Novel integration of governmental data sources using machine learning to identify super-utilization among U.S. counties.

Background: Super-utilizers consume the greatest share of resource intensive healthcare (RIHC) and reducing their utilization remains a crucial challenge to healthcare systems in the United States (U.S.). The objective of this study was to predict RIHC among U.S. counties, using routinely collected data from the U.S. government, including information on consumer spending, offering an alternative method for identifying super-utilization among population units rather than individuals. Methods: Cross-sectional data from 5 governmental sources in 2017 were used in a machine learning pipeline, where target-prediction features were selected and used in 4 distinct algorithms. Outcome metrics of RIHC utilization came from the American Hospital Association and included yearly: (1) emergency rooms visit, (2) inpatient days, and (3) hospital expenditures. Target-prediction features included: 149 demographic characteristics from the U.S. Census Bureau, 151 adult and child health characteristics from the Centers for Disease Control and Prevention, 151 community characteristics from the American Community Survey, and 571 consumer expenditures from the Bureau of Labor Statistics. SHAP analysis identified important target-prediction features for 3 RIHC outcome metrics. Results: 2475 counties with emergency rooms and 2491 counties with hospitals were included. The median yearly emergency room visits per capita was 0.450 [IQR:0.318, 0.618], the median inpatient days per capita was 0.368 [IQR: 0.176, 0.826], and the median hospital expenditures per capita was $2104 [IQR: $1299.93, 3362.97]. The coefficient of determination (R2), calculated on the test set, ranged between 0.267 and 0.447. Demographic and community characteristics were among the important predictors for all 3 RIHC outcome metrics. Conclusions: Integrating diverse population characteristics from numerous governmental sources, we predicted 3-outcome metrics of RIHC among U.S. counties with good performance, offering a novel and actionable tool for identifying super-utilizer segments in the population. Wider integration of routinely collected data can be used to develop alternative methods for predicting RIHC among population units.

Quantifying differences in packaged food and drink purchases among households with diet-related cardiometabolic multi-morbidity: a cross-sectional analysis.

BACKGROUND: Diet is important for chronic disease management, with limited research understanding dietary choices among those with multi-morbidity, the state of having 2 or more chronic conditions. The objective of this study was to identify associations between packaged food and drink purchases and diet-related cardiometabolic multi-morbidity (DRCMM). METHODS: Cross-sectional associations between packaged food and drink purchases and household DRCMM were investigated using a national sample of U.S. households participating in a research marketing study. DRCMM households were defined as household head(s) self-reporting 2 or more diet-related chronic conditions. Separate multivariable logistic regression models were used to model the associations between household DRCMM status and total servings of, and total calories and nutrients from, packaged food and drinks purchased per month, as well as the nutrient density (protein, carbohydrates, and fat per serving) of packaged food and drinks purchased per month, adjusted for household size. RESULTS: Among eligible households, 3795 (16.8%) had DRCMM. On average, households with DRCMM versus without purchased 14.8 more servings per capita, per month, from packaged foods and drinks (p < 0.001). DRCMM households were 1.01 times more likely to purchase fat and carbohydrates in lieu of protein across all packaged food and drinks (p = 0.002, p = 0.000, respectively). DRCMM households averaged fewer grams per serving of protein, carbohydrates, and fat per month across all food and drink purchases (all p < 0.001). When carbonated soft drinks and juices were excluded, the same associations for grams of protein and carbohydrates per serving per month were seen (both p < 0.001) but the association for grams of fat per serving per month attenuated. CONCLUSIONS: DRCMM households purchased greater quantities of packaged food and drinks per capita than non-DRCMM households, which contributed to more fat, carbohydrates, and sodium in the home. However, food and drinks in DRCMM homes on average were lower in nutrient-density. Future studies are needed to understand the motivations for packaged food and drink choices among households with DRCMM to inform interventions targeting the home food environment.

Can diverse population characteristics be leveraged in a machine learning pipeline to predict resource intensive healthcare utilization among hospital service areas?

