Clinical Classification Software Research Articles

Pandemic effect on national hospitalizations for acute hyperglycemic complications

Objective This study aimed to describe the effect of the pandemic on epidemiologic trends and disparities in outcomes for patients hospitalized with acute hyperglycemic complications (AHC). Methods This was a retrospective study of the National Inpatient Sample (NIS) database from 2016 to 2020. The population included adults hospitalized with AHCs as a principal diagnosis using the Clinical Classifications Software Refined code. Results There was a decrease in the AHC hospitalization rate per 100,000 admissions for type 1 diabetes (T1D) during the pandemic (577 vs 600). However, there was an increase for type 2 diabetes (T2D) (117 vs 125). The mean age during the pandemic versus prepandemic was 34.8 ± 14.1 vs 34.7 ± 14.2 (P = 0.41) and 59.1 ± 14.4 vs 58.8 ± 14.7 (P = 0.51) for T1D and T2D, respectively. No statistically significant difference was observed in mortality in T1D (0.20 vs 0.23; P = 0.42) or T2D (1.1 vs 0.8; P = 0.09). There was no difference in mortality after stratifying results by gender, race, median household income, or hospital region. During the pandemic, COVID-19 was the principal diagnosis in 5.5% of those with AHC in T1D and 9.1% in those with AHC in T2D. Conclusion The pandemic had a significant impact on the hospitalization rate for both T1D and T2D.

The impact of comorbid psychiatric disorders on chronic obstructive pulmonary disease (COPD) hospitalizations: a nationwide retrospective study.

To characterize the register of a secondary diagnosis of mental illnesses in all chronic obstructive pulmonary disease (COPD) hospitalizations registered in Portugal from 2008 to 2015 and explore their impact on hospitalization outcomes. A retrospective observational study was conducted. Hospitalizations of patients with at least 40 years old, discharged between 2008 and 2015 with a primary diagnosis of COPD (ICD-9-CM codes 491.x, 492.x and 496) were retrieved from a national administrative database. Comorbid psychiatric diagnoses were identified and defined by the HCUP Clinical Classification Software (CCS) category codes 650-670 (excluding 662). Length of hospital stay (LoS), admission type, in-hospital mortality, and estimated hospital charges were analyzed according to psychiatric diagnostic categories using sex and age-adjusted models. Of 66,661 COPD hospitalizations, 25,869 (38.8%) were episodes with a registered psychiatric comorbidity. These were more likely to correspond to younger inpatients (OR = 2.16, 95%CI 2.09-2.23; p < 0.001), to stay longer at the hospital (aOR = 1.08, 95%CI 1.05-1.12; p < 0.001), to incur in higher estimated hospital charges (aOR = 1.37, 95%CI 1.33-1.42; p < 0.001) and to be urgently admitted (aOR = 1.33, 95%CI 1.23-1.44; p < 0.001). After adjustment for age, in-hospital mortality was lower for episodes with psychiatric diagnoses (aOR = 0.90; 95%CI 0.84-0.96; p < 0.001), except for organic and neurodegenerative diseases category and developmental disorders, intellectual disabilities and disorders usually diagnosed in infancy, childhood, or adolescence category. These findings corroborate the additional burden placed by psychiatric disorders on COPD hospitalizations, highlighting the importance of individualizing care to address these comorbidities and minimize their impact on treatment outcomes.

Factors Predicting 90-Day Readmissions for US Older Adult Burn Patients From the 2016-2018 Nationwide Readmissions Database.

Older adults with burn injuries have a high risk of readmission after the initial burn admission. Frailty is associated with poor outcomes from the initial burn injury, however, it remains unknown if frailty impacts readmission in older adults after the initial index burn admission. Our study aims to examine patient, frailty, burn, surgical, and hospital factors that contribute to 90-day readmissions in older adults with an acute burn. Patients ≥ 50 years of age with an acute burn diagnosis in the Nationwide Readmissions Database (2016-2018) were included. Patients who died during index admissions were excluded from the readmission analysis. Patient, burn, hospital stay, and hospital characteristics were analyzed. Using variables significant from univariate logistic regression analysis, we performed a multivariable logistic regression analysis to identify factors associated with 90-day readmissions. A total of 41,500 weighted index burn admissions were included. The 90-day readmission rate was 25.6% (10,641/41,500). Almost all the readmissions were unplanned (96.5%). The most common primary readmission diagnoses based on Clinical Classifications Software Refined were burn injury and septicemia. Multivariable logistic regression analysis demonstrated an increased hospital frailty risk and lower-body-only burns were associated with an increased risk of readmission. Notably, surgical treatment of the burn injury during the index admission was associated with a decreased risk of readmission. This highlights the importance and challenges of surgical decision-making for older adults with high comorbidity burden. Future studies are needed to develop decision guides to help clinicians identify patients who would benefit from surgical treatment versus more conservative strategies in older adult patients with burns.

