Publications

Sankey diagram of agreement between dischareg summary, discharge instructions, and patient provided reasoning for chronic medication changes made during hospitalization.

BACKGROUND: Opioids are a widely prescribed class of medication for pain management. However, they have variable efficacy and adverse effects among patients, due to the complex interplay between biological and clinical factors. Pharmacogenetic testing can be used to match patients' genetic profiles to individualize opioid therapy, improving pain relief and reducing the risk of adverse effects. Despite its potential, the pharmacogenetic testing uptake (use of pharmacogenetic testing) remains low due to a range of barriers at the patient, health care provider, infrastructure, and financial levels. Since testing typically involves a shared decision between the provider and patient, predicting the likelihood of a patient undergoing pharmacogenetic testing and understanding the factors influencing that decision can help optimize resource use and improve outcomes in pain management.

OBJECTIVE: This study aimed to develop machine learning (ML) models, identifying patients' likelihood of pharmacogenetic uptake based on their demographics, clinical variables, medication use, and social determinants of health.

METHODS: We used electronic health record data from a single center health care system to identify patients prescribed opioids. We extracted patients' demographics, clinical variables, medication use, and social determinants of health, and developed and validated ML models, including a neural network, logistic regression, random forest, extreme gradient boosting (XGB), naïve Bayes, and support vector machines for pharmacogenetic testing uptake prediction based on procedure codes. We performed 5-fold cross-validation and created an ensemble probability-based classifier using the best-performing ML models for pharmacogenetic testing uptake prediction. Various performance metrics, uptake stratification analysis, and feature importance analysis were used to evaluate the performance of the models.

RESULTS: The ensemble model using XGB and support vector machine-radial basis function classifiers had the highest C-statistics at 79.61%, followed by XGB (78.94%), and neural network (78.05%). While XGB was the best-performing model, the ensemble model achieved a high accuracy (32,699/48,528, 67.38%), recall (537/702, 76.50%), specificity (32,162/47,826, 67.25%), and negative predictive value (32,162/32,327, 99.49%). The uptake stratification analysis using the ensemble model indicated that it can effectively distinguish across uptake probability deciles, where those in the higher strata are more likely to undergo pharmacogenetic testing in the real world (320/4853, 6.59% in the highest decile compared to 6/4853, 0.12% in the lowest). Furthermore, Shapley Additive Explanations value analysis using the XGB model indicated age, hypertension, and household income as the most influential factors for pharmacogenetic testing uptake prediction.

CONCLUSIONS: The proposed ensemble model demonstrated a high performance in pharmacogenetic testing uptake prediction among patients using opioids for pain. This model can be used as a decision support tool, assisting clinicians in identifying patients' likelihood of pharmacogenetic testing uptake and guiding appropriate decision-making.

BACKGROUND: Transitions of care (ToC) programs are important for patient safety, but their implementation and success remain highly variable across US hospitals, particularly for patients with multimorbidity and health-related social needs (HRSNs). Hospitalists, as key decision-makers at discharge, encounter firsthand the factors that hinder the success of ToC programs.

OBJECTIVE: To explore hospitalists' perspectives on successes, shortcomings, and implementation barriers in ToC programs, particularly during transitions from hospital to community settings.

METHODS: Rapid qualitative study featuring virtual focus groups with participants from the Hospital Medicine Reengineering Network (HOMERuN). Data were analyzed using a mixed inductive-deductive framework to identify key themes.

RESULTS: Twenty-two individuals from 19 different organizations participated in focus groups. None of the organizations offered comprehensive ToC programs to all patients. Four major themes emerged: (1) Diagnosis-specific ToC programs are effective but contribute to care fragmentation, particularly for patients with multimorbidity; (2) postdischarge follow-up is hindered by limited appointment availability, insurance barriers, and geographic challenges; (3) ToC programs often fail to address patient preferences, HRSNs, and health literacy, and lack adequate resources and leadership support; (4) successful programs require institutional commitment, dedicated funding, interprofessional collaboration, and community engagement. Participants emphasized the need to prioritize patient-centered care over financial return on investment.

