Joseph Finkelstein

Potential utility of precision medicine for older adults with polypharmacy: a case series study

Pharmacogenomic (PGx) testing has been increasingly used to optimize drug regimens; however, its potential in older adults with polypharmacy has not been systematically studied. In this hypothesis-generating study, we employed a case series design to explore potential utility of PGx testing in older adults with polypharmacy and to highlight barriers in implementing this methodology in routine clinical practice. Three patients with concurrent chronic heart and lung disease aged 74, 78, and 83 years and whose medication regimen comprised 26, 17, and 18 drugs, correspondingly, served as cases for this study. PGx testing identified major genetic polymorphisms in the first two cases. The first case was identified as “CYP3A4/CYP3A5 poor metabolizer”, which affected metabolism of eleven prescribed drugs. The second case had “CYP2D6 rapid metabolizer” status affecting three prescribed medications, two of which were key drugs for managing this patient’s chronic conditions. Both these patients also had VKORC1 allele *A, resulting in higher sensitivity to warfarin. All cases demonstrated a significant number of potential drug–drug interactions. Both patients with significant drug–gene interactions had a history of frequent hospitalizations (six and 23, respectively), whereas the person without impaired cytochrome P450 enzyme activity had only two acute episodes in the last 5 years, although he was older and had multiple comorbidities. Since all patients received guideline-concordant therapy from the same providers and were adherent to their drug regimen, we hypothesized that genetic polymorphism may represent an additional risk factor for higher hospitalization rates in older adults with polypharmacy. However, evidence to support or reject this hypothesis is yet to be established. Studies evaluating clinical impact of PGx testing in older adults with polypharmacy are warranted. For practical implementation of pharmacogenomics in routine clinical care, besides providing convincing evidence of its clinical effectiveness, multiple barriers must be addressed. Introduction of intelligent clinical decision support in electronic medical record systems is required to address complexities of simultaneous drug–gene and drug–drug interactions in older adults with polypharmacy. Physician training, clear clinical pathways, evidence-based guidelines, and patient education materials are necessary for unlocking full potential of pharmacogenomics into routine clinical care of older adults.

Using a Mobile App to Promote Smoking Cessation in Hospitalized Patients

Background The potential of interactive health education for preventive health applications has been widely demonstrated. However, use of mobile apps to promote smoking cessation in hospitalized patients has not been systematically assessed. Objective This study was conducted to assess the feasibility of using a mobile app for the hazards of smoking education delivered via touch screen tablets to hospitalized smokers. Methods Fifty-five consecutive hospitalized smokers were recruited. Patient sociodemographics and smoking history was collected at baseline. The impact of the mobile app was assessed by measuring cognitive and behavioral factors shown to promote smoking cessation before and after the mobile app use including hazards of smoking knowledge score (KS), smoking attitudes, and stages of change. Results After the mobile app use, mean KS increased from 27(3) to 31(3) (P<0.0001). Proportion of patients who felt they “cannot quit smoking” reduced from 36% (20/55) to 18% (10/55) (P<0.03). Overall, 13% (7/55) of patients moved toward a more advanced stage of change with the proportion of patients in the preparation stage increased from 40% (22/55) to 51% (28/55). Multivariate regression analysis demonstrated that knowledge gains and mobile app acceptance did not depend on age, gender, race, computer skills, income, or education level. The main factors affecting knowledge gain were initial knowledge level (P<0.02), employment status (P<0.05), and high app acceptance (P<0.01). Knowledge gain was the main predictor of more favorable attitudes toward the mobile app (odds ratio (OR)=4.8; 95% confidence interval (CI) (1.1, 20.0)). Attitudinal surveys and qualitative interviews identified high acceptance of the mobile app by hospitalized smokers. Over 92% (51/55) of the study participants recommended the app for use by other hospitalized smokers and 98% (54/55) of the patients were willing to use such an app in the future. Conclusions Our results suggest that a mobile app promoting smoking cessation is well accepted by hospitalized smokers. The app can be used for interactive patient education and counseling during hospital stays. Development and evaluation of mobile apps engaging patients in their care during hospital stays is warranted.

Machine learning approaches to personalize early prediction of asthma exacerbations

Patient telemonitoring results in an aggregation of significant amounts of information about patient disease trajectory. However, the potential use of this information for early prediction of exacerbations in adult asthma patients has not been systematically evaluated. The aim of this study was to explore the utility of telemonitoring data for building machine learning algorithms that predict asthma exacerbations before they occur. The study dataset comprised daily self‐monitoring reports consisting of 7001 records submitted by adult asthma patients during home telemonitoring. Predictive modeling included preparation of stratified training datasets, predictive feature selection, and evaluation of resulting classifiers. Using a 7‐day window, a naive Bayesian classifier, adaptive Bayesian network, and support vector machines were able to predict asthma exacerbation occurring on day 8, with sensitivity of 0.80, 1.00, and 0.84; specificity of 0.77, 1.00, and 0.80; and accuracy of 0.77, 1.00, and 0.80, respectively. Our study demonstrated that machine learning techniques have significant potential in developing personalized decision support for chronic disease telemonitoring systems. Future studies may benefit from a comprehensive predictive framework that combines telemonitoring data with other factors affecting the likelihood of developing acute exacerbation. Approaches implemented for advanced asthma exacerbation prediction may be extended to prediction of exacerbations in patients with other chronic health conditions.