Multimorbidity in Psoriasis as a Risk Factor for Psoriatic Arthritis: A Population-Based Study.

To examine multimorbidity in psoriasis and its association with the development of PsA. A retrospective cohort study was performed using the Rochester Epidemiology Project. Population-based incidence (2000-2009) and prevalence (Jan 1, 2010) cohorts of psoriasis were identified by manual chart review. A cohort of individuals without psoriasis (comparators) were identified (1:1 matched on age, sex, and county). Morbidities were defined using ≥2 Clinical Classification Software codes ≥30 days apart within prior five years. PsA was defined using ClASsification of Psoriatic ARthritis (CASPAR) criteria. χ2 and rank-sum tests were used to compare morbidities, and age-, sex-, and race-adjusted Cox models to examine the association of baseline morbidities in psoriasis with development of PsA. Among 817 incident psoriasis patients, the mean age was 45.2 years with 52.0% females, and 82.0% moderate/severe psoriasis. No multimorbidity differences were found between incident psoriasis patients and comparators. However, in the 1,088 prevalent psoriasis patients, multimorbidity was significantly more common compared with 1,086 comparators (OR : 1.35 and OR : 1.48 for ≥2 and ≥5 morbidities, respectively). Over a median 13.3-year follow-up, 23 patients (cumulative incidence: 2.9% by 15 years) developed PsA. Multimorbidity (≥2 morbidities) was associated with a 3-fold higher risk of developing PsA. Multimorbidity was more common in the prevalent but not incident cohort of psoriasis compared with the general population, suggesting patients with psoriasis may experience accelerated development of multimorbidity. Moreover, multimorbidity at psoriasis onset significantly increased the risk of developing PsA, highlighting the importance of monitoring multimorbid psoriasis patients for the development of PsA.

ED Visits for Schizophrenia Spectrum Disorders During the COVID-19 Pandemic at 5 Campus Health Systems.

Although substantial research has reported grave population-level psychiatric sequelae of the COVID-19 pandemic, evidence pertaining to temporal changes in schizophrenia spectrum disorders in the US following the pandemic remains limited. To examine the monthly patterns of emergency department (ED) visits for schizophrenia spectrum disorders after the onset of the COVID-19 pandemic. This observational cohort study used time-series analyses to examine whether monthly counts of ED visits for schizophrenia spectrum disorders across 5 University of California (UC) campus health systems increased beyond expected levels during the COVID-19 pandemic. Data included ED visits reported by the 5 UC campuses from 2016 to 2021. Participants included persons who accessed UC Health System EDs had a diagnosis of a psychiatric condition. Data analysis was performed from March to June 2023. The exposures were binary indicators of initial (March to May 2020) and extended (March to December 2020) phases of the COVID-19 pandemic. The primary outcome was monthly counts of ED visits for schizophrenia spectrum disorders. International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnosis codes, categorized within Clinical Classification Software groups, were used to identify ED visits for schizophrenia spectrum disorders and all other psychiatric ED visits, from the University of California Health Data Warehouse database, from January 2016 to December 2021. Time-series analyses controlled for autocorrelation, seasonality, and concurrent trends in ED visits for all other psychiatric conditions. The study data comprised a total of 377 872 psychiatric ED visits, with 37 815 visits for schizophrenia spectrum disorders. The prepandemic monthly mean (SD) number of ED visits for schizophrenia spectrum disorders was 519.9 (38.1), which increased to 558.4 (47.6) following the onset of the COVID-19 pandemic. Results from time series analyses, controlling for monthly counts of ED visits for all other psychiatric conditions, indicated 70.5 additional ED visits (95% CI, 11.7-129.3 additional visits; P = .02) for schizophrenia spectrum disorders at 1 month and 74.9 additional visits (95% CI, 24.0-126.0 visits; P = .005) at 3 months following the initial phase of the COVID-19 pandemic in California. This study found a 15% increase in ED visits for schizophrenia spectrum disorders within 3 months after the initial phase of the pandemic in California across 5 UC campus health systems, underscoring the importance of social policies related to future emergency preparedness and the need to strengthen mental health care systems.

