Deprescribing, intentional medication discontinuation or dose reduction, can reduce potentially inappropriate medication use and medication-related harms. Engaging patients in deprescribing discussions may increase likelihood of deprescribing and promote shared decision-making. To examine the impact of patient-directed educational brochures on patient engagement and deprescribing discussions with primary care providers (PCPs). We mailed medication-specific brochures 2 weeks prior to each patient’s PCP appointment (4/12/2021–10/7/2022), followed by a mailed survey 2 weeks after scheduled PCP visits. Patients from three Veterans Affairs facilities with scheduled PCP appointments eligible for one of three medication-based cohorts (proton pump inhibitor, gabapentin, diabetes-hypoglycemia risk). Our primary outcome was patient-reported deprescribing discussions with their PCP (yes/no). Descriptive statistics characterized engagement with and reactions to the brochure. Multivariable logistic regression models determined associations of patient characteristics, attitudes, and brochure-engagement with reported deprescribing discussions. Adjusting only for patient characteristics, discussions were less likely if respondents were Black (vs. White: OR 0.47, 95

BACKGROUND:More than one in three older adults use potentially inappropriate medications (PIMs), and those with chronic conditions are more likely to be on multiple PIMs. Deprescribing, defined as stopping or dose-reducing a medication, can avoid harms from PIMs. Despite its benefits, deprescribing is rarely adopted into clinical practice. OBJECTIVES:We examined the effectiveness, sustainability, and safety of a patient-engagement strategy in a pragmatic trial at three facilities. METHODS:Subjects were mailed brochures for one of three PIMs (proton pump inhibitors, high-dose gabapentin, and diabetes agents with hypoglycemia risk) if they had chronic active prescriptions before a primary care provider (PCP) visit. Control subjects received usual care. RESULTS:There were 2448 patients in the control group (n = 2448) and 2480 patients in the implementation strategy cohort (n = 2480). In a mixed effect multivariable logistic regression model with PCP and site treated as random factors and controlling for patient and PCP characteristics, implementation strategy patients were significantly more likely to have deprescribing at 12-months compared with controls (odds ratio [OR] = 1.19; 95% confidence interval [CI] = 1.04, 1.35; p = 0.012). In a multinomial logistic regression model controlling for patient characteristics, implementation strategy patients were significantly more likely to exhibit sustained deprescribing (deprescribing at 6 and 12 months, OR = 1.32 [95% CI = 1.13, 1.55]) but not short-term (6 months only, OR = 0.96 [95% CI = 0.79, 1.16]) or delayed (12 months only, OR = 0.99 [95% CI = 0.84, 1.17]) deprescribing. There were five potential severe adverse drug withdrawals (0.2% incidence rate) possibly related to deprescribing. CONCLUSIONS:Despite low intensity, patient-directed education materials are an effective and safe implementation strategy to promote and sustain deprescribing. TRIAL REGISTRATION:ClinicalTrials.gov, NCT0429490, NCT04294901.

Introduction: Evidence has grown around the relationship between lipoprotein(a) [Lp(a)] and cardiovascular disease risk. The 2024 focused update on Lp(a) from the National Lipid Association recognizes elevated plasma Lp(a) as an important independent, causal risk factor for atherosclerotic cardiovascular disease and aortic valve stenosis, as well as delineates risk classifications. It is now recommended for patients to have Lp(a) measured at least once in their adult life, with Lp(a) levels <75 nmol/L (30 mg/dL) classified as low risk, individuals with Lp(a) levels ≥125 nmol/L (50 mg/dL) as high risk, and individuals with Lp(a) levels between 75 and 125 nmol/L (30–50 mg/dL) as intermediate risk. Aims: With Lp(a) lab testing recommendations changing and leading to a steady increase in patients who have had Lp(a) drawn, we aimed to understand these patients’ relative risk of cardiovascular comorbidities and events, in addition to current management. Methods: Within a university-based health system, a large data extraction model was used to identify all patients who had undergone Lp(a) testing. In a system of roughly 4,553,100 patients, 1,976 unique patients have had Lp(a) drawn since 2013. This population was then subclassified by risk groups, as outlined above. Results: Testing thus far, using new recommendations for classifications of patients, places 1491 patients in the low-risk category, 241 in the intermediate-risk category, and 211 in the high-risk category, summarized in Table 1. In this population, patients in the high risk Lp(a) group, tended to have higher rates of CAD, PAD, ACS and CVA events. Notably, no significant differences exist regarding more aggressive management of these patients with either escalation of lipid lowering therapy including high-intensity statins, ezetimibe and PCSK9-i or with subspecialist referrals. There are, however, 7 patients undergoing apheresis from the high-risk group. Conclusions: Consistent with current literature, patients stratified to the higher risk lp(a) group had higher rates of cardiovascular comorbid conditions and events, though do not seem to be managed more aggressively than those in lower risk groups. Opportunity exists in escalating therapy and referring to subspecialists for consideration of therapies like PCSK9-inhibitors, lipoprotein apheresis, and soon, direct Lp(a) lowering medications.

