Search

Objective: Machine learning (ML) may enhance prediction of outcomes such as mortality or acute kidney injury (AKI) among cardiac patients after coronary artery bypass graft (CABG). In this study, we used meta-analyses of reported ML models to assess what ML has been able to accomplish in this field, by evaluating the model performance in studies with CABG patients. Methods: We performed a literature search using Google Scholar and included studies that reported AUC and 95% CI for various models in our analysis. In addition, total participants, year of publication, type of analytical method (gradient boosting, random forest, etc.) and type of outcome (mortality or AKI) were extracted. We combined effect sizes using random effects model, and tested for heterogeneity, and publication bias. Results: 5 models from 5 studies were included in the analysis (patients= 35,152; with outcome mortality =3,080, AKI=933). Combined mean AUC was 0.796 (95% CI: 0.776, 0.815). Test of heterogeneity showed high variation between studies (I 2 = 66.7%). Egger’s test intercept was -1.03 (95% CI: -7.22, 5.17, p > .25) indicating no small study bias. Meta regression showed newer publications had a positive association ( coef = 0.003) and number of variables in the study had a negative association with higher AUC values ( coef = -0.0002). In subgroup analysis, the pooled AUC values for mortality and AKI groups were 0.795 and 0.805 respectively. The highest individual AUC was from ensemble model predicting AKI with AUC 0.84 and lowest was from gradient boosting model predicting mortality with AUC 0.77. Conclusion: Among the presented models for CABG ensemble methods performed well, but surprisingly methods with lesser number of variables tended to have higher predictive power. In near future, ML-based models may form the basis to build intelligent decision support systems for patient selection and risk stratification prior to CABG and could be applied to other cardiac surgeries.

Developing and implementing analytical models for predicting mortality or readmission related outcomes among heart failure (HF) patients are challenging. In this study, we used meta-analyses of reported predictive models to assess what machine learning (ML) has been able to accomplish in this field, by evaluating the ML model performance for studies in HF. We performed a literature search using Google Scholar, Web of Science and PubMed. The studies reporting AUC and 95% CI for various models were included. In addition, total participants, year of publication, type of analytical method (logistic regression, RF, etc.) and type of outcome (mortality, readmission, etc.) were extracted. We combined effect sizes using random effects (RF) model, and tested for heterogeneity, and publication bias. 12 studies were included in the analysis (patients= 123,832; AUC=15, with outcome mortality =17,471, readmission=15,703, hospitalization=67,523). Combined mean AUC was 0.77 (95% CI: 0.72, 0.82). Test of heterogeneity showed high variation between studies (I2=98.9%). Egger’s test intercept was 5.2 (95% CI: -4.2, 14.7, p > .25) indicating no small study effects/bias. Meta regression showed newer publications provide better AUC values (p < 0.03). In subgroup analysis, the pooled AUC for readmission, hospitalization, and mortality groups were 0.71, 0.80, and 0.78 respectively. The highest individual AUC was from neural networks (NN) predicting hospitalization with AUC 0.96 and lowest was from RF predicting readmission with AUC 0.65. Presented models were diverse, ranking in quality from fair to very good, and being varied for different clinical outcomes among HF patients. Situation known from studies using classical statistical methods holds also for methods using ML, with better predictive values for hospitalizations, and lower for other outcomes. Methods using NN and methods using higher numbers of variables performed very well and had highest predictive power.

Objective: Identifying patients at high risk of AE after TAVR is essential to prolong their survival. Current prediction models for AE after TAVR suffer from a lack of accuracy and external validation. Modern ML approaches can account for higher-dimensional relationships among variables, potentially improving the prediction of outcomes. We performed a systematic review and meta-analysis to estimate the discriminative ability of recently developed ML-based models, which predict various AE after TAVR. Methods: We searched Pubmed, Google Scholar, and Web of Science for studies (Jan 2019 to Jan 2022) that used ML approaches to predict AE after TAVR. Inputs in the meta-analysis were study-reported c-index values and 95% CI. Subgroup analyses separated models by outcome (mortality or clinical AE). Combined effect sizes using a random-effects model, test for heterogeneity, and Egger’s test to assess publication bias were considered. Results: Eight studies were included in the systematic review (patients = 26,023; outcomes = 1,014), of which five models had sufficient data for the meta-analysis. The number of features included in each model ranged from 6 to 107. The two most common models were random forest (n=2) and logistic regression (n=2). The most common outcome was mortality (n=5). The meta-analysis showed that models predicting mortality performed better (0.90; 95% CI: 0.81, 1.01) than models predicting clinical AE (0.80; 95% CI: 0.79, 0.95). The combined mean c-index was 0.87 (95% CI: 0.79-0.95). Test of heterogeneity showed high variation among studies ( I 2 =98.5%). Egger’s test did not indicate publication bias ( β = 1.48; 95% CI: –18.14, 21.09, p = 0.848). Conclusion: Although relatively few studies have applied ML for predicting AE after TAVR, the results are very promising. The time of complex sophisticated models has arrived with improved predictive accuracy through advanced ML methods able to help identify patients who are at risk for clinical AE early in their care.

Background: The prevalence of obesity among people of reproductive age has steadily increased. The rates of both maternal/parental and fetal adverse outcomes increased significantly with higher BMI. In this study, we examined the national estimates of adverse maternal/parental and fetal outcomes among pregnant individuals with pre-pregnancy obesity using a nationally representative database. Methods: This was a retrospective observational analysis of data retrieved from the Nationwide Inpatient Sample database (2010-2014). The primary outcomes of this study were various adverse maternal/parental and fetal outcomes. Delivery related hospitalizations and all outcomes were identified using the ICD-9 codes. Additionally, propensity-score matching analysis was conducted. Results: There were a total of 18,687,217 delivery-related hospitalizations, of which 1,048,323 were among people with obesity. The mean age of pregnant persons with obesity was 28.5 years while among those without obesity was 28.0 years. Between both groups, most pregnant persons were White, followed by Hispanic and Black. Pregnant persons with obesity were more likely to have labor inductions (AOR: 1.51; 95% CI: 1.42, 1.60), cesarean deliveries (1.70; 1.62, 1.79), and greater length of stay after both cesarean deliveries (1.14; 1.08, 1.36) and vaginal deliveries (1.48; 1.23, 1.77). They were also more likely to have risk factors for adverse obstetrical outcomes such as gestational hypertension (2.17; 2.06, 2.29), preeclampsia (2.06; 1.42, 2.99), gestational diabetes (2.75; 2.60, 2.90), premature rupture of membranes (1.17; 1.08, 1.27), chorioamnionitis (1.39; 1.25, 1.55), and venous thromboembolism (1.63; 1.34, 1.99). Additionally, pregnant persons with obesity were more likely to have adverse fetal outcomes such as excessive fetal growth (3.18; 2.96, 3.43) and fetal distress (1.28; 1.21, 1.35). Conclusion: Pregnant persons with obesity had significantly greater risk for adverse obstetrical outcomes. Risk stratification of pregnant individuals based on obesity can inform obstetrical providers’ clinical decision-making and potentially improve patient outcomes and decrease costs. There is an opportunity to address these risks with pre-conception interventions. Future studies could examine the effect of pre-conception interventions such as counseling on pre-pregnancy weight management, lifestyle modifications addressing nutrition and activity, and pre-pregnancy bariatric surgery.

