Temporal Trends and Patient Characteristics Associated with 30-Day Hospital Readmission Rates after a First Acute Myocardial Infarction.

BACKGROUND: Limited data exist about relatively recent trends in the magnitude and characteristics of patients who are re-hospitalized after hospital admission for an acute myocardial infarction. This study examined trends in the frequency and sociodemographic and clinical characteristics of patients readmitted to the hospital within 30 days after an initial acute myocardial infarction. METHODS: We reviewed the medical records of 3116 individuals who were hospitalized for a validated first acute myocardial infarction in 6 study periods between 2003 and 2015 at the 3 major medical centers in central Massachusetts. RESULTS: The median age of our population was 67 years, and 42% were women. The risk of being readmitted to the hospital within 30 days after an initial acute myocardial infarction increased slightly during the most recent study years after controlling for potentially confounding factors. Overall, older adults and patients with previously diagnosed atrial fibrillation, heart failure, diabetes, chronic kidney disease, stroke, and peripheral vascular disease were at higher risk for being readmitted to the hospital than respective comparison groups. For those hospitalized in the most recent study years of 2011/2015, a higher risk of rehospitalization was associated with a previous diagnosis of chronic kidney disease, peripheral vascular disease, the presence of 3 or more chronic conditions, and having developed atrial fibrillation or heart failure during the patient's hospitalization for a first acute myocardial infarction. CONCLUSIONS: We identified several groups at higher risk for hospital readmission in whom enhanced surveillance efforts as well as tailored educational and treatment approaches remain needed.

Trends in the magnitude of chronic conditions in patients hospitalized with a first acute myocardial infarction.

BACKGROUND: Among adults with heart disease, there is a high prevalence of concomitant chronic medical conditions. We studied patients with a first acute myocardial infarction to describe: sample population characteristics; trends of the most prevalent pairs of chronic conditions; and differences in hospital management according to burden of these morbidities. METHODS AND RESULTS: Patients (n = 1,564) hospitalized with an incident AMI at the 3 major medical centers in central Massachusetts during 2005, 2011, and 2015 comprised the study population. Hospital medical records were reviewed to identify 11 more prevalent chronic conditions. The median age of this population was 68 years and 56% were men. The median number of previously diagnosed chronic conditions was 2. Patients hospitalized during 2015 were more likely to be younger than those hospitalized in the earliest study cohorts. The most common pairs of chronic conditions for those hospitalized in 2005 were: anemia-chronic kidney disease (31%), chronic kidney disease-heart failure (30%), and stroke-atrial fibrillation (27%). Among patients hospitalized during 2011, chronic kidney disease-heart failure (29%), hypertension-hyperlipidemia (27%), and hypertension-diabetes (27%) were the most common pairs whereas hypertension-hyperlipidemia (43%), diabetes-heart failure (30%), and chronic kidney disease-diabetes (23%) were the most frequent pairs recorded in 2015. There was a significant decrease in the odds of undergoing cardiac catheterization and a percutaneous coronary intervention in those with higher chronic disease burden in the most recent as compared to earliest study years. CONCLUSIONS: Our findings highlight the magnitude of chronic conditions in patients with AMI and the challenges of caring for this vulnerable population.

Learning Latent Space Representations to Predict Patient Outcomes: Model Development and Validation.

BACKGROUND: Scalable and accurate health outcome prediction using electronic health record (EHR) data has gained much attention in research recently. Previous machine learning models mostly ignore relations between different types of clinical data (ie, laboratory components, International Classification of Diseases codes, and medications). OBJECTIVE: This study aimed to model such relations and build predictive models using the EHR data from intensive care units. We developed innovative neural network models and compared them with the widely used logistic regression model and other state-of-the-art neural network models to predict the patient's mortality using their longitudinal EHR data. METHODS: We built a set of neural network models that we collectively called as long short-term memory (LSTM) outcome prediction using comprehensive feature relations or in short, CLOUT. Our CLOUT models use a correlational neural network model to identify a latent space representation between different types of discrete clinical features during a patient's encounter and integrate the latent representation into an LSTM-based predictive model framework. In addition, we designed an ablation experiment to identify risk factors from our CLOUT models. Using physicians' input as the gold standard, we compared the risk factors identified by both CLOUT and logistic regression models. RESULTS: Experiments on the Medical Information Mart for Intensive Care-III dataset (selected patient population: 7537) show that CLOUT (area under the receiver operating characteristic curve=0.89) has surpassed logistic regression (0.82) and other baseline NN models (<0.86). In addition, physicians' agreement with the CLOUT-derived risk factor rankings was statistically significantly higher than the agreement with the logistic regression model. CONCLUSIONS: Our results support the applicability of CLOUT for real-world clinical use in identifying patients at high risk of mortality.

Age Differences in the Chief Complaint Associated With a First Acute Myocardial Infarction and Patient's Care-Seeking Behavior.

