Distinctive evolution of alveolar T cell responses is associated with clinical outcomes in unvaccinated patients with SARS-CoV-2 pneumonia.

The evolution of T cell molecular signatures in the distal lung of patients with severe pneumonia is understudied. Here, we analyzed T cell subsets in longitudinal bronchoalveolar lavage fluid samples from 273 patients with severe pneumonia, including unvaccinated patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) or with respiratory failure not linked to pneumonia. In patients with SARS-CoV-2 pneumonia, activation of interferon signaling pathways, low activation of the NF-κB pathway and preferential targeting of spike and nucleocapsid proteins early after intubation were associated with favorable outcomes, whereas loss of interferon signaling, activation of NF-κB-driven programs and specificity for the ORF1ab complex late in disease were associated with mortality. These results suggest that in patients with severe SARS-CoV-2 pneumonia, alveolar T cell interferon responses targeting structural SARS-CoV-2 proteins characterize individuals who recover, whereas responses against nonstructural proteins and activation of NF-κB are associated with poor outcomes.

Local and systemic responses to SARS-CoV-2 infection in children and adults.

It is not fully understood why COVID-19 is typically milder in children1-3. Here, to examine the differences between children and adults in their response to SARS-CoV-2 infection, we analysed paediatric and adult patients with COVID-19 as well as healthy control individuals (total n = 93) using single-cell multi-omic profiling of matched nasal, tracheal, bronchial and blood samples. In the airways of healthy paediatric individuals, we observed cells that were already in an interferon-activated state, which after SARS-CoV-2 infection was further induced especially in airway immune cells. We postulate that higher paediatric innate interferon responses restrict viral replication and disease progression. The systemic response in children was characterized by increases in naive lymphocytes and a depletion of natural killer cells, whereas, in adults, cytotoxic T cells and interferon-stimulated subpopulations were significantly increased. We provide evidence that dendritic cells initiate interferon signalling in early infection, and identify epithelial cell states associated with COVID-19 and age. Our matching nasal and blood data show a strong interferon response in the airways with the induction of systemic interferon-stimulated populations, which were substantially reduced in paediatric patients. Together, we provide several mechanisms that explain the milder clinical syndrome observed in children.

Microbiology of Severe Community-Acquired Pneumonia and the Role of Rapid Molecular Techniques.

The microbiology of severe community acquired pneumonia (SCAP) has implications on management, clinical outcomes and public health policy. Therefore, knowledge of the etiologies of SCAP and methods to identify these microorganisms is key. Bacteria including Streptococcus pneumoniae, Staphylococcus aureus and Enterobacteriaceae continue to be important causes of SCAP. Viruses remain the most commonly identified etiology of SCAP. Atypical organisms are also important etiologies of SCAP and are critical to identify for public health. With the increased number of immunocompromised individuals, less common pathogens may also be found as the causative agent of SCAP. Traditional diagnostic tests, including semi-quantitative respiratory cultures, blood cultures and urinary antigens continue to hold an important role in the evaluation of patients with SCAP. Many of the limitations of the aforementioned tests are addressed by rapid, molecular diagnostic tests. Molecular diagnostics utilize culture-independent technology to identify species-specific genetic sequences. These tests are often semi-automated and provide results within hours, which provides an opportunity for expedient antibiotic stewardship. The existing literature suggests molecular diagnostic techniques may improve antibiotic stewardship in CAP, and future research should investigate optimal methods for implementation of these assays into clinical practice.

Obesity Is Associated with Attenuated Tissue Immunity in COVID-19.