Introduction: Prior research has shown that hospital outcomes such as mortality among COVID-19 patients receiving methylprednisolone and dexamethasone differed despite having similar anti-inflammatory effect. In this study, we compared mortality rate among three groups (usual care with no steroids [NS], methylprednisolone [MP], or dexamethasone [DE]) of COVID-19 patients. Methods: Data was collected retrospectively from a large healthcare system in the South Florida region. All eligible COVID-19 patients from April 1 to Dec 31, 2020 were included in this study. Propensity scores (PS) was utilized to match patients on their demographics and acuity scores (Age Group, Ethnicity, Gender, APR risk of mortality, and APR severity of illness, etc.). All-cause mortality was compared among the three groups after accounting for demographics, length of stay (hospital and ICU), days on MP or DE, ventilator status, days on ventilator, other medication history (HCQ, Tocilizumab, etc.), pO2, and lab values (hCRP, Ferritin, Glucose, etc.). Multivariate Logistic regression and Survival analysis was utilized for the analysis. Results: There were a total of 2445 eligible COVID-19 cases. After PS matching, 292 cases in each group (n = 876) were included in the analysis. There were 42% females and the majority of participants were White Hispanic (81%). The average age (SD) of the PS matched cohort was 64 (17.1) years, and there was a significant difference between LOS among the three groups with MP being the highest (Mean: 12.4; SD: 12.1). All-cause mortalities in the NS, MP, and DE groups were about 6%, 19% and 7% (p < 0.0001) respectively. Logistic regression showed that the odds of mortality among NS (AOR: 1.9, 95% CI: 0.10, 41.8) group were higher, and DE (AOR: 0.15, 95% CI: 0.04, 0.57) group were significantly lower than MP group respectively. Similar results were observed in the survival analysis after accounting for total hospital length of stay. The adjusted hazard ratios suggest that patients who were in the DE group had lower mortality as compared to MP group but, the effect was not significant (HR: 0.70, p < 0.18). In a subgroup survival analysis among patients who were ventilated (n = 69) in the matched cohort, mortality was higher in DE group compared to MP group however the effect was not significant (HR: 1.8, 95% CI: 0.82, 4.0). Conclusions: Among hospitalized COVID-19 cases, preliminary results show that patients who received DE had lower mortality as compared to those who received MP. Among COVID-19 patients who require mechanical ventilation, sufficiently dosed MP may lead to a decreased mortality as compared to DE. However more in depth analysis with higher number of patients and better characterization of variables such as dosage of the drugs, are needed to confirm these findings.

OBJECTIVE The objective of this study was to understand recent trends in direct health care expenditures among cancer survivors using novel cost-estimation methods and a nationally representative database. MATERIALS AND METHODS This study was a retrospective analysis of 193,003 adults, ≥18 years of age, using the Medical Expenditure Panel Survey during the years 2009-2016. Manning and Mullahy two-part model was used to calculate adjusted mean and incremental medical expenditures after adjusting for covariates. RESULTS The mean direct annual health care expenditure among cancer survivors ($13,025.0 [$12,572.0 to $13,478.0]) was nearly 3 times greater than noncancer participants ($4689.3 [$4589.2 to $4789.3]) and were mainly spent on inpatient services, office-based visits, and prescription medications. Cancer survivors had an additional health care expenditure of $4407.6 ($3877.6, $4937.6) per person per year, compared with noncancer participants after adjusting for covariates (P

Objective: There are several challenges in implementing models for predicting stroke or stroke related outcomes. Most of these models have average concordance, and several of the important variables cannot be modified. In this study, we updated and performed a meta-analysis of commonly utilized models to predict stroke related outcomes. Our primary aim was to evaluate the discriminative ability of the concordance statistic by adding additional studies. Methods: Studies reporting c-index and SE (or 95% CI) for predicting stroke or related outcomes were included in our analysis. In addition to the c-index, total participants, year of publication, type of analytical method (survival, logistic regression, neural network, etc.) and type of outcome (predicting stroke or composite outcome) were utilized. Combined effect sizes with the random model, test for heterogeneity, and publication bias were considered. Egger’s test was used to assess funnel asymmetry. Results: Twenty-seven models were included (patients= 1762461; c-index=14, Harrell’s c-index= 13; only stroke =21, composite=6) in the analysis. Combined mean c-index was 0.76 (95% CI: 0.71, 0.81; 95% predictive interval: 0.59, 0.93). Combined mean Harrell’s c-index was 0.65 (95% CI: 0.61, 0.69; 95% predictive interval: 0.56, 0.74). Test of heterogeneity showed high variation between studies reporting c-index and Harrell’s c-index (I2=97.49% and 80.0% respectively). Egger’s test intercept was -2.1 (95% CI: -7.2, 3.0, P > .40) for c-statistic and 1.2 (95% CI: -1.2, 3.5, P > .32) for Harrell’s c-index studies. Conclusion: Current studies have not improved the prediction interval significantly as compared to our previous meta-analysis for predictive or explanatory model available for stroke risk. However, recent studies were found to be more inclusive of non-traditional biomarkers (e.g., genetic, or polygenic scores) and utilized various machine learning methods that were not used before.

Background: Acute ischemic stroke (AIS) patients show a good prognosis with Carotid endarterectomy (CEA). However, lowering short-term readmissions after CEA for AIS cases remains a significant challenge. Several machine learning (ML) models predicting readmissions fail to explain the predictors. Hence the utility of such ML models in clinical settings is limited. This study used interpretable machine learning (ML) methods to identify the predictors associated with 30-day readmission. Methods: We utilized the National Surgical Quality Improvement Program registry (2014-2017) for this study. Patients aged >18 years and who underwent CEA for AIS were included. ICD-9, ICD-10, and CPT codes were used to identify AIS and CEA cases. Decision Trees and Random Forest classification techniques were utilized to identify predictors of 30-day readmission. Results: A total of 22,373 AIS patients underwent CEA during the study period. The mean (SD) age of the patients was 70.7 (9.4) years, and the majority (61%) were males. About 80% were Non-Hispanic White, followed by non-Hispanic Black (4.6%). About 7% of AIS patients who underwent CEA had 30-day readmission. Random Forest classification and Decision Tree were able to provide clinically relevant predictors and cut-off values. For example, one of the top predictors was pre-operative Hematocrit, and its cut-off value of ≤33% with Diabetes showed a higher risk for readmission. Conclusion: Our study showed that interpretable ML models could be helpful for clinicians to stratify patients based on their pre-operative risk for a 30-day readmission and could help optimize management strategies for improving patient outcomes. Incorporating these cut-off values in EMR could help clinical decision-making and plan interventions to reduce readmissions by managing risk factors. This would improve the quality of care and save high healthcare costs.

Background: Carotid endarterectomy (CEA) is associated with improved overall clinical outcomes in patients with acute ischemic stroke (AIS). However, studies on rates and factors associated with readmission following CEA for AIS are scarce. In this study, we used machine learning (ML) methods to identify the factors associated with readmission using a large-scale national database. Methods: We used National Surgical Quality Improvement Program (NSQIP) registry (2014-2017) and included patients 18 years or older, who underwent CEA for AIS. AIS and CEA were identified using ICD-9 and ICD-10 diagnosis and CPT procedure codes, respectively. We used Naïve Bayes, Boosted Decision Trees, and Bootstrapped Random Forest classification techniques to explore the predictors of 30-day readmission using demographics, past medical history, and preoperative variables. Results: There were a total of 22,373 AIS patients who underwent CEA. Mean (SD) age of the patients was 70.7 (9.4) years, and 61% were men. Majority were non-Hispanic White (80%), followed by non-Hispanic Black (4.6%). During the study period, 1 in 15 AIS patients who underwent CEA experienced 30-day readmission. Bootstrapped Random Forest classification performed best and Naïve Bayes worst with an AUROC of 92% and 59% respectively. The top 5 predictors of 30-day readmission after CEA were Hematocrit, BUN, Creatinine, WBC count, and Platelet count, all collected pre-operatively. Conclusion: Our study showed that ML techniques could accurately predict 30-day readmission using pre-operative risk factors. This ML model could be incorporated in EMR as a potential clinical decision support system. Implementing this system could help in early identification of patients who are at high risk for readmission following CEA. This could help physicians to plan and intervene effectively and prevent short-term readmissions; thereby improving quality of care and saving healthcare costs.

