Faculty Bibliography

OBJECTIVE:While timely interventions can improve medication adherence, it is challenging to identify which patients are at risk of nonadherence at point-of-care. We aim to develop and validate flexible machine learning (ML) models to predict a continuous measure of adherence to guideline-directed medication therapies (GDMTs) for heart failure (HF). MATERIALS AND METHODS/METHODS:We utilized a large electronic health record (EHR) cohort of 34,697 HF patients seen at NYU Langone Health with an active prescription for ≥1 GDMT between April 01, 2021 and October 31, 2022. The outcome was adherence to GDMT measured as proportion of days covered (PDC) at 6 months following a clinical encounter. Over 120 predictors included patient-, therapy-, healthcare-, and neighborhood-level factors guided by the World Health Organization's model of barriers to adherence. We compared performance of several ML models and their ensemble (superlearner) for predicting PDC with traditional regression model (OLS) using mean absolute error (MAE) averaged across 10-fold cross-validation, % increase in MAE relative to superlearner, and predictive-difference across deciles of predicted PDC. RESULTS:Superlearner, a flexible nonparametric prediction approach, demonstrated superior prediction performance. Superlearner and quantile random forest had the lowest MAE (mean [95% CI] = 18.9% [18.7%-19.1%] for both), followed by MAEs for quantile neural network (19.5% [19.3%-19.7%]) and kernel support vector regression (19.8% [19.6%-20.0%]). Gradient boosted trees and OLS were the 2 worst performing models with 17% and 14% higher MAEs, respectively, relative to superlearner. Superlearner demonstrated improved predictive difference. CONCLUSION/CONCLUSIONS:This development phase study suggests potential of linked EHR-pharmacy data and ML to identify HF patients who will benefit from medication adherence interventions. DISCUSSION/CONCLUSIONS:Fairness evaluation and external validation are needed prior to clinical integration.

PURPOSE/OBJECTIVE:Chemotherapy-induced alopecia (CIA) affects approximately 65% of patients receiving chemotherapy and has a negative impact on quality of life (QoL). Scalp cooling (SC) is the only FDA-cleared intervention for CIA. This systematic review and meta-analysis evaluated SC adverse events (AEs), reasons for discontinuation, and scalp metastasis incidence. METHODS:Meta-analyses using random-effects models estimated pooled prevalences of SC AEs, SC discontinuation, and reasons for discontinuation. A generalized linear mixed model was used to estimate the incidence of scalp metastasis. RESULTS:Sixty-seven studies met the inclusion criteria. The most common AEs were generalized chills (42%, 95% confidence interval (CI) 26-58%), cap heaviness (35%, 95% CI 18-52%), and headache (30%, 95% CI 21-39%). The SC discontinuation rate was 18% (95% CI 13-23%). The most common reasons for discontinuation were progressive alopecia (15%, 95% CI 10-20%) and reasons unrelated to SC (9%, 95% CI 5-13%). The most frequent AEs leading to SC discontinuation were headache (4%, 95% CI 2-6%), cold intolerance (4%, 95% CI 3-5%), and general discomfort (4%, 95% CI 2-7%). Secondary analysis of scalp metastases yielded an incidence of 0.15% (95% CI 0.05-0.47%). Analysis of FDA Manufacturer and User Facility Device Experience (MAUDE) database medical device reports revealed that user error contributed to cold thermal injuries. Prevalence estimates were limited by significant heterogeneity between studies, reflecting variations in study methodology and real-world SC practices. CONCLUSION/CONCLUSIONS:SC is generally well tolerated with minimal safety concerns. Clinical comfort strategies like supportive medications and improved patient education could enhance SC tolerability and support its implementation.