Introduction: Lower low-density lipoprotein (LDL) leads to greater cardiovascular disease risk reduction over the lifespan. Despite this, women are less likely to reach guideline-recommended LDL goals, perhaps due to bias in statin prescription or statin intensity. and disparities in adherence. There are known sex differences in cardiovascular disease and statin response, but little is known about how these differences manifest under real-world conditions. Aim: We aim to characterize long-term LDL between men and women who have been prescribed statin therapy as a function of age at initial statin therapy. Methods: We retrospectively identified patients within the UCHealth electronic health record among adults from 2012 to 2023 prescribed a statin with at least 3 LDL levels. Patients prescribed concomitant lipid-lowering agents were excluded. We performed a multivariate linear regression model to predict long-term average LDL between genders, adjusting for age at the time of statin prescription, race, ethnicity, BMI, interval from statin prescription, creatinine, and the following diagnoses: diabetes mellitus, ischemic heart disease, ischemic stroke, and smoking history. Results: Among 75,276 encounters, 14,902 patients had at least 3 LDL-C measurements and an interval statin initiation. Post-treatment LDL was obtained 212 (± 177) days after statin initiation. The cohort was 49% (7,303 patients) female, with 86% (12,819) white, 4% (648) African American, and 8% (1,249) Hispanic patients. At age 30, men started on statin therapy had a predicted LDL of 117 (95% CI: 114-121 mg/dL), whereas women began with a predicted LDL of 115 (95% CI: 112-119 mg/dL). At age 70, that value was 94.1 (95% CI: 93.4-94.8 mg/dL) among men vs 105.6 (95% CI: 104.9-106.3 mg/dL) in women (Figure). Discussion: In this single-center, retrospective observational study, we found that women had a higher predicted LDL compared to men across the lifespan, with a greater difference in those whose first prescription was at an older age. More work is needed to determine if this reflects less aggressive treatment of women than men. Further research needs to be done to bridge this gap in therapy.

ImportancePatient-directed educational materials are a promising implementation strategy to expand deprescribing reach and adoption, but little is known about the impact across medication groups with potentially different perceived risks.ObjectiveTo examine the impact of a patient-directed education intervention on clinician deprescribing of potentially low-benefit (proton pump inhibitors) or high-risk medications (high-dose gabapentin, diabetes agents with hypoglycemia risks).Design, Setting, and ParticipantsThis pragmatic multisite nonrandomized clinical trial took place at 3 geographically distinct US Veterans Affairs (VA) medical centers from April 2021 to October 2022. The total study sample was composed of the intervention cohort and the historical control cohort cared for by 103 primary care practitioners (PCPs).InterventionThe primary intervention component was a medication-specific brochure, mailed during the intervention time frame to all eligible patients 2 to 3 weeks prior to upcoming primary care appointments. Patients seen by the same PCPs at the same sites 1 year prior to the study intervention served as controls.Main Outcome and MeasuresThe primary binary outcome variable was deprescribing 6 months after the intervention, defined as complete cessation or any dose reduction of the target medication using VA pharmacy dispensing data.ResultsThe total study sample included 5071 patients. The overall rate of deprescribing among the intervention cohort (n = 2539) was 29.5% compared with 25.8% among the controls (n = 2532). In an unadjusted model, the intervention cohort was statistically significantly more likely to have deprescribing (odds ratio [OR], 1.17 [95% CI, 1.03-1.33]; P = .02). In a multivariable logistic regression model nesting patients within PCPs within sites and controlling for patient and PCP characteristics, the odds of deprescribing in the intervention cohort were 1.21 times that of the control cohort (95% CI, 1.05-1.38; P = .008). The difference in deprescribing prevalence between the intervention and control cohorts (proton pump inhibitors: 29.4% vs 25.4%; gabapentin: 40.2% vs 36.2%; hypoglycemia risk: 27.3% vs 25.1%) did not statistically significantly differ by medication group (P = .90).Conclusion and RelevanceThis nonrandomized clinical trial found that patient-directed educational materials provided prior to scheduled primary care appointments can effectively promote deprescribing for potentially low-benefit and high-risk medication groups.Trial RegistrationClinicalTrials.gov Identifier: NCT0429490