Background: Studies report that acute ischemic stroke (AIS) affects males and females differently. For example, the treatment outcomes of intra-arterial thrombolysis differ between males and females. In this study, we examined mortality and 30-day readmission differences by sex among AIS patients who had carotid endarterectomy (CEA). Methods: We used data from National Surgical Quality Improvement Program (NSQIP) registry (2014-2017). Patients ≥18 years of age, with CEA for AIS were included. AIS and CEA were identified using ICD-9 and ICD-10 diagnosis, and CPT codes, respectively. Using machine learning methods such as Hierarchical clustering, we grouped patients (low, medium, and high-risk clusters) based on their demographics, past medical history, and preoperative variables. Differences in means, and differences in proportions were calculated. Logistic regression was conducted for 30-day readmission and survival analysis for mortality, accounting for cluster groups and sex. Results: There were a total of 22,373 AIS patients who received CEA treatment. Mean (SD) age of the sample was 70.7 (9.4) years, and 61% were males while 39% were females. Mortality rates were 0.8% and 0.7% for men and women (p=0.113), respectively. Thirty-day readmission rates were 6.3% and 7.6% for men and women (p Conclusion: Our study found sex related disparities in short-term readmissions as it was higher among females. This could be because of underlaying sex specific pathophysiology of AIS. Healthcare providers should consider sex-specific management to improve post-stroke recovery for women and reduce their excess burden.

Introduction: Statins use for secondary cardiovascular disease (CVD) prevention is associated with better health outcomes and lower cost among adults with ASCVD. We aimed to examine the prevalence of statin use among adults with ASCVD in the US and to assess for gender and racial differences in statin use. Methods: We identified adults aged 40-75 years with ASCVD (defined using ICD codes and self-reported history) from the 2008-17 Medical Expenditure Panel Survey (MEPS) data. We estimated the trends in statin use among them, and used joint point regression (JPR) to estimate the annual percentage change in statin use. Logistic regression was used to assess the racial variation in statin use. We applied the weights in the MEPS complex survey design to estimate population sizes, so all results are nationally representative. Results: Between 2008 and 2017, 24896 adults aged 40-75 years with ASCVD were identified. In 2017, there were 2445 patients with ASCVD (equivalent to 24.5 million adults). In 2017, the mean age [SE] was 62.2 [0.3] years, and 42.9% were females. The overall proportion of adults with ASCVD using statin increased from 50.0% in 2008 to 58.7% in 2017, with an average annual percentage change of 0.95% between 2010 and 2017 ( p= 0.01). As shown in Figure 1a , among those with ASCVD, statin use was persistently lower in women than in men. In 2017, compared to White adults with ASCVD, Blacks (OR=0.69; 95% CI: 0.51-0.92) and Hispanics (OR=0.62; 95% CI: 0.45-0.85) were less likely to be on statin. Each year, over 3 million women with ASCVD were not on statin, and majority of them were Blacks and Hispanics. (Figure 1b) Conclusion: Millions of adults with ASCVD are not on statins in the US while gender and racial inequalities in statin use remains, despite persistent guideline recommendations. Herein lies an opportunity to improve CVD-related outcomes and cost by intensifying efforts to use statins for the secondary prevention of CVD and closing gender and racial gaps, especially with the 2030 sustainable development goals in view.

Background: Major adverse cardiovascular and cerebrovascular events (MACCE) are significant causes of perioperative morbidity and mortality but, the incidence and effects following cancer surgeries are unknown. The aims of this study were to evaluate national trends in MACCE after major cancer surgeries and to identify cancer types associated with cardiovascular events using a large national database. Methods: Patients who had major cancer surgeries from 2005 to 2014 were identified from the National Inpatient Sample database. Hospitalizations for surgeries for cancer of prostate, bladder, esophagus, pancreas, lung, liver, breast, colon and rectum were identified by ICD9 diagnosis and procedure codes. The main outcome was perioperative MACCE, defined as in-hospital, all-cause death, acute myocardial infarction (AMI), or acute ischemic stroke, and was evaluated over time. Results: Among 2,854,810 hospitalizations for major cancer surgeries, perioperative MACCE occurred in 67,316 hospitalizations (2.4%). Mean (SE) age of patients was 65.4 (0.07) years and 54.2% were male patients. MACCE occurred most frequently in patients undergoing surgeries for lung (6.8%), pancreatic (4.5%), and colorectal (3.3%) cancers. Between 2005 and 2014, the frequency of MACCE declined from 2.7% to 2.2% ( P P P = 0.002). However, no significant changes were observed for acute ischemic stroke ( P = 0.6) during the study period. Conclusion: Perioperative MACCE occurs in 1 out of every 42 hospitalizations for major cancer surgeries. Despite reductions in the rate of death and AMI among patients undergoing major cardiac surgeries, perioperative ischemic stroke remained constant over time. The lack of improvements in perioperative ischemic stroke rate is concerning and requires additional interventions. Significant efforts should be directed towards improving cardiovascular care during the perioperative period of cancer surgeries.

Aims: The Cardiac Lipid Panel (CLP) is a newly discovered panel of metabolite-based biomarkers that has previously shown to improve the diagnostic value of NT-proBNP. Little is known about its usefulness in predicting outcomes. In this study, the aim was to develop a risk score for 4-year cardiovascular death in elderly CHF patients. Methods and Results: We included 280 patients >65 years of age with CHF randomly selected from the CIBIS-ELD trial. A targeted metabolomic analysis of the three CLP biomarkers was performed on baseline serum samples. Cox proportional hazards regression was used to determine the relationship of the biomarkers with the outcome and adjusted for established risk factors (age, sex, BMI, creatinine, LDL, triglycerides, and LVEF). The CLP risk score was calculated by counting the number of biomarkers above a cut-off determined using the Youden’s indexes. During the follow-up period, mean 50 months [±8 months], 35 (18%) subjects met the primary endpoint of cardiovascular death. The area under the receiver operating curve (AUROC) for the model based on only clinical variables was 0.72, the second model with clinical variables plus NT-proBNP was 0.79, and the final model with clinical variables, NT-proBNP, plus the CLP was 0.83. Conclusions: A risk score based on a panel of 3 metabolite-based biomarkers was found to improve the prognostic value of traditional risk factors and NT-proBNP in a cohort of CHF subjects. Our next steps include validating the prognostic value of the CLP biomarker kit in order to provide a better clinical risk assessment of CHF patients.

Introduction: Cardiometabolic syndrome consists of a cluster of metabolic dysfunctions such as impaired glucose tolerance, insulin resistance, dyslipidemia, central adiposity, and hypertension. According to the latest estimates, globally, nearly 25% of all adults have cardiometabolic syndrome. Both cardiometabolic syndrome and cancer pathophysiology commonly involve inflammation and oxidative stress. The objective of this systematic review was to evaluate existing evidences that support the association between cardiometabolic syndrome and risk of developing cancer. Methods: We searched MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials and Scopus for relevant articles published from the database inception until October 2019. We followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines for this review. Using the Oxford Center for Evidence-Based Medicine guidelines individual studies were evaluated. A total of 59 articles were included in this study. Results: Our review showed that cardiometabolic syndrome was associated with increased risk for colorectal, hepatic, endometrial, breast, and bladder cancers. These associations showed variations for sex and geographical locations. For example, the associations were stronger for pancreatic and rectal cancers among women. The strength of these associations was also stronger for sex specific cancers such as breast and endometrial cancers. Studies on European populations showed that these associations were stronger for colorectal cancer among women. However, one study showed that presence of cardiometabolic syndrome contributed protective effects to prostate cancer among American men. In general, strongest associations were found for colorectal cancer among both men and women and hepatic cancer among men. Among cardiometabolic factors, impaired glucose tolerance and central adiposity were the greatest contributors towards increased risk for cancers. Conclusion: Given these results, there should be greater focus on primary prevention to identify and treat cardiometabolic risk factors. In addition, patients with greater cardiometabolic risk factors should be screened earlier and more frequently for cancers. A number of gender and geographical gaps identified in this review could be targeted for improvements as per the goals of 2030 sustainable development initiatives. Future studies should consider cardiometabolic syndrome and cancer together and develop effective interventions for decreasing the incidence and morbidity associated with both the conditions.