National Mortality Outcomes of Acute Chest Syndrome Hospitalization during the COVID-19 Pandemic

Introduction Sickle cell disease (SCD) is an hereditary hemoglobin disorder characterized by repeated episodes of severe acute pain, acute chest syndrome (ACS) and other complications. The average life expectancy of SCD patients remains 20 years less than the general population, and ACS is the leading cause of mortality. This retrospective study describes the effect of the COVID-19 pandemic on epidemiologic trends and outcomes of Acute Chest Syndrome. Methods The National Inpatient Sample (NIS) database from 2016 to 2020 was searched for hospitalizations with ACS as the principal diagnosis using Clinical Classifications Software Refined codes. The ACS admission rate for each calendar year was obtained as admission per 10,000 SCD adults hospitalized. The primary outcome was a comparison of inpatient mortality, and the secondary outcomes were the length of hospital stay (LOS) and total hospital charge (THC) between prepandemic (2019) and pandemic years (2020). Results In 2019 (prepandemic), there were 5,160 hospitalizations for ACS, accounting for 624 per 10,000 adult SCD hospitalizations. In 2020 (pandemic), there were 4,285 patients hospitalized for ACS, accounting for 540 per 10,000 adult SCD hospitalizations. The mean age of admission was similar for both years (31.5 vs 32 years, P=0.143), and there was no significant difference in age categories when both years were compared. The gender distribution was similar during both periods (females: 45.6 vs 46.7%, p=0.68). The racial distribution was predominantly blacks with no significant difference during both year (92 vs 92%, p=0.3). There was no significant difference in hospitalizations with regard to insurance status, hospital region, or hospital size between both years. There was a three-fold increase in overall proportion of ACS inpatient mortality in the pandemic year of 2020 compared to the prepandemic level (0.6% vs 1.9%, P=0.009). Proportion of mortality among women had a five-fold increase (0.4% vs 2.3%, P=0.02). There was a four-fold increase in mortality proportions among the Black population (0.5% vs 2%; p=0.007). Mortality outcomes were similar over both years when stratified for income quartiles and hospital regions. The LOS during the pandemic was similar to the prepandemic level (6.5. vs 7.4days, P=0.563). Also, there was no statistically significant difference in the total hospital charge ($61,515 vs $80,191, P=0.108). Conclusion This study depicts a statistically significant increase in inpatient mortality outcomes of Acute chest syndrome during the pandemic period. However, the length of hospital stay and total hospital charge were similar. Further studies need be conducted to evaluate the long-term effect of the COVID-19 pandemic on SCD.

2473. Rates of Antibiotic Resistant Infections in Hospitalized Patients with Hematologic and Non-Hematologic Cancers, 2019

Abstract Background Patients with cancer have risk factors for developing both hospital-onset (HO) and antibiotic resistant (AR) infections such as high antibiotic use, long lengths of stay, and device utilization. Understanding the epidemiology of AR infections in this population can inform clinical management and prevention strategies. We aimed to describe the rates and proportions of AR infections in hospitalized patients with cancer. Methods Using the PINC AI Healthcare Database, we included adult discharges from 2019 among facilities reporting microbiology data. We identified patients with at least one cancer diagnosis from the Healthcare Cost and Utilization Project’s Clinical Classifications Software Refined categories and Elixhauser Comorbidity Software Refined for ICD-10-CM. We calculated age-standardized rates of AR infections among cancer patients per 10,000 discharges, stratified by community-onset (CO: ≤ day 3 of hospitalization) and HO (≥ day 4 of hospitalization), and hematologic and non-hematologic cancer types. Hematologic cancers included lymphoma, leukemia, myeloma and myelodysplastic syndrome. We also examined the proportion of AR infections in patients with cancer among all AR infections in the cohort. Results Of &amp;gt;2 million discharges from 277 facilities, 11.4% were patients with cancer. Compared to all patients, cancer patients had higher rates of infection with CO and HO vancomycin resistant Enterococcus (VRE) and carbapenem resistant Enterobacterales (CRE), and HO Enterobacterales with AR phenotypes suggestive of extended-spectrum β-lactamase (ESBL) production (Table 1). Compared to patients with non-hematologic cancer, patients with hematologic cancer had higher rates of all AR infections, except for CO carbapenem resistant Acinetobacter spp. (CRAsp) infections (Table 1). For most AR infections, cancer patients made up a greater proportion of all HO AR infections compared to CO AR infections (Table 2). Conclusion Patients with cancer had higher rates of several AR infections compared to all patients. To decrease the burden of AR infections in this population, infection prevention strategies may focus on VRE, ESBL, and CRE pathogens, as well as HO infections, in all cancer patients, particularly those with hematologic cancer. Disclosures All Authors: No reported disclosures

Emergency department visits in Parkinson's disease: The impact of comorbid conditions