Importance:Patient-directed educational materials are a promising implementation strategy to expand deprescribing reach and adoption, but little is known about the impact across medication groups with potentially different perceived risks. Objective:To examine the impact of a patient-directed education intervention on clinician deprescribing of potentially low-benefit (proton pump inhibitors) or high-risk medications (high-dose gabapentin, diabetes agents with hypoglycemia risks). Design, Setting, and Participants:This pragmatic multisite nonrandomized clinical trial took place at 3 geographically distinct US Veterans Affairs (VA) medical centers from April 2021 to October 2022. The total study sample was composed of the intervention cohort and the historical control cohort cared for by 103 primary care practitioners (PCPs). Intervention:The primary intervention component was a medication-specific brochure, mailed during the intervention time frame to all eligible patients 2 to 3 weeks prior to upcoming primary care appointments. Patients seen by the same PCPs at the same sites 1 year prior to the study intervention served as controls. Main Outcome and Measures:The primary binary outcome variable was deprescribing 6 months after the intervention, defined as complete cessation or any dose reduction of the target medication using VA pharmacy dispensing data. Results:The total study sample included 5071 patients. The overall rate of deprescribing among the intervention cohort (n = 2539) was 29.5% compared with 25.8% among the controls (n = 2532). In an unadjusted model, the intervention cohort was statistically significantly more likely to have deprescribing (odds ratio [OR], 1.17 [95% CI, 1.03-1.33]; P = .02). In a multivariable logistic regression model nesting patients within PCPs within sites and controlling for patient and PCP characteristics, the odds of deprescribing in the intervention cohort were 1.21 times that of the control cohort (95% CI, 1.05-1.38; P = .008). The difference in deprescribing prevalence between the intervention and control cohorts (proton pump inhibitors: 29.4% vs 25.4%; gabapentin: 40.2% vs 36.2%; hypoglycemia risk: 27.3% vs 25.1%) did not statistically significantly differ by medication group (P = .90). Conclusion and Relevance:This nonrandomized clinical trial found that patient-directed educational materials provided prior to scheduled primary care appointments can effectively promote deprescribing for potentially low-benefit and high-risk medication groups. Trial Registration:ClinicalTrials.gov Identifier: NCT0429490.

Drug-induced QT prolongation (diLQTS), and subsequent risk of torsade de pointes, is a major concern with use of many medications, including for non-cardiac conditions. The possibility that genetic risk, in the form of polygenic risk scores (PGS), could be integrated into prediction of risk of diLQTS has great potential, although it is unknown how genetic risk is related to clinical risk factors as might be applied in clinical decision-making. In this study, we examined the PGS for QT interval in 2500 subjects exposed to a known QT-prolonging drug on prolongation of the QT interval over 500ms on subsequent ECG using electronic health record data. We found that the normalized QT PGS was higher in cases than controls (0.212±0.954 vs. -0.0270±1.003, P = 0.0002), with an unadjusted odds ratio of 1.34 (95%CI 1.17-1.53, P<0.001) for association with diLQTS. When included with age and clinical predictors of QT prolongation, we found that the PGS for QT interval provided independent risk prediction for diLQTS, in which the interaction for high-risk diagnosis or with certain high-risk medications (amiodarone, sotalol, and dofetilide) was not significant, indicating that genetic risk did not modify the effect of other risk factors on risk of diLQTS. We found that a high-risk cutoff (QT PGS ≥ 2 standard deviations above mean), but not a low-risk cutoff, was associated with risk of diLQTS after adjustment for clinical factors, and provided one method of integration based on the decision-tree framework. In conclusion, we found that PGS for QT interval is an independent predictor of diLQTS, but that in contrast to existing theories about repolarization reserve as a mechanism of increasing risk, the effect is independent of other clinical risk factors. More work is needed for external validation in clinical decision-making, as well as defining the mechanism through which genes that increase QT interval are associated with risk of diLQTS.

BACKGROUND:The U.S. Department of Veterans Affairs (VA) is undergoing an enterprise-wide transition from a homegrown electronic health record (EHR) system to a commercial off-the-shelf product. Because of the far-reaching effects of the EHR transformation through all aspects of the healthcare system, VA Health Services Research and Development identified a need to develop a research agenda that aligned with health system priorities so that work may inform evidence-based improvements in implementation processes and outcomes. OBJECTIVE:The purpose of this paper is to report on the development of a research agenda designed to optimize the EHR transition processes and implementation outcomes in a large, national integrated delivery system. DESIGN:We used a sequential mixed-methods approach (portfolio assessment, literature review) combined with multi-level stakeholder engagement approach that included research, informatics, and healthcare operations experts in EHR transitions in and outside the VA. Data from each stage were integrated iteratively to identify and prioritize key research areas within and across all stakeholder groups. PARTICIPANTS:VA informatics researchers, regional VA health system leaders, national VA program office leaders, and external informatics experts with EHR transition experience. KEY RESULTS:Through three rounds of stakeholder engagement, priority research topics were identified that focused on operations, user experience, patient safety, clinical outcomes, value realization, and informatics innovations. CONCLUSIONS:The resulting EHR-focused research agenda was designed to guide development and conduct of rigorous research evidence aimed at providing actionable results to address the needs of operations partners, clinicians, clinical staff, patients, and other stakeholders. Continued investment in research and evaluation from both research and operations divisions of VA will be critical to executing the research agenda, ensuring its salience and value to the health system and its end users, and ultimately realizing the promise of this EHR transition.