Background: In August 2019, FDA approved TAVR for low-risk patients, a paradigm shift in the field of Cardiology. This approval will pave the way for TAVR to rapidly become mainstream, and it will be a preferred procedure for patients with symptomatic aortic stenosis in the coming years. In this study, we considered creating a useful tool to assess the risk of 30-day readmission after TAVR. Methods: We utilized data from ACS National Surgical Quality Improvement Program (ACS NSQIP) between 2015 and 2017. We selected 35 past medical history (such as history of CHF, COPD, etc.) and preoperative lab (such as BUN, WBC count, etc.) variables. A binary variable for 30-day readmission was created. Elastic net, random forest, and gradient boosted classification trees machine learning models were run and best performing model was selected based on the AUROC. Variables were plotted according to the importance. We created a nomogram using logistic regression by including important variables. Methods: About 594 patients underwent TAVR between 2015 and 2017; 56% patients were 80 years or older; and 45% were females. Majority of patients (62.8%) were Non-Hispanic White, had BMI ≥ 25 (71%), and took antihypertensive medication (82%). History of CHF was reported by 26%, and COPD by 17%. There were 68 (11.4%) patients who were readmitted within 30-days for any cause. Gradient boosted classification performed best. Top 15 variables were selected for logistic regression analysis to construct a nomogram. Five of the top variables were Total hospital LOS, pre-op platelet count, pre-op serum sodium, general anesthesia, pre-op BUN. Conclusion: In the US, 1 in 6 patients with TAVR could be readmitted within 30-days due to any cause after TAVR. We presented an interpretable model based on past medical history and pre-operative variables, which can be used to identify high risk patients so that their hospital readmission risk and can be reduced. This will also lower financial burden among patients.

Stigma is a primary barrier to care and support for people living with HIV (PLWH). We explored relationships between HIV-related stigma and treatment adherence and the effects of psychological and structural factors on these relationships. HIV treatment adherence, stigma, and coping strategies were measured with questionnaires. Participants included 285 PLWH in Haiti. Multivariable linear regression was used to estimate predictors of treatment adherence. Structural equation modeling was used to determine whether relationships between stigma and treatment adherence variables were mediated by coping variables. Mean adherence was 93.1%; 72.3% of participants reported ≥ 95% adherence. Perceived stigma and quality-of-care satisfaction scores significantly predicted treatment adherence. Maladaptive coping did not act as a mediator between perceived stigma and treatment adherence, which could be due to stronger effects of perceived stigma on treatment adherence. Our study may help to improve treatment adherence and the care and quality of life for PLWH.

Objective: Maternal pre-pregnancy hypertension is an established risk factor for preterm births and infant deaths. We sought to investigate the effects of maternal pre-pregnancy hypertension and tobacco use on preterm and infant deaths. Methods: We analyzed National Vital Statistics System’s 2013 and 2015 linked birth/ infant death data from a total sample of 6,629,652 live births. There were 11,782 preterm deaths due to “Newborn affected by maternal factors and by complications.” Maternal factors and complications were defined as deaths at less than 1 year of age with ICD-10 codes P00-P04. Regression model was accounted for tobacco use, cigarette use before and during pregnancy, prenatal care, maternal age and demographics. Results: Disease specific preterm mortality due to pre-pregnancy hypertension was 1.9 per/1000 live births. Preterm risk increased by two folds (adjusted odds ratio (AOR)=2.12; 95% confidence interval (CI) 1.92-2.18) with maternal pre-pregnancy hypertension history. Presence of pre-pregnancy hypertension and tobacco use doubled the risk of preterm (AOR=3.82 (95% CI: 3.67-4.82). African American mothers with pre-pregnancy hypertension were 1.72 time more likely to have preterm mortality than White females. Conclusion: Black race, maternal risk factor pre-pregnancy hypertension and tobacco use during pregnancy were associated with higher risk of preterm mortality. Further investigation is crucial to better understand the risk factors for the disparities in preterm so that women who are at risk pre-pregnancy can be identified and provided risk-specific services.

Background: Studies have shown that implantation of cardiac resynchronization therapy (CRT) devices could reduce morbidity and mortality in heart failure patients. Aims of this study are to examine the temporal trends and outcomes in CRT device implanted cancer survivors. Method: This study was a retrospective analysis of Nationwide Inpatient Sample during the period 2005-2014. Patients ≥18 years and who underwent CRT device implantation were included in the study. CCS codes 11-44 were used to identify cancer survivors. The International Classification of Diseases-9th Revision-Clinical Modification (ICD-9-CM) primary procedure codes 00.50 and 00.51 were used to identify patients with device implantation. Association between cancer diagnosis and in-hospital mortality and complications were determined using multivariable analyses. Result: During 2005 to 2014, a total of 392,758 inpatient CRT implantations were performed in the United States. Nearly 8.3% of these procedures were performed among patients with history of cancer. Number of procedures among cancer survivors decreased from 3127 in 2005 to 2115 in 2015. However, proportion of cancer patients undergoing CRT implantations increased from 6.8% in 2005 to 10.3% in 2014 (relative increase: 51.4%, P Discussion: Our study found that proportion of CRT implantations among cancer survivors increased from 6.8% to 10.3% during the study period. In addition, CRT implantations among cancer survivors were associated higher in-hospital complications. Strategies like prehabilitation and close collaboration between cardiologists and oncologists could improve outcome and general well-being of cancer survivors undergoing CRT implantations.

Background: Atherosclerotic cardiovascular disease (ASCVD) remains a nation-wide crisis. Optimal cardiovascular risk factor (CRF) profile has been associated with lower mortality and morbidity, as well as lower healthcare expenditure and resource utilization. In this study, we examined the association of optimal CRF profile to financial hardship and other cost-related barriers to care, both with and without ASCVD. Methods: We used a nationally representative sample of non-elderly adults aged between 18-64 years from the National Health Interview Survey (2013-2017). We assessed atherosclerotic cardiovascular disease (ASCVD) status, CRF profile, financial hardship from medical bills, unable to pay bill at all, cost-related medication non-adherence, forgone/delayed care, and perceived financial distress from self-reported questionnaire. Results: A total of 119,388 individuals were included in the study. Individuals with ASCVD had financial hardship and inability to pay bills at 45% and 19%, respectively, significantly higher than those without ASCVD and optimal CRF profile at 24% and 6%, respectively. Similarly, the prevalence of other cost-related barriers to care was significantly higher among ASCVD (>2-fold) compared to non-ASCVD with optimal CRF profile. Individuals without ASCVD and optimal CRF profile among low/poor income and uninsured had higher prevalence of all the outcomes when compared to ASCVD among high income and insured. Individuals without ASCVD and optimal CRF profile had lowest odds of financial hardship (OR 0.44, 95% CI 0.41, 0.48), inability paying bills (OR 0.30, 95% CI 0.26, 0.34), cost-related medication non-adherence (OR 0.42, 95% CI 0.38, 0.48), foregone/delayed care (OR 0.41, 95% CI 0.37, 0.45), and high financial distress (OR 0.52, 95% CI 0.47, 0.58) (Table). There were no significant differences in these outcomes when comparing individuals without ASCVD and poor CRF profile to those with ASCVD. Conclusion: Optimal CRF profile is strongly associated with lower prevalence of financial hardship and other cost-related barriers to care. Upfront investment of preventive health programs, especially among vulnerable populations, may improve overall health and decrease financial hardship.