Rationale: Obesity affects 40% of U.S. adults, is associated with a proinflammatory state, and presents a significant risk factor for the development of severe coronavirus disease (COVID-19). To date, there is limited information on how obesity might affect immune cell responses in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Objectives: To determine the impact of obesity on respiratory tract immunity in COVID-19 across the human lifespan. Methods: We analyzed single-cell transcriptomes from BAL in three ventilated adult cohorts with (n = 24) or without (n = 9) COVID-19 from nasal immune cells in children with (n = 14) or without (n = 19) COVID-19, and from peripheral blood mononuclear cells in an independent adult COVID-19 cohort (n = 42), comparing obese and nonobese subjects. Measurements and Main Results: Surprisingly, we found that obese adult subjects had attenuated lung immune or inflammatory responses in SARS-CoV-2 infection, with decreased expression of IFN-α, IFN-γ, and TNF-α (tumor necrosis factor α) response gene signatures in almost all lung epithelial and immune cell subsets, and lower expression of IFNG and TNF in specific lung immune cells. Peripheral blood immune cells in an independent adult cohort showed a similar but less marked reduction in type-I IFN and IFNγ response genes, as well as decreased serum IFNα, in obese patients with SARS-CoV-2. Nasal immune cells from obese children with COVID-19 also showed reduced enrichment of IFN-α and IFN-γ response genes. Conclusions: These findings show blunted tissue immune responses in obese patients with COVID-19, with implications for treatment stratification, supporting the specific application of inhaled recombinant type-I IFNs in this vulnerable subset.

Clinical Features of COVID-19 and Differentiation from Other Causes of CAP.

Community-acquired pneumonia (CAP) is a significant cause of morbidity and mortality, one of the most common reasons for infection-related death worldwide. Causes of CAP include numerous viral, bacterial, and fungal pathogens, though frequently no specific organism is found. Beginning in 2019, the COVID-19 pandemic has caused incredible morbidity and mortality. COVID-19 has many features typical of CAP such as fever, respiratory distress, and cough, and can be difficult to distinguish from other types of CAP. Here, we highlight unique clinical features of COVID-19 pneumonia such as olfactory and gustatory dysfunction, lymphopenia, and distinct imaging appearance.

Anthropomorphic Character Animations Versus Digital Chalk Talks in a Resident Diabetes Pharmacotherapy Curriculum: a Randomized Controlled Trial.

BACKGROUND: Animation in medical education has boomed over the past two decades, and demand for distance learning technologies will likely continue in the context of the COVID-19 pandemic. However, experimental data guiding best practices for animation in medical education are scarce. OBJECTIVE: To compare the efficacy of two animated video styles in a diabetes pharmacotherapy curriculum for internal medicine residents. DESIGN: Learners were randomized to receive one of two versions of the same multimodal didactic curriculum. They received identical lectures, group activities, and quizzes, but were randomized to either digital chalk talk (DCT) videos or Sugar-Coated Science (SCS). SCS is an animated series using anthropomorphic characters, stories, and mnemonics to communicate knowledge. PARTICIPANTS: Ninety-two internal medicine residents at a single academic medical center received the curriculum within ambulatory medicine didactics. MAIN MEASURES: Knowledge was measured at multiple time points, as was residents' self-reported comfort using each medication class covered. Surveys assessed video acceptability and telepresence. Key themes were identified from open-ended feedback. KEY RESULTS: Baseline knowledge was low, consistent with prior needs assessments. On immediate posttest, mean scores were higher with SCS than DCT (74.8% versus 68.4%), but the difference was not statistically significant, p = 0.10. Subgroup analyses revealed increased knowledge in the SCS group for specific medication classes. Delayed posttest showed significant knowledge gains averaging 17.6% across all participants (p < 0.05); these gains were similar between animation types. SCS achieved significantly higher telepresence, entertainment, and acceptability scores than DCT. Qualitative data suggested that residents prioritize well-designed, multimodal curricula over specific animation characteristics. CONCLUSION: SCS and DCTs both led to learning within a multimodal curriculum, but SCS significantly enhanced learner experience. Animation techniques exemplified by both SCS and DCTs have roles in the medical educator toolkit. Selection between them should incorporate context, learner factors, and production resources.

Bacterial Superinfection Pneumonia in Patients Mechanically Ventilated for COVID-19 Pneumonia.