BACKGROUND:Data on patient-centered outcomes of mobile health cardiac rehabilitation (mHealth-CR) for older adults with ischemic heart disease are limited. The RESILIENT (Rehabilitation at Home Using Mobile Health in Older Adults After Hospitalization for Ischemic Heart Disease) trial, the largest randomized study of mHealth-CR in this population, found no significant improvements in functional capacity, health status, angina, or disability compared with usual care. OBJECTIVES/OBJECTIVE:The purpose of this study was to evaluate whether mHealth-CR affects personalized goal attainment-a prespecified secondary endpoint of RESILIENT-using goal attainment scaling (GAS). METHODS:A total of 400 patients (≥65 years) with ischemic heart disease were randomized to mHealth-CR or usual care. Participants specified goals for CR at baseline using the five-category goal attainment scale: much-less-than-expected (-2), less-than-expected (-1), expected (0), better-than-expected (+1), and much-better-than-expected (+2). Goal attainment was assessed at 3 months. RESULTS:Of 400 patients (median age, 71.0 years [range 65.0-91.0]; 72.8% male; 65.2% prefrail/frail) randomized to mHealth-CR (n = 298) or usual care (n = 102), 353 (88.3%) completed GAS. Most goals addressed physical activity (54.0% mHealth-CR vs 59.0% usual care), health care behaviors (14.4% vs 11.9%), or symptom management (13.1% vs 9.0%). Rates of attaining or exceeding goals (GAS ≥0) were similar between groups (80.5% vs 77.6%; P = 0.492). However, in the intervention arm, there was a higher rate of exceeding expected level of goal attainment (GAS +1, +2) compared with usual care (52.6% vs 34.2%; P = 0.006). CONCLUSIONS:In a trial that did not demonstrate differences on traditional endpoints, those receiving mHealth-CR were more likely to exceed personalized CR goals. These findings suggest the intervention facilitated greater progress toward individualized goals and underscore the importance of patient-centered outcomes in CR.

BACKGROUND:Despite the widespread use of electronic health records, a standardized approach to identify heart failure patients is lacking. We sought to create and validate definitions for identifying patients with heart failure using electronic health record data. METHODS:To define heart failure, we developed 8 distinct definitions created from combinations of heart failure diagnosis based on ICD-10 codes listed in the clinical encounter, problem list or past medical history, and/or ejection fraction ≤40%. To validate our definitions, we used stratified sampling and physician chart review guided by the Universal Definition of Heart Failure as our gold standard and compared their performance using sensitivity and positive predictive value. RESULTS:We identified 41,392 patients who met at least one of our eight definitions for heart failure, plus an additional 2,692 patients with an ICD-10 diagnosis of heart failure outside of a standard clinical setting and 696,896 patients with a cardiovascular diagnosis other than heart failure. Using these groups, we randomly sampled a total of 528 charts for physician chart review. Sensitivities of the eight definitions of heart failure ranged from 10.3% to 42.0%, and positive predictive values ranged from 80.7% to 98.6%. CONCLUSIONS:We found that patients meeting EHR-based definitions of heart failure likely represented true clinical cases of disease. Nevertheless, the definitions captured less than half of the patients with heart failure, thus severely underestimating the prevalence of disease and underlining a need for more comprehensive methods to effectively use this data to understand the epidemiological burden of heart failure.

BACKGROUND:Between 2022 and 2023, overdose mortality decreased among non-Hispanic (NH) white people but stayed the same or increased among people of colour in the USA. County racialised economic segregation may contribute to overdose mortality. METHODS:measures, one for higher-income NH white and lower-income black residents and another for higher-income NH white and lower-income Hispanic residents. Models included random effects for county, year and county-year interaction, and fixed effects for proportion male, proportion aged 25-44, land area, state and year. We estimated relative risk (RR) by quintile (least vs most privileged) and the difference in overdose mortality per 100 000 (RD) had all counties shifted to the risk of the most advantaged counties (Q5). RESULTS:Counties with the highest proportion of lower-income racially minoritised residents (Q1) had an increased RR of overdose deaths compared with Q5 counties, both overall (aRRs 1.64 (1.51-1.78); 1.40 (1.29-1.52)), and among subgroups. Had all counties experienced the risk of Q5 counties, we estimated an average reduction in overdose deaths overall (RDs per 100 000: -7.20 (-8.25 to -6.10); -6.37 (-7.38 to -5.25)) and among subgroups. CONCLUSION/CONCLUSIONS:County racialised economic segregation was associated with overdose mortality risk in 2018-2022. Investment in evidence-based strategies to reduce overdose risk in places experiencing harms related to racialised economic segregation is critical.

