Correlating mammographic and pathologic findings in clinical decision support using natural language processing and data mining methods.

BACKGROUND: A key challenge to mining electronic health records for mammography research is the preponderance of unstructured narrative text, which strikingly limits usable output. The imaging characteristics of breast cancer subtypes have been described previously, but without standardization of parameters for data mining. METHODS: The authors searched the enterprise-wide data warehouse at the Houston Methodist Hospital, the Methodist Environment for Translational Enhancement and Outcomes Research (METEOR), for patients with Breast Imaging Reporting and Data System (BI-RADS) category 5 mammogram readings performed between January 2006 and May 2015 and an available pathology report. The authors developed natural language processing (NLP) software algorithms to automatically extract mammographic and pathologic findings from free text mammogram and pathology reports. The correlation between mammographic imaging features and breast cancer subtype was analyzed using one-way analysis of variance and the Fisher exact test. RESULTS: The NLP algorithm was able to obtain key characteristics for 543 patients who met the inclusion criteria. Patients with estrogen receptor-positive tumors were more likely to have spiculated margins (P = .0008), and those with tumors that overexpressed human epidermal growth factor receptor 2 (HER2) were more likely to have heterogeneous and pleomorphic calcifications (P = .0078 and P = .0002, respectively). CONCLUSIONS: Mammographic imaging characteristics, obtained from an automated text search and the extraction of mammogram reports using NLP techniques, correlated with pathologic breast cancer subtype. The results of the current study validate previously reported trends assessed by manual data collection. Furthermore, NLP provides an automated means with which to scale up data extraction and analysis for clinical decision support. Cancer 2017;114-121. © 2016 American Cancer Society.

Atrial Fibrillation as a Prognostic Indicator in Patients With Orthostatic Hypotension: Nationwide Inpatient Sample Analysis.

Background: Orthostatic hypotension and atrial fibrillation have common etiology and a bidirectional relationship with several cardiovascular conditions. Despite both conditions being highly prevalent in hospitalized patients, prior research has primarily evaluated adverse outcomes due to orthostatic hypotension and atrial fibrillation independent of each other. In this study, we aim to assess if the presence of atrial fibrillation exacerbates in-hospital outcomes of patients with orthostatic hypotension. Methods: Adult patients hospitalized in 2019 with a primary diagnosis of orthostatic hypotension with or without pre-existing atrial fibrillation were identified using the International Classification of Diseases, Tenth Revision (ICD-10) code. The primary outcome of interest was in-patient mortality and cardiac arrest. Secondary outcomes of interest were the length of stay and total hospital charges. Adjusted and unadjusted analysis was performed on appropriate variables of interest. Results: Among 10,630 hospitalizations with orthostatic hypotension, 2,987 (median (interquartile range (IQR)) age: 78.5 (68.5 - 88.5) years; 1,197 women (40.1%)) comprised the atrial fibrillation cohort. Mean Charlson comorbidity index was noted to be significantly higher in orthostatic hypotension and atrial fibrillation patients (mean (standard deviation (SD)): 3.1 (2.1) vs. 2.5 (2.1), P < 0.001).Compared to orthostatic hypotension patients without atrial fibrillation, the prevalence of congestive heart failure (1,263 (42.3%) vs. 1,367 (17.9%)), coronary artery disease (1,432 (47.9%) vs. 2,481 (32.5%)), history of percutaneous coronary intervention or graft (443 (14.83%) vs. 860 (11.3%)), chronic obstructive pulmonary disease (644 (21.6%) vs. 1,131 (14.8%)) , chronic kidney disease (1,182 (39.6%) vs. 2,216 (29.0%)), and hyperlipidemia (1,828 (61.2%) vs. 4,087 (53.5%); all P < 0.05), were significantly higher in orthostatic hypotension patients with atrial fibrillation. Following multivariable analysis of orthostatic hypotension patients, atrial fibrillation was associated with 5.0 times greater odds for cardiac arrest (adjusted odds ratio (aOR) = 5.0 (95% confidence interval (CI): 1.4 - 18.2), P = 0.014), without increased risk of in-hospital mortality (aOR = 2.1 (95% CI: 0.9 - 5.0), P = 0.090). Conclusions: Atrial fibrillation is an independent predictor for cardiac arrest but not in-hospital mortality in patients with orthostatic hypotension. The short- and long-term prognostic value of atrial fibrillation in orthostatic hypotension patients must be confirmed in future prospective trials to improve patient outcomes.