Rationale: Current guidelines recommend patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pneumonia receive empirical antibiotics for suspected bacterial superinfection on the basis of weak evidence. Rates of ventilator-associated pneumonia (VAP) in clinical trials of patients with SARS-CoV-2 pneumonia are unexpectedly low. Objectives: We conducted an observational single-center study to determine the prevalence and etiology of bacterial superinfection at the time of initial intubation and the incidence and etiology of subsequent bacterial VAP in patients with severe SARS-CoV-2 pneumonia. Methods: Bronchoscopic BAL fluid samples from all patients with SARS-CoV-2 pneumonia requiring mechanical ventilation were analyzed using quantitative cultures and a multiplex PCR panel. Actual antibiotic use was compared with guideline-recommended therapy. Measurements and Main Results: We analyzed 386 BAL samples from 179 patients with SARS-CoV-2 pneumonia requiring mechanical ventilation. Bacterial superinfection within 48 hours of intubation was detected in 21% of patients. Seventy-two patients (44.4%) developed at least one VAP episode (VAP incidence rate = 45.2/1,000 ventilator days); 15 (20.8%) initial VAPs were caused by difficult-to-treat pathogens. The clinical criteria did not distinguish between patients with or without bacterial superinfection. BAL-based management was associated with significantly reduced antibiotic use compared with guideline recommendations. Conclusions: In patients with SARS-CoV-2 pneumonia requiring mechanical ventilation, bacterial superinfection at the time of intubation occurs in <25% of patients. Guideline-based empirical antibiotic management at the time of intubation results in antibiotic overuse. Bacterial VAP developed in 44% of patients and could not be accurately identified in the absence of microbiologic analysis of BAL fluid.

Circadian Biology and Its Importance to Intensive Care Unit Care and Outcomes.

Circadian rhythms are an integral part of life on earth. Circadian rhythms play a fundamental role in homeostasis as they ensure coordination between the environment and an organism's behavior and physiology. This coordination is called entrainment. Entrainment depends on environmental cues known as zeitgebers. Human zeitgebers include light (primary zeitgeber), sleep, eating, exercise, and activity. Circadian rhythms are disrupted in critically-ill patients due to both critical illness and current intensive care unit (ICU) practices. Disruptions in circadian rhythms are tightly linked with ICU sleep disruption. Together these entities potentiate numerous adverse outcomes including delirium, metabolic derangements, cardiovascular instability, and immune compromise. Herein, we will highlight potential areas for care improvement via chronobundles. We suggest bright light during the day, maintaining darkness, and protecting sleep at night, intermittent rather than continuous feeds, and activity via mobilization during the day. Optimizing circadian rhythms is a low-risk intervention that is underutilized in current ICU practice. This optimization could be a powerful tool in helping to improve outcomes in the critically-ill patient.

Neutrophil percentages in bronchoalveolar lavage fluid: Implications for diagnosing bacterial pneumonia in patients with immunocompromise and neutropenia.

Background: Pneumonia is the leading cause of infectious deaths and the most common infection identified in ICU patients. Assessment of bronchoalveolar lavage fluid (BALF) cellularity can aid in pneumonia diagnosis. Low percentages (<50%) of BALF neutrophils have a high negative predictive value for bacterial pneumonia in a general medical ICU population, but unclear operating characteristics in patients with immunocompromise and neutropenia remain unknown. Methods: We analyzed a large cohort of BALF specimens obtained for routine care for suspected pneumonia in mechanically ventilated patients and enrolled in the single-center Successful Clinical Response In Pneumonia Therapy (SCRIPT) study. BALF neutrophils were reported as a percentage of leukocytes by the clinical laboratory. The etiology of each episode of suspected pneumonia was adjudicated by a committee of critical care physicians using a predefined protocol. Immunocompromise was defined using predetermined criteria by the study research team. Neutropenia was defined here as a peripheral ANC <1500 cells/µl. Data are expressed as median [Quartile (Q) 1, Q3] and compared using the Mann-Whitney U test. Results: 688 mechanically ventilated patients with suspected pneumonia were included. 409 (59.4%) were male; median age was 62 [51,71]. 461 patients (67.0%) were immunocompetent, 149 (21.7%) were immunocompromised without neutropenia and 78 (11.3%) were neutropenic at some point during their admission. A total of 1746 BALs were performed. Fifty-seven BALs were obtained on a day where the patient's ANC<1500. Amongst pneumonia episodes classified as bacterial, no difference was found amongst BALF percent neutrophils taken patients who were immunocompetent and those who were immunocompromised but not neutropenic on day of sampling: 84.0% [69.0, 93.0] vs 87.0% [68.3, 93.0], p = 0.878 (Figure 1B). However, BALF percent neutrophils were significantly lower in patients neutropenic on day of sampling, with median BALF percent neutrophils of only 65.0% [22.3, 70.5] (p=0.016 compared with immunocompromised group, p=0.0096 compared with immunocompetent group). Conclusion: Among patients with bacterial pneumonia, BALF neutrophil percentage was not significantly decreased by a spectrum of immunocompromise. However, the subset of patients who were acutely neutropenic at the time of BAL sampling had significantly lower BALF % neutrophils. A traditional approach using BALF<50% to suggest against bacterial pneumonia may be inaccurate in this particular population.

