LeMahieu, Allison M

Background/Objective: To externally validate our previously published machine learning model for identifying hospitalized patients with community-acquired pneumonia (CAP) at low risk of respiratory deterioration or death using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Methods: This is a retrospective cohort study of adult patients (≥18 years) who were admitted with CAP and acute hypoxemic respiratory failure using the publicly available MIMIC-IV critical care dataset (Beth Israel Deaconess Medical Center, Boston, MA). We conducted an external validation study of a previously developed gradient boosting machine (GBM) model without recalibration using data available within the first 6 h of hospital admission. Results: For the primary composite outcome (need for advanced respiratory support [high flow nasal cannula (HFNC), non-invasive mechanical ventilation (NIMV), invasive mechanical ventilation (IMV)] or in-hospital death), the gradient boosting model demonstrated comparable performance in the derivation and external validation cohorts. The area under the receiver operating characteristic curve (AUC) was 0.713 in the Mayo cohort (n = 4379) and 0.689 in the MIMIC-IV cohort. Accuracy was 0.612 (95% confidence interval [CI], 0.595–0.628) versus 0.606 (95% CI 0.600–0.611), specificity 0.574 versus 0.523, sensitivity 0.723 versus 0.754, NPV 0.860 versus 0.842, and PPV 0.364 versus 0.401, respectively. For secondary outcomes, model discrimination was comparable between cohorts. The AUC for in-hospital mortality was 0.727 in the Mayo Cohort versus 0.733 in MIMIC-IV; for IMV, 0.736 versus 0.724; and for NIMV, 0.732 versus 0.708. Conclusions: Our machine learning algorithm demonstrated good discrimination and a high negative predictive value for identifying low-risk hospitalized CAP patients for respiratory deterioration or death in the external MIMIC-IV dataset, supporting its potential utility for prognostic enrichment in pneumonia clinical trials.

Objective Asking students to answer 30 + questions in an optional student evaluation of teaching (SET) increases the likelihood that instructors may not receive critical feedback due to survey fatigue. A considerable body of educational research on how to incentivize student completion of SETs offers a variety of suggestions to combat fatigue, perceptions of value, and other issues. Methods This retrospective cohort study from January 2021 to March 2024 investigates the impact of a shortened, 4-question mandatory SET on response rates and qualitative feedback within a Clinical and Translational Science (CTS) program. Unadjusted linear mixed-effects regression models assessed the relationships of SET type (optional SET vs experimental SET) and SET response outcomes. Results Over the course of the study, 844 students completed optional (N = 473) and experimental (N = 371) SETs. Requiring SET completion to access assessments increased both participation (Estimate 62.43, 95% CI: 52.41-72.45; P < .001) and qualitative responses (Est. 43.65, 95% CI: 32.24-55.05; P < .001). Relative to the optional SET, the experimental SET was associated with more blank qualitative responses (Est. 7.28, 95% CI: 4.70-9.86; P < .001) and a higher total word count (Est. 989.59, 95% CI: 513.78-1465.40; P < .001). Course satisfaction scores did not differ significantly between survey types (Est. 0.05, 95% CI: −0.13-0.22; P = .60). Conclusion This study's findings demonstrate that short, mandatory, learning management system-integrated SETs effectively increase qualitative feedback and can offer a feasible model for institutions aiming to gather actionable items to inform course design and delivery.

Background: While criteria for the diagnosis of nosocomial pneumonias exist, objective definitions are a challenge and there is no gold standard for diagnosis. We analyzed the impact of the implementation of a logical, consensus-based diagnostic and treatment protocol for managing nosocomial pneumonias in the cardiovascular surgery intensive care unit (CVS-ICU). Methods: We conducted a quasi-experimental, interrupted time series analysis to evaluate the impact of a diagnostic and treatment protocol for nosocomial pneumonias in the CVS-ICU. Impacts were measured relative to patient outcomes, diagnostic processes, and antimicrobial stewardship improvement. Descriptive statistics were used to analyze results. Results: Overall, 35 pre-protocol and 39 post-protocol patients were included. Primary clinical variables suggesting pneumonia in pre- and post-protocol patients were new lung consolidation (50% vs. 71%), new leukocytosis (59% vs. 64%), and positive culture (32% vs. 55%). Appropriate diagnostic testing improved (23% vs. 54%, p = 0.008) after protocol implementation. The proportion of patients meeting the criteria for nosocomial pneumonia (77% vs. 87%) was not statistically significant, though more patients in the post-protocol group met probable diagnostic criteria (51% vs. 77%). Duration of therapy was not significantly different (6 days [IQR = 5.0, 10.0] vs. 7 days [IQR = 6.0, 9.0]). Conclusions: The implementation of a diagnostic and treatment protocol for management of nosocomial pneumonias in the CVS-ICU resulted in improved diagnostic accuracy, advanced antimicrobial and diagnostic stewardship efforts, and laboratory cost savings without an adverse impact on patient-centered outcomes.

Objective This study investigated the occurrence of postictal agitation (PA) in patients undergoing an acute series of electroconvulsive therapy (ECT) and further explored patient and treatment variables associated with PA. Methods Charts were retrospectively searched for patients undergoing an acute series of ECT. Postictal agitation was identified by the administration of a sedative after ECT. Demographic, diagnostic, medication, and ECT variables that could also be associated with PA were collected and accounted for in statistical analysis. Results In this population, 22 of 156 patients experienced PA. Associations that reached statistical significance included sex, weight, active substance use disorder, seizure duration (as observed by motor movements), and waking time. Only seizure duration and waking time maintained significance after multivariable analysis. Conclusions These data identify clinical factors that could help predict PA. Patients with greater weight, male sex, or an active substance use disorder ought to be carefully monitored for PA, and staff in the recovery suite should be especially vigilant about such patients with longer seizures and waking times.