Screening, brief intervention, and referral for alcohol misuse during primary care appointments is recommended to address high rates of unhealthy alcohol use. However, implementation of screening and referral practices into primary care remains difficult. Computerized Relational Agents programmed to provide alcohol screening, brief intervention, and referral can effectively reduce the burden on clinical staff by increasing screening practices. As part of a larger clinical trial, we aimed to solicit input from patients about the design and development of a Relational Agent for alcohol brief intervention. We also solicited input from patients who interacted with the implemented version of the Relational Agent intervention after they finished the trial. A two-part development and evaluation study was conducted. To begin, a user-centered design approach was used to customize the intervention for the population served. A total of 19 participants shared their preferences on the appearance, setting, and preferences of multiple Relational Agents through semi-structured interviews. Following the completion of the study one interviews, a Relational Agent was chosen and refined for use in the intervention. In study two, twenty participants who participated in the clinical trial intervention were invited back to participate in a semi-structured interview to provide feedback about their experiences in interacting with the intervention. Study one results showed that participants preferred a female Relational Agent located in an office-like setting, but the mechanical and still movements of the Relational agent decreased feelings of authenticity and human trustworthiness for participants. After refinements to the Relational Agent, post-intervention results in study two showed that participants ( n = 17, 89%) felt comfortable interacting and discussing their drinking habits with the Relational Agent and participants ( n = 10, 53%) believed that the intervention had a positive impact on the way participants thought about drinking or on their actual drinking habits. Despite variability in the preferences of participants during the development stage of the intervention, incorporating the feedback of participants during the design process resulted in optimized comfort levels for individuals interacting with the Relational Agent. clinicaltrials.gov, NCT02030288, https://clinicaltrials.gov/ct2/home

BACKGROUND:Many machine learning approaches are limited to classification of outcomes rather than longitudinal prediction. One strategy to use machine learning in clinical risk prediction is to classify outcomes over a given time horizon. However, it is not well-known how to identify the optimal time horizon for risk prediction.OBJECTIVE:In this study, we aim to identify an optimal time horizon for classification of incident myocardial infarction (MI) using machine learning approaches looped over outcomes with increasing time horizons. Additionally, we sought to compare the performance of these models with the traditional Framingham Heart Study (FHS) coronary heart disease gender-specific Cox proportional hazards regression model.METHODS:We analyzed data from a single clinic visit of 5201 participants of a cardiovascular health study. We examined 61 variables collected from this baseline exam, including demographic and biologic data, medical history, medications, serum biomarkers, electrocardiographic, and echocardiographic data. We compared several machine learning methods (eg, random forest, L1 regression, gradient boosted decision tree, support vector machine, and k-nearest neighbor) trained to predict incident MI that occurred within time horizons ranging from 500-10,000 days of follow-up. Models were compared on a 20% held-out testing set using area under the receiver operating characteristic curve (AUROC). Variable importance was performed for random forest and L1 regression models across time points. We compared results with the FHS coronary heart disease gender-specific Cox proportional hazards regression functions.RESULTS:There were 4190 participants included in the analysis, with 2522 (60.2%) female participants and an average age of 72.6 years. Over 10,000 days of follow-up, there were 813 incident MI events. The machine learning models were most predictive over moderate follow-up time horizons (ie, 1500-2500 days). Overall, the L1 (Lasso) logistic regression demonstrated the strongest classification accuracy across all time horizons. This model was most predictive at 1500 days follow-up, with an AUROC of 0.71. The most influential variables differed by follow-up time and model, with gender being the most important feature for the L1 regression and weight for the random forest model across all time frames. Compared with the Framingham Cox function, the L1 and random forest models performed better across all time frames beyond 1500 days.CONCLUSIONS:In a population free of coronary heart disease, machine learning techniques can be used to predict incident MI at varying time horizons with reasonable accuracy, with the strongest prediction accuracy in moderate follow-up periods. Validation across additional populations is needed to confirm the validity of this approach in risk prediction.

Background Drug-induced long-QT syndrome (diLQTS) is a major concern among patients who are hospitalized, for whom prediction models capable of identifying individualized risk could be useful to guide monitoring. We have previously demonstrated the feasibility of machine learning to predict the risk of diLQTS, in which deep learning models provided superior accuracy for risk prediction, although these models were limited by a lack of interpretability. Objective In this investigation, we sought to examine the potential trade-off between interpretability and predictive accuracy with the use of more complex models to identify patients at risk for diLQTS. We planned to compare a deep learning algorithm to predict diLQTS with a more interpretable algorithm based on cluster analysis that would allow medication- and subpopulation-specific evaluation of risk. Methods We examined the risk of diLQTS among 35,639 inpatients treated between 2003 and 2018 with at least 1 of 39 medications associated with risk of diLQTS and who had an electrocardiogram in the system performed within 24 hours of medication administration. Predictors included over 22,000 diagnoses and medications at the time of medication administration, with cases of diLQTS defined as a corrected QT interval over 500 milliseconds after treatment with a culprit medication. The interpretable model was developed using cluster analysis (K=4 clusters), and risk was assessed for specific medications and classes of medications. The deep learning model was created using all predictors within a 6-layer neural network, based on previously identified hyperparameters. Results Among the medications, we found that class III antiarrhythmic medications were associated with increased risk across all clusters, and that in patients who are noncritically ill without cardiovascular disease, propofol was associated with increased risk, whereas ondansetron was associated with decreased risk. Compared with deep learning, the interpretable approach was less accurate (area under the receiver operating characteristic curve: 0.65 vs 0.78), with comparable calibration. Conclusions In summary, we found that an interpretable modeling approach was less accurate, but more clinically applicable, than deep learning for the prediction of diLQTS. Future investigations should consider this trade-off in the development of methods for clinical prediction.