Background: Self-insured employers, which are majority in US, face an increasing financial burden as health care costs have increased relative to savings. By applying machine learning (ML) techniques, we may decrease unnecessary hospitalizations by identifying low risk patients, who are on the path to become high risk, and eventually high cost. In this complex simulation, we applied and compared several ML techniques on data gathered from medical insurance claims. Methods: The analysis was limited to employees and their spouses only. We identified about 8000 employees and spouses who were covered by employer sponsored health insurance plan between 2011 and 2016. De-identified data were used to predict high cost claimants. High cost was defined as annual spending of over $10,000 in medical claims. We utilized and compared methods namely: bootstrapped random forest, gradient boosted tree, neural networks, Naïve Bayes classifier, and logistic regression with LASSO. Variables in the model included yearly inpatient and ER visits, CCS categories, CPT codes, and demographic information. Results: The mean (SD) age of eligible population was 49.7 (11.0) years. Approximately, the employee to spouse ratio each year was 3.5:1. Results show that bootstrapped random forest performed better than other techniques (AUROC: 80.33%). All other algorithms also performed extremely well. Future progress of these models will be oriented towards including employee health fair data and also strategies to avoid over fitting. Conclusion: Healthcare is way behind other industries in successfully using ML to improve processes and disrupt one industry that is admittedly in need of positive disruption. Models allowing successful preventive actions will both increase revenue and lower costs. Application of modern ML techniques can identify high cost claimants earlier. Our approach is one of the first applications of significant recent advances in ML algorithms and computer technologies, opening new opportunities for the future analysis of high healthcare costs.

Objective: The accuracy and applicability of various cardiovascular disease (CVD) risk calculators may not be same in different populations. Wecompared 5 RFS for CVD risk on patients from Bosnia and Herzegovina to explore relations between different RFS and estimate use of theircombination. Methods: We utilized demographic, medical history, and lipid profile data gathered from patients seen in primary care clinic in 2013. FiveRFS namely, 10 year Framingham CHD score, QRISK 2, AHA/ACC ASCVD risk score, Framingham ATP III score, and EU SCORE were calculatedusing demographic, history of diabetes and hypertension, taking any hypertension medication, and lipid profile variables. Additionally, we entered these scores as dimensions in hierarchical cluster analysis to group people based on their risk of developing CVD. Results: There were 1277 patients in thisstudy and majority (65%) were women. The mean (SD) age of the sample was 56.2 (11.4) years. Correlation between these scores are presented inTable 1. We obtained 4 clusters with significant different cluster centers. Clusters were ordered from lowest to highest risk; cluster 1 containing patients with lowest and cluster 4 with highest mean RFS. Cluster centers, which represents mean values for all the RFS are given in Table 2. Conclusion: All risk scores performed well in this population. Only EU SCORE correlated less with other RFS. Obtained clusters are more homogenous and were ableto classify patients better. This novel method of using calculated RFCs as dimensions in clustering produced very good estimates of patient’s risk ofdeveloping CVD by combining several RFS, even in a case where originally RFS were created for and from different population.

Background: Hospital Readmissions Reduction Program requires CMS to reduce payments to hospitals with excess readmissions. Trend in 30 days readmission among patients with index hospitalization for stroke in the United States is largely unknown. In this study, national trend was examined to estimate quality of care in the US population. Methods: Patients, ≥40 years, with index hospitalization for stroke were extracted from the Nationwide Readmission Database (2010-2015). They were identified using ICD9 and ICD10 codes (433.x, 434.x, 435.x, 436.x, I63.x). Readmissions were defined as all cause readmission during 30 days post index hospitalization discharge, and were divided into planned and unplanned according to AHRQ guidelines. Weighted percentages and Joinpoint analysis were used to evaluate 30 days readmission trends and the annual percentage change (APC). Results: Older patients (≥75 years) and those with a higher comorbidity burden were readmitted frequently. Overall, there was a downward trend in unplanned 30 days readmissions between 2010 and 2015; the lowest rate (95% CI), was seen in 2014 (10.7%; 10.5% - 10.8%) and the highest in 2010 (11.6%; 11.4% - 11.8%). From 2010 to 2015, reduction in APC was 1.3% ( P > .08). Conclusions: Around 10%-12% of patients with stroke get readmitted within 30 days post discharge. Not only there is a high clinical and economic burden associated with readmissions due to stroke, it is also associated with poor survival. This highlights the need for interventions ensuring appropriate level of service and early outpatient follow-up after discharge. Reducing postoperative complications; managing behavioral risk factors (smoking, alcohol, diabetes management); or choosing appropriate carotid revascularization technique such as endarterectomy over stenting may reduce 30 days unplanned readmission and the high risk of mortality among stroke patients. Efforts should be made among elderly patients so that there is a decrease in trend.

Objective: Models predicting stroke suffer from several problems: they are not accurate, contain mostly non-modifiable factors, and they explain more than they predict. We performed a meta-analysis of frequently used stroke models. Our goal was to estimate the discriminative ability of the concordance statistic by establishing confidence and predictive intervals. Methods: Studies with most representative predictive methods were used in our analysis. Inputs in analysis were study-reported c-index values and corresponding 95% CI. Subgroup analysis, separating survival and logistic regression models, and separating models by outcome (predicting stroke or composite outcome) were executed. Combined effect sizes with the random model, test for heterogeneity, and publication bias were considered. Egger’s test was used to assess for funnel asymmetry. Results: Fifteen models were included (patients= 13177; LR=7, Cox= 8; only stroke =9, composite=6) in the analysis. Models predicting composite outcomes performed worse (0.62; 95% CI: 0.59, 0.65) than model predicting only stroke (0.70; 95% CI: 0.64, 0.75); combined mean c-index was 0.66 (95% CI: 0.62, 0.70; 95% predictive interval: 0.55, 0.78). Test of heterogeneity showed high variation between studies (I 2 =74.9%). Egger’s test intercept was 1.59 (95% CI: –0.80, 3.98, P > .17). Conclusion: At the moment, there is no good quality predictive or explanatory model available for stroke risk. Current models do not indicate towards risk factors exclusive only to stroke. Because of this, models predicting stroke mostly have non-modifiable risk factors. In addition to this, models orienting toward explanation are borrowing heavily from models developed for predicting composite cardiovascular diseases.

Introduction: Every year more than 75 million adults are diagnosed with HTN in the US. Despite spending $50 billion/ year to lower mortality related to HTN, only about 54% of patients have this condition managed. In 2014, mortality due to HTN or related complications was 410,000. AMI is a more severe manifestation of coronary artery disease with 735,000 people diagnosed yearly. It is estimated that hospitals lose $4493-$7940 per patient due to AMI. We studied costs for events associated with HTN and AMI in MEPS. Methods: Individuals aged ≥20 in MEPS (2014) were included. Hierarchical clustering was used for analysis. Age, sex, education status, race, Hispanic ethnicity, US citizenship status, family income, insurance, people who reported HTN and/ or AMI events were entered as dimensions. Cluster and descriptive analyses were adjusted for survey weights. Results: About 236 million weighted individuals were eligible for the analysis (Female: 51.7%; White: 79.1%; High school or above: 61%; Any private insurance: 66%; and Income≥ 400% of poverty line: 41%). Out of these, about 4.9 million participants reported costs/ events related to AMI and 61.8 million related to HTN. Five groups were identified based on similarity within each cluster (Table 1). The maximum number of participants were present in cluster 1 (weighted n = 101 million) with the most number of participants (21.5%) who reported HTN among all cluster groups. In general, cluster 5 had the lowest annual total of direct health care payments and events. Selected medical history is described in Table 2. Conclusion: Results show insights into the cost of HTN and AMI events, and their relationship with other comorbid conditions. Being computationally demanding, clustering methodologies are rarely utilized in survey data analysis. This could be the first example of clustering approach applied to data from MEPS. Once clustering methods are improved in weighted survey data, estimates will be more reliable.

