Computer Alerts for Peripheral Arterial Disease

(PAD-ALERT Trial)

Overseen By Candrika D Kharaini

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: Brigham and Women's Hospital

Must not be taking: Statins, Ezetimibe, Bempedoic acid, others

Disqualifiers: None

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: Unknown

Duration: 3 months

400 Participants Needed

Trial Summary

What is the purpose of this trial?

This single-center, 400-patient, randomized controlled trial assesses the impact of a patient- and provider-facing EPIC Best Practice Advisory (BPA; alert-based computerized decision support tool) to increase guideline-directed utilization of statin and statin-alternative oral LDL-C lowering therapies in patients with PAD who are not being prescribed LDL-C-lowering therapy.

Will I have to stop taking my current medications?

If you are currently taking any LDL-C-lowering medications like statins, ezetimibe, bempedoic acid, PCSK9 inhibitors, or inclisiran, you cannot participate in this trial. Otherwise, the protocol does not specify if you need to stop other medications.

What data supports the effectiveness of the treatment Alert-Based Computerized Decision Support for Peripheral Arterial Disease?

Research shows that Best Practice Alerts (BPAs) in electronic health records can improve patient care by encouraging appropriate use of healthcare resources and reducing costs. Tailoring these alerts to specific patient needs can also reduce the burden of excessive alerts, making them more effective. ^{1 2 3 4 5}

Is the alert-based computerized decision support system safe for humans?

The safety of alert-based computerized decision support systems, like the EPIC Best Practice Advisory, is not well-documented in terms of direct human safety. However, these systems can lead to 'alert fatigue,' where too many alerts cause healthcare providers to ignore them, potentially putting patients at risk. ^{1 4 5 6 7}

How is the Alert-Based Computerized Decision Support treatment unique for peripheral arterial disease?

This treatment is unique because it uses computer alerts to assist doctors in making better decisions for patients with peripheral arterial disease, potentially improving outcomes by optimizing treatment plans based on individual patient data. ^{8 9 10 11 12}

Research Team

Gregory Piazza

Principal Investigator

BWH

Eligibility Criteria

This trial is for patients with Peripheral Artery Disease (PAD) who are not currently taking medication to lower LDL cholesterol. It's designed to see if a computer alert can help improve the use of recommended treatments.

Inclusion Criteria

I have been diagnosed with peripheral artery disease (PAD).

I am not on any medication to lower my LDL cholesterol.

Patients must be seen in Cardiovascular Medicine Clinic, Primary Care, Podiatry, Vascular Surgery, and Diabetology

Exclusion Criteria

I am not currently taking any cholesterol-lowering medications.

Timeline

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Treatment

Participants are exposed to an alert-based computerized decision support tool to increase utilization of LDL-C lowering therapies

3 months

Follow-up

Participants are monitored for safety and effectiveness after treatment, including assessment of major adverse cardiovascular and limb events

6 months

Treatment Details

Interventions

• Alert-Based Computerized Decision Support

Trial Overview The study tests an electronic alert system that reminds healthcare providers and patients about the benefits of statins or alternative therapies for lowering cholesterol in PAD patients not on treatment.

Participant Groups

Group I: Alert Experimental Treatment 1 Intervention

Alert-based CDS will consist of an on-screen electronic alert that will notify the clinician that the patient has an indication for LDL-C-lowering therapy but is not prescribed any. The clinician will have the opportunity to proceed to an order template through which appropriate lipid-lowering can be prescribed. The clinician could also elect to learn more about current evidence-based recommendations for LDL-C lowering in the PAD population. Finally, the clinician could elect to proceed without ordering oral LDL-C-lowering therapy or reading evidence-based recommendations for LDL-C lowering but would have to provide a rationale for not doing so.

Group II: No Alert Active Control 1 Intervention

No on-screen notification will be issued to the clinician

Who Is Running the Clinical Trial?

Brigham and Women's Hospital

Lead Sponsor

Trials

1,694

Recruited

14,790,000+

Esperion Therapeutics, Inc.

Industry Sponsor

Trials

26

Recruited

21,900+

Founded

2008

Headquarters

Ann Arbor, USA

Known For

Cholesterol Therapies

Top Products

NEXLETOL, NEXLIZET, NILEMDO, NUSTENDI

Findings from Research

The study at Jurong Health Campus showed a 59.6% reduction in interruptive Best Practice Advisory (BPA) alerts after implementing optimization strategies, which significantly improved clinician response rates to alerts.