An observational cohort study of bronchoalveolar lavage fluid galactomannan and Aspergillus culture positivity in patients requiring mechanical ventilation.

Rationale: Critically ill patients who develop invasive pulmonary aspergillosis (IPA) have high mortality rates despite antifungal therapy. Diagnosis is difficult in these patients. Bronchoalveolar lavage (BAL) fluid galactomannan (GM) is a helpful marker of infection, although the optimal cutoff for IPA is unclear. We aimed to evaluate the BAL fluid GM and fungal culture results, demographics, and outcomes among a large cohort of mechanically ventilated patients with suspected pneumonia. Methods: A single-center cohort study of patients enrolled in the Successful Clinical Response in Pneumonia Therapy (SCRIPT) study from June 2018 to March 2023. Demographics, BAL results, and outcomes data were extracted from the electronic health record and compared between groups of patients who grew Aspergillus on a BAL fluid culture, those who had elevated BAL fluid GM levels (defined as >0.5 or >0.8) but did not grow Aspergillus on BAL fluid culture, and those with neither. Results: Of over 1700 BAL samples from 688 patients, only 18 BAL samples grew Aspergillus. Patients who had a BAL sample grow Aspergillus (n=15) were older (median 71 vs 62 years, p=0.023), had more days intubated (29 vs 11, p=0.002), and more ICU days (34 vs 15, p=0.002) than patients whose BAL fluid culture was negative for Aspergillus (n=672). The BAL fluid galactomannan level was higher from samples that grew Aspergillus on culture than those that did not (median ODI 7.08 vs 0.11, p<0.001), though the elevation of BAL fluid GM varied across BAL samples for patients who had serial sampling. Patients who grew Aspergillus had a similar proportion of underlying immunocompromise compared with the patients who did not, and while no statistically significant difference in overall unfavorable outcome, had longer duration of ventilation and longer ICU stays. Conclusions: In this large cohort of critically ill patients with a high number of BAL samples with GM levels, we found a relatively low rate of Aspergillus growth. Patients who eventually grew Aspergillus had inconsistently elevated BAL fluid GM, and many patients with elevated BAL fluid GM did not grow Aspergillus. These data suggest that the pre-test probability of invasive pulmonary aspergillosis should be considered low in a general ICU population undergoing BAL evaluation to define the etiology of pneumonia. Improved scoring systems are needed to enhance pre-test probability for diagnostic test stewardship purposes.

Antibiotic De-escalation Patterns and Outcomes in Critically Ill Patients with Suspected Pneumonia as Informed by Bronchoalveolar Lavage Results.

