities in diabetes and emphasize the need to understand reasons for nonuse to ultimately improve guideline adherence in a disease that represents a major public health burden.”