IntroductionOlder adults have complex, often overlapping, medical conditions requiring careful management that may lead to increased emergency department usage compared to younger adults. Parkinson's disease (PD), a progressive neurodegenerative disorder characterized by distinct motor and nonmotor features, frequently occurs with additional comorbid disease. Classifying comorbid conditions into clinical subgroups allows for further understanding of the heterogeneity in outcomes in patients with PD. The current study examines the reasons for emergency department (ED) visits in a cohort of patients with PD and identifies comorbidities that are potential risk factors for specific ED presenting conditions. MethodsUsing data from Optum's de-identified Integrated Claims-Clinical dataset years 2010–2018, patients with PD were identified based on ICD-9/10 diagnosis codes. We identified all ED visits occurring after the first observed diagnosis code for PD. Comorbid conditions were classified using the AHRQ Clinical Classification Software (CCS). We classified patients using Latent Class Analysis (LCA) and conducted multiple logistic regression models with the outcome of reason-for-visit to examine the associations with comorbidity-profile class, patient demographics, and socio-economic characteristics. ResultsThe most common reasons for ED admission were injuries such as fractures and contusions, diseases of the circulatory system, and general signs and symptoms, including abdominal pain, malaise, and fatigue. Comorbid medical conditions often observed in this patient population include depression, diabetes mellitus, and chronic pulmonary disease. Patients in the “Poorest Health” classification of the LCA had greater odds for ED admission for diseases related to the gastrointestinal system, musculoskeletal system, and injury/poisoning categories and reduced odds for admission for diseases of the circulatory system. DiscussionPatients with PD who present to the emergency department with injuries are more likely to be in poor health overall with a high comorbidity burden. Clarifying the complex medical needs of patients with PD is the first step to further individualize care, which may reduce ED visits in this population, improve quality of life, and lessen the footprint on the healthcare system.

Epilepsy hospitalizations and mental disorders: A Portuguese population-based observational retrospective study (2008–2015)

BackgroundPsychiatric comorbidities are highly frequent in people with epilepsy and were found to be markers of poorer prognosis. These comorbidities increase the use of healthcare resources, including emergency department visits and inpatient care. Despite this, there is little information on healthcare utilization associated with a wide range of comorbid mental disorders in people with epilepsy (PWE). ObjectiveTo characterize registered mental disorders among all hospitalizations with a primary diagnosis of epilepsy and to analyze their association with crucial hospitalization outcomes. MethodsAn observational retrospective study was performed using administrative data from hospitalization episodes with epilepsy as the primary diagnosis discharged between 2008 and 2015. Mental disorder categories 650 to 670 from Clinical Classification Software were selected as secondary diagnoses. Mann-Whitney U, Kruskall-Wallis, and Chi-squared tests were used to establish comparisons. For each episode, data regarding hospitalization outcomes was retrieved, including length of stay (LoS), in-hospital mortality (IHM), 8-year period readmissions, and total estimated charges. ResultsOverall, 27,785 hospitalizations were analyzed and 33.9% had registered mental disorders, with alcohol-related disorders being the most prevalent (11.7%). For episodes with a concomitant register of a mental disorder, LoS was significantly longer (5.0 vs. 4.0 days, P <0.001), and IHM was higher (2.8% vs. 2.2%, P <0.001), as were readmissions (25.5% vs. 23.7%, P <0.001), and median episodes’ charges (1,578.7 vs. 1,324.4 euros, P <0.001). ConclusionEpilepsy-related hospitalizations with registered mental disorders heightened the utilization of healthcare resources, stressing the importance of diagnosing and treating mental disorders in PWE.

Comparing emergency department visits 10-year apart at a tertiary care center in Lebanon.

Presentations to the emergency department (ED) are growing worldwide. With the increasing risk factors of non-communicable disease (NCD) and communicable diseases (CD) in low- and middle-income countries, it is crucial to understand how ED presentations are changing with time to meet patients' needs and allocate acute care resources. The aim of this study is to compare the changes in patient and diseases characteristics over 2 time periods 10 year apart at the largest tertiary care center in Lebanon. This was a retrospective descriptive study of patients presenting to the ED at a large tertiary care center in 2009/2010 and 2018/2019. The discharge diagnoses were coded into Clinical Classification Software codes. We used descriptive statistics, odds ratios (OR), and non-parametric test to compare the different diagnoses. The total number of ED visits increased by 33% from 2009/2010 to 2018/2019. The highest increase rate was among patients older than 65 years (2.6%), whereas the percentage of pediatric patients decreased from 30.8% to 25.3%. ED presentations shifted from NCD to CD. A shift in the discharge diagnoses was also noted within age groups, specifically a shift in cardiovascular diseases to a younger age. Our study suggests that the role of the ED is changing and moving towards treating the aging population and CD. There is a need to invest and mitigate CD, better allocate resources to accommodate the aging population, focus on awareness campaigns targeting early detection of cardiovascular diseases and modifying its risk factors.