Background: Each year in the USA, more than 75 million adults are diagnosed with hypertension (HTN), but less than 54% have it condition under control. Mortality due to HTN or related complications was 410,000 in 2014 and resulted in close to $50 billion spent. Acute Myocardial Infarction (AMI) is the most dramatic manifestation of coronary artery disease. Every year, about 735,000 people are diagnosed with AMI and it is estimated that hospitals lose $4493-$7940 per AMI patient on hospitalization. We studied interplays of expenditures for all events associated with HTN and AMI in the US in 2014. Methods: We used all available subsets from Medical Expenditure Panel Survey (MEPS; 2014) to extract all cost and utilization events related to HTN and AMI, following procedures for complex surveys as described in MEPS specifications to define population influenced by HTN and AMI, number of events, and related costs. Patients were included in this analysis if they were ≥20 years. Results: In 2014, there were about 236 million weighted participants in MEPS. Out of these, about 4.9 million participants reported costs related to AMI and 61.8 million related to HTN. About 3.9 million people reported both. Out of 61 million with HTN, 4 million reported AMI, whereas out of 174 million without HTN, only 1 million had AMI (OR: 11.1; 7.9-15.5). Those who had events associated with AMI, with an average cost of $5328/event, adding to total expenditures of $26.6 billion. Cost of AMI, HTN and both were $6.8 billion, $4.1 billion, and $2.7 billion, respectively. The total cost of hospitalization due to both HTN and AMI was 31.4% of the total healthcare expenditure among these participants. Conclusion: Burden of AMI includes almost 5 million people who live and die with AMI, and pay a high cost to manage this condition. Evidence from this study suggests that there is a significant disparity between the cost of hospitalization due to AMI between people with HTN and without HTN, which has potential implications on outcomes, quality of care, and hospital sustainability for AMI population. Symptom management at primary care settings may decrease the number of AMI events resulting in lower spending and cost savings by hospitals.

Introduction: According to the National Institute on Drug Abuse, lifetime cocaine use has remained stable since 2009. An estimated 1.5 million individuals aged ≥12 years reported cocaine use in 2014. Cocaine use is a significant risk factor for acute myocardial infarction (AMI), especially in the age group 18-45 years. Hence, we examined trends in hospitalizations due to cocaine-induced AMI in the United States. Methods: The current study was a retrospective analysis of the National Inpatient Sample, collected during the period 2005-2014. Participants between 18 and 45 years were included in the analysis. Cocaine-induced AMI hospitalizations were identified using ICD-9-CM codes 304.2X and 305.6X for cocaine dependence or abuse, and 410.XX for AMI diagnosis. We used Cochran Armitage test for categorical variables and Cuzick nonparametric test for trends for continuous variables. Results: A total of 49,715 weighted cocaine-induced AMI hospitalizations were reported during the study period. Cocaine-induced AMI hospitalizations increased from 4,619 in 2005 to 5,065 in 2014 (relative increase, 9.7%, P trend =0.124). Majority of the patients were non-White (55.3%), men (79.3%), in the age group 41-45 years (83.1%), and with Medicare/Medicaid coverage (51.2%). Highest change in trend was observed for non-White population and Medicare/Medicaid payers. Conclusion: Our study showed that cocaine-induced AMI hospitalizations remained stable during the years 2005-2014. Initial results point towards structural and racial disparities in such hospitalizations, especially among non-White men and population with Medicare/Medicaid coverage. Effective interventions to reduce cocaine use are needed, and treatment must be based on 2008 AHA guidelines for cocaine-associated chest pain and myocardial infarction.

Background: Lipoproteins-rich triglycerides are associated with markers of coronary atherosclerosis such as coronary artery calcification (CAC) in young high-risk population. This study examines differences between sexes associated with lipoprotein sub-fractions and atherosclerosis. Methods: The study presents analysis from baseline data of a randomized trial of employees of Baptist Health South Florida with metabolic syndrome or diabetes. Participants completed lipoprotein sub-fraction analysis at baseline using ion mobility technique. Subjects above 35 years had a CAC test completed. We tested for differences between correlational matrices of the original variables. The same analysis was repeated for sets of 3 principal components (PC) that were computed for the combination of all lipoprotein subclasses. Results: The study population (N=170) was largely female (128) with a mean age of 58 years. Three PCs accounted for 83% variation in the sample. Distribution of atherosclerostic PC2 by sex was entirely different between sexes (Wishart test p=.004). In addition, a significant Jennrich’s test (p < .0045) implied that the inter-correlations among all dependent variables for different groups were dissimilar; with the finding of significant angles between PC components by sex. PC2, with main contributions from VLDL particles, was in the positive direction; however PC3 with large LDL particles was in the negative direction. Atherogenic lipoprotein profiles were significant only for males, with PC keeping their independent predictability. Conclusion: In a relatively young high-risk population, there were sex-specific differences in lipoprotein sub-fractions profiles that are associated with increased risk of atherosclerosis. The relationship was much weaker for males compared to females (see graph). These findings points towards a possible sex-linked lipoprotein sub-fraction association with the atherosclerosis pathogenesis.

Background: Health insurance is effective in preventing financial hardship from unexpected major healthcare events. However, to determine whether vulnerable patients are adequately protected from longitudinal health care costs, it is also essential to assess the financial impact of cumulative out of-pocket (OOP) healthcare expenses for common chronic conditions like ASCVD, particularly for low-income families. Methods: Using the Medical Expenditure Panel Survey (2006-2015), we assessed the annual inflation-adjusted OOP expenses, inclusive of insurance premiums, for all patients with ASCVD (coronary, cerebral, or peripheral vascular disease). We assessed these expenses at a family-level, i.e. for families with ≥1 members with ASCVD, and compared against annual family incomes. Low-income families were defined by an income 20% and >40% of family income defined high, and catastrophic health expenses, respectively. Survey methods were used to obtain national estimates. Results: We identified 22521 adults (≥18y) with ASCVD, represented in 20600 families in MEPS. They correspond to an annual estimated 23 million or 9.9% of US adults, and 21 million or 15% of all US families. Of these, 39% were low-income. The median annual family OOP expenses were $2450 (IQR 725, 5343), and income was $39765 (IQR 19080, 77635). Overall, OOP expenses represented a median 6.4% (IQR: 1.3, 19.1) of income in low-income, and 5.5% (IQR 2.6, 10.4) in mid/high income families. Low-income families had over 3-fold and 9-fold higher odds than mid/high income families of expenses of >20% of income (24.1% vs 8.1%, OR 3.6 [3.2, 4.1]), and >40% of income (11.2% vs 1.4%, OR 9.2 [7.5, 11.2]) (Fig A). While more low-income families gained insurance coverage (90% in 2006 to 93% in 2015), those with insurance had higher rates of OOP expenses than those without insurance (Fig B/C) or any mid/high income insurance subgroups. Conclusion: 1 in 4 low-income families with a member with ASCVD experience a high financial burden, and 1 in 10 experience a catastrophic healthcare expense, even with insurance coverage. Underinsurance of low-income families has improved minimally. To alleviate economic disparities, policy interventions must extend focus to improving the quality of coverage for low-income families.