Despite increasing the number of unique BPAs from 54 to 360, the optimized alerts led to a 74% reduction in alerts from seven specific BPAs, saving an estimated 3600 hours of provider time annually and enhancing overall alert compliance.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Optimizing Best Practice Advisory alerts in electronic medical records with a multi-pronged strategy at a tertiary care hospital in Singapore. Ng, HJH., Kansal, A., Abdul Naseer, JF., et al.[2023]

The implementation of a formal alert management process at NewYork-Presbyterian Hospital has improved the quality of clinical decision support by requiring active involvement from hospital committees and departments in alert development.

In the first year after implementation, the hospital received 10 alert requests, indicating a successful model that has reduced the burden on medical informatics resources and enhanced knowledge management activities.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Managing the alert process at NewYork-Presbyterian Hospital. Kuperman, GJ., Diamente, R., Khatu, V., et al.[2018]

An AI screening tool using the BioMed-RoBERTa model effectively identified patients suitable for deep vein thrombosis (DVT) prophylaxis, achieving high accuracy with a precision-recall area under the curve of 0.82 and a receiver operator curve area of 0.89.

By selectively applying Best Practice Alerts (BPAs), the tool reduced unnecessary alerts by 20% and increased the relevance of alerts by 14.8%, demonstrating the potential of large language models to improve healthcare efficiency and patient safety.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

A Large Language Model Screening Tool to Target Patients for Best Practice Alerts: Development and Validation. Savage, T., Wang, J., Shieh, L.[2023]

In a study analyzing 1012 medication regimens in a nephrology clinic, more specific alerting strategies significantly reduced the number of alerts generated, with the most advanced strategy (D) resulting in alerts for only 7.81% of regimens compared to 87.5% with the least specific strategy (A).

The findings suggest that using patient and drug-specific information in alert systems can effectively minimize alert fatigue among physicians, potentially improving the safety of prescribing medications to patients with renal impairment.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Tailoring of alerts substantially reduces the alert burden in computerized clinical decision support for drugs that should be avoided in patients with renal disease. Czock, D., Konias, M., Seidling, HM., et al.[2017]

A survey of 98 healthcare providers revealed that 69% approved of the electronic alert system for acute kidney injury (AKI), indicating a generally positive reception among physicians, pharmacists, and non-physician providers.

Approval of the alert system was strongly linked to the belief that it improved patient care, but notably, approval decreased by 20% for every additional 30 days of trial duration, highlighting the importance of perceived efficacy over time.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Provider acceptance of an automated electronic alert for acute kidney injury. Oh, J., Bia, JR., Ubaid-Ullah, M., et al.[2020]

Computerized decision support systems can help healthcare providers choose the best treatment for patients with complex cardiovascular conditions, but they often focus more on effectiveness than on the safety of therapies for individual patients.

Websites like CredibleMeds.org and BrugadaDrugs.org provide valuable safety information about drugs, which could enhance electronic prescribing systems and promote safer, personalized medicine by addressing the risk of drug-induced arrhythmias.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Assessing cardiovascular drug safety for clinical decision-making. Woosley, RL., Romero, K.[2022]

A study involving 14 pediatric clinicians revealed that many prescribers are unaware of drug-drug interactions (DDIs) for medications they frequently prescribe, highlighting a gap in DDI knowledge that could impact patient safety.

Clinicians preferred receiving DDI alerts during medication selection rather than at order submission, and suggested improvements such as standardized text, risk information, and actionable alerts to enhance the effectiveness and usability of these safety reminders.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Clinician Perceptions of Timing and Presentation of Drug-Drug Interaction Alerts. Humphrey, KE., Mirica, M., Phansalkar, S., et al.[2021]

In a study of 375 procedures involving 345 patients with peripheral arterial disease, the greatest improvement in calculated quality of life (cQOL) was observed in patients who underwent endovascular or open revascularization compared to those who had primary amputation or medical therapy.