BACKGROUND: Super-utilizers represent approximately 5% of the population in the United States (U.S.) and yet they are responsible for over 50% of healthcare expenditures. Using characteristics of hospital service areas (HSAs) to predict utilization of resource intensive healthcare (RIHC) may offer a novel and actionable tool for identifying super-utilizer segments in the population. Consumer expenditures may offer additional value in predicting RIHC beyond typical population characteristics alone. METHODS: Cross-sectional data from 2017 was extracted from 5 unique sources. The outcome was RIHC and included emergency room (ER) visits, inpatient days, and hospital expenditures, all expressed as log per capita. Candidate predictors from 4 broad groups were used, including demographics, adults and child health characteristics, community characteristics, and consumer expenditures. Candidate predictors were expressed as per capita or per capita percent and were aggregated from zip-codes to HSAs using weighed means. Machine learning approaches (Random Forrest, LASSO) selected important features from nearly 1,000 available candidate predictors and used them to generate 4 distinct models, including non-regularized and LASSO regression, random forest, and gradient boosting. Candidate predictors from the best performing models, for each outcome, were used as independent variables in multiple linear regression models. Relative contribution of variables from each candidate predictor group to regression model fit were calculated. RESULTS: The median ER visits per capita was 0.482 [IQR:0.351-0.646], the median inpatient days per capita was 0.395 [IQR:0.214-0.806], and the median hospital expenditures per capita was $2,302 [1$,544.70-$3,469.80]. Using 1,106 variables, the test-set coefficient of determination (R2) from the best performing models ranged between 0.184-0.782. The adjusted R2 values from multiple linear regression models ranged from 0.311-0.8293. Relative contribution of consumer expenditures to model fit ranged from 23.4-33.6%. DISCUSSION: Machine learning models predicted RIHC among HSAs using diverse population data, including novel consumer expenditures and provides an innovative tool to predict population-based healthcare utilization and expenditures. Geographic variation in utilization and spending were identified.

Information Extraction From Electronic Health Records to Predict Readmission Following Acute Myocardial Infarction: Does Natural Language Processing Using Clinical Notes Improve Prediction of Readmission?

Background Social risk factors influence rehospitalization rates yet are challenging to incorporate into prediction models. Integration of social risk factors using natural language processing (NLP) and machine learning could improve risk prediction of 30-day readmission following an acute myocardial infarction. Methods and Results Patients were enrolled into derivation and validation cohorts. The derivation cohort included inpatient discharges from Vanderbilt University Medical Center between January 1, 2007, and December 31, 2016, with a primary diagnosis of acute myocardial infarction, who were discharged alive, and not transferred from another facility. The validation cohort included patients from Dartmouth-Hitchcock Health Center between April 2, 2011, and December 31, 2016, meeting the same eligibility criteria described above. Data from both sites were linked to Centers for Medicare & Medicaid Services administrative data to supplement 30-day hospital readmissions. Clinical notes from each cohort were extracted, and an NLP model was deployed, counting mentions of 7 social risk factors. Five machine learning models were run using clinical and NLP-derived variables. Model discrimination and calibration were assessed, and receiver operating characteristic comparison analyses were performed. The 30-day rehospitalization rates among the derivation (n=6165) and validation (n=4024) cohorts were 15.1% (n=934) and 10.2% (n=412), respectively. The derivation models demonstrated no statistical improvement in model performance with the addition of the selected NLP-derived social risk factors. Conclusions Social risk factors extracted using NLP did not significantly improve 30-day readmission prediction among hospitalized patients with acute myocardial infarction. Alternative methods are needed to capture social risk factors.

Adaptation of an NLP system to a new healthcare environment to identify social determinants of health.

Social determinants of health (SDoH) are increasingly important factors for population health, healthcare outcomes, and care delivery. However, many of these factors are not reliably captured within structured electronic health record (EHR) data. In this work, we evaluated and adapted a previously published NLP tool to include additional social risk factors for deployment at Vanderbilt University Medical Center in an Acute Myocardial Infarction cohort. We developed a transformation of the SDoH outputs of the tool into the OMOP common data model (CDM) for re-use across many potential use cases, yielding performance measures across 8 SDoH classes of precision 0.83 recall 0.74 and F-measure of 0.78.