Use of clinical classifications software to address ICD coding transition in large healthcare databases analyzed via high-dimensional propensity scores.

The EUPAS26595 study characterized the rate of acute renal failure (ARF) in patients exposed to levetiracetam versus other antiepileptic drugs using healthcare claims data and a high-dimensional propensity score (hd-PS) for confounding adjustment. The data contained several coding systems by design and an update in International Classification of Diseases (ICD) coding dictionary. Such coding heterogeneity can affect the performance of hd-PS, and manually coding harmonization is not feasible. Our objective was to explore the impact of code aggregation via Clinical Classifications Software (CCS) on the analysis of a large claims-based database using hd-PS. Patients with epilepsy, who were new-users of an antiepileptic drug, were identified from the IBM® MarketScan® Research Databases. We used CCS categories to harmonize coding and compared the results with other alternatives. Incidence rate ratios (IRRs) were computed using modified Poisson regression model with a robust variance estimator. For January 2008-October 2015 (before ICD update), 34 833 eligible patients initiated levetiracetam and 52 649 initiated a comparator drug; IRR (95% CI) for ARF for the hd-PS analysis was 1.34 (0.72-2.50) without CCS categories and 1.30 (0.71-2.39) with CCS categories. For January 2008-December 2017 (including ICD coding change), 45 672 eligible patients initiated levetiracetam and 64 664 initiated a comparator drug; IRR (95% CI) for the hd-PS analysis was 1.34 (0.78-2.29) without CCS categories and 1.37 (0.80-2.34) with CCS categories. Using single-level CCS categories to overcome differences in coding provides consistent results and can be used in studies that use large claims data and hd-PS for adjustment.

Using Comorbidity Statistical Modeling to Predict Inpatient Mortality: Insights Into the Burden on Hospitalized Patients.

Background The expenditures of the United States for healthcare are the highest in the world. Assessment of inpatient disease classifications associated with death can provide useful information for risk stratification, outcome prediction, and comparative analyses to understand the most resource-intensive chronic illnesses. This project aims to adapt a comorbidity index model to the National Inpatient Sample (NIS) database of 2020 to predict one-year mortality for patients admitted with select International Classification of Diseases, 10th Edition (ICD-10) codes of diagnoses. Methodology A retrospective cohort study analyzed mortality with comorbidity using the Charlson comorbidity index model (CCI) in a sample population of an estimated 5,533,477 adult inpatients (individuals aged ≥18 years) obtained from the National Inpatient Database for 2020. A multivariate logistic regression model was constructed with in-hospital mortality as the outcome variable and identifying predictor variables as defined by the Clinical Classifications Software Refined Variables (CCSR) codes for selected ICD-10 diagnoses. Descriptive statistics and the base logistic regression analyses were conducted using SAS statistical software version 9.4 (SAS Institute, Cary, NC, USA). To avoid overpowering, a subsample (n= 100,000) was randomly selected from the original dataset. The initial CCI assigned weights to ICD-10 diagnoses based on the associated risk of death, and conditions with the greatest collective weights were included in a subsequent backward stepwise logistic regression model. Results The results of the base CCI regression analysis revealed 16 chronic conditions with P-values <0.20. Anemia (1,567,081, 28.32%), pulmonary disease (asthma, chronic obstructive pulmonary disease [COPD], pneumoconiosis; 1,210,892, 21.88%), and diabetes without complications (1,077,239, 19.47%) were the three most prevalent conditions associated with inpatient mortality. Results of the backward stepwise regression analysis revealed that severe liver disease/hepatic failure (adjusted odds ratio [aOR] 10.50; 95% confidence interval [CI] 10.40-10.59), acute myocardial infarction (aOR 2.85; 95% CI 2.83-2.87)and malnutrition (aOR 2.15, 95% CI 2.14-2.16)were three most important risk factors and had the highest impact on inpatient mortality (P-value <0.0001). The concordance statistic (c-statistic) or the area under the curve (AUC) for the final model was 0.752. Conclusions The CCI model proved to be a valuable approach in categorizing morbidity classifications associated with the greatest risk of death using a national sample of hospitalized patients in 2020.Study findings provide an objective approach to compare patient populations that bear important implications for healthcare system improvements, clinician treatment approaches, and ultimately decision decision-makers poised to influence advanced models of care and prevention strategies that limit disease progression and improve patient outcomes.