BACKGROUND:Patients with heart failure (HF) often have difficulty obtaining life-saving medications due to coverage barriers, such as prior authorizations and high out-of-pocket costs. To promote better coverage for high value therapies and inform policymakers about cost effectiveness, the American Heart Association/American College of Cardiology/Heart Failure Society of America added Value Statements to HF guidelines. We assessed whether these guidelines influenced Medicare drug coverage policies for 2 life-saving, costly HF medications: angiotensin receptor neprilysin inhibitors (ARNI-guideline "high value") and sodium glucose cotransporter-2 inhibitors (SGLT2i-guideline "intermediate value"). METHODS:We performed an observational study using Medicare prescription drug plan formulary files from April 2020 to April 2023 to separately assess for changes in coverage barriers to ARNI and SGLT2i after Value Statement publication (April 2022), and subsequent Medicare plan online update (October 2022). The primary outcome was the percentage of plans each month with any barrier to drug coverage (prior authorizations, tier ≥3 out-of-pocket cost-sharing, step therapy, or no coverage). Analyses used interrupted time series and difference-in-differences approaches. Difference-in-differences analyses used direct oral anticoagulants as a control due to their comparable cost and use as ARNI and SGLT2i, but without a Value Statement. RESULTS:Among 7396 Medicare drug plans, monthly rates of any coverage barrier ranged from 94.3% to 97.4% for ARNI and 93.2% to 96.6% for SGLT2i. Most barriers were due to tier ≥3 out-of-pocket cost-sharing requirements (ARNI: 94.3%-95.8%; SGLT2i: 93.2%-95.6%). Coverage barriers remained stable in April 2022 and declined slightly in October 2022. In difference-in-differences analyses, the presence of a Value Statement was associated with a ~1 percentage point decline in coverage barriers for both ARNI (difference-in-differences estimate, -1.07% [95% CI, -1.44% to -0.70%]) and SGLT2i (-1.32% [95% CI, -1.63% to -1.00%]). CONCLUSIONS:Coverage barriers to ARNI and SGLT2i were common and changed only slightly after publication of Value Statements in HF guidelines. There is a critical need for robust strategies to improve access to life-saving HF medications.

BACKGROUND:Sarcopenia has been associated with adverse postoperative outcomes in older age cohorts, but has not been assessed in older adults with inflammatory bowel disease (IBD). Further, current assessments of sarcopenia among all aged individuals with IBD have used various measures of muscle mass as well as cutoffs to define its presence, leading to heterogeneous findings. METHODS:In this single-institution, multihospital retrospective study, we identified all patients aged 60 years and older with IBD who underwent disease-related intestinal resection between 2012 and 2022. Skeletal Muscle Index (SMI) and Total Psoas Index (TPI) were measured at the superior L3 endplate on preoperative computed tomography scans and compared through receiver operating characteristic curve. We then performed multivariable logistic regression to assess risk factors associated with an adverse 30-day postoperative outcome. Our primary outcome included a 30-day composite of postoperative mortality and complications, including infection, bleeding, cardiac event, cerebrovascular accident, acute kidney injury, venous thromboembolism, reoperation, all-cause rehospitalization, and need for intensive care unit-level care. RESULTS:A total of 120 individuals were included. Overall, 52% were female, 40% had ulcerative colitis, 60% had Crohn's disease, and median age at time of surgery was 70 years (interquartile range: 65-75). Forty percent of older adults had an adverse 30-day postoperative outcome, including infection (23%), readmission (17%), acute kidney injury (13%), bleeding (13%), intensive care unit admission (10%), cardiac event (8%), venous thromboembolism (7%), reoperation (6%), mortality (5%), and cerebrovascular accident (2%). When evaluating the predictive performance of SMI vs TPI for an adverse 30-day postoperative event, SMI had a significantly higher area under the curve of 0.66 (95% CI, 0.56-0.76) as compared to 0.58 (95% CI, 0.48-0.69) for TPI (P = .02). On multivariable logistic regression, prior IBD-related surgery (adjusted odds ratio [adjOR] 6.46, 95% CI, 1.85-22.51) and preoperative sepsis (adjOR 5.74, 95% CI, 1.36-24.17) significantly increased the odds of adverse postoperative outcomes, whereas increasing SMI was associated with a decreased risk of an adverse postoperative outcome (adjOR 0.88, 95% CI, 0.82-0.94). CONCLUSIONS:Sarcopenia, as measured by SMI, is associated with an increased risk of postoperative complications among older adults with IBD. Measurement of SMI from preoperative imaging can help risk stratify older adults with IBD undergoing intestinal resection.