BACKGROUND: This study set out to describe age differences in patient's chief complaint related to a first myocardial infarction and how the "typicality" of patient's acute symptoms relates to extent of prehospital delay. METHODS: The medical records of 2586 residents of central Massachusetts hospitalized at 11 greater Worcester medical centers with a first myocardial infarction on a biennial basis between 2001 and 2011 were reviewed. RESULTS: The average age of the study population was 66.4 years, 39.6% were women, 40.2% were diagnosed with a ST-elevation myocardial infarction (STEMI), and 72.0 % presented with typical symptoms of myocardial infarction, namely acute chest pain or pressure. Patients were categorized into 5 age strata: >55 years (23%), 55-64 years (20%), 65-74 years (19%), 75-84 years (22%), and ≥85 years (16%). The lowest proportion (11%) of atypical symptoms of myocardial infarction was observed in patients <55 years, increasing to 17%, 28%, 40%, and 51% across the respective age groups. The most prevalent chief complaint reported at the time of hospitalization was chest pain, but the proportion of patients reporting this symptom decreased from the youngest (83%) to the oldest patient groups (45%). There was a slightly increased risk of prehospital delay across the different age groups (higher in the oldest old) in those who presented with atypical, rather than typical, symptoms of myocardial infarction. CONCLUSIONS: The present results provide insights to the presenting chief complaint of patients hospitalized with a first myocardial infarction according to age and the relation of symptom presentation to patient's care-seeking behavior.

Detecting Hypoglycemia Incidents Reported in Patients' Secure Messages: Using Cost-Sensitive Learning and Oversampling to Reduce Data Imbalance.

BACKGROUND: Improper dosing of medications such as insulin can cause hypoglycemic episodes, which may lead to severe morbidity or even death. Although secure messaging was designed for exchanging nonurgent messages, patients sometimes report hypoglycemia events through secure messaging. Detecting these patient-reported adverse events may help alert clinical teams and enable early corrective actions to improve patient safety. OBJECTIVE: We aimed to develop a natural language processing system, called HypoDetect (Hypoglycemia Detector), to automatically identify hypoglycemia incidents reported in patients' secure messages. METHODS: An expert in public health annotated 3000 secure message threads between patients with diabetes and US Department of Veterans Affairs clinical teams as containing patient-reported hypoglycemia incidents or not. A physician independently annotated 100 threads randomly selected from this dataset to determine interannotator agreement. We used this dataset to develop and evaluate HypoDetect. HypoDetect incorporates 3 machine learning algorithms widely used for text classification: linear support vector machines, random forest, and logistic regression. We explored different learning features, including new knowledge-driven features. Because only 114 (3.80%) messages were annotated as positive, we investigated cost-sensitive learning and oversampling methods to mitigate the challenge of imbalanced data. RESULTS: The interannotator agreement was Cohen kappa=.976. Using cross-validation, logistic regression with cost-sensitive learning achieved the best performance (area under the receiver operating characteristic curve=0.954, sensitivity=0.693, specificity 0.974, F1 score=0.590). Cost-sensitive learning and the ensembled synthetic minority oversampling technique improved the sensitivity of the baseline systems substantially (by 0.123 to 0.728 absolute gains). Our results show that a variety of features contributed to the best performance of HypoDetect. CONCLUSIONS: Despite the challenge of data imbalance, HypoDetect achieved promising results for the task of detecting hypoglycemia incidents from secure messages. The system has a great potential to facilitate early detection and treatment of hypoglycemia.

Trends in Length of Hospital Stay and the Impact on Prognosis of Early Discharge After a First Uncomplicated Acute Myocardial Infarction.

Few studies have examined recent trends in the length of stay (LOS) among patients hospitalized with a first uncomplicated acute myocardial infarction (AMI) and the impact of early hospital discharge on various short-term outcomes in these low-risk patients. We used data from 1,501 residents hospitalized with a first uncomplicated AMI from all central Massachusetts medical centers on a biennial basis between 2001 and 2011. The association between hospital LOS and subsequent hospital readmission or death was examined using logistic regression modeling. The average age of the study population was 63.7 years, 63.0% were men, and 91.4% were non-Hispanic whites. The average hospital LOS declined from 4.1 days in 2001 to 2.9 days in 2011. During the years under study, the average 30-day hospital readmission rate was 11.9%, whereas the 30- and 90-day death rates were 1.5% and 2.9%, respectively. The multivariable adjusted odds ratio of a 30-day hospital readmission (odds ratio [OR] = 0.81, 95% confidence interval [CI] = 0.52 to 1.41), or 30-day (OR = 0.93, 95% CI = 0.29 to 2.98) and 90-day (OR = 0.89, 95% CI = 0.36 to 2.20) death rates were not significantly different between patients who were discharged from central Massachusetts medical centers during the first 2 days as compared with those discharged thereafter. In conclusion, the average LOS in patients with a first uncomplicated AMI declined during the years under study, and early discharge from the hospital at day 2 or sooner of these low-risk patients does not appear to be associated with an increased risk of adverse events post discharge compared with those discharged at a later time.