Introduction: Previous studies have linked depression and cardiovascular diseases, however gender differences in cost of hospitalization and care associated with events related to depression and myocardial infarction (MI) is not studied in detail. We utilized data from 2014 Medical Expenditure Panel Survey (MEPS) to evaluate national estimates of such costs. Hypothesis: Proportion of depression and MI and corresponding healthcare expenditures are high in general population and differ between genders. Also, payments are appropriated differently between payers. Methods: Participants from 2014 MEPS with events attributed to MI and/ or depression were included in this study. Mean (95% CI) event related cost, and total cost of health care was calculated using survey methods. Expenditure and utilization cost was grouped as related to ambulatory, emergency room, inpatient, home visits and medications. By source, payments were grouped as paid by family, MEDICARE, MEDICAID, private insurance, VA, Tricare and other. Results: There were 23486 participants in the study, representing 242,628,543 individuals in the US who were 20 years or above. Total health expenditure in 2014 among these was $1.5 trillion (Males: $696,940,498,022; Females: $837,486,094,699) with $27,937,582,549 attributed to depression (Males: $10,991,761,342; Females: $16,945,821,207) and $51,142,260,003 to MI (Males: $40,676,887,518; Females: $10,465,372,485). There were around 6,174,408 (2.5%) and 27,269,837 (11.2%) events associated with MI and depression respectively. Approximately 901,762 individuals reported both events. Among males, MI was 4,189,696 (3.6%) and among females, 1,984,711 (1.6%). Depression was reported 8,755,276 (7.5%) among males and 18,514,560 (14.6%) among females. Among females who were depressed, 2.3% reported MI, and 1.4% among those who were not depressed. Among males who were depressed, 5.4% reported MI and 3.4% among those who were not depressed. Among depressed males, AMI hospitalization was 0.31% whereas among depressed females, AMI hospitalization was 0.24%. Conclusion: Among both depressed and non-depressed populations, males had significantly higher proportion of MI and hospitalizations related to AMI when compared to females. But, out of total MI costs for males, less than 1% cost was accrued by depressed; whereas, of total MI cost for females, 9.6% of cost was accrued by depressed females showing gender based disparities in healthcare cost and utilization. With Medicare paying between 78%-83% of all MI costs, treating depression can result in significant savings.

Background: Web-based platforms have been proposed as tools to facilitate lifestyle improvement, however, their efficacy in individuals with high cardiometabolic risk has not been adequately tested. The Baptist Employee Healthy Heart Study (BEHHS) was designed to assess the addition of a personalized, interactive, web-based, lifestyle-management program to the existing health-expertise web platform available to BHSF employees with metabolic syndrome (METS) or type 2 diabetes (DM2) Methods: In this 1:1 randomized, non-blinded trial, the intervention arm was provided access to a web-based personalized and interactive lifestyle program that provided targeted and personalized dietary, weight management and physical activity counseling. The intervention was in addition to access to an online wellness program, a non-interactive website that provided information on healthy diet and physical activity. The control group only had access to the online wellness program. At baseline, each participant had their demographic data collected via questionnaire. At each study visit (baseline, 4 months and 12 months) participants completed questionnaires on lifestyle indices such as diet and physical activity, had their weight, height, waist circumference, body fat (by plethysmography) and blood pressure measured. Laboratory testing was done for traditional lipids and glucose at each visit. Results: Of the 182 participants that were randomized, 163 (82 in the intervention arm and 81 in the control arm) completed the baseline survey and had complete laboratory data at baseline. Loss to follow-up was 12% at 4 months 34% at 12 months. As shown the table, intention to treat analysis using both single imputation (last observation carried forward) and multiple imputation techniques showed no difference in BMI, other measures of adiposity, blood pressure, lipids, physical activity and diet scores. When analyses were restricted to completers alone, no significant change in the results were observed. Conclusion: The addition of web-based, personalized lifestyle program to an already existing lifestyle educational platform did not significantly impact healthy lifestyle promotion and cardiometabolic risk in employees with MetS or DM2.

Background: An age, sex, and blood gene expression score (ASGES) has been previously validated to detect obstructive coronary artery disease (CAD) in non-diabetic patients presenting with stable chest pain in the outpatient setting. However, the diagnostic performance of this test in ruling out obstructive CAD in patients presenting with acute chest pain (ACP) to the emergency department (ED) is unknown. Methods: In an ongoing study, 371 low-intermediate risk patients with ACP and no prior history of CAD (TIMI risk score ≤ 2, negative troponins and normal/non-diagnostic ECG) underwent coronary CT angiography (CCTA) using institutional protocols. Patients were classified based on severity of stenosis (obstructive CAD, >50%; high grade stenosis, >70%) and ASGES. The ASGES blood test sample was drawn before ED discharge and analyzed in a commercial reference laboratory (Redwood City, CA). We excluded 23 (6%) patients with unreportable ASGES and 47 (13%) diabetics from this primary analysis. Results: 301 (53±10 years, 45% males, 78% Hispanics) non-diabetic ACP patients undergoing CCTA in an ED setting were included in this analysis. No plaque was detected in 183 (60%) patients, and 22 (7%) patients had obstructive CAD. In this population, 51% of patients had scores below the previously defined threshold of ASGES≤ 15. This threshold yielded sensitivity, specificity, NPV, and PPV of 71% (52-86%), 53% (47-59%), 97% (93-98%), and 12% (9-14%) for obstructive CAD. Furthermore, ASGES≤15 yielded a 100% sensitivity and NPV for patients with high grade stenosis (n=7, 2%). In a multivariable analysis including patient demographics and clinical covariates, ASGES ≤15 was significantly associated with obstructive CAD (OR: 0.15, 95% CI: 0.04-0.62). As a continuous variable, increasing ASGES was positively correlated with the presence of obstructive CAD and CCTA-defined plaque burden (p Conclusions: This is the first study validating the use of this blood-based precision medicine test to rule out obstructive CAD among low-intermediate risk non-diabetic patients presenting with ACP in ED setting. 30-day follow-up is underway to evaluate the prognostic implications of these findings.

Background: The CANTOS trial supports the use of canakinumab to target residual inflammatory risk for secondary cardiovascular prevention. To better understand the potential impact of these findings, we use nationally representative data to estimate the number of US adults who qualify for therapy with canakinumab and the projected cost burden on the US health system based on its current pricing. Methods: In National Health and Nutrition Examination Survey (NHANES) 2005-10, we identified all adults with a prior MI and elevated hsCRP ≥2 mg/L, and fulfilling all eligibility criteria for CANTOS. Further, we also assessed the number of individuals not receiving statins who might be potentially eligible. Statin non-users were considered potentially eligible if they would have had hsCRP ≥2mg/L even after assuming a maximal hsCRP lowering of 50% with statin therapy in all patients. We used survey methods to obtain weighted estimates for the eligible adults in the US population. Annual costs were estimated at $60000/year for canakinumab therapy. Results: Of the 17689 adults sampled in NHANES during 2005-2010, 16608 (94%) had hsCRP testing. Of these, 684 (3.0%), representing an estimated 6.8 million (95% CI 5.7-7.8 million) US adults, had a history of MI (Figure 1). Among those with a prior MI, 42% did not report using a statin. Among statin-users, 44% had an hsCRP Conclusion: Nearly 1 in 3 or 2.1 million US adults with a prior MI are potential treatment candidates. Assuming current canakinumab prices would incur an annual financial burden of up to 124 billion dollars in the United States. Further work is needed to define the population that achieves favorable value with treatment.

Introduction: As number of elderly Americans is projected to double by 2050, population aged ≥65 years will have the highest impact on Medicare spending. The prevalence of elderly population with hypertension (HTN) is around 65% and if patients receive good primary care or an early intervention, these HTN related hospitalizations could be prevented saving millions in Medicare spending. Our objective was to examine the trends in hospitalizations related to HTN and total cost over 10 years among aged ≥65 years. Methods: Using data from Nationwide Inpatient Sample database from 2005-2014, we explored the existence of trends in HTN admission rates and corresponding hospitalization costs among adults ≥65 years vs other age groups. Weighted estimates for rates and mean total cost were reported using SUDAAN after adjusting for complex survey design. Results: Overall, hospitalizations due to HTN increased by 26% in 2014 from 2005. Hospitalizations were highest among patients ≥65 years (57.8%) followed by 40-64 (32.6%) and 18-39 years (9.6%). During 2005-2014, hospitalization rates among ≥65 years increased significantly (10.7%, p p Conclusion: Hospitalizations and costs associated with HTN, especially Medicare, have increased over the past 10 years among elderly. Although these trends are alarming, these events could be prevented through symptom management at primary care, medication adherence, care coordination, or modifying care-seeking behavior resulting in potential savings of 851M over a year. This would reduce burden of Medicare spending, which is projected to surpass the rate of growth in federal revenues.