Professional Fees for U.S. Hospital Care, 2016-2020.

The latest comprehensive diagnosis-specific estimates of hospital professional fees relative to facility fees are from 2004 to 2012. Update professional fee ratio (PFR) estimates to improve cost analysis opportunities with hospital discharge data sources and compare them with previous PFR estimates. 2016-2020 MarketScan inpatient admissions and emergency department (ED) treat and release claims. PFR was calculated as total admission or ED visit payment divided by facility-only payment. This measure can be multiplied by hospital facility costs to yield a total cost estimate. Generalized linear regression models controlling for selected patient and service characteristics were used to calculate adjusted mean PFR per admission or ED visit by health payer type (commercial or Medicaid) and by selected diagnostic categories representing all clinical diagnoses (Major Diagnostic Category, Diagnostic Related Group, and Clinical Classification Software Revised). Mean 2016-2020 PFR was 1.224 for admissions with commercial payers (n = 6.7 million admissions) and 1.178 for Medicaid (n = 4.2 million), indicating professional payments on average increased total payments by 22.4% and 17.8%, respectively, above facility-only payments. This is a 9% and 3% decline in PFR, respectively, compared with 2004 estimates. PFR for ED visits during 2016-2020 was 1.283 for commercial payers (n = 22.2 million visits) and 1.415 for Medicaid (n = 17.7 million). This is a 12% and 5% decline in PFR, respectively, compared with 2004 estimates. Professional fees comprise a declining proportion of hospital-based care costs. Adjustments for professional fees are recommended when hospital facility-only financial data are used to estimate hospital care costs.

Automatic International Classification of Diseases Coding via Note-Code Interaction Network with Denoising Mechanism.

Clinical notes are comprehensive files containing explicit information about a patient's visit. However, accurately assigning medical codes from clinical documents can be a persistent challenge due to the complexity of clinical data and the vast range of medical codes. Moreover, the large volume of medical records, the noisy medical records, and the uneven quality of coders all negatively impact the quality of the final codes. Deep learning technology has recently been integrated into automatic International Classification of Diseases (ICD) coding tasks to improve accuracy. Nevertheless, the imbalanced class problem, the complexness of code associations, and the noise in lengthy records still restrict the advancement of ICD coding tasks in deep learning. Thus, we present the Note-code Interaction Denoising Network (NIDN) that employs the self-attention mechanism to pull critical semantic features in electronic medical records (EMRs). Our model utilizes the label attention mechanism for retaining code-specific text expression. We introduce Clinical Classifications Software coding for multitask learning, capturing the functional relationships of medical coding to oblige in model prediction. To minimize the impact of noise on model prediction and improve the label distribution imbalance, a denoising module is introduced to filter noise. Our practical consequences indicate that the model NIDN exceeds competitive models on a third version of Medical Information Mart for Intensive Care data set.

Associations Between Psychiatric Diagnoses on Length of Stay and Mortality After Rib Fracture: A Retrospective Analysis

IntroductionConcurrent psychiatric diagnoses adversely impact outcomes in surgical patients, but their relationship to patients with rib fracture after trauma is less understood. We hypothesized that psychiatric comorbidity would be associated with increases in hospital length of stay (LOS) and mortality risk after rib fracture. Materials and methodsThe 2017 National Inpatient Sample was queried for adult patients who were admitted with rib fracture after trauma. Mental health disorders were categorized into 34 psychiatric diagnosis groups (PDGs) using clinical classifications software refined for International Classification of Diseases-10. Outcomes of interest were LOS and mortality. Bivariable analysis determined associations between PDGs, patient demographics, hospital characteristics, and outcomes. Logistic regression was performed to identify adjusted effects on mortality, and linear regression was performed to identify effects on LOS. ResultsOf 32,801 patients, median age was 61 y (IQR 46-76), and median LOS was 5 d (IQR 3-9). No PDGs were associated with increased odds of mortality. Concurrent diagnosis of schizophrenia spectrum (Coeff. 3.5, 95% CI 2.7-4.4, P < 0.001) or trauma- or stressor-related (Coeff. 1.6, 95% CI 0.9-2.5, P < 0.001) disorders demonstrated the greatest association with prolonged LOS. Increased odds of death and prolonged hospital stay were also associated with male sex, non-White patient race, and surgery occurring at urban and public hospitals. ConclusionsPsychiatric comorbidities are associated with death after rib fracture but are associated with increased LOS. These findings may help promote multidisciplinary patient management in trauma.