The development of the acetylcholinesterase (AChE) texture of the cortex of the rat brain was studied during the first three weeks of life. The Tago technique enables visualization of both AChE+ cells and fibers with both shown in exquisite detail making quantification possible. At each age--0 (birth), 7, 14, 21 and 60 days (adult)--four brain areas were studied (cingulate, dorsal neocortex, lateral neocortex and olfactory) at each of three coronal planes in the brain (anterior, intermediate, posterior). Fiber density reached adult levels by Day 21 in cingulate cortex in intermediate and posterior planes. In other areas fiber density reached adult levels by Day 14 indicating a high rate of fiber growth during the first two weeks of life since at birth rat cortex is innervated only by a sparse AChE+ fiber invasion into neocortex in the anterior plane. Fiber density did not regress after adult levels were reached, however, cell staining showed a different pattern. At birth many lightly stained cells were seen in the olfactory cortex in all three planes, but other areas were devoid of cells. In all areas there was a peak at Day 7 in number of cells stained and in intensity of cells staining with a gradual decline in cell staining until by Day 21 very few stained cells were seen in the cortex (typical adult pattern).

Background Medication discrepancies can lead to adverse drug events and patient harm. Medication reconciliation is a process intended to reduce medication discrepancies. We developed a Secure Messaging for Medication Reconciliation Tool (SMMRT), integrated into a web-based patient portal, to identify and reconcile medication discrepancies during transitions from hospital to home. Objective We aimed to characterize patients’ perceptions of the ease of use and effectiveness of SMMRT. Methods We recruited 20 participants for semistructured interviews from a sample of patients who had participated in a randomized controlled trial of SMMRT. Interview transcripts were transcribed and then qualitatively analyzed to identify emergent themes. Results Although most patients found SMMRT easy to view at home, many patients struggled to return SMMRT through secure messaging to clinicians due to technology-related barriers. Patients who did use SMMRT indicated that it was time-saving and liked that they could review it at their own pace and in the comfort of their own home. Patients reported SMMRT was effective at clarifying issues related to medication directions or dosages and that SMMRT helped remove medications erroneously listed as active in the patient’s electronic health record. Conclusions Patients viewed SMMRT utilization as a positive experience and endorsed future use of the tool. Veterans reported SMMRT is an effective tool to aid patients with medication reconciliation. Adoption of SMMRT into regular clinical practice could reduce medication discrepancies while increasing accessibility for patients to help manage their medications. Trial Registration ClinicalTrials.gov NCT02482025; https://clinicaltrials.gov/ct2/show/NCT02482025

Background Deprescribing, or the intentional discontinuation or dose-reduction of medications, is an approach to reduce harms associated with inappropriate medication use. We sought to determine how direct-to-patient educational materials impacted patient-provider discussion about and deprescribing of potentially inappropriate medications. Methods We conducted a pre-post pilot trial, using an historical control group, at an urban VA medical center. We included patients in one of two cohorts: 1) chronic proton pump inhibitor users (PPI), defined as use of any dose for 90 consecutive days, or 2) patients at hypoglycemia risk, defined by diabetes diagnosis; prescription for insulin or sulfonylurea; hemoglobin A1c < 7%; and age ≥ 65 years, renal insufficiency, or cognitive impairment. The intervention consisted of mailing medication-specific patient-centered EMPOWER (Eliminating Medications Through Patient Ownership of End Results) brochures, adapted to a Veteran patient population, two weeks prior to scheduled primary care appointments. Our primary outcome – deprescribing – was defined as clinical documentation of target medication discontinuation or dose-reduction. Our secondary outcome was documentation of a discussion about the target medication (yes/possible vs. no/absent). Covariates included age, sex, race, specified comorbidities, medications, and utilization. We used chi-square tests to examine the association of receiving brochures with each outcome. Results The 348 subjects (253 intervention, 95 historical control) were primarily age ≥ 65 years, white, and male. Compared to control subjects, intervention subjects were more likely to have deprescribing (36 [14.2%] vs. 4 [4.2%], p = 0.009) and discussions about the target medication (31 [12.3%] vs. 1 [1.1%], p = 0.001). Conclusions Targeted mailings of EMPOWER brochures temporally linked to a scheduled visit in primary care clinics are a low-cost, low-technology method associated with increases in both deprescribing and documentation of patient-provider medication discussions in a Veteran population. Leveraging the potential for patients to initiate deprescribing discussions within clinical encounters is a promising strategy to reduce drug burden and decrease adverse drug effects and harms.

