Treatment, clinical outcomes, and predictors of mortality among a national cohort of admitted patients with Acinetobacter baumannii infection

Objectives: To analyze treatment, clinical outcomes, and predictors of mortality in hospitalized patients with Acinetobacter baumannii infection. Methods: Retrospective cohort study of inpatients with A. baumannii cultures and treatment from 2010-2019. Patients who died during admission were compared to those who survived to identify predictors of inpatient mortality, using multivariable unconditional logistic regression models. Results: We identified 4,599 inpatients with A. baumannii infection; 13.6% died during admission. Fluoroquinolones (26.8%), piperacillin/tazobactam (24%) and carbapenems (15.6%) were used for treatment. Tigecycline (3%) and polymyxins (3.7%) were not used often. Predictors of inpatient mortality included current acute respiratory failure (adjusted odds ratio [aOR] 3.94), shock (aOR 3.05), and acute renal failure (aOR 2.01); blood (aOR 1.94) and respiratory (aOR 1.64) infectious source; multidrug-resistant A. baumannii (MDRAB) infection (aOR 1.66); liver disease (aOR 2.15); and inadequate initial treatment (aOR 1.30). Inpatient mortality was higher in those with MDRAB vs. non-MDRAB (aOR 1.61) and in those with CRAB vs. non-CRAB infection (aOR 1.68). Length of stay >10 days was higher among those with MDRAB vs. non-MDRAB (aOR 1.25) and in those with CRAB vs. non-CRAB infection (aOR 1.31). Conclusions: In our national cohort of inpatients with A. baumannii infection, clinical outcomes were worse among those with MDRAB and/or CRAB infection. Predictors of inpatient mortality included several current conditions associated with severity, infectious source, underlying illness, and inappropriate treatment. Our study may assist healthcare providers in the early identification of admitted patients with A. baumannii infection who are at higher risk of death.

Impact of atrial fibrillation on in-hospital outcomes among hospitalizations for cardiac surgery: an analysis of the National Inpatient Sample

The short-term impact of atrial fibrillation (AF) on cardiac surgery hospitalizations has been previously reported in cohorts of various sizes, but results have been variable. Using the 2005–2014 National Inpatient Sample, we identified all adult hospitalizations for cardiac surgery using the International Classification of Diseases, Ninth Revision, Clinical Modification as any procedure code and AF as any diagnosis code. We estimated the impact of AF on inpatient mortality, length of stay (LOS), and cost of hospitalization using survey-weighted, multivariable logistic, accelerated failure-time log-normal, and log-transformed linear regressions, respectively. Additionally, we exact-matched AF to non-AF hospitalizations on various confounders for the same outcomes. A total of 1,269,414 hospitalizations were noted for cardiac surgery during the study period. Coexistent AF was found in 44.9% of these hospitalizations. Overall mean age was 65.6 years, 40.9% were female, mean LOS was 11.6 days, and inpatient mortality was 4.5%. Stroke rate was lower in AF hospitalizations (1.8% vs 2.1%, p<0.001). Mortality was lower in the AF (3.9%) versus the non-AF (5%) group (exact-matched OR or emOR=0.48, 95% CI 0.29 to 0.80, p<0.001; 987 matched pairs, n=2423), with similar results after procedural stratification: isolated valve replacement/repair (emOR=0.38, p<0.001), isolated coronary artery bypass graft (CABG) (emOR=0.33, p<0.001), and CABG with valve replacement/repair (emOR=0.55, p<0.001). A 12% increase was seen in LOS in the AF subgroup (exact-matched time ratio=1.12, 95% CI 1.10 to 1.14, p<0.001) among hospitalizations which underwent valve replacement/repair with or without CABG. Hospitalizations for cardiac surgery which had coexistent AF were found to have lower inpatient mortality risk and stroke prevalence but higher LOS and hospitalization costs compared with hospitalizations without AF.

Enhancing Robustness of Machine Learning Integration With Routine Laboratory Blood Tests to Predict Inpatient Mortality After Intracerebral Hemorrhage

