Objectives: To develop and internally validate a multivariable prediction model for survival to hospital discharge after pediatric extracorporeal cardiopulmonary resuscitation (ECPR) for in-hospital cardiac arrest (IHCA), and to use this model to derive risk-standardized survival rates (RSSRs) and quantify between-hospital variation in outcomes. Design: Retrospective multicenter cohort study. Setting: Hospitals contributing to the Extracorporeal Life Support Organization (ELSO) registry between 2020 and 2023. Patients: Children (< 18 yr) treated with ECPR for IHCA. Interventions: None. Measurements and Main Results: Eligible patients were identified from the ELSO Registry and split into derivation ( N = 1060) and validation ( N = 317) cohorts using a hospital-stratified split to preserve within-hospital clustering. Survival prediction models were derived using penalized logistic regression with the least absolute shrinkage and selection operator. A full-event model incorporating 14 routinely available clinical variables demonstrated fair discrimination (area under the curve, 0.728 in the derivation cohort and 0.678 in the validation cohort) and good calibration (observed-to-expected ratio, 0.995; Brier score, 0.222 in the validation cohort). This model was embedded in a multilevel logistic regression model to generate hospital-level RSSRs for centers reporting five or more ECPR cases. Between-hospital variation was quantified using the median odds ratio, which was 1.30 (95% CI, 1.11–1.93), indicating significant differences in survival likelihood between centers after adjusting for patient-level risk. Conclusions: In a large, international cohort of pediatric ECPR patients, we developed and internally validated a clinically interpretable model for survival to hospital discharge. The model performance as a discriminator likely limits its use as a standalone tool for clinical decision-making. However, strong internal calibration allowed for determination of hospital-level RSSRs, which revealed significant inter-hospital variability. These findings demonstrate the feasibility of developing a risk-adjustment model capable of estimating hospital-level RSSRs. Additional external validation and evaluation of residual case-mix differences are needed before such models can be used for benchmarking across pediatric ECPR centers.
More than 400 million people have chronic obstructive pulmonary disease (COPD), with exacerbations representing a major health burden. Although the overall incidence of invasive pulmonary aspergillosis (IPA) in patients with COPD with hospitalized exacerbation is only 1% to 4%, certain factors substantially increase this frequency. Risk factors compromising defenses against Aspergillus spp. in COPD patients include systemic or high-dose inhaled corticosteroids; comorbidities including bronchiectasis, diabetes, and cardiovascular disease; and prolonged courses of antibiotics. An international group of experts met to develop criteria for diagnosing IPA based on existing literature and consensus in nonventilated COPD patients. The preliminary diagnostic recommendations were further evaluated by additional experts using the Delphi methodology. A hospitalized exacerbation of COPD with 2 or more of the above clinical risk factors should prompt (1) a computed tomography scan of the chest; (2) sending a respiratory sample (sputum, induced sputum or bronchoscopy sample) for direct microscopy for fungi, high-volume fungal culture, and preferably Aspergillus PCR and, if a bronchoscopy sample is obtained, then also Aspergillus antigen (galactomannan); and (3) a serum sample for galactomannan and Aspergillus IgG. The combination of a high-risk COPD patient, with compatible imaging abnormalities and any 2 positive tests (2 samples or different tests on the same respiratory sample) for Aspergillus is sufficient to establish the diagnosis of IPA with enough confidence to initiate antifungal therapy and/or enroll the patient in a clinical or epidemiological study of IPA in COPD. Additional studies are required to augment performance data for most assays in COPD and validate the proposed diagnostic criteria.
Objectives: To determine the contemporary incidence of tracheal intubation–associated cardiac arrest (TI-CA) in critically ill children and identify potentially modifiable factors to prevent TI-CA in PICUs and pediatric cardiac ICUs (PCICUs). Design: Retrospective cohort study Setting: Seventy international PICUs and PCICUs Patients: Critically ill children requiring tracheal intubation in PICUs and PCICUs Interventions: Tracheal intubation data were prospectively collected for all initial intubations in 68 PICUs and PCICUs from January 2015 to December 2023 using the National Emergency Airway Registry for Children registry. Measurements and Main Results: TI-CA was defined as CA occurrence or receipt of chest compressions for at least 1 minute within 20 minutes of endotracheal tube placement. TI during ongoing cardiopulmonary resuscitation was not considered TI-CA. A total of 27,773 TIs were evaluated, with TI-CA reported in 1.2% ( n = 331). Pre-TI oxygenation failure and shock were associated with higher odds of TI-CA (adjusted odds ratio [aOR], 2.29; 95% CI, 1.81–2.89 and aOR, 4.67; 95% CI, 3.68–5.92, respectively). Each additional TI attempt required was associated with increased odds of TI-CA (aOR, 1.70; 95% CI, 1.56–1.85, for each attempt). Use of ketamine (aOR, 0.64; 95% CI, 0.49–0.82), propofol (aOR, 0.44; 95% CI, 0.27–0.72), neuromuscular blockade (aOR, 0.67; 95% CI, 0.49–0.91), and video laryngoscopy (aOR, 0.76; 95% CI, 0.6–0.97) were associated with lower odds of TI-CA. Mediation analysis indicated that 13% of video laryngoscopy’s protective effect was due to a reduction in TI attempt counts. The protective effects of ketamine, propofol, and neuromuscular blockade were minimally mediated by reduced TI attempt counts (5%, 4%, and 7%). Conclusions: TI-CA occurred in 1.2% of TIs performed in the PICUs. Odds of TI-CA increased incrementally with successive TI attempts. Video laryngoscopy was protective against TI-CA. Mediation analysis revealed that reduced TI attempts accounted for 13% of the protective benefit associated with video laryngoscopy.