Background Despite the numerous studies evaluating various rhythm control strategies for atrial fibrillation (AF), determination of the optimal strategy in a single patient is often based on trial and error, with no one-size-fits-all approach based on international guidelines/recommendations. The decision, therefore, remains personal and lends itself well to help from a clinical decision support system, specifically one guided by artificial intelligence (AI). QRhythm utilizes a 2-stage machine learning (ML) model to identify the optimal rhythm management strategy in a given patient based on a set of clinical factors, in which the model first uses supervised learning to predict the actions of an expert clinician and identifies the best strategy through reinforcement learning to obtain the best clinical outcome—a composite of symptomatic recurrence, hospitalization, and stroke. Objective We qualitatively evaluated a novel, AI-based, clinical decision support system (CDSS) for AF rhythm management, called QRhythm, which uses both supervised and reinforcement learning to recommend either a rate control or one of 3 types of rhythm control strategies—external cardioversion, antiarrhythmic medication, or ablation—based on individual patient characteristics. Methods Thirty-three clinicians, including cardiology attendings and fellows and internal medicine attendings and residents, performed an assessment of QRhythm, followed by a survey to assess relative comfort with automated CDSS in rhythm management and to examine areas for future development. Results The 33 providers were surveyed with training levels ranging from resident to fellow to attending. Of the characteristics of the app surveyed, safety was most important to providers, with an average importance rating of 4.7 out of 5 (SD 0.72). This priority was followed by clinical integrity (a desire for the advice provided to make clinical sense; importance rating 4.5, SD 0.9), backward interpretability (transparency in the population used to create the algorithm; importance rating 4.3, SD 0.65), transparency of the algorithm (reasoning underlying the decisions made; importance rating 4.3, SD 0.88), and provider autonomy (the ability to challenge the decisions made by the model; importance rating 3.85, SD 0.83). Providers who used the app ranked the integrity of recommendations as their highest concern with ongoing clinical use of the model, followed by efficacy of the application and patient data security. Trust in the app varied; 1 (17%) provider responded that they somewhat disagreed with the statement, “I trust the recommendations provided by the QRhythm app,” 2 (33%) providers responded with neutrality to the statement, and 3 (50%) somewhat agreed with the statement. Conclusions Safety of ML applications was the highest priority of the providers surveyed, and trust of such models remains varied. Widespread clinical acceptance of ML in health care is dependent on how much providers trust the algorithms. Building this trust involves ensuring transparency and interpretability of the model.

