Automated Patient-Specific Reminders to Clinicians Help Improve Statin Prescribing Practices
Dr. Salim Virani, a tenured professor of cardiology and cardiovascular research at Baylor College of Medicine, physician investigator at Michael E. DeBakey Department of Veterans Affairs Medical Center, an alumnus and staff at The Texas Heart Institute, and vice provost for research and professor at The Aga Khan University, and his research team developed an automated reminder system for clinicians about patients’ cardiovascular conditions that can improve guideline-recommended statin therapy prescribing practices. This machine learning-based innovative approach would revolutionize the outcome of statin therapy. This study, funded by the U.S. Department of Veterans Affairs Health Services Research & Development, was presented at ACC.23 Together With WCC, the annual scientific meeting of the American College of Cardiology, jointly with the World Heart Federation, in March 2023 in New Orleans, Louisiana. Also, this work has been published in Circulation.
Statins, the HMG-CoA reductase inhibitors, are prescription medicines people take to lower their blood cholesterol levels. Statins decrease the level of low-density lipoprotein (LDL) cholesterol, which leads to the building up of cholesterol inside arteries. The cholesterol that builds up inside arteries makes those narrow (atherosclerosis), making it hard for blood to circulate and increasing the likelihood of a heart attack or stroke. Statins get in the way when the liver is trying to make cholesterol and help the liver get rid of excess cholesterol. Thus, statins decrease the risk of having a heart attack or stroke and are highly effective at preventing premature death in people with established cardiovascular disease related to clogged arteries, such as a previous heart attack. Statins may cause side effects, such as muscle aches, and high blood sugar. Although medical guidelines recommend high-intensity statin therapy for nearly all people with established cardiovascular disease, guideline compliance is low, often because of these side effects.
This study, involving over 36,000 patients, is one of the largest to date on using reminders to clinicians to influence prescribing practices. Dr. Virani and team found that patients with cardiovascular disease (eg, heart disease, stroke, or peripheral arterial disease) were significantly more likely to be prescribed guideline-recommended high-intensity statin therapy if their clinicians were sent an automated reminder with information about their cardiovascular condition (including disease history, prior statin use, and statin-associated side effects) alongside guideline recommendations. The reminder system was generated by using machine learning algorithms and had patient-specific summaries and general statin guidelines.
The research team first used machine learning algorithms to analyze clinician notes on statin-associated side effects and generate summaries of patients’ cardiovascular disease history, history of statin use, and side effects. They then interviewed patients and clinicians to determine the circumstances in which reminders would be most beneficial. The reminder system was tested for a 15-month period in randomly assigned 14 clinics with 117 treating clinicians implementing the reminder system (intervention arm) versus 13 clinics with 128 clinicians continuing their usual care practices (control arm), involving approximately over 18,000 patients in each arm. In the intervention arm, about 27% of the reminders were sent synchronously, ie, those reminders were sent within 2 to 7 days before the patients’ clinic visit, and the remaining were sent at other times during the study period. However, clinicians who had more than 3 unsigned reminders in a row were not sent reminders, and protocols were designed to minimize clinicians’ alert fatigue.
The proportion of patients who were prescribed high-intensity statin therapy increased by nearly 4% across all the reminder system-implementing clinics over those with the usual care practice and by 10% among the subset of patients whose clinicians received an alert about them. A 9% increase in high-intensity statin use was observed even in patients with statin-associated side effects among the patients in the intervention arm over those in the control arm, a remarkable finding because those patients tend to be tough to treat with statin therapy, given their history of statin-associated side effects. Moreover, statin adherence, measured by data on prescription refills, was 2.8% higher among patients in the intervention group. According to Dr. Virani, for every 10 reminders, healthcare systems can expect 1 patient with cardiovascular disease to be initiated on or titrated to high-intensity statin therapy, and the results are encouraging in improving guideline-based statin prescribing practices.
The study algorithms limited the number of alerts sent to each clinician to minimize alert fatigue, and because of this restriction, a substantial number of clinicians did not receive alerts on all patients they could have otherwise. Still, ~30% of clinicians opted out of the study. Thus, it remains a persistent challenge to send alerts to clinicians that can attract their attention for a prompt response without overwhelming them. The study was conducted during the COVID-19 pandemic when workflows fluctuated rapidly, which might have affected the results. More work is warranted so clinicians can initiate a prescription directly from the reminder, which would establish this system as a decision-support tool in clinics. The results from this study underscore the importance of how reminders can fit in with the clinical workflow while avoiding clinicians’ alert fatigue.
Related Stories
ACC March 5, 2023
Targeted Reminders Increase Prescriptions for High-Intensity Statins
Automated reminders found effective for improving guideline-based statin prescribing practices