OBJECTIVE:To compare anonymized and non-anonymized approaches for imputing race and ethnicity in descriptive studies of chronic disease burden using electronic health record (EHR)-based datasets. STUDY SETTING AND DESIGN/METHODS:In this New York City-based study, we first conducted simulation analyses under different missing data mechanisms to assess the performance of Bayesian Improved Surname Geocoding (BISG), single imputation using neighborhood majority information, random forest imputation, and multiple imputation with chained equations (MICE). Imputation performance was measured using sensitivity, precision, and overall accuracy; agreement with self-reported race and ethnicity was measured with Cohen's kappa (κ). We then applied these methods to impute race and ethnicity in two EHR-based data sources and compared chronic disease burden (95% CIs) by race and ethnicity across imputation approaches. DATA SOURCES AND ANALYTIC SAMPLE/UNASSIGNED:Our data sources included EHR data from NYU Langone Health and the INSIGHT Clinical Research Network from 3/6/2016 to 3/7/2020 extracted for a parent study on older adults in NYC with multiple chronic conditions. PRINCIPAL FINDINGS/RESULTS: = 0.33). When these methods were applied to the NYU and INSIGHT cohorts, however, racial and ethnic distributions and chronic disease burden were consistent across all imputation methods. Slight improvements in the precision of estimates were observed under all imputation approaches compared to a complete case analysis. CONCLUSIONS:BISG imputation may provide a more accurate racial and ethnic classification than single or multiple imputation using anonymized covariates, particularly if the missing data mechanism is MNAR. Descriptive studies of disease burden may not be sensitive to methods for imputing missing data.

BACKGROUND:Disruption of ambulatory healthcare in New York City (NYC) during the COVID-19 pandemic was common, but the impact on the cardiometabolic health of vulnerable patient groups is unknown. Therefore, we estimated the effect of total care disruption (TCD) on blood pressure (BP) control among older NYC residents with hypertension and at least one other chronic condition, and examined whether neighborhood poverty moderated this impact. METHODS:From the INSIGHT Clinical Research Network, we identified NYC residents ≥50 years of age with hypertension and at least one other chronic condition. TCD was defined as no ambulatory or telehealth visit during the pandemic. We contrasted the change in prevalence of controlled BP (BP <140/90) before and after the pandemic among those with and without TCD via an inverse probability weighted (IPW) difference-in-difference regression model. RESULTS:Among 212,673 eligible individuals, mean age was 69.5 years (SD: 10.2 years) and 15.1% experienced TCD. BP control declined from 52.4% to 45.9% among those with TCD and from 53.6% to 48.9% among those without TCD. After IPW adjustment, a larger decline in BP control was noted among those with TCD (adjusted difference-in-difference = 1.13 percentage points (95% CI 0.32-1.94, p-value=0.0058)). There was no consistent difference in the relationship between TCD and post-pandemic BP control across neighborhood poverty levels. CONCLUSION/CONCLUSIONS:COVID-19-related TCD was associated with a modest decline in BP control among older adults with hypertension in NYC; this was not moderated by neighborhood poverty level.

OBJECTIVE:This study examined associations between access to leisure-time physical activity (LTPA) facilities and parks and repeated measures of glycated hemoglobin (A1C) over time, using follow-up tests among United States Veterans with newly diagnosed type-2 diabetes (T2D). METHODS:Data were analyzed from 274,463 patients in the Veterans Administration Diabetes Risk cohort who were newly diagnosed with T2D between 2008 and 2018 and followed through 2023. Generalized estimating equations with a logit link function and binomial logistic regression were used to examine associations. RESULTS:Patients were on average 60.5 years of age, predominantly male (95.0 %) and white (66.9 %), and had an average of 11.7 A1C tests during the study follow-up period. In high- and low-density urban communities, a one-unit higher LTPA facility density score was associated with 1 % and 3 % greater likelihood of in-range A1C tests during follow-up, respectively, but no association was observed among patients living in suburban/small town and rural communities. Across community types, closer park distance was not associated with subsequent greater odds of in-range A1C tests. Unexpectedly, in low-density urban areas, the likelihood of in-range A1C tests was 1 % lower at farther park distances. CONCLUSIONS:These results suggest that broader access to LTPA facilities, but not park proximity, may contribute in small ways to maintaining glycemic control after T2D diagnosis in urban communities. Tailored interventions may be needed to promote patients' effective use of LTPA facilities and parks.