Clinical Characteristics and Predictors of Mortality in Minority Patients Hospitalized with COVID-19 Infection.

OBJECTIVES: To identify the early mortality predictors in minority patients hospitalized with coronavirus disease 2019 (COVID-19). DESIGN: Demographics, presenting characteristics, admission laboratory data, ICU admission, and mortality data were collected from 200 consecutively hospitalized patients with COVID-19. RESULTS: The mean (SD) age was 58.9 (15.1) years, 121(60.5%) were men, 143 (71.5%) were African Americans, and 33 (16.5%) were Latino. Common presenting symptoms were cough 130 (65.0%), shortness of breath 129 (64.5%), and fever 121 (60.5%). One or more comorbid illness occurred in 171 (85.5%) and common comorbidities were hypertension (130 (65.2%)), diabetes (100 (50.0%)) and chronic kidney disease (60 (30.0%)). Of the 200 patients, 71 (35.5%) were treated in the ICU, 47 (24.2%) received mechanical ventilation, 45 (22.5%) died, and 155(77.5%) patients discharged home alive. The non-survivors were significantly older and had elevated markers of inflammation, coagulation, and acute organ damage on presentation. Age ≥ 65 years (odds ratio (OR), 3.78; 95% CI, 1.74-8.22; P = .001), lactate dehydrogenase level > 400 IU/L (OR, 9.1; 95% CI, 2.97-28.1; p < 0.001), C-reactive protein > 20 mg/dl (OR, 5.56; 95%CI, 1.84-16.8; p < 0.001), ferritin > 2000 ng/ml (OR, 5.42; 95%CI, 1.63-17.9; p = 0.006), creatinine kinase > 1000 iu/l (OR, 3.57; 95% CI, 1.23 10.3; p = 0.019), procalcitonin > 2.5 ng/ml (OR, 4.21; 95% CI, 1.47-12.0; p = 0.007), D-dimer level > 3.0 µg/ml (OR,10.9; 95% CI, 3.33-36.2; p = < 0.001), creatinine > 2 mg/dl (OR, 4.5; 95% CI, 1.29-15.8; P = 0.018) at admission were associated independently with increases risk of in-hospital mortality. CONCLUSION: Patients of advanced age that present with elevated biomarkers of inflammation, coagulation, and end-organ damage were at higher risk of mortality.

Impact of Stent Types on In-Hospital Outcomes of Patients With Cancer Undergoing Percutaneous Coronary Intervention: A Nationwide Analysis.

BACKGROUND: Cardiovascular disease and cancer frequently coexist, and patients with cancer are at increased risk of cardiovascular events, including myocardial infarction and stroke. However, the impact of stent types on in-hospital outcomes of patients with malignancy is largely unknown. METHODS: Patients with concomitant diagnosis of cancer undergoing PCI between January 2005 and December 2014 were identified in the National Inpatient Sample. They were then categorized into those who have undergone coronary stenting with bare-metal stent (BMS) or drug-eluting stent (DES). Primary outcomes were in-hospital mortality and stent thrombosis. Adjusted and unadjusted analysis was employed on appropriate variables of interest. RESULTS: 8755 patients were included in the BMS group and 11,611 patients in the DES group. Following propensity matching, 4313 patients were randomly selected in both groups using a 1:1 ratio. There was high use of BMS stent in cancer patient (BMS 43.0%, DES 57.0%) compared to general population (BMS 23.2%, DES 76.8%). When comparing BMS to DES group, there was no statistically significant difference in mortality (4.7% vs. 3.8%, p = 0.097), acute kidney injury (11.3% vs. 10.6%, p = 0.425), bleeding complications (3.50% vs. 3.45%, p = 0.914), and length of hospital stay (5.4% vs. 5.2%, p = 0.119). However, an increased incidence of stent thrombosis was observed in the DES group (4.26% vs. 3.01%, p = 0.002). CONCLUSION: A higher incidence of BMS placement was noted in patients with cancer than in the general population. Paradoxically there was a high incidence of stent thrombosis in the DES group without increasing mortality.