Background: Antibiotic stewardship in critically ill pneumonia patients is crucial yet challenging, partly due to the limited diagnostic yield of noninvasive infectious tests. In this study, we report an antibiotic prescription pattern informed by bronchoalveolar lavage (BAL) results, where clinicians de-escalate antibiotics based on the combination of quantitative cultures and multiplex PCR rapid diagnostic tests. Methods: We analyzed data from SCRIPT, a single-center prospective cohort study of mechanically ventilated patients who underwent a BAL for suspected pneumonia. We used the novel Narrow Antibiotic Therapy (NAT) score to quantify day-by-day antibiotic prescription pattern for each suspected pneumonia episode etiology (bacterial, viral, mixed bacterial/viral, microbiology-negative, and non-pneumonia control). We also analyzed and compared clinical outcomes for each pneumonia etiology, including unfavorable outcomes (a composite of in-hospital mortality, discharge to hospice, or requiring lung transplantation during hospitalization), duration of ICU stay, and duration of intubation. Clinical outcomes were compared with the Mann-Whitney U test and Fisher's exact test. Results: We included 686 patients with 927 pneumonia episodes. NAT score analysis indicated that an antibiotic de-escalation pattern was evident in all pneumonia etiologies except resistant bacterial pneumonia. Microbiology-negative pneumonia was treated similarly to susceptible bacterial pneumonia in terms of antibiotic spectrum. Over a quarter of the time in viral pneumonia episodes, antibiotics were completely discontinued. Unfavorable outcomes were comparable across all pneumonia etiologies. Patients with viral and mixed bacterial/viral pneumonia had longer durations of ICU stay and intubation. Conclusions: BAL quantitative cultures and multiplex PCR rapid diagnostic tests resulted in prompt antibiotic de-escalation in critically ill pneumonia patients. There was no evidence of increased incidence of unfavorable outcomes. Key points: We found that bronchoalveolar lavage-based diagnostics is associated with antibiotic de-escalation, including to no antibiotics in more than a quarter of cases where only a virus is recovered.

Antibiotic use during influenza infection augments lung eosinophils that impair immunity against secondary bacterial pneumonia.

A leading cause of mortality after influenza infection is the development of a secondary bacterial pneumonia. In the absence of a bacterial superinfection, prescribing antibacterial therapies is not indicated but has become a common clinical practice for those presenting with a respiratory viral illness. In a murine model, we found that antibiotic use during influenza infection impaired the lung innate immunologic defenses toward a secondary challenge with methicillin-resistant Staphylococcus aureus (MRSA). Antibiotics augment lung eosinophils, which have inhibitory effects on macrophage function through the release of major basic protein. Moreover, we demonstrated antibiotic treatment during influenza infection causes a fungal dysbiosis that drive lung eosinophilia and impair MRSA clearance. Finally, we evaluated three cohorts of hospitalized patients and found eosinophils positively correlated with antibiotic use, systemic inflammation, and worsened outcomes. Altogether, our work demonstrates a detrimental effect of antibiotic treatment during influenza infection that has harmful immunologic consequences via recruitment of eosinophils to the lungs thereby increasing the risk of developing a secondary bacterial infection.

A Common Longitudinal Intensive Care Unit data Format (CLIF) to enable multi-institutional federated critical illness research.

Background: Critical illness, or acute organ failure requiring life support, threatens over five million American lives annually. Electronic health record (EHR) data are a source of granular information that could generate crucial insights into the nature and optimal treatment of critical illness. However, data management, security, and standardization are barriers to large-scale critical illness EHR studies. Methods: A consortium of critical care physicians and data scientists from eight US healthcare systems developed the Common Longitudinal Intensive Care Unit (ICU) data Format (CLIF), an open-source database format that harmonizes a minimum set of ICU Data Elements for use in critical illness research. We created a pipeline to process adult ICU EHR data at each site. After development and iteration, we conducted two proof-of-concept studies with a federated research architecture: 1) an external validation of an in-hospital mortality prediction model for critically ill patients and 2) an assessment of 72-hour temperature trajectories and their association with mechanical ventilation and in-hospital mortality using group-based trajectory models. Results: We converted longitudinal data from 94,356 critically ill patients treated in 2020-2021 (mean age 60.6 years [standard deviation 17.2], 30% Black, 7% Hispanic, 45% female) across 8 health systems and 33 hospitals into the CLIF format, The in-hospital mortality prediction model performed well in the health system where it was derived (0.81 AUC, 0.06 Brier score). Performance across CLIF consortium sites varied (AUCs: 0.74-0.83, Brier scores: 0.06-0.01), and demonstrated some degradation in predictive capability. Temperature trajectories were similar across health systems. Hypothermic and hyperthermic-slow-resolver patients consistently had the highest mortality. Conclusions: CLIF facilitates efficient, rigorous, and reproducible critical care research. Our federated case studies showcase CLIF's potential for disease sub-phenotyping and clinical decision-support evaluation. Future applications include pragmatic EHR-based trials, target trial emulations, foundational multi-modal AI models of critical illness, and real-time critical care quality dashboards.