Maternal morbidity and mortality among patients with cancer at time of delivery

Individuals with cancer during pregnancy are a medically complex patient population that is anticipated to grow. A better understanding of this population and patterns of risk at time of delivery would offer an opportunity for providers to mitigate maternal morbidity. This study aimed to estimate the prevalence in the United States of concurrent cancer diagnoses at time of delivery by cancer type and associated maternal morbidity and mortality. Using the National Inpatient Sample, we identified delivery-associated hospitalizations between 2007 and 2018. Concurrent cancer diagnoses were classified using the Clinical Classifications Software. Main outcomes included severe maternal morbidity, as defined by the Centers for Disease Control and Prevention indicators, and mortality during delivery hospitalization. We calculated adjusted rates for cancer diagnosis at time of delivery and adjusted odds ratios of severe maternal morbidity and maternal death during hospitalization using survey-weighted multivariable logistic regression models. In this sample of 9,418,761 delivery-associated hospitalizations, 63 per 100,000 deliveries had a concurrent cancer diagnosis (95% confidence interval, 60-66; national weighted estimate, 46,654,042). The most common cancer types were breast cancer (8.4 per 100,000 deliveries), leukemia (8.4 per 100,000 deliveries), Hodgkin lymphoma (7.4 per 100,000 deliveries), non-Hodgkin lymphoma (5.4 per 100,000 deliveries), and thyroid cancer (4.0 per 100,000 deliveries). Patients with cancer were at significantly higher risk for any severe maternal morbidity (adjusted odds ratio, 5.25; 95% confidence interval, 4.73-5.83) and maternal death (adjusted odds ratio, 67.5; 95% confidence interval, 45.1-101.4). Risks of hysterectomy (adjusted odds ratio, 16.92; 95% confidence interval, 13.96-20.52), acute respiratory distress (adjusted odds ratio, 12.76; 95% confidence interval, 9.92-16.42), sepsis (adjusted odds ratio, 11.91; 95% confidence interval, 8.68-16.32), and embolism (adjusted odds ratio, 11.12; 95% confidence interval, 6.94-17.82) were particularly heightened among patients with cancer. Patients with leukemia, specifically, had the highest risk of adverse maternal outcomes (adjusted rate, 113 per 1000 deliveries; 95% confidence interval, 91-135 per 1000) when evaluating risk by cancer type. Patients with cancer are at markedly increased risk of maternal morbidity and all-cause mortality during delivery-associated hospitalization. Risk is distributed unevenly within this population, with certain cancer types carrying unique risks for specific morbidity events.

Comparison of Machine Learning Algorithms for Predicting Hospital Readmissions and Worsening Heart Failure Events in Patients With Heart Failure With Reduced Ejection Fraction: Modeling Study.

Heart failure (HF) is highly prevalent in the United States. Approximately one-third to one-half of HF cases are categorized as HF with reduced ejection fraction (HFrEF). Patients with HFrEF are at risk of worsening HF, have a high risk of adverse outcomes, and experience higher health care use and costs. Therefore, it is crucial to identify patients with HFrEF who are at high risk of subsequent events after HF hospitalization. Machine learning (ML) has been used to predict HF-related outcomes. The objective of this study was to compare different ML prediction models and feature construction methods to predict 30-, 90-, and 365-day hospital readmissions and worsening HF events (WHFEs). We used the Veradigm PINNACLE outpatient registry linked to Symphony Health's Integrated Dataverse data from July 1, 2013, to September 30, 2017. Adults with a confirmed diagnosis of HFrEF and HF-related hospitalization were included. WHFEs were defined as HF-related hospitalizations or outpatient intravenous diuretic use within 1 year of the first HF hospitalization. We used different approaches to construct ML features from clinical codes, including frequencies of clinical classification software (CCS) categories, Bidirectional Encoder Representations From Transformers (BERT) trained with CCS sequences (BERT + CCS), BERT trained on raw clinical codes (BERT + raw), and prespecified features based on clinical knowledge. A multilayer perceptron neural network, extreme gradient boosting (XGBoost), random forest, and logistic regression prediction models were applied and compared. A total of 30,687 adult patients with HFrEF were included in the analysis; 11.41% (3184/27,917) of adults experienced a hospital readmission within 30 days of their first HF hospitalization, and nearly half (9231/21,562, 42.81%) of the patients experienced at least 1 WHFE within 1 year after HF hospitalization. The prediction models and feature combinations with the best area under the receiver operating characteristic curve (AUC) for each outcome were XGBoost with CCS frequency (AUC=0.595) for 30-day readmission, random forest with CCS frequency (AUC=0.630) for 90-day readmission, XGBoost with CCS frequency (AUC=0.649) for 365-day readmission, and XGBoost with CCS frequency (AUC=0.640) for WHFEs. Our ML models could discriminate between readmission and WHFE among patients with HFrEF. Our model performance was mediocre, especially for the 30-day readmission events, most likely owing to limitations of the data, including an imbalance between positive and negative cases and high missing rates of many clinical variables and outcome definitions. We predicted readmissions and WHFEs after HF hospitalizations in patients with HFrEF. Features identified by data-driven approaches may be comparable with those identified by clinical domain knowledge. Future work may be warranted to validate and improve the models using more longitudinal electronic health records that are complete, are comprehensive, and have a longer follow-up time.