An AI-driven clinical care pathway to reduce 30-day readmission for chronic obstructive pulmonary disease (COPD) patients.

Healthcare regulatory agencies have mandated a reduction in 30-day hospital readmission rates and have targeted COPD as a major contributor to 30-day readmissions. We aimed to develop and validate a simple tool deploying an artificial neural network (ANN) for early identification of COPD patients with high readmission risk. Using COPD patient data from eight hospitals within a large urban hospital system, four variables were identified, weighted and validated. These included the number of in-patient admissions in the previous 6 months, the number of medications administered on the first day, insurance status, and the Rothman Index on hospital day one. An ANN model was trained to provide a predictive algorithm and validated on an additional dataset from a separate time period. The model was implemented in a smartphone app (Re-Admit) incorporating four input risk factors, and a clinical care plan focused on high-risk readmission candidates was then implemented. Subsequent readmission data was analyzed to assess impact. The areas under the curve of receiver operating characteristics predicting readmission with ANN is 0.77, with sensitivity 0.75 and specificity 0.67 on the separate validation data. Readmission rates in the COPD high-risk subgroup after app and clinical intervention implementation saw a significant 48% decline. Our studies show the efficacy of ANN model on predicting readmission risks for COPD patients. The AI enabled Re-Admit smartphone app predicts readmission risk on day one of the patient's admission, allowing for early implementation of medical, hospital, and community resources to optimize and improve clinical care pathways.

Impact of Obstructive Sleep Apnea On In-Hospital Outcomes in Patients With Atrial Fibrillation: A Retrospective Analysis of the National Inpatient Sample.

Background Obstructive sleep apnea (OSA) is frequently seen with atrial fibrillation (AF) and is associated with increased cardiovascular morbidity, including hypertension, congestive heart failure, ischemic heart disease, and stroke. However, the impact of OSA on in-hospital outcomes of patients with AF is unclear. Methodology All patients aged ≥18 admitted primarily for AF between January 2016 and December 2017 were identified in the National Inpatient Sample database. They were then categorized into those with OSA and those without OSA. The primary outcome was in-hospital mortality. Unadjusted and adjusted analysis was performed on appropriate variables of interest. Results Of 156,521 primary AF hospitalizations, 15% of the patients had OSA. Baseline characteristics revealed no race disparity between the two groups. However, compared to those without OSA, the OSA group was younger and had a significantly higher proportion of males, obesity, heart failure, hypertension, chronic obstructive pulmonary disease, diabetes, and hyperlipidemia. Long-term anticoagulation and inpatient cardioversion were also higher in the OSA group. Following propensity matching, inpatient mortality was similar between the two groups [0.54% in OSA vs. 0.51% in non-OSA; adjusted odds ratio = 1.06 (95% confidence interval = 0.82-1.35)]. Similarly, OSA was not significantly associated with acute kidney injury, cardiac arrest, gastrointestinal bleed, acute stroke, or length of stay. However, the OSA group was less anemic and required fewer in-hospital blood transfusions. Conclusions Although OSA is highly prevalent in AF patients, inpatient mortality and cardiovascular outcomes such as cardiac arrest, stroke, or major bleeding were similar in AF patients with or without concomitant OSA with no significant differences in length of stay.

Riociguat in the Treatment of Chronic Thromboembolic Pulmonary Hypertension: An Evidence-Based Review of Its Place in Therapy.

Chronic thromboembolic pulmonary hypertension (CTEPH) is classified as group-4 pulmonary hypertension caused by organized thrombi in pulmonary arteries and vasculopathy in nonoccluded areas leading to right heart failure and death. In addition to chronic anticoagulation therapy, each patient with CTEPH should receive treatment assessment starting with evaluation for pulmonary endarterectomy (PEA), which is the guideline recommended treatment. There is increasing experience with balloon pulmonary angioplasty (BPA) for inoperable patients; this option, like PEA, is reserved for specialized centers with expertise in this treatment method. Inoperable patients are candidates for targeted drug therapy. Riociguat remains the only approved medical therapy for CTEPH patients deemed inoperable or with persistent pulmonary hypertension after PEA. The role of riociguat therapy preoperatively or in tandem with BPA is currently under investigation. The purpose of this review is to evaluate the safety and efficacy of riociguat in the treatment of CTEPH.