Pharmacogenetic polymorphism as an independent risk factor for frequent hospitalizations in older adults with polypharmacy: a pilot study

Pharmacogenetic testing identifies genetic biomarkers that are predictive of individual sensitivity to particular drugs. A significant proportion of medications that are widely prescribed for older adults are metabolized by enzymes that are encoded by highly polymorphic genes. Pharmacogenetic testing is increasingly used to optimize the medication regimen; however, its potential in older adults with polypharmacy has not been systematically explored. Following the initial case–series study, this study hypothesized that frequently hospitalized older adults with polypharmacy have higher frequency of pharmacogenetic polymorphism as compared to older adults with polypharmacy who are rarely admitted to a hospital. To test this hypothesis, a nested case–control study was conducted with pharmacogenetic polymorphism as an exposure and hospitalization rate as an outcome. In this study, frequently hospitalized older adults (≥65 years of age) with polypharmacy were matched with rarely hospitalized older adults with poly-pharmacy by age, gender, race, ethnicity, and chronic disease score. Average age and number of prescription drugs did not differ in cases and controls (77.2±5.0 and 78.3±5.1 years, 14.3±5.3 and 14.0±2.9 medications, respectively). No statistically significant difference in sociodemographic, clinical, and behavioral characteristics that are known to affect hospitalization risk was found between the cases and controls. Major pharmacogenetic polymorphism defined as presence of at least one allelic combination resulting in poor or rapid metabolizer status was identified in all the cases. No major pharmacogenetic polymorphisms were detected in controls. Based on the exact McNemar’s test, the difference in major pharmacogenetic polymorphism frequency between cases and controls was statistically significant (p<0.05). In 50% of cases, more than one major pharmacogenetic polymorphism was found. The frequency of CYP2C19 rapid metabolizer, CYP3A4/5 poor metabolizer, VKORC1 low sensitivity, and CYP2D6 rapid metabolizer status in cases was 67%, 33%, 33%, and 17%, respectively, which significantly exceeded respective prevalence in general population. The mean number of major gene–drug interactions found in cases was 2.8±2.2, whereas no major drug–gene interactions were identified in controls. The difference in the number of major drug–gene interactions between cases and controls was statistically significant (p<0.05). The pilot data supported the hypothesis that pharmacogenetic polymorphism may represent an independent risk factor for frequent hospitalizations in older adults with polypharmacy. Due to small sample size, the results of this proof-of-concept study cannot be conclusive. Further work on the utility of pharmacogenetic testing for optimization of medication regimens in this vulnerable group of older adults is warranted.

Using CART for advanced prediction of asthma attacks based on telemonitoring data

Advanced prediction of asthma exacerbations may significantly improve patient quality of life and reduce costs of urgent care delivery. Majority of current algorithms predict who is likely to experience asthma exacerbation rather than when it is about to occur. We used data from asthma home-based telemonitoring for advanced prediction of asthma exacerbation. The goal of this project was to develop an algorithm that predicts asthma exacerbation one day in advance based on previous 7-day window. CART was used for predictive modeling. Resulting algorithm had specificity 0.971, sensitivity of 0.647, and accuracy of 0.809. We concluded that machine learning has great potential for advanced prediction of chronic disease exacerbations based on home telemonitoring data.

A Comprehensive Informatics Framework to Increase Breast Cancer Risk Assessment and Chemoprevention in the Primary Care Setting

The U.S. Preventive Services Task Force recommends that clinicians engage in shared decision making with women at high-risk for breast cancer about medications to reduce their risk, also known as chemoprevention. However, uptake has been low (<;5%) in the primary care setting. The goal of this project was to implement a comprehensive informatics framework to increase breast cancer risk assessment and chemoprevention in the primary care setting. We built a novel breast cancer risk navigation (BNAV) tool, which incorporates the Gail breast cancer risk model into the electronic health record. To address patient-related barriers to chemoprevention, an interactive decision aid, RealRisks, was developed that allows participants to experience risk through an activity. An automated system of theory-based tailored messaging was built to alert providers to engage in guideline-concordant chemoprevention practices. This innovative technology has high potential to facilitate evidence-based decision support for womens health in primary care setting.