Objectives: Extensive data suggests that lesbian, gay and bisexual (LGB) adults are more likely to experience adverse cardiovascular outcomes relative to heterosexuals. However, evidence regarding cardiovascular health (CVH) disparities and sexual orientation is scarce. The aim of this study was to examine the distribution of CVH metrics in a US nationally representative population of heterosexual (HT), and LGB adults. Methods: This cross-sectional study analyzed 2445 participants (representing 115 million) adults aged 18 or over years in the 2011-2012 NHANES survey. The CVH factors of smoking, body mass index (BMI), physical activity (PA), diet, blood pressure (BP), total cholesterol (TC) and glucose (GLU) were measured. Each CVH factor was then classified as ideal; intermediate; or poor. Ideal CVH was defined as presence of >=5 ideal CVH metrics. Results: 95.1% of the weighted sample self-identified as HT (95% CI: 93.5%, 96.6%) compared to 4.9% (95%: 3.3%, 6.5%) LGB. The figure illustrates the distribution of each of the 7 CVH categories according to sexual orientation. In age, gender, and race adjusted analysis, LGB individuals were 36% (AOR: 0.64; 95%: 0.29, 1.4; p > 0.05) less likely to have ideal CVH compared to HT. These proportions go higher after adjusting for age. Conclusions: The results suggest that LGB individuals face a higher risk of being in the category for poor cardiovascular health compared to heterosexuals. Evidence suggests that there are sexual orientation disparities among adults. If confirmed in other studies, results point towards disproportionately higher risk for cardiovascular disease among sexual-minority populations. Figure

Introduction: Higher education and income can modify behavioral risk factors for lifetime cardiovascular disease (CVD) risk. The aim of this study was to observe ethnic differences in the CV health, as measured by inadequate cardiovascular health index (CVHI) category, among a cohort of females with high family income, education, and BMI. We also examined differences in perception about self-reported weight and attitudes toward weight loss and exercise. Methods: Adult non-pregnant females aged 20 years or older from the 2011-2012 NHANES were included in the study. Participants were at least college educated, had a poverty to income ratio of 350% or above, and BMI (≥ 30). Weighted regression was adjusted for demographics, PHQ-9 score, perception about weight, if participant tried to lose weight last year, and history of CVD, angina, or heart attack. Results: The sample represented 2,990,456 eligible females in the US. About 13.2% Hispanic perceived their weight to be about right as compared to 10% Black, and 3.3% White. About 89% Hispanics tried to lose weight in past year compared to 71% White and Black. Weighted regression showed that Hispanic (β: -2.47, 95% CI: -4.44, -0.50) and Black (β: -0.74, 95% CI: -1.93, 0.45) had low CVHI score as compared to White. Higher PHQ9 score (β: -0.12, p > 0.05) and those who did not try to lose weight (β: -1.55, p < 0.01) were associated with lower, and those with positive attitude towards their weight were associated with higher CVHI score (β: 0.38, p > 0.05). Our study suggests huge variations in these characteristics by country of birth and citizenship with corroborated ambiguity in definitions. Conclusion: Previous research is inconclusive about existence of Hispanic paradox in cardiovascular health domain. Our study suggests that Hispanic ethnicity was a significant predictor of negative CVHI, when socioeconomic status (SES) was high among all obese females. There was higher prevalence of history of weight management and low CVHI score among Hispanic females; Hispanic females belonging to higher SES may be at a higher risk for CVD related adverse events. In our study, effect of education and income diminished and SES was not associated with participant’s willingness to lose weight despite being obese. Health disparities expressed as differences in CVHI exist despite high education and family income.

Background: Each year in the USA, more than 75 million adults are diagnosed with hypertension (HTN), but less than 54% have this condition under control. Due to poor management, mortality due to HTN or related complications was 410,000 in 2014 and resulted in close to $50 billion spent. We sought to examine disparities in the proportion of events and related expenditure due to HTN between 54 million Hispanics, representing 17% in the USA population and non Hispanics. Methods and Population: We used data from the Medical Expenditure Panel Survey (MEPS), the most complete source of data on the cost and use of health care and health insurance coverage for 2013 and 2014. Cost was grouped as related to ambulatory, emergency room, inpatient, home visits and medications. By source, payments were grouped as paid by family, MEDICARE, MEDICAID, private insurance, VA, Tricare and other. Results: Overall, there were 61.2 and 61.9 million total events associated with HTN in 2013 and 2014 respectively; Hispanics accounted for 5.8 (9.5%) and 5.4 (8.7%) million events each year. On an average, HTN events involving Hispanics were costlier up to $90 - $300 more than non Hispanics ($1053 vs. $ 746 in 2013; and $890 vs. $804 in 2014). For Hispanics, payments were mainly covered by MEDICAID (42.1%) and MEDICARE (27.5%), compared to MEDICARE (39.3%) and private insurance (23.7%) for non-Hispanic population. Hispanics HTN expenditures were $6.1 billion (12.9%) in 2013 and 5.3 billion (10.3%) in 2014 and Hispanics had disproportionately fewer number of events than expected 17%, and the structure of their costs for those events was not different from non-Hispanics. In regression model, accounting for demographics and type of insurance, being Hispanic was a significant predictor of the total, ambulatory and inpatient cost, but not emergency room or medication cost. Conclusion and Discussion: Hispanics participate disproportionately less in HTN events and costs compared to their proportion in population, even when age, demographic and socioeconomic factors are accounted for. They also have on average higher and more complex events compared with non Hispanics. Almost 70% of HTN expenditure for Hispanics in 2013-2014 was covered by MEDICAID and MEDICARE indicating socioeconomic disparities.

Objective: Carotid intima media thickness (CIMT) is well-known marker of cerebrovascular & CVD outcomes. Recent literature has discussed association of sleep duration with stroke &CVD, but still limited evidence exists regarding the true relationship of sleep duration with CIMT. The aim of this study is to determine association of short& long sleep duration with CIMT. Method: Baptist Health South Florida, a not for profit organization, conducted a randomized, non-blinded controlled trial in 2014.This study examined effect of web based interventions on reducing CVD risk in employees. The inclusion criteria were physician diagnosed T2DM and/or Metabolic Syndrome. We used cross sectional data for analysis. Per CDC.gov guidelines, we categorized self-reported sleep duration (hrs) as short ( Result: Study population (n=183; 74% female, 49% Hispanic) with mean age 51±10 years. Mean CIMT(mm) in females [0.879±0.15] and males [0.911±0.19] was not different (p>0.05). Atherosclerotic plaque was defined as any obvious focal luminal encroachment > 1.2 mm. In multivariate logistic regression model, per hour increase in sleep duration was associated with twice the odds of increase in CIMT >1.2mm [OR 2.15;95% CI (1.15-4.02)]. However, once we compared the reference sleep with short and long sleep duration categories, we determined, as compared to 7-9 hrs (ref) of sleep, the odds of CIMT >1.2 in those sleeping Conclusion: Although we observed that per hour increase in sleep was related to increase in CIMT >1.2mm (risk of plaque), but we did not find any significant increased risk of plaque in either short or long sleep duration. Longitudinal studies with larger sample size are needed to clarify this association.