Patients with cardiovascular conditions, such as heart disease, stroke, or peripheral arterial disease, were significantly more likely to be prescribed guideline-recommended high-intensity statin therapy if their clinician received an automated reminder containing information regarding their history of cardiovascular disease, prior statin use, and history of statin-associated side effects, according to research presented at the American College of Cardiology’s Annual Scientific Session with the World Congress of Cardiology . The reminders were generated using machine learning algorithms alongside the guideline recommendations.
Statins are cholesterol-lowering drugs that are very effective in preventing premature death in people with established cardiovascular disease related to clogged arteries, such as a previous heart attack, as well as reducing the likelihood of serious cardiac events in people at high risk of heart disease. or stroke. Although medical guidelines recommend high-intensity statins for almost all people with established cardiovascular disease, previous studies have shown that guideline compliance rates are low.
We have found that if you are able to send reminders with personally relevant information to the patient, it works. Our data also shows that it is important to be aware of how reminders fit into the clinical workflow to avoid creating alert fatigue. »
Salim Virani, MD, PhD, cardiologist at the Michael E. DeBakey VA Medical Center in Houston, vice-president for research and professor at the Aga Khan University in Karachi, Pakistan, and staff at the Texas Heart Institute/Baylor College of Medicine
To develop the reminder system, the researchers first used machine learning algorithms to analyze clinicians’ notes for evidence of statin-associated side effects (a factor often cited as contributing to poor compliance with statins). guidelines) and generate summaries of patients’ cardiovascular disease history, history of statin use, and any reported side effects. They then conducted interviews with patients and clinicians to understand the contexts in which reminders featuring patient-specific summaries alongside general statin guidelines would be most useful.
To test the reminder system, researchers randomly assigned 14 clinics (with a total of 117 clinicians treating 18,427 patients) to implement reminders for a 15-month period and 13 clinics (128 clinicians treating 18,214 patients) to continue their usual care practices. All clinics received basic training on statin guidelines before the start of the study, and all clinicians had access to a dashboard with detailed information regarding statin use in their patients with cardiovascular disease throughout the study.
For clinics implementing reminders, approximately 27% of reminders were sent synchronously, meaning they were sent within two to seven days before the clinician saw the patient, and the remainder were sent asynchronously at other times during the study. The system was designed not to send reminders to clinicians who had more than three unsigned reminders in a row.
Overall, the study, which is one of the largest to date to use machine learning-generated reminders to influence clinicians’ prescribing practices, involved more than 36,000 patients with more than 18 000 in each arm. In the intervention group, a total of 4,928 reminders were sent to 4,532 unique patients. Only regarding half of the cohort was eligible for reminders due to protocols built into the study algorithm to minimize alert fatigue. The proportion of patients who received high-intensity statin therapy increased by 3.8% in all clinics in the intervention group compared to the usual care group, and this proportion increased by 10.1% among the under – set of patients who received a booster in the intervention group.
“A 10% increase in the use of high-intensity statins is very significant at the health system level given the challenges associated with this issue, as shown in multiple studies,” Virani said. “Our results show that for every 10 reminders sent, healthcare systems can expect a patient with cardiovascular disease to be initiated or titrated toward high-intensity statin therapy. »
Patients presented in synchronous recalls were slightly more likely to be prescribed high-intensity statin therapy, as were patients who had not previously reported statin-associated side effects. However, the study showed a 9% increase in high-intensity statin use, even in patients with statin-associated side effects, among the subset of patients who were in a recall group. ‘intervention. “This is important because these patients are the most difficult to treat given their history of side effects associated with statins,” Virani said.
In addition, adherence to statins (measured by prescription refill data) was 2.8% higher among patients in the intervention group. The total proportion of patients who received statins (not just high-intensity statin therapy) decreased slightly in both groups, but this attrition was lower in patients in the intervention group.
Despite the benefits of alerts, Virani said regarding 30% of clinicians in the study withdrew from the trial, highlighting the ongoing challenge of creating alerts that can grab clinicians’ attention without overwhelming them. He said the study was conducted at a time when workflows were changing rapidly due to iterative waves of COVID-19, which may have affected the results. The researchers plan to further analyze the data to understand why some clinicians opted out of receiving the reminders. Additionally, clinicians were unable to click directly from the reminder to initiate a prescription, which may have limited its use as a clinical decision support tool.
The study was funded by the Department of Veterans Affairs Health Services Research and Development.
This study was simultaneously published online in the journal Circulation at the time of submission.