CONCLUSIONS: Current ToC programs are fragmented, undermining safe and equitable transitions. Addressing HRSNs, fostering leadership support, and prioritizing patient-centered care over short-term financial metrics are essential for improving ToC outcomes.

BACKGROUND: The Medicare Part D Medication Therapy Management (MTM) program uses comprehensive medication reviews (CMRs) and targeted medication reviews to address medication-related problems in older adults. One such problem is central nervous system (CNS)-active polypharmacy, which is associated with impaired cognition and falls in older adults. To date, little is known about the prevalence of CNS-active polypharmacy among MTM enrollees and how they might differ from MTM enrollees without CNS-active polypharmacy; such insights would be helpful to MTM programs developing interventions to reduce CNS-active polypharmacy.

OBJECTIVE: To (1) estimate the prevalence of MTM enrollees with CNS-active polypharmacy in a nationwide cohort and (2) compare patient characteristics and MTM service use of MTM enrollees with CNS-active polypharmacy vs without.

METHODS: Cross-sectional, observational study of 2019-2021 Medicare 5% fee-for-service data linked to 2020-2021 MTM data. Patient characteristics and MTM use were compared between MTM enrollees with and without CNS-active polypharmacy.

RESULTS: Among 38,733 MTM enrollees in 2021, 4,144 (10.7%) experienced CNS-active polypharmacy. Compared with those without CNS-active polypharmacy, the CNS-active polypharmacy cohort was more likely to be male (72% vs 55%, standardized mean difference [SMD] = 34.7%), be dually enrolled in Medicaid (42% vs 24%, SMD = 41.1%), and have greater comorbidity burden (Charlson Comorbidity Score of 6.0 vs 5.0, SMD = 13.2%). The CNS-active polypharmacy cohort also had more prior-year health care utilization, such as being more likely to have an inpatient stay (37.5% vs 29.0%, SMD = 18.1%) or emergency department visit (53.7% vs 43.0%, SMD = 21.6%), as well as number of outpatient visits (7.0 vs 5.0, SMD = 20.8%) and number of unique prescription drugs (21.0 vs 15.0, SMD = 91.9%). The number of targeted medication reviews received was greater in the CNS-active polypharmacy cohort, but a lower proportion (35% vs 39%) participated in a CMR.

CONCLUSIONS: More than 1 in 10 MTM enrollees experience CNS-active polypharmacy, which is higher than the general Medicare fee-for-service beneficiary population. MTM enrollees with CNS-active polypharmacy are more likely to be male, dually enrolled in Medicaid, and with greater comorbidity burden and prior-year use of health care and medications, suggesting that interventions for this population may need to account for additional clinical and socioeconomic needs. Despite being at greater risk of adverse drug events including impaired cognition and falls, just over one-third of MTM enrollees with CNS-active polypharmacy participate in a CMR, suggesting opportunity for more targeted outreach and intervention by MTM programs, Part D plan sponsors, and Centers for Medicare & Medicaid Services.

BACKGROUND: In 2020, the US Multi-Society Task Force (USMSTF) on Colorectal Cancer revised its recommended interval for surveillance of 1 to 2 subcentimeter tubular adenomas from 5 to 10 years to 7 to 10 years. This study assessed the impact of retroactively applying these guidelines on endoscopy center capacity.

METHODS: Colonoscopy reports were reviewed for patients with an upcoming colonoscopy reminder at the Veterans Affairs Pittsburgh Healthcare System (VAPHS). Patients with incomplete records, high-risk indications, or inadequate bowel preparation were excluded. Previous surveillance recommendations were compared with the 2020 USMSTF guideline to determine whether colonoscopy intervals could be extended.

RESULTS: Of 583 patients, 331 (56.7%) had adequate data to determine extendibility. Using the 2020 USMSTF guidelines, colonoscopy interval could be extended by ≥ 1 year for 64 patients (11%) for a total of 153 years extended. For 36 patients, the ability to extend the colonoscopy interval was directly attributable to the updated 2020 USMSTF guideline.

CONCLUSIONS: Retroactive application of USMSTF surveillance guidelines could modestly increase endoscopic capacity, but overall impact may be limited by fragmentation of care.