Objective: The accurate evaluation of outcomes at a personalized level in patients with intracerebral hemorrhage (ICH) is critical clinical implications. This study aims to evaluate how machine learning integrates with routine laboratory tests and electronic health records (EHRs) data to predict inpatient mortality after ICH.Methods: In this machine learning-based prognostic study, we included 1,835 consecutive patients with acute ICH between October 2010 and December 2018. The model building process incorporated five pre-implant ICH score variables (clinical features) and 13 out of 59 available routine laboratory parameters. We assessed model performance according to a range of learning metrics, such as the mean area under the receiver operating characteristic curve [AUROC]. We also used the Shapley additive explanation algorithm to explain the prediction model.Results: Machine learning models using laboratory data achieved AUROCs of 0.71–0.82 in a split-by-year development/testing scheme. The non-linear eXtreme Gradient Boosting model yielded the highest prediction accuracy. In the held-out validation set of development cohort, the predictive model using comprehensive clinical and laboratory parameters outperformed those using clinical alone in predicting in-hospital mortality (AUROC [95% bootstrap confidence interval], 0.899 [0.897–0.901] vs. 0.875 [0.872–0.877]; P &lt;0.001), with over 81% accuracy, sensitivity, and specificity. We observed similar performance in the testing set.Conclusions: Machine learning integrated with routine laboratory tests and EHRs could significantly promote the accuracy of inpatient ICH mortality prediction. This multidimensional composite prediction strategy might become an intelligent assistive prediction for ICH risk reclassification and offer an example for precision medicine.

Effect of COVID-19 on presentations of decompensated liver disease in Scotland

Background and aimsSARS-CoV-2 and consequent pandemic has presented unique challenges. Beyond the direct COVID-related mortality in those with liver disease, we sought to determine the effect of lockdown on people with liver disease in Scotland. The effect of lockdown on those with alcohol-related disease is of interest; and whether there were associated implications for a change in alcohol intake and consequent presentations with decompensated disease.MethodsWe performed a retrospective analysis of patients admitted to seven Scottish hospitals with a history of liver disease between 1 April and 30 April 2020 and compared across the same time in 2017, 2018 and 2019. We also repeated an intermediate assessment based on a single centre to examine for delayed effects between 1 April and 31 July 2020.ResultsWe found that results and outcomes for patients admitted in 2020 were similar to those in previous years in terms of morbidity, mortality, and length of stay. In the Scotland-wide cohort: admission MELD (Model for End-stage Liver Disease) (16 (12–22) vs 15 (12–19); p=0.141), inpatient mortality ((10.9% vs 8.6%); p=0.499) and length of stay (8 days (4–15) vs 7 days (4–13); p=0.140). In the Edinburgh cohort: admission MELD (17 (12–23) vs 17 (13–21); p=0.805), inpatient mortality ((13.7% vs 10.1%; p=0.373) and length of stay (7 days (4–14) vs 7 days (3.5–14); p=0.525)).ConclusionThis assessment of immediate and medium-term lockdown impacts on those with chronic liver disease suggested a minimal effect on the presentation of decompensated liver disease to secondary care.

Inpatient Mortality Among Patients With Acute Respiratory Distress Syndrome at ECMO and Non-ECMO Centers in the United States

Background A 2009 randomized control trial found patients with severe acute respiratory distress syndrome (ARDS) who transferred to an extra-corporeal membrane oxygenation therapy (ECMO) center had better survival, even if they did not receive ECMO. This study aimed to use a national US database to determine if care at ECMO centers offer a survival advantage in patients with ARDS with mechanical ventilation only. Methods Hospitalizations of patients 18-64 years old who had ARDS and mechanical ventilation in the 2010-2016 Health care Cost and Utilization Project National Readmission Database were included. ECMO centers performed at least 1 veno-venous ECMO hospitalization annually; or >5, >20, and >50 on sensitivity analysis. Multivariable logistic regression compared inpatient mortality, after adjusting for timing of hospitalization, patient demographics, comorbidities, and hospital characteristics. Results Of the 1 224 447 ARDS hospitalizations and mechanical ventilation, 41% were at ECMO centers. ECMO centers were more likely to be larger, private, non-profit, teaching hospitals. ARDS at admission was more common at non-ECMO centers (31% vs 23%, P < .0001); however, other patient demographics and comorbidities did not differ. After adjustment, no difference in inpatient mortality was seen between ECMO and non-ECMO centers (OR 0.99, 95% CI: 0.97, 1.02). This relationship did not change in sensitivity analyses. Discussion Adult patients with ARDS requiring mechanical ventilation may not have improved outcomes if treated at an ECMO center and suggest that early transfer of all ARDS patients to ECMO centers may not be warranted. Further evaluation of ECMO center volume and illness severity is needed.