Development of Electronic Health Record-Based Prediction Models for 30-Day Readmission Risk Among Patients Hospitalized for Acute Myocardial Infarction.

Importance: In the US, more than 600 000 adults will experience an acute myocardial infarction (AMI) each year, and up to 20% of the patients will be rehospitalized within 30 days. This study highlights the need for consideration of calibration in these risk models. Objective: To compare multiple machine learning risk prediction models using an electronic health record (EHR)-derived data set standardized to a common data model. Design, Setting, and Participants: This was a retrospective cohort study that developed risk prediction models for 30-day readmission among all inpatients discharged from Vanderbilt University Medical Center between January 1, 2007, and December 31, 2016, with a primary diagnosis of AMI who were not transferred from another facility. The model was externally validated at Dartmouth-Hitchcock Medical Center from April 2, 2011, to December 31, 2016. Data analysis occurred between January 4, 2019, and November 15, 2020. Exposures: Acute myocardial infarction that required hospital admission. Main Outcomes and Measures: The main outcome was thirty-day hospital readmission. A total of 141 candidate variables were considered from administrative codes, medication orders, and laboratory tests. Multiple risk prediction models were developed using parametric models (elastic net, least absolute shrinkage and selection operator, and ridge regression) and nonparametric models (random forest and gradient boosting). The models were assessed using holdout data with area under the receiver operating characteristic curve (AUROC), percentage of calibration, and calibration curve belts. Results: The final Vanderbilt University Medical Center cohort included 6163 unique patients, among whom the mean (SD) age was 67 (13) years, 4137 were male (67.1%), 1019 (16.5%) were Black or other race, and 933 (15.1%) were rehospitalized within 30 days. The final Dartmouth-Hitchcock Medical Center cohort included 4024 unique patients, with mean (SD) age of 68 (12) years; 2584 (64.2%) were male, 412 (10.2%) were rehospitalized within 30 days, and most of the cohort were non-Hispanic and White. The final test set AUROC performance was between 0.686 to 0.695 for the parametric models and 0.686 to 0.704 for the nonparametric models. In the validation cohort, AUROC performance was between 0.558 to 0.655 for parametric models and 0.606 to 0.608 for nonparametric models. Conclusions and Relevance: In this study, 5 machine learning models were developed and externally validated to predict 30-day readmission AMI hospitalization. These models can be deployed within an EHR using routinely collected data.

Computable Phenotype Implementation for a National, Multicenter Pragmatic Clinical Trial: Lessons Learned From ADAPTABLE.

BACKGROUND: Many large-scale cardiovascular clinical trials are plagued with escalating costs and low enrollment. Implementing a computable phenotype, which is a set of executable algorithms, to identify a group of clinical characteristics derivable from electronic health records or administrative claims records, is essential to successful recruitment in large-scale pragmatic clinical trials. This methods paper provides an overview of the development and implementation of a computable phenotype in ADAPTABLE (Aspirin Dosing: a Patient-Centric Trial Assessing Benefits and Long-Term Effectiveness)-a pragmatic, randomized, open-label clinical trial testing the optimal dose of aspirin for secondary prevention of atherosclerotic cardiovascular disease events. METHODS AND RESULTS: A multidisciplinary team developed and tested the computable phenotype to identify adults ≥18 years of age with a history of atherosclerotic cardiovascular disease without safety concerns around using aspirin and meeting trial eligibility criteria. Using the computable phenotype, investigators identified over 650 000 potentially eligible patients from the 40 participating sites from Patient-Centered Outcomes Research Network-a network of Clinical Data Research Networks, Patient-Powered Research Networks, and Health Plan Research Networks. Leveraging diverse recruitment methods, sites enrolled 15 076 participants from April 2016 to June 2019. During the process of developing and implementing the ADAPTABLE computable phenotype, several key lessons were learned. The accuracy and utility of a computable phenotype are dependent on the quality of the source data, which can be variable even with a common data model. Local validation and modification were required based on site factors, such as recruitment strategies, data quality, and local coding patterns. Sustained collaboration among a diverse team of researchers is needed during computable phenotype development and implementation. CONCLUSIONS: The ADAPTABLE computable phenotype served as an efficient method to recruit patients in a multisite pragmatic clinical trial. This process of development and implementation will be informative for future large-scale, pragmatic clinical trials. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02697916.