An exploratory survey about using ChatGPT in education, healthcare, and research.

OBJECTIVE: ChatGPT is the first large language model (LLM) to reach a large, mainstream audience. Its rapid adoption and exploration by the population at large has sparked a wide range of discussions regarding its acceptable and optimal integration in different areas. In a hybrid (virtual and in-person) panel discussion event, we examined various perspectives regarding the use of ChatGPT in education, research, and healthcare. MATERIALS AND METHODS: We surveyed in-person and online attendees using an audience interaction platform (Slido). We quantitatively analyzed received responses on questions about the use of ChatGPT in various contexts. We compared pairwise categorical groups with a Fisher's Exact. Furthermore, we used qualitative methods to analyze and code discussions. RESULTS: We received 420 responses from an estimated 844 participants (response rate 49.7%). Only 40% of the audience had tried ChatGPT. More trainees had tried ChatGPT compared with faculty. Those who had used ChatGPT were more interested in using it in a wider range of contexts going forwards. Of the three discussed contexts, the greatest uncertainty was shown about using ChatGPT in education. Pros and cons were raised during discussion for the use of this technology in education, research, and healthcare. DISCUSSION: There was a range of perspectives around the uses of ChatGPT in education, research, and healthcare, with still much uncertainty around its acceptability and optimal uses. There were different perspectives from respondents of different roles (trainee vs faculty vs staff). More discussion is needed to explore perceptions around the use of LLMs such as ChatGPT in vital sectors such as education, healthcare and research. Given involved risks and unforeseen challenges, taking a thoughtful and measured approach in adoption would reduce the likelihood of harm.

Outcomes of Percutaneous Tracheostomy for Patients With SARS-CoV-2 Respiratory Failure.

BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can cause severe respiratory failure leading to prolonged mechanical ventilation. Data are just emerging about the practice and outcomes of tracheostomy in these patients. We reviewed our experience with tracheostomies for SARS-CoV-2. METHODS: We retrospectively reviewed the demographics, comorbidities, timing of mechanical ventilation, tracheostomy, and intensive care unit and hospital lengths of stay in SARS-CoV-2 patients who received tracheostomies performed by the interventional pulmonary team. A tertiary care, teaching hospital in Chicago, Illinois. From March 2020 to April 2021, our center had 473 patients intubated for SARS-CoV-2, and 72 (15%) had percutaneous bedside tracheostomy performed by the interventional pulmonary team. RESULTS: Median time from intubation to tracheostomy was 20 (interquartile range: 16 to 25) days. Demographics and comorbidities were similar between early and late tracheostomy, but early tracheostomy was associated with shorter intensive care unit lengths of stay and a shorter total duration of ventilation. To date, 39 (54%) patients have been decannulated, 17 (24%) before hospital discharge; median time to decannulation was 22 (IQR: 18 to 36) days. Patients that were decannulated were younger (56 vs. 69 y). The rate of decannulation for survivors was 82%. No providers developed symptoms or tested positive for SARS-CoV-2. CONCLUSION: Tracheostomy enhances care for patients with prolonged respiratory failure from SARS-CoV-2 since early tracheostomy is associated with shorter duration of critical care, and decannulation rates are high for survivors. It furthermore appears safe for both patients and operators.

An exploratory survey about using ChatGPT in education, healthcare, and research.