Patients whose actual treatment matched the surgeon's recommendations or predictions from decision analysis models experienced significantly better outcomes, highlighting the importance of treatment concordance for improving patient quality of life.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Failure of patients with peripheral arterial disease to accept the recommended treatment results in worse outcomes. Brothers, TE.[2022]

The natural language processing (NLP) algorithm demonstrated superior accuracy (91.8%) in identifying cases of lower extremity peripheral arterial disease (PAD) from clinical notes compared to traditional billing code algorithms, which had accuracies of 81.8% and 83%.

The NLP system also showed higher positive predictive value (92.9%) and specificity (92.5%), indicating its effectiveness in accurately identifying PAD cases, which could enhance clinical decision-making and patient care.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Mining peripheral arterial disease cases from narrative clinical notes using natural language processing. Afzal, N., Sohn, S., Abram, S., et al.[2022]

In a study of 206 patients with peripheral arterial disease, a decision analysis model suggested different treatment plans than those initially proposed by surgeons, indicating that bypass surgery was recommended more often than necessary, while angioplasty and medical management were underutilized.

Surgeons showed greater comfort in their decisions when their initial treatment plans aligned with the model's predictions, suggesting that decision analysis can enhance decision-making processes without significantly altering final treatment choices.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Prospective decision analysis modeling indicates that clinical decisions in vascular surgery often fail to maximize patient expected utility. Brothers, TE., Cox, MH., Robison, JG., et al.[2022]

Machine learning models were developed using data from over 150,000 patients hospitalized for peripheral artery disease (PAD) to predict in-hospital mortality, showing an in-hospital death rate of 1.8%.

The models demonstrated strong predictive performance, with an area under the curve of 0.83 to 0.85, indicating their potential to identify high-risk patients and guide personalized treatment interventions.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Machine Learning Approach to Predict In-Hospital Mortality in Patients Admitted for Peripheral Artery Disease in the United States. Zhang, D., Li, Y., Kalbaugh, CA., et al.[2022]

A model-based billing code algorithm demonstrated high accuracy in identifying patients with peripheral arterial disease (PAD), achieving a c statistic of 0.91 in both training and validation datasets, indicating its effectiveness in clinical settings.

The simpler algorithm, while reasonably accurate for patients in the vascular laboratory, showed significantly lower sensitivity in a community-based sample, highlighting the importance of using more sophisticated methods for accurate PAD identification in broader populations.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Billing code algorithms to identify cases of peripheral artery disease from administrative data. Fan, J., Arruda-Olson, AM., Leibson, CL., et al.[2022]

References

1. United States pubmed.ncbi.nlm.nih.gov

Optimizing Best Practice Advisory alerts in electronic medical records with a multi-pronged strategy at a tertiary care hospital in Singapore. [2023]

2. United States pubmed.ncbi.nlm.nih.gov

Managing the alert process at NewYork-Presbyterian Hospital. [2018]

3. Canada pubmed.ncbi.nlm.nih.gov

A Large Language Model Screening Tool to Target Patients for Best Practice Alerts: Development and Validation. [2023]

4. England pubmed.ncbi.nlm.nih.gov

Tailoring of alerts substantially reduces the alert burden in computerized clinical decision support for drugs that should be avoided in patients with renal disease. [2017]

5. England pubmed.ncbi.nlm.nih.gov

Provider acceptance of an automated electronic alert for acute kidney injury. [2020]

6. England pubmed.ncbi.nlm.nih.gov

Assessing cardiovascular drug safety for clinical decision-making. [2022]

7. Germany pubmed.ncbi.nlm.nih.gov

Clinician Perceptions of Timing and Presentation of Drug-Drug Interaction Alerts. [2021]

8. Netherlands pubmed.ncbi.nlm.nih.gov

Failure of patients with peripheral arterial disease to accept the recommended treatment results in worse outcomes. [2022]

9. United States pubmed.ncbi.nlm.nih.gov

Mining peripheral arterial disease cases from narrative clinical notes using natural language processing. [2022]

10. United States pubmed.ncbi.nlm.nih.gov

Prospective decision analysis modeling indicates that clinical decisions in vascular surgery often fail to maximize patient expected utility. [2022]

11. England pubmed.ncbi.nlm.nih.gov

Machine Learning Approach to Predict In-Hospital Mortality in Patients Admitted for Peripheral Artery Disease in the United States. [2022]

12. England pubmed.ncbi.nlm.nih.gov

Billing code algorithms to identify cases of peripheral artery disease from administrative data. [2022]