Medication discrepancies, defined as unintended differences between medication lists,1 occur in up to 60% of patients' electronic health records (EHRs).2-4 They are associated with adverse drug events and increased healthcare utilization5-8; thus, the Joint Commission recommends medication reconciliation to resolve discrepancies between patient-reported medications and those documented within the record.9 Medication reconciliation tools to identify and correct discrepancies can provide patient safety and cost-related benefits.7 Discrepancy types within classification systems differ, but common types include Commission (i.e., medication present in EHR but patient not taking), Omission (i.e., medication absent from EHR), and Drug-dose (i.e., dose missing or incorrect).10 The Medication Discrepancy Taxonomy (MedTax) is a universal classification system developed to advance research and evaluations to reduce discrepancies.1 MedTax was designed and validated for use by pharmacists only. Exclusively relying on pharmacists' or clinicians' expertise to classify medication discrepancies may be limited by costs and time availability. Using nonclinicians in medication reconciliation processes could provide many benefits, especially in research and quality improvement (QI) initiatives. However, little is known about the role of nonclinical personnel in the classification subprocess. We determined if research assistants (RAs) without formal clinical education can classify identified medication discrepancies with accuracy comparable to pharmacists. A dataset containing 1024 discrepancies (e.g., omissions, commissions) was derived from medication lists of 179 patients (5.7 discrepancies/patient) collected during care transitions as part of a larger trial from December 2019 to October 2020.11 All study procedures were approved by the VA Boston Healthcare System Institutional Review Board. Three coders participated in the classification process. The pharmacist had a Doctor of Pharmacy degree and was completing a postgraduate pharmacy residency. There were two RAs; RA-1 had a Master of Arts in Psychology, and RA-2 had a Master of Public Health. We used a modified version of the validated MedTax1 to classify potential medication discrepancies. To modify MedTax, two physicians (Steven R. Simon and Amy M. Linsky) used clinical and content expertise to select medication discrepancy types most applicable to the healthcare setting of the study (see Text Box 1). Drug omission Drug commission (or addition) Drug duplication The Discrepancy in the strength and/or frequency and/or number of units of dosage form and/or total daily dose Computer system expiration Discrepancy in the dosage, form, and/or route of administration Other No discrepancy A 2-h training session on the application of the modified MedTax was conducted using 79 medication discrepancies. Participants included three coders: pharmacist (Megan Nowak) and RAs (Kate Yeksigian and Julianne E. Brady) – and two physicians (Steven R. Simon and Amy M. Linsky). During Round 1, all five participants reviewed 25 discrepancies sequentially and collaboratively discussed each classification. After Round 1, the pharmacists, RAs, and physicians independently classified the remaining 54 discrepancies in five training rounds (Rounds 2–6; 13.5 discrepancies/round; range 7–14). Classification between the three coders and the two physicians (i.e., the “gold standard”) was assessed. Discordant classifications were discussed as a group, with the physicians providing additional guidance. The training aimed to achieve acceptable a priori agreement between the RAs and pharmacist (>75%).12 After the training, the pharmacist and each RA independently coded the remaining 945 discrepancies. The pharmacist reviewed all discrepancies (n = 945), and the RAs divided the discrepancies for review (RA-1: n = 516; RA-2: n = 429). To determine the agreement between the pharmacist and two RAs, Krippendorff's α was calculated and interpreted using the following benchmarks of agreement: 0–0.667: poor; 0.668–0.799: possibly acceptable; and 0.800–1: acceptable.13 To assess individual coder accuracy, Cohen's κ was calculated to compare each of the pharmacist and RA coders to the gold standard and interpreted using the following benchmarks of agreement: 0–0.20: poor; 0.21–0.40: fair; 0.41–0.60: moderate; 0.61–0.80: substantial; and 0.81–1.00: almost perfect.14 Percent agreement was also calculated. In the primary analysis, we calculated Krippendorff's α to measure interrater reliability (IRR) between the three coders. In secondary analyses, we calculated Cohen's κ and percent agreement to assess the IRR of each RA to the pharmacist. In supplemental analyses, all IRR statistics for each discrepancy type were calculated. All analyses were conducted using R 4.015 and Stata 16.16 The RAs and pharmacist showed moderate agreement with each other during the training rounds (α = 0.73; 95% confidence interval [CI]: 0.59, 0.83). Across all three coders, the overall percent agreement was 69.8%, just below the a priori goal of 75%. In the primary analysis, RAs classified discrepancies using the modified MedTax with acceptable accuracy compared to the pharmacist (945 discrepancies, α = 0.81, 95% CI: 0.78–0.84). In secondary analyses, both RAs individually coded with acceptable performance compared to the pharmacist, with almost perfect concordance between the pharmacist and both RA-1 (516 discrepancies; ƙ = 0.82, 95% CI: 0.79–0.86; 86.6% agreement) and RA-2 (429 discrepancies; ƙ = 0.80, 95% CI: 0.75–0.84; 84.6% agreement). In supplemental analyses, the pharmacist and RAs classified five of eight (63%) MedTax medication discrepancy types with acceptable concordance: Omission (α = 0.86), Commission (α = 0.83), Duplication (α = 0.92), Strength (α = 0.85), and Computer System Expiration (α = 0.92). The pharmacist and RAs had possibly acceptable concordance classifying Drug Dose/Form/Route (α = 0.67) and poor concordance classifying No Discrepancy (α = 0.65) and Other discrepancies (α = 0.31). A post hoc review of the discordance classifying Other discrepancies found these were all classified as Other by the pharmacist and omission by the RAs. In post hoc blinded physician adjudication of all discordant classifications, the physician agreed with RA classifications for 100% (12/12) of the Other disagreements and 74% (71/96) of the No Discrepancy disagreements. RAs without formal clinical education can successfully classify medication discrepancies with accuracy comparable to a clinical pharmacist. These findings that RAs can be trained to reliably use a modified medication discrepancy classification taxonomy (MedTax) support engagement of nonclinical individuals to classify discrepancies within research and QI settings (i.e., nonclinical settings) where frequency and types of medication discrepancies are often used to assess the effectiveness of medication reconciliation interventions.17 While direct contribution by clinical pharmacists in such efforts is beneficial, competing clinical responsibilities18 may preclude their involvement and require identification of nonclinical individuals (e.g., RAs) to perform specific aspects of medication reconciliation. Employing clinical pharmacists in research and QI projects is difficult due to direct and opportunity costs19, 20 Cost and time savings, combined with our finding that RA classification is reliable and valid, would yield even greater net benefit when combined with the clinical value of high-quality medication reconciliation.21, 22 The RAs classified five of eight discrepancy types (Omission, Commission, Duplication, Strength, Computer System Expiration) with near perfect concordance. Classifying changes in Dosage Form or Route of Administration may be more complex and require additional training. However, post hoc analyses supported an RA's ability to accurately classify these two discrepancy types, despite lower concordance with the pharmacist. Findings should be interpreted in the context of the following limitations. There were only a small number of individuals; findings should be replicated including coders with more diverse clinical and nonclinical backgrounds. The pharmacist was completing a pharmacy residency; one with more professional experience may yield different findings. This study evaluated the ability to train RAs on one classification system. Nevertheless, the modified taxonomy was based upon the validated MedTax and included classifications for the most common medication discrepancies. Training RAs to classify medication discrepancies reliably has many benefits in both research and QI settings. Greater availability and lower overall cost of individuals without formal clinical education optimizes the use of limited time and financial resources while maintaining confidence that nonclinical personnel can be trained to classify previously identified medication discrepancies. Julianne E. Brady: Data curation and writing – original draft preparation. Steven R. Simon: Funding acquisition, conceptualization, methodology, and writing – reviewing and editing. Kate Yeksigian: Data curation, and writing – reviewing and editing. Alan J. Zillich: Conceptualization, methodology, and writing – reviewing and editing. Jonathan Moyer: Formal analysis, and writing – reviewing and editing. Amy M. Linsky: Conceptualization, methodology, and writing – original draft preparation. All authors have read and approved the final version of the manuscript. The authors would like to thank late Megan Nowak, PharmD (1994–2021) for her work on the data coding and analysis. This study was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Services (IIR 14-059; PI Steven R. Simon). The authors declare no conflict of interest. The lead author Julianne E. Brady affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. Julianne E. Brady had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis.

