Assessing ChatGPT's ability to emulate human reviewers in scientific research: A descriptive and qualitative approach.

BACKGROUND: ChatGPT is an AI platform whose relevance in the peer review of scientific articles is steadily growing. Nonetheless, it has sparked debates over its potential biases and inaccuracies. This study aims to assess ChatGPT's ability to qualitatively emulate human reviewers in scientific research. METHODS: We included the first submitted version of the latest twenty original research articles published by the 3rd of July 2023, in a high-profile medical journal. Each article underwent evaluation by a minimum of three human reviewers during the initial review stage. Subsequently, three researchers with medical backgrounds and expertise in manuscript revision, independently and qualitatively assessed the agreement between the peer reviews generated by ChatGPT version GPT-4 and the comments provided by human reviewers for these articles. The level of agreement was categorized into complete, partial, none, or contradictory. RESULTS: 720 human reviewers' comments were assessed. There was a good agreement between the three assessors (Overall kappa >0.6). ChatGPT's comments demonstrated complete agreement in terms of quality and substance with 48 (6.7 %) human reviewers' comments, partially agreed with 92 (12.8 %), identifying issues necessitating further elaboration or recommending supplementary steps to address concerns, had no agreement with a significant 565 (78.5 %), and contradicted 15 (2.1 %). ChatGPT comments on methods had the lowest proportion of complete agreement (13 comments, 3.6 %), while general comments on the manuscript displayed the highest proportion of complete agreement (17 comments, 22.1 %). CONCLUSION: ChatGPT version GPT-4 has a limited ability to emulate human reviewers within the peer review process of scientific research.

Association of anaesthesia provider sex with perioperative complications: a two-centre retrospective cohort study.

BACKGROUND: Previous studies suggested that surgeon sex is associated with differential patient outcomes. Whether this also applies to anaesthesia providers is unclear. We hypothesised that female sex of the primary anaesthesia provider is associated with lower risk of perioperative complications. METHODS: The first case for all adult patients undergoing anaesthesia care between 2008 and 2022 at two academic healthcare networks in the USA was included in this retrospective cohort study. The primary exposure was the sex of the anaesthesia provider who spent the most time in the operating theatre during the case. The primary outcome was intraoperative complications, defined as hypotension (mean arterial blood pressure <55 mm Hg for ≥5 cumulative minutes) or hypoxaemia (oxygen saturation <90% for >2 consecutive minutes). The co-primary outcome was 30-day adverse postoperative events (including complications, readmission, and mortality). Analyses were adjusted for a priori defined confounders. RESULTS: Among 364,429 included patients, 57,550 (15.8%) experienced intraoperative complications and 55,168 (15.1%) experienced adverse postoperative events. Care by female compared with male anaesthesia providers was associated with lower risk of intraoperative complications (adjusted odds ratio [aOR] 0.95, 95% confidence interval [CI] 0.94-0.97, P<0.001), which was magnified among non-trainees (aOR 0.84, 95% CI 0.82-0.87, P-for-interaction <0.001). Anaesthesia provider sex was not associated with the composite of adverse postoperative events (aOR 1.00, 95% CI 0.98-1.02, P=0.88). CONCLUSIONS: Care by a female anaesthesia provider was associated with a lower risk of intraoperative complications, which was magnified among non-trainees. Future studies should investigate underlying mechanisms.

Increased Postoperative Opioid Consumption in the Presence of Co-administration of 5-HT3 Antagonists with Acetaminophen: A Hospital Registry Study.