Timing of Percutaneous Coronary Intervention in Non-ST Elevation Acute Coronary Syndrome - Meta-Analysis and Systematic Review of Literature.

BACKGROUND: Percutaneous coronary intervention (PCI) is the standard procedure of care for most patients with non-ST elevation acute coronary syndrome (NSTE-ACS). However, the timing of PCI remains unclear. We performed this meta-analysis with available randomized controlled trials (RCTs) to compare early versus late coronary intervention in patients with NSTE-ACS. METHOD: A total of 13 RCTs were selected through PubMed/MEDLINE via OVID, EMBASE via OVID and Cochrane Central Register of Controlled Trials (inception to October 2018) search. Outcomes were analyzed using the relative risk (RR) and 95% CI. Pooled RRs were determined using M-H random-effects model, which can account for between study heterogeneity. RESULTS: We included 13 RCTs with 11,972 patients were included. There were 7101 patients were randomized into early invasive group and 4871 in late invasive group. There was a significant decrease in myocardial infarction with long-term follow up in early invasive group compared to the delayed invasive group (RR 0.847 [95% CI 0.74-0.95], p = 0.009) with no difference in mortality between early and late invasive group (5.41% vs 6.49%, RR 0.882 [95% CI, 0.76-1.02]). On subgroup analysis, data was available from 6 RCTs for GRACE (Global Registry of Acute Coronary Events) score and 8 RCTs for elevated troponin. Early intervention led to decrease in adverse events in patients with elevated GRACE score > 140 (Mantel-Haenszel pooled RR 0.88 [95% CI 0.82-0.95], p-value 0.002) but no difference was seen in patients with elevated troponin. CONCLUSION: It can be postulated from these results that early invasive strategy leads to decrease in myocardial infarction but without significant decrease in mortality. In patients with elevated GRACE score (>140), early intervention did show a trend towards decrease in major adverse cardiac events, whereas in patients with elevated troponin alone, similar association was not observed. However, adequately powered randomized controlled trial is necessary to validate these findings.

In-hospital Outcomes and Characteristics of Heart Failure in Sickle Cell Disease.

Sickle cell disease (SCD) predominantly affects African-Americans (AAs) in the United States (US). Due to increasing life expectancy in developed countries, SCD-associated cardiomyopathy is typically seen in adults. The aim of this study was to distinguish hospitalization for this phenotype from traditional heart failure (HF) in AAs. We used the National Inpatient Sample (NIS) database to identify HF hospitalizations in AAs between 2005 and 2014 and stratified them according to SCD status. We compared the characteristics and outcomes before and after matching in a 1:3 ratio for age, gender, insurance, smoking status and admission year. Amongst the 1,195,718 HF admissions in AAs, SCD accounted for 7835. The age (mean ± SD) in the SCD cohort was significantly younger (45.66 ± 13.2) vs non-SCD (64.8 ± 15.2), p<0.001. SCD adults had significantly higher rates of pulmonary hypertension (PH), deep vein thrombosis, and pulmonary embolism while non-SCD adults had higher rates of cardiogenic shock and respiratory failure requiring intubation. The national hospitalization rate for HF in AAs increased from 151 to 257 per million between 2005 and 2011 before declining to 241 per million in 2014. There was a decrease in in-hospital mortality in AAs from 4.8% in 2005 to 3.6% in 2014. We also identified independent predictors of in-hospital mortality in SCD with HF. In conclusion, we described hospitalizations for an emerging heart failure phenotype in AAs. Although there is a national decreasing rate of HF hospitalizations in the US, this may not be reflective of the AA population.

Comparison of Direct Patient Care Costs and Quality Outcomes of the Teaching and Nonteaching Hospitalist Services at a Large Academic Medical Center.