Patients Undergoing Elective Inpatient Major Therapeutic Procedures in Florida Had No Significant Change in Hospital Mortality or Mortality-Related Comorbidities Between 2007 and 2019.

In a recent study, rapid response team implementation at 1 hospital was associated with only a 0.1% reduction in inpatient mortality from 2005 to 2018, characterized in the accompanying editorial as a "tepid" improvement. The editorialist postulated that an increase in the degree of illness of hospitalized patients might have masked a larger reduction that otherwise might have occurred. Impressions of greater patient acuity during the studied period might have been an artifact of efforts to document more comorbidities and complications, possibly facilitated by the change in diagnosis coding from the International Classification of Diseases , Ninth Revision ( ICD-9 ) to the Tenth Revision ( ICD-10 ). We used inpatient data from every nonfederal hospital in Florida from the last quarter of 2007 through 2019. We studied hospitalizations for major therapeutic surgical procedures with lengths of stay ≥2 days. Using logistic regression with clustering by the Clinical Classification Software (CCS) code of the primary surgical procedure, we evaluated the trends for decreased mortality, changes in the prevalence of Medicare Severity Diagnosis Related Groups (MS-DRG) with complications or comorbidities (CC) or major complications or major comorbidities (MCC), and changes in the van Walraven index (vWI), a measure of patient comorbidities associated with increased inpatient mortality. Also incorporated in the modeling was the change from ICD-9 to ICD-10 . There were 3,151,107 hospitalizations comprising 130 distinct CCS codes and 453 MS-DRG groups among 213 hospitals. Despite a progressive increase of 4.1% per year in the odds of a CC or MCC ( P = .001), there were no significant changes in the marginal estimates of in-house mortality over time (net estimated decrease, 0.036%; 99% confidence interval [CI], -0.168% to 0.097%; P = .49). There was also absence of a significantly greater fraction of discharges with vWI >0 attributable to the year of the study (odds ratio, 1.017 per year; 99% CI, 0.995-1.041). The changes in MS-DRG to those with CC or MCC were not increased significantly from either the ICD-10 coding change or the number of years after the change. Consistent with the previous study, there was at most a small decrease in the mortality rate over a 12-year period. We found no reliable evidence that patients undergoing elective inpatient surgical procedures were any sicker in 2019 than in 2007. There were substantively more comorbidities and complications documented over time, but this was unrelated to the change to ICD-10 coding.

Integrating Environmental Data with Medical Data in a Records-Linkage System to Explore Groundwater Nitrogen Levels and Child Health Outcomes.

Background: The Rochester Epidemiology Project (REP) medical records-linkage system offers a unique opportunity to integrate medical and residency data with existing environmental data, to estimate individual-level exposures. Our primary aim was to provide an archetype of this integration. Our secondary aim was to explore the association between groundwater inorganic nitrogen concentration and adverse child and adolescent health outcomes. Methods: We conducted a nested case-control study in children, aged seven to eighteen, from six counties of southeastern Minnesota. Groundwater inorganic nitrogen concentration data were interpolated, to estimate exposure across our study region. Residency data were then overlaid, to estimate individual-level exposure for our entire study population (n = 29,270). Clinical classification software sets of diagnostic codes were used to determine the presence of 21 clinical conditions. Regression models were adjusted for age, sex, race, and rurality. Results: The analyses support further investigation of associations between nitrogen concentration and chronic obstructive pulmonary disease and bronchiectasis (OR: 2.38, CI: 1.64-3.46) among boys and girls, thyroid disorders (OR: 1.44, CI: 1.05-1.99) and suicide and intentional self-inflicted injury (OR: 1.37, CI: >1.00-1.87) among girls, and attention deficit conduct and disruptive behavior disorders (OR: 1.34, CI: 1.24-1.46) among boys. Conclusions: Investigators with environmental health research questions should leverage the well-enumerated population and residency data in the REP.

Have we been underestimating the burden of heart failure?

Have we been underestimating the burden of heart failure?