Trends in the Magnitude of, and Patient Characteristics Associated With, Multiple Hospital Readmissions After Acute Myocardial Infarction.

There are limited contemporary data available describing recent trends in the magnitude and characteristics of patients who are rehospitalized multiple times after hospital discharge for an acute myocardial infarction (AMI). We reviewed the medical records of 4,480 residents of the Worcester, Massachusetts, metropolitan area, who were discharged from 3 Central Massachusetts medical centers after an AMI in 6 biennial periods from 2001 to 2011 and were followed for all-cause and cause-specific hospital readmissions over the subsequent 6 months. The average age of our study population was 68 years, 89% were white, and 41% were women. Overall, ∼1 of every 3 patients had a readmission to the hospital within 6 months after hospital discharge for an AMI. The proportion of patients who were readmitted to the hospital 1, 2, or 3 times for any cause within 6 months remained unchanged during the years under study (20%, 8%, and 6%, respectively); 59% of these readmissions were cardiac related. Women, elderly patients, those with multiple chronic conditions, patients with a prolonged index hospitalization, and those who developed heart failure and/or atrial fibrillation during hospitalization were at higher risk for being readmitted multiple times compared with those who were readmitted once. Six-month hospital readmission rates after hospital discharge for an AMI remained stable during the years under study. In conclusion, we identified several groups at higher risk for multiple hospital readmissions who might be targeted for intensified monitoring efforts and tailored educational and treatment approaches.

Thirty-day Hospital Readmissions in Patients with Non-ST-segment Elevation Acute Myocardial Infarction.

BACKGROUND: Limited data exist about relatively recent trends in the magnitude and characteristics of patients who are rehospitalized shortly after admission for a non-ST-segment elevation acute myocardial infarction (NSTEMI). This observational study describes decade-long trends (1999-2009) in the magnitude and characteristics of patients readmitted to the hospital within 30 days of hospitalization for an incident (initial) episode of NSTEMI. METHODS: We reviewed the medical records of 2249 residents of the Worcester (Mass) metropolitan area who were hospitalized for an initial NSTEMI in 6 biennial periods between 1999 and 2009 at 3 central Massachusetts medical centers. RESULTS: The average age of our study population was 72 years, 90% were white, and 46% were women. The proportion of patients who were readmitted to the hospital for any cause within 30 days after discharge for an NSTEMI remained unchanged between 1999 and 2009 (approximately 15%) in both crude and multivariable adjusted analyses. Slight declines were observed for cardiovascular disease-related 30-day readmissions over the 10-year study period. Women, elderly patients, those with multiple chronic comorbidities or a prolonged index hospitalization, and patients who developed heart failure during their index hospitalization were at higher risk for being readmitted within 30 days than respective comparison groups. CONCLUSION: Thirty-day hospital readmission rates after hospital discharge for a first NSTEMI remained stable between 1999 and 2009. We identified several groups at higher risk for hospital readmission; further surveillance efforts and/or tailored educational and treatment approaches remain needed for these groups.

Changing trends (1986-2003) in the use of lipid lowering medication in patients hospitalized with acute myocardial infarction: a community-based perspective.

OBJECTIVES: The objectives of this community-wide observational study were to describe nearly two decade long (1986-2003) trends in the use of lipid lowering therapy in patients hospitalized with acute myocardial infarction (AMI) and clinical and demographic factors associated with underutilization of this treatment regimen. METHODS: A total of 9429 greater Worcester (MA) residents hospitalized with confirmed AMI at all metropolitan Worcester medical centers in 10 annual periods between 1986 and 2003 comprised the study population. Hospital medical records were reviewed to ascertain the prescribing of lipid lowering agents during hospitalization for AMI. RESULTS: The mean age of the study sample was 70 years, 58% were men, and the average total and LDL serum cholesterol levels were 203 and 114 mg/dL, respectively. There was a marked increase in the use of lipid lowering therapy in greater Worcester residents hospitalized with AMI between 1986 (<1%) and 2003 (76%). Increasing use of lipid lowering medication was observed both with regards to the maintenance of this therapy in patients who were already on this treatment regimen and in the new initiation of lipid lowering medication in patients who had not been previously treated with this therapy. Several patient demographic and clinical factors, including advanced age and an initial AMI, were associated with the failure to be prescribed lipid lowering therapy during hospitalization for AMI. CONCLUSIONS: The results of this study suggest encouraging increases over time in the use of lipid lowering therapy in patients hospitalized with AMI. Despite these encouraging trends, several high-risk patient groups remain suboptimally treated.