Controlled Study of Decision-Making Algorithms for Kidney Replacement Therapy Initiation in Acute Kidney Injury

Background: Acute kidney injury (AKI) requiring kidney replacement therapy (KRT) is associated with high mortality and utilization. We evaluated the use of an AKI-Standardized Clinical Assessment and Management Plan (SCAMP) on patient outcomes including mortality, hospital and ICU length of stay. Methods: We conducted a 12-month controlled study in the ICUs of a large academic tertiary medical center. We alternated use of the AKI-SCAMP with use of a "sham" control form in 4-6-week blocks. The primary outcome was risk of inpatient mortality. Pre-specified secondary outcomes included 30-day mortality, 60-day mortality and hospital and ICU length of stay. Generalized estimating equations were used to estimate the impact of the AKI-SCAMP on mortality and length of stay. Results: There were 122 patients in the AKI-SCAMP group and 102 patients in the control group. There was no significant difference in inpatient mortality associated with AKI-SCAMP use (41% vs 47% control). AKI-SCAMP use was associated with significantly reduced ICU length of stay (mean 8 (95% CI 8-9) vs 12 (95% CI 10-13) days; p = <0.0001) and hospital length of stay (mean 25 (95% CI 22-29) vs 30 (95% CI 27-34) days; p = 0.02). Patients in the AKI-SCAMP group less likely to receive KRT in the context of physician-perceived treatment futility than those in the control group (2% vs 7%, p=0.003). Conclusions: Use of the AKI-SCAMP tool for AKI-KRT was not significantly associated with inpatient mortality but was associated with reduced ICU and hospital length of stay and use of KRT in cases of physician-perceived treatment futility.

Does Frailty influence Inhospital Management and Outcomes of COVID-19 in Older Adults in the US?

Older age has been consistently associated with adverse COVID-19 outcomes. Frailty, a syndrome characterized by declining function across multiple body systems is common in older adults and may increase vulnerability to adverse outcomes among COVID-19 patients. However, the impacts of frailty on COVID-19 management, severity, or outcomes have not been well characterized in a large, representative US population. Using the National COVID Cohort Collaborative, a multi-institutional US repository for COVID-19 research, we calculated the Hospital Frailty Risk Score (HFRS), a validated EHR-based frailty score, among COVID-19 inpatients age ≥ 65. We examined patient demographics and comorbidities, length of stay (LOS), systemic corticosteroid and remdesivir use, ICU admission, and inpatient mortality across subgroups by HFRS score. Among 58,964 inpatients from 53 institutions (51% male, 65% White, 18% Black, 9% Hispanic, mean age 75, mean Charlson comorbidity count 3.0, and median LOS 7 days), 38,692 (66%), 4,180 (7%), 3,531 (6%), 3,525 (6%) and 7,862 (13%) had HFRS scores of 0-1, 2, 3, 4, and &gt;=5 , respectively. Frailty was only moderately correlated with age and comorbidity (□=0.178 and 0.348, respectively, p&lt;0.001). Overall, 34% received systemic corticosteroid and 19% received remdesivir. We observed 4% ICU admissions and 16% inpatient death. Among non-ICU admissions, after adjusting for demographics and comorbidities, frailty (HFRS ≥ 2) was associated with 79% greater systemic corticosteroid use and 22% greater remdesivir use, whereas a higher HRFS score was marginally associated with higher rates of severe COVID disease, inpatient death, or ICU admission.

Increased Mortality in Hip Fracture Patients With Aortic Stenosis and Pulmonary Hypertension Compared to Aortic Stenosis Alone

Introduction Hip fractures are one of the most common traumatic injuries in the United States, secondary to an aging population. Multiple comorbidities are found in patients who present to trauma centers (TCs) with isolated hip fractures (IHFs) including significant cardiac disease. Aortic stenosis (AS) among these patients has been recently shown to increase mortality. However, factors leading to death from AS are unknown. We hypothesize that pulmonary hypertension (PH) is a significant mechanism of death among IHF patients with AS. Methods This is a multicenter retrospective cohort study examining IHF patients treated at Level I and II TCs within a large hospital system from 2015 to 2019. Patients who had IHFs and AS were compared to those who had IHFs, AS, and PH. Multivariable logistic regression was used to risk adjust by age, race, insurance status, and comorbidities. The primary outcome was inpatient mortality. The secondary outcomes were hospital-acquired complications. Results A total of 1388 IHF patients with AS were included in the study. Eleven percent of these patients also had PH. The crude mortality rate was higher if IHF patients had both AS and PH compared to IHF with AS alone (9% vs 3.7%, P-value .003). After risk adjustment, a higher risk of mortality was still significant (aOR 2.56 [95% CI 1.28, 5.11]). In addition, IHF patients with both AS and PH had higher complication rates; the exposure group had higher percentage of pulmonary embolism (1.4% vs .2%, adjusted P-value .03), new-onset congestive heart failure (4.1% vs 1%, adjusted P-value .01), and sepsis/septicemia (3.5% vs 1.4%, adjusted P-value .05). Conclusion In patients with IHFs, PH and AS increase the likelihood of inpatient mortality by 2.5 times compared to AS alone. Pulmonary hypertension among IHF patients with AS is an important risk factor to identify in the preoperative period. Early identification may lead to better perioperative management and counseling of patients at higher risk of complications.