Measuring the Effect of Environmental Tobacco Smoke on Lung Function: Results From a Small Observational Investigation of Acute Exposure.

OBJECTIVE: Exposure to environmental tobacco smoke (ETS) in smoky venues puts patrons and employees at risk for immediate respiratory symptoms. Although much literature focuses on outcomes associated with chronic ETS exposure, the current study assesses changes in lung function after acute exposure. METHODS: Ninety-six nonsmoking, healthy adults were exposed to ETS at a bar. Lung function [eg, forced vital capacity (FVC), forced expiratory volume in 1 second (FEV1)] was assessed at baseline, immediately after 3 hours of ETS exposure, and 2 hours after exiting the bar. PM2.5 recordings were also measured. RESULTS: Repeated-measures analysis of variance found significant decreases in FEV1, FVC and FEF25-75%, and peak expiratory flow after ETS exposure compared with baseline that remained significantly decreased after a 2-hour recovery period. CONCLUSIONS: Acute exposure to ETS in a natural environment significantly attenuates lung function. A subgroup experienced heightened reductions in lung function.

Tobacco Policies in Louisiana: Recommendations for Future Tobacco Control Investment from SimSmoke, a Policy Simulation Model.

Despite the presence of tobacco control policies, Louisiana continues to experience a high smoking burden and elevated smoking-attributable deaths. The SimSmoke model provides projections of these health outcomes in the face of existing and expanded (simulated) tobacco control polices. The SimSmoke model utilizes population data, smoking rates, and various tobacco control policy measures from Louisiana to predict smoking prevalence and smoking-attributable deaths. The model begins in 1993 and estimates are projected through 2054. The model is validated against existing Louisiana smoking prevalence data. The most powerful individual policy measure for reducing smoking prevalence is cigarette excise tax. However, a comprehensive cessation treatment policy is predicted to save the most lives. A combination of tobacco control policies provides the greatest reduction in smoking prevalence and smoking-attributable deaths. The existing Louisiana excise tax ranks as one of the lowest in the country and the legislature is against further increases. Alternative policy measures aimed at lowering prevalence and attributable deaths are: cessation treatments, comprehensive smoke-free policies, and limiting youth access. These three policies have a substantial effect on smoking prevalence and attributable deaths and are likely to encounter more favor in the Louisiana legislature than increasing the state excise tax.

Mental Health Screening in North Sulawesi, Indonesia: Kessler 6 pilot data and needs assessment results from the LearnToLive Indonesian Health Initiative.

INTRODUCTION: Communities around the world are increasing their focus on mental health and substance use disorders. However, the struggle to identify and treat patients remains great. The sequelae of these disorders, including severe chronic disability and suicide, are significant, and its impact is felt most in lower and middle-income countries. In the rural and underserved region of North Sulawesi, Indonesia, there are limited data published regarding the prevalence of depression, anxiety, and other symptoms of psychological distress. METHODS: In order to characterize and quantify some specific areas of psychological distress, the LearnToLive Indonesian Health Initiative completed a retroactive review of Kessler 6 data from 697 people in rural communities of North Sulawesi. RESULTS: Our results demonstrate a rate of near 10% for psychological distress, particularly with anxiety and depressive symptoms. We also found that the village of Sapa scored higher on most of the subcomponents of the screen compared with the other villages in the study. DISCUSSION: While the Kessler 6 screening tool is not diagnostic, our results suggest significant mental health issues in need of further exploration and research. We found that these results exist in an environment with high stigma, limited education regarding mental illness, and limited outpatient services. The results from this analysis will hopefully guide future mental health education in the region and will ultimately assist in the development of the clinical infrastructure needed to effectively identify, treat, and manage mental health conditions.