BACKGROUND: Acetaminophen and 5-hydroxytryptamine-type-3 (5-HT3) receptor antagonists are administered as standard prophylaxes for postoperative pain, nausea, and vomiting. Preclinical studies however suggest that 5-HT3 antagonists may compromise acetaminophen's analgesic effect. This hospital registry study investigates whether 5-HT3 antagonists mitigate the analgesic effect of prophylactic acetaminophen in a perioperative setting. METHODS: This study included 55,016 adult patients undergoing general anesthesia for ambulatory procedures at a tertiary healthcare center in Massachusetts, United States of America, from 2015 to 2022. Using binary exposure variables and a comprehensive selection of pre-planned patient- and procedure-related covariates for confounder control, we investigated whether intraoperative 5-HT3 antagonists affected the association between pre- or intraoperative acetaminophen and postoperative opioid consumption, gauged by opioid dose in mg oral morphine equivalents (OME) administered in the post-anesthesia care unit (PACU). A multivariable, zero-inflated negative binomial regression model was applied. RESULTS: 3,166 (5.8%) patients received only acetaminophen, 15,438 (28.1%) only 5-HT3 antagonists, 31,850 (57.9%) both drugs, and 4,562 (8.3%) neither drug. The median PACU opioid dose was 7.5 mg OME (interquartile range 7.5 to 14.3 mg OME) among 16,640/55,016 (30.3%) patients who received opioids and the average opioid dose was 3.2 mg OME across all patients (maximum cumulative dose: 20.4 mg OME). Acetaminophen administration was associated with a 5.5% (95%CI -9.6% to -1.4%;p=0.009; adjusted absolute difference -0.19 mg OME;95%CI -0.33 to -0.05;p=0.009) reduction in opioid consumption among patients who did not receive a 5-HT3 antagonist, while there was no effect in patients who received a 5-HT3 antagonist (adjusted absolute difference 0.00 mg OME; 95%CI -0.06 to 0.05;p=0.93,p-for-interaction=0.012). CONCLUSION: A dose-dependent association of pre- or intraoperative acetaminophen with decreased postoperative opioid consumption was not observed when 5-HT3 antagonists were co-administered, suggesting that physicians might consider reserving 5-HT3 antagonists as rescue medication for postoperative nausea or vomiting when acetaminophen is administered for pain prophylaxis.

Intraoperative Opioid Waste and Association of Intraoperative Opioid Dose with Postoperative Adverse Outcomes: A Hospital Registry Study.

INTRODUCTION: Perioperative opioid use has been associated with adverse clinical outcomes. Additionally, opioid disposal carries significant costs, due to the waste of pharmaceutical products and the time needed by skilled labor to report the waste. In this study, we aimed to estimate costs and predict factors of opioid-associated intraoperative product waste, as well as to evaluate whether higher intraoperative opioid doses are associated with increased risk of adverse postoperative outcomes. METHODS: We included 170,607 patients undergoing general anesthesia and receiving intraoperative fentanyl, hydromorphone, or morphine at Beth Israel Deaconess Medical Center, Boston, MA, USA, between January 2010 and June 2020. We estimated product waste-associated costs based on various opioid syringe sizes and determined predictors of opioid waste. Further, we evaluated whether higher opioid doses were associated with postoperative adverse events according to the severity-indexed, incident report-based medication error-reporting program classification. The primary outcome included post-extubation desaturation, postoperative nausea or vomiting, or postoperative somnolence or sedation. RESULTS: The use of the smallest syringe sizes (50 mcg for fentanyl, 0.2 mg for hydromorphone, and 2 mg for morphine) resulted in the lowest product waste-associated costs. The main predictor of opioid waste was the administration of more than one intraoperative opioid (adjusted odds ratio [aOR] = 7.64, 95% CI 7.40-7.89, P < 0.001). Intraoperative doses of fentanyl > 50-100 mcg (aOR = 1.17 [1.10-1.25], P < 0.001, adjusted risk difference [ARD] 2%) and > 100 mcg (aOR = 1.24 [1.16-1.33], P < 0.001, ARD 3%), hydromorphone > 1 mg (aOR = 1.13 [1.06-1.20], P < 0.001, ARD 2%), and morphine > 2-4 mg (aOR = 1.26 [1.02-1.56], P = 0.04, ARD 3%) and > 4 mg (aOR = 1.45 [1.18-1.77], P < 0.001, ARD 5%) were associated with higher risk of the primary outcome. CONCLUSION: Smaller syringe sizes of intraoperative opioids may help to reduce product waste and associated costs, as well postoperative adverse events through utilization of lower intraoperative opioid doses.

Adjustments of Ventilator Parameters during Operating Room-to-ICU Transition and 28-Day Mortality.