Background and Objectives: During its monthly morbidity and mortality conference (MMC), the University of Colorado Division of Cardiology reviewed a "near-miss" patient safety event involving the delayed completion of a Stat-priority (ie, statim, meaning high priority) electrocardiogram (ECG). Because critical and interprofessional stakeholders participated in the conference, we hypothesized that the MMC would be associated with reduced ECG completion times. Methods: Data were collected for in-hospital ECGs performed at the University of Colorado Hospital between January 1, 2017, and June 30, 2018. An interrupted time series analysis was used to estimate the immediate and ongoing impact of the MMC (held on February 28, 2018) on ECG completion times, stratified by order priority (Stat, Now, or Routine). The percentage of delayed Stat-priority ECGs was analyzed as a secondary outcome. Results: Before the MMC, ECG completion times were stable for all order priorities (P > .2), but the proportion of delayed Stat-priority ECGs increased from 5% in January 2017 to 20% in February 2018 (P < .01). The MMC was associated with an immediate reduction in average daily ECG completion times for Routine (-18.4 minutes, P = .03) and Now (-8 minutes, P = .024) priority ECGs. No reduction was seen for Stat ECGs (P = .97), though the percentage of delayed Stat ECGs stopped increasing (P = .63). In the post-MMC period, completion times for Routine-priority ECGs increased and approached pre-MMC levels. Conclusions: The MMC was associated with an immediate, but temporary, improvement in ECG completion times. Although the observed clinical benefit of the MMC is novel, these data support the need for more durable reforms to sustain initial improvements.

Abstract Cardiovascular disease is the leading cause of death globally. While pharmacological advancements have improved the morbidity and mortality associated with cardiovascular disease, non-adherence to prescribed treatment remains a significant barrier to improved patient outcomes. A variety of strategies to improve medication adherence have been tested in clinical trials, and include the following categories: improving patient education, implementing medication reminders, testing cognitive behavioral interventions, reducing medication costs, utilizing healthcare team members, and streamlining medication dosing regimens. In this review, we describe specific trials within each of these categories and highlight the impact of each on medication adherence. We also examine ongoing trials and future lines of inquiry for improving medication adherence in patients with cardiovascular diseases.

Background: Drug-induced QT prolongation is a potentially preventable cause of morbidity and mortality, however there are no widespread clinical tools utilized to predict which individuals are at greatest risk. Machine learning (ML) algorithms may provide a method for identifying these individuals, and could be automated to directly alert providers in real time. Objective: This study applies ML techniques to electronic health record (EHR) data to identify an integrated risk-prediction model that can be deployed to predict risk of drug-induced QT prolongation. Methods: We examined harmonized data from the UCHealth EHR and identified inpatients who had received a medication known to prolong the QT interval. Using a binary outcome of the development of a QTc interval >500 ms within 24 hours of medication initiation or no ECG with a QTc interval >500 ms, we compared multiple machine learning methods by classification accuracy and performed calibration and rescaling of the final model. Results: We identified 35,639 inpatients who received a known QT-prolonging medication and an ECG performed within 24 hours of administration. Of those, 4,558 patients developed a QTc > 500 ms and 31,081 patients did not. A deep neural network with random oversampling of controls was found to provide superior classification accuracy (F1 score 0.404; AUC 0.71) for the development of a long QT interval compared with other methods. The optimal cutpoint for prediction was determined and was reasonably accurate (sensitivity 71%; specificity 73%). Conclusions: We found that deep neural networks applied to EHR data provide reasonable prediction of which individuals are most susceptible to drug-induced QT prolongation. Future studies are needed to validate this model in novel EHRs and within the physician order entry system to assess the ability to improve patient safety.

BACKGROUND:Unintentional medication discrepancies due to inadequate medication reconciliation pose a threat to patient safety. Skilled nursing facilities (SNFs) are an important care setting where patients are vulnerable to unintentional medication discrepancies due to increased medical complexity and care transitions. This study describes a quality improvement (QI) approach to improve medication reconciliation in an SNF setting as part of the Multi-Center Medication Reconciliation Quality Improvement Study 2 (MARQUIS2).METHODS:This study was conducted at a 112-bed US Department of Veterans Affairs SNF. The researchers used several QI methods, including data benchmarking, stakeholder surveys, process mapping, and a Healthcare Failure Mode and Effect Analysis (HFMEA) to complete comprehensive baseline assessments.RESULTS:Baseline assessments revealed that medication reconciliation processes were error-prone, with high rates of medication discrepancies. Provider surveys and process mapping revealed extremely labor-intensive and highly complex processes lacking standardization. Factors contributing were polypharmacy, limited resources, electronic health record limitations, and patient exposure to multiple care transitions. HFMEA enabled a methodical approach to identify and address challenges. The team validated the best possible medication history (BPMH) process for hospital settings as outlined by MARQUIS2 for the SNF setting and found it necessary to use additional medication lists to account for multiple care transitions.CONCLUSION:SNFs represent a critical setting for medication reconciliation efforts due to challenges completing the reconciliation process and the concomitant high risk of adverse drug events in this population. Initial baseline assessments effectively identified existing problems and can be used to guide targeted interventions.