Search

Background: Large language models (LLMs) have demonstrated impressive performances in various medical domains, prompting an exploration of their potential utility within the high-demand setting of emergency department (ED) triage. This study evaluated the triage proficiency of different LLMs and ChatGPT, an LLM-based chatbot, compared to professionally trained ED staff and untrained personnel. We further explored whether LLM responses could guide untrained staff in effective triage., Objective: This study aimed to assess the efficacy of LLMs and the associated product ChatGPT in ED triage compared to personnel of varying training status and to investigate if the models' responses can enhance the triage proficiency of untrained personnel., Methods: A total of 124 anonymized case vignettes were triaged by untrained doctors; different versions of currently available LLMs; ChatGPT; and professionally trained raters, who subsequently agreed on a consensus set according to the Manchester Triage System (MTS). The prototypical vignettes were adapted from cases at a tertiary ED in Germany. The main outcome was the level of agreement between raters' MTS level assignments, measured via quadratic-weighted Cohen κ. The extent of over- and undertriage was also determined. Notably, instances of ChatGPT were prompted using zero-shot approaches without extensive background information on the MTS. The tested LLMs included raw GPT-4, Llama 3 70B, Gemini 1.5, and Mixtral 8x7b., Results: GPT-4-based ChatGPT and untrained doctors showed substantial agreement with the consensus triage of professional raters (κ=mean 0.67, SD 0.037 and κ=mean 0.68, SD 0.056, respectively), significantly exceeding the performance of GPT-3.5-based ChatGPT (κ=mean 0.54, SD 0.024; P<.001). When untrained doctors used this LLM for second-opinion triage, there was a slight but statistically insignificant performance increase (κ=mean 0.70, SD 0.047; P=.97). Other tested LLMs performed similar to or worse than GPT-4-based ChatGPT or showed odd triaging behavior with the used parameters. LLMs and ChatGPT models tended toward overtriage, whereas untrained doctors undertriaged., Conclusions: While LLMs and the LLM-based product ChatGPT do not yet match professionally trained raters, their best models' triage proficiency equals that of untrained ED doctors. In its current form, LLMs or ChatGPT thus did not demonstrate gold-standard performance in ED triage and, in the setting of this study, failed to significantly improve untrained doctors' triage when used as decision support. Notable performance enhancements in newer LLM versions over older ones hint at future improvements with further technological development and specific training., (©Lars Masanneck, Linea Schmidt, Antonia Seifert, Tristan Kölsche, Niklas Huntemann, Robin Jansen, Mohammed Mehsin, Michael Bernhard, Sven G Meuth, Lennert Böhm, Marc Pawlitzki. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 14.06.2024.)

Background: Currently, the data regarding the impact of prehospital postcardiac arrest anesthesia on target hemodynamic and ventilatory parameters of early postresuscitation care and recommendations on its implementation are rare. The present study examines the incidence and impact of prehospital postcardiac arrest anesthesia on hemodynamic and ventilatory target parameters of postresuscitation care., Methods: In this multicentre observational study between 2019 and 2021 unconscious adult patients after out-of-hospital-cardiac arrest with the presence of a return-of-spontaneous circulation until hospital admission were included. Primary endpoint was the application of postarrest anesthesia. Secondary endpoints included the medication group used, predisposing factors to its implementation, and its influence on achieving target parameters of postresuscitation care (systolic blood pressure: ≥ 100 mmHg, etCO 2 :35-45 mmHg, SpO 2 : 94-98%) at hospital handover., Results: During the study period 2,335 out-of-hospital resuscitations out of 391,305 prehospital emergency operations (incidence: 0.58%; 95% CI 0.54-0.63) were observed with a return of spontaneous circulation to hospital admission in 706 patients (30.7%; 95% CI 28.8-32.6; female: 34.3%; age:68.3 ± 14.2 years). Postcardiac arrest anesthesia was performed in 482 patients (68.3%; 95% CI 64.7-71.7) with application of hypnotics in 93.4% (n = 451), analgesics in 53.7% (n = 259) and relaxants in 45.6% (n = 220). Factors influencing postcardiac arrest sedation were emergency care by an anesthetist (odds ratio: 2.10; 95% CI 1.34-3.30; P < 0.001) and treatment-free interval ≤ 5 min (odds ratio: 1.59; 95% CI 1.01-2.49; P = 0.04). Although there was no evidence of the impact of performing postcardiac arrest anesthesia on achieving a systolic blood pressure ≥ 100 mmHg at the end of operation (odds ratio: 1.14; 95% CI 0.78-1.68; P = 0.48), patients with postcardiac arrest anesthesia were significantly more likely to achieve the recommended ventilation (odds ratio: 1.59; 95% CI 1.06-2.40; P = 0.02) and oxygenation (odds ratio:1.56; 95% CI 1.04-2.35; P = 0.03) targets. Comparing the substance groups, the use of hypnotics significantly more often enabled the target values for etCO2 to be reached alone (odds ratio:2.79; 95% CI 1.04-7.50; P = 0.04) as well as in combination with a systolic blood pressure ≥ 100 mmHg (odds ratio:4.42; 95% CI 1.03-19.01; P = 0.04)., Conclusions: Postcardiac arrest anesthesia in out-of-hospital cardiac arrest is associated with early achievement of respiratory target parameters in prehospital postresuscitation care without evidence of more frequent hemodynamic complications., (© 2024. The Author(s).)

Background: An out-of-hospital cardiac arrest (OHCA) with return of spontaneous circulation (ROSC) may need to be treated with airway management, emergency ventilation, invasive interventions, and post-arrest sedation. We investigated the influence of the use of midazolam for post-arrest sedation on achieving postresuscitation care targets and the associated risk of hemodynamic complications., Methods: All emergency rescue missions of the Dresden, Gütersloh, and Lippe medical rescue services in the years 2019-2021 were reviewed to identify adult patients who had OHCA, unconsciousness, and sustained ROSC with spontaneous circulation until arrival at the hospital; the findings were supplemented with data from the German Resuscitation Registry. Patients who received midazolam (alone or in combination with other anesthetic agents) for post-arrest sedation were compared with those who did not. The endpoints were the regaining of a systolic blood pressure ≥ 100 mmHg, end-tidal pCO2 35-45 mmHg, and oxygen saturation (SpO2) 94-98%. A propensity score analysis was used to adjust for age, sex, and variables potentially affecting hemodynamic status or the targets for oxygenation and ventilation., Results: There were 2335 cases of OHCA among 391 305 emer - gency rescue missions. 571 patients had ROSC before arrival in the hospital (24.5%; female, 33.6%; age, 68 ± 14 years). Of the 395 among them (69.2%) who were treated with postarrest sedation, 249 (63.0%) received midazolam. Patients who received midazolam reached the guideline- recommended targets for oxygenation, ventilation, and blood pressure more frequently than those who were not sedated: the respective odds ratios and 95% confidence intervals were 2.00 [1.20; 3.34], 1.57 [0.99; 2.48], and 1.41 [0.89; 2.21]., Conclusion: The pre-hospital administration of midazolam leads to more frequent pre-hospital attainment of the oxygenation and ventilation targets in post-resuscitation care, without any evidence of an elevated risk of hemodynamic complications.