Rationale: Lung-protective mechanical ventilation strategies have been proven beneficial in the operating room (OR) and the ICU. However, differential practices in ventilator management persist, often resulting in adjustments of ventilator parameters when transitioning patients from the OR to the ICU. Objectives: To characterize patterns of ventilator adjustments during the transition of mechanically ventilated surgical patients from the OR to the ICU and assess their impact on 28-day mortality. Methods: Hospital registry study including patients undergoing general anesthesia with continued, controlled mechanical ventilation in the ICU between 2008 and 2022. Ventilator parameters were assessed 1 hour before and 6 hours after the transition. Measurements and Main Results: Of 2,103 patients, 212 (10.1%) died within 28 days. Upon OR-to-ICU transition, VT and driving pressure decreased (-1.1 ml/kg predicted body weight [IQR, -2.0 to -0.2]; P < 0.001; and -4.3 cm H2O [-8.2 to -1.2]; P < 0.001). Concomitantly, respiratory rates increased (+5.0 breaths/min [2.0 to 7.5]; P < 0.001), resulting overall in slightly higher mechanical power (MP) in the ICU (+0.7 J/min [-1.9 to 3.0]; P < 0.001). In adjusted analysis, increases in MP were associated with a higher 28-day mortality rate (adjusted odds ratio, 1.10; 95% confidence interval, 1.06-1.14; P < 0.001; adjusted risk difference, 0.7%; 95% confidence interval, 0.4-1.0, both per 1 J/min). Conclusion: During transition of mechanically ventilated patients from the OR to the ICU, ventilator adjustments resulting in higher MP were associated with a greater risk of 28-day mortality.

Affiliation Bias in Peer Review of Abstracts by a Large Language Model.

This study assesses affiliation bias in peer review of medical abstracts by a commonly used large language model.

Inflammatory subphenotypes in patients at risk of ARDS: evidence from the LIPS-A trial.

PURPOSE: Latent class analysis (LCA) has identified hyper- and non-hyper-inflammatory subphenotypes in patients with acute respiratory distress syndrome (ARDS). It is unknown how early inflammatory subphenotypes can be identified in patients at risk of ARDS. We aimed to test for inflammatory subphenotypes upon presentation to the emergency department. METHODS: LIPS-A was a trial of aspirin to prevent ARDS in at-risk patients presenting to the emergency department. In this secondary analysis, we performed LCA using clinical, blood test, and biomarker variables. RESULTS: Among 376 (96.4%) patients from the LIPS-A trial, two classes were identified upon presentation to the emergency department (day 0): 72 (19.1%) patients demonstrated characteristics of a hyper-inflammatory and 304 (80.9%) of a non-hyper-inflammatory subphenotype. 15.3% of patients in the hyper- and 8.2% in the non-hyper-inflammatory class developed ARDS (p = 0.07). Patients in the hyper-inflammatory class had fewer ventilator-free days (median [interquartile range, IQR] 28[23-28] versus 28[27-28]; p = 0.010), longer intensive care unit (3[2-6] versus 0[0-3] days; p < 0.001) and hospital (9[6-18] versus 5[3-9] days; p < 0.001) length of stay, and higher 1-year mortality (34.7% versus 20%; p = 0.008). Subphenotypes were identified on day 1 and 4 in a subgroup with available data (n = 244). 77.9% of patients remained in their baseline class throughout day 4. Patients with a hyper-inflammatory subphenotype throughout the study period (n = 22) were at higher risk of ARDS (36.4% versus 10.4%; p = 0.003). CONCLUSION: Hyper- and non-hyper-inflammatory subphenotypes may precede ARDS development, remain identifiable over time, and can be identified upon presentation to the emergency department. A hyper-inflammatory subphenotype predicts worse outcomes.

Endovascular Perforation Model for Subarachnoid Hemorrhage Combined with Magnetic Resonance Imaging (MRI).

The endovascular filament perforation model to mimic subarachnoid hemorrhage (SAH) is a commonly used model - however, the technique can cause a high mortality rate as well as an uncontrollable volume of SAH and other intracranial complications such as stroke or intracranial hemorrhage. In this protocol, a standardized SAH mouse model is presented, induced by endovascular filament perforation, combined with magnetic resonance imaging (MRI) 24 h after operation to ensure the correct bleeding site and exclude other relevant intracranial pathologies. Briefly, C57BL/6J mice are anesthetized with an intraperitoneal ketamine/xylazine (70 mg/16 mg/kg body weight) injection and placed in a supine position. After midline neck incision, the common carotid artery (CCA) and carotid bifurcation are exposed, and a 5-0 non-absorbable monofilament polypropylene suture is inserted in a retrograde fashion into the external carotid artery (ECA) and advanced into the common carotid artery. Then, the filament is invaginated into the internal carotid artery (ICA) and pushed forward to perforate the anterior cerebral artery (ACA). After recovery from surgery, mice undergo a 7.0 T MRI 24 h later. The volume of bleeding can be quantified and graded via postoperative MRI, enabling a robust experimental SAH group with the option to perform further subgroup analyses based on blood quantity.