Objective ChatGPT is the first large language model (LLM) to reach a large, mainstream audience. Its rapid adoption and exploration by the population at large has sparked a wide range of discussions regarding its acceptable and optimal integration in different areas. In a hybrid (virtual and in-person) panel discussion event, we examined various perspectives regarding the use of ChatGPT in education, research, and healthcare. Materials and Methods We surveyed in-person and online attendees using an audience interaction platform (Slido). We quantitatively analyzed received responses on questions about the use of ChatGPT in various contexts. We compared pairwise categorical groups with Fisher's Exact. Furthermore, we used qualitative methods to analyze and code discussions. Results We received 420 responses from an estimated 844 participants (response rate 49.7%). Only 40% of the audience had tried ChatGPT. More trainees had tried ChatGPT compared with faculty. Those who had used ChatGPT were more interested in using it in a wider range of contexts going forwards. Of the three discussed contexts, the greatest uncertainty was shown about using ChatGPT in education. Pros and cons were raised during discussion for the use of this technology in education, research, and healthcare. Discussion There was a range of perspectives around the uses of ChatGPT in education, research, and healthcare, with still much uncertainty around its acceptability and optimal uses. There were different perspectives from respondents of different roles (trainee vs faculty vs staff). More discussion is needed to explore perceptions around the use of LLMs such as ChatGPT in vital sectors such as education, healthcare and research. Given involved risks and unforeseen challenges, taking a thoughtful and measured approach in adoption would reduce the likelihood of harm.

Comparing scientific abstracts generated by ChatGPT to real abstracts with detectors and blinded human reviewers.

Large language models such as ChatGPT can produce increasingly realistic text, with unknown information on the accuracy and integrity of using these models in scientific writing. We gathered fifth research abstracts from five high-impact factor medical journals and asked ChatGPT to generate research abstracts based on their titles and journals. Most generated abstracts were detected using an AI output detector, 'GPT-2 Output Detector', with % 'fake' scores (higher meaning more likely to be generated) of median [interquartile range] of 99.98% 'fake' [12.73%, 99.98%] compared with median 0.02% [IQR 0.02%, 0.09%] for the original abstracts. The AUROC of the AI output detector was 0.94. Generated abstracts scored lower than original abstracts when run through a plagiarism detector website and iThenticate (higher scores meaning more matching text found). When given a mixture of original and general abstracts, blinded human reviewers correctly identified 68% of generated abstracts as being generated by ChatGPT, but incorrectly identified 14% of original abstracts as being generated. Reviewers indicated that it was surprisingly difficult to differentiate between the two, though abstracts they suspected were generated were vaguer and more formulaic. ChatGPT writes believable scientific abstracts, though with completely generated data. Depending on publisher-specific guidelines, AI output detectors may serve as an editorial tool to help maintain scientific standards. The boundaries of ethical and acceptable use of large language models to help scientific writing are still being discussed, and different journals and conferences are adopting varying policies.

An Adjudication Protocol for Severe Pneumonia.

Background: Clinical end points that constitute successful treatment in severe pneumonia are difficult to ascertain and vulnerable to bias. The utility of a protocolized adjudication procedure to determine meaningful end points in severe pneumonia has not been well described. Methods: This was a single-center prospective cohort study of patients with severe pneumonia admitted to the medical intensive care unit. The objective was to develop an adjudication protocol for severe bacterial and/or viral pneumonia. Each episode of pneumonia was independently reviewed by 2 pulmonary and critical care physicians. If a discrepancy occurred between the 2 adjudicators, a third adjudicator reviewed the case. If a discrepancy remained after all 3 adjudications, consensus was achieved through committee review. Results: Evaluation of 784 pneumonia episodes during 593 hospitalizations achieved only 48.1% interobserver agreement between the first 2 adjudicators and 78.8% when agreement was defined as concordance between 2 of 3 adjudicators. Multiple episodes of pneumonia and presence of bacterial/viral coinfection in the initial pneumonia episode were associated with lower interobserver agreement. For an initial episode of bacterial pneumonia, patients with an adjudicated day 7-8 clinical impression of cure (compared with alternative impressions) were more likely to be discharged alive (odds ratio, 6.3; 95% CI, 3.5-11.6). Conclusions: A comprehensive adjudication protocol to identify clinical end points in severe pneumonia resulted in only moderate interobserver agreement. An adjudicated end point of clinical cure by day 7-8 was associated with more favorable hospital discharge dispositions, suggesting that clinical cure by day 7-8 may be a valid end point to use in adjudication protocols.