PURPOSE: To compare costs of care and quality outcomes between teaching and nonteaching hospitalist services, while testing the assumption that resident-driven care is more expensive. METHOD: Records of inpatients with the top 20 Medicare Severity Diagnosis-Related Groups admitted to the University Teaching Service (UTS) and nonteaching hospitalist service (NTHS) at Houston Methodist Hospital from 2014-2015 were analyzed retrospectively. Direct costs of care, length of stay (LOS), in-hospital mortality (IHM), 30-day readmission rate (30DRR), and consultant utilization were compared between the UTS and NTHS. Propensity score matching and case mix index (CMI) were used to mitigate differences in baseline characteristics. To compare outcomes between matched groups, the Wilcoxon rank sum test and chi-square test were used. A sensitivity analysis was conducted using multivariable regression analysis. RESULTS: From the overall study population of 8,457 patients, 1,041 UTS and 3,123 NTHS patients were matched. CMI was 1.07 for each group. The UTS had lower direct costs of care per case ($5,028 vs. $5,502, P = .006), lower LOS (4.7 vs. 5.2 days, P = .0002), and lower consultant utilization (1.0 vs. 1.6, P ≤ .0001) versus the NTHS. The UTS and NTHS 30DRR (17.2% vs. 19.3%, P = .110) and IHM (2.9% vs. 3.7%, P = .206) were comparable. The multivariable regression analysis validated the matched data and identified an incremental cost savings of $333/UTS patient. CONCLUSIONS: Patients of an academic hospitalist service had significantly shorter LOS, fewer consultants, and lower direct care costs than comparable patients of a nonteaching service.

Prevalence of in-hospital nonsteroidal antiinflammatory drug exposure in patients with a primary diagnosis of heart failure.

AIM: To determine the prevalence of in-hospital nonsteroidal antiinflammatory drug (NSAID) exposure and associated outcomes in patients admitted with a primary diagnosis of heart failure. METHODS: We performed a propensity-matched cohort analysis of patients admitted to Houston Methodist Hospital System with a primary diagnosis of heart failure according to the International Classification of Diseases-9-Clinical Modification (ICD-9-CM) from January 1, 2011 to December 31, 2014. RESULTS: Of the 9742 patients admitted with a primary diagnosis of heart failure, 384 patients (3.9%) were exposed to NSAID. After applying propensity scores we matched 305 NSAID exposed with 915 unexposed patients. Patients with in-hospital NSAID exposure had a longer length of stay (7.0±8.8 days vs 6.1±8.5; P=.003) and increased prevalence of worsening renal function (34.4% vs 27.9%; P=.030). There were not statically significant differences in in-hospital mortality rate or 30-day all-cause readmission rate. CONCLUSION: Exposure to NSAID in patients admitted with a primary diagnosis of heart failure was low but was associated with adverse outcomes including longer length of stay and higher prevalence or worsening renal function.

METEOR: An Enterprise Health Informatics Environment to Support Evidence-Based Medicine.

GOAL: The aim of this paper is to propose the design and implementation of next-generation enterprise analytics platform developed at the Houston Methodist Hospital (HMH) system to meet the market and regulatory needs of the healthcare industry. METHODS: For this goal, we developed an integrated clinical informatics environment, i.e., Methodist environment for translational enhancement and outcomes research (METEOR). The framework of METEOR consists of two components: the enterprise data warehouse (EDW) and a software intelligence and analytics (SIA) layer for enabling a wide range of clinical decision support systems that can be used directly by outcomes researchers and clinical investigators to facilitate data access for the purposes of hypothesis testing, cohort identification, data mining, risk prediction, and clinical research training. RESULTS: Data and usability analysis were performed on METEOR components as a preliminary evaluation, which successfully demonstrated that METEOR addresses significant niches in the clinical informatics area, and provides a powerful means for data integration and efficient access in supporting clinical and translational research. CONCLUSION: METEOR EDW and informatics applications improved outcomes, enabled coordinated care, and support health analytics and clinical research at HMH. SIGNIFICANCE: The twin pressures of cost containment in the healthcare market and new federal regulations and policies have led to the prioritization of the meaningful use of electronic health records in the United States. EDW and SIA layers on top of EDW are becoming an essential strategic tool to healthcare institutions and integrated delivery networks in order to support evidence-based medicine at the enterprise level.