Impact of major bleeding and thrombosis on 180-day survival in patients with severe COVID-19 supported with veno-venous extracorporeal membrane oxygenation in the United Kingdom: a multicentre observational study.

Bleeding and thrombosis are major complications in patients supported with extracorporeal membrane oxygenation (ECMO). In this multicentre observational study of 152 consecutive patients (≥18 years) with severe COVID-19 supported by veno-venous (VV) ECMO in four UK commissioned centres during the first wave of the COVID-19 pandemic (1 March to 31 May 2020), we assessed the incidence of major bleeding and thrombosis and their association with 180-day mortality. Median age (range) was 47 years (23-65) and 75% were male. Overall, the 180-day survival was 70·4% (107/152). The rate of major bleeding was 30·9% (47/152), of which intracranial bleeding (ICH) was 34% (16/47). There were 96 thrombotic events (63·1%) consisting of venous 44·7% [68/152 of which 66·2% were pulmonary embolism (PE)], arterial 18·6% (13/152) and ECMO circuit thrombosis 9·9% (15/152). In multivariate analysis, only raised lactate dehydrogenase (LDH) at the initiation of VV ECMO was associated with an increased risk of thrombosis [hazard ratio (HR) 1·92, 95% CI 1·21-3·03]. Major bleeding and ICH were associated with 3·87-fold (95% CI 2·10-7·23) and 5·97-fold [95% confidence interval (CI) 2·36-15·04] increased risk of mortality and PE with a 2·00-fold (95% CI1·09-3·56) risk of mortality. This highlights the difficult balancing act often encountered when managing coagulopathy in COVID-19 patients supported with ECMO.

A pilot feasibility study comparing large language models in extracting key information from ICU patient text records from an Irish population.

BACKGROUND: Artificial intelligence, through improved data management and automated summarisation, has the potential to enhance intensive care unit (ICU) care. Large language models (LLMs) can interrogate and summarise large volumes of medical notes to create succinct discharge summaries. In this study, we aim to investigate the potential of LLMs to accurately and concisely synthesise ICU discharge summaries. METHODS: Anonymised clinical notes from ICU admissions were used to train and validate a prompting structure in three separate LLMs (ChatGPT, GPT-4 API and Llama 2) to generate concise clinical summaries. Summaries were adjudicated by staff intensivists on ability to identify and appropriately order a pre-defined list of important clinical events as well as readability, organisation, succinctness, and overall rank. RESULTS: In the development phase, text from five ICU episodes was used to develop a series of prompts to best capture clinical summaries. In the testing phase, a summary produced by each LLM from an additional six ICU episodes was utilised for evaluation. Overall ability to identify a pre-defined list of important clinical events in the summary was 41.5 ± 15.2% for GPT-4 API, 19.2 ± 20.9% for ChatGPT and 16.5 ± 14.1% for Llama2 (p = 0.002). GPT-4 API followed by ChatGPT had the highest score to appropriately order a pre-defined list of important clinical events in the summary as well as readability, organisation, succinctness, and overall rank, whilst Llama2 scored lowest for all. GPT-4 API produced minor hallucinations, which were not present in the other models. CONCLUSION: Differences exist in large language model performance in readability, organisation, succinctness, and sequencing of clinical events compared to others. All encountered issues with narrative coherence and omitted key clinical data and only moderately captured all clinically meaningful data in the correct order. However, these technologies suggest future potential for creating succinct discharge summaries.

Thombosis, major bleeding, and survival in COVID-19 supported by veno-venous extracorporeal membrane oxygenation in the first vs second wave: a multicenter observational study in the United Kingdom.

BACKGROUND: Bleeding and thrombosis are major complications of veno-venous (VV) extracorporeal membrane oxygenation (ECMO). OBJECTIVES: To assess thrombosis, major bleeding (MB), and 180-day survival in patients supported by VV-ECMO between the first (March 1 to May 31, 2020) and second (June 1, 2020, to June 30, 2021) waves of the COVID-19 pandemic. METHODS: An observational study of 309 consecutive patients (aged ≥18years) with severe COVID-19 supported by VV-ECMO was performed in 4 nationally commissioned ECMO centers in the United Kingdom. RESULTS: Median age was 48 (19-75) years, and 70.6% were male. Probabilities of survival, thrombosis, and MB at 180 days in the overall cohort were 62.5% (193/309), 39.8% (123/309), and 30% (93/309), respectively. In multivariate analysis, an age of >55 years (hazard ratio [HR], 2.29; 95% CI, 1.33-3.93; P = .003) and an elevated creatinine level (HR, 1.91; 95% CI, 1.19-3.08; P = .008) were associated with increased mortality. Correction for duration of VV-ECMO support, arterial thrombosis alone (HR, 3.0; 95% CI, 1.5-5.9; P = .002) or circuit thrombosis alone (HR, 3.9; 95% CI, 2.4-6.3; P < .001) but not venous thrombosis increased mortality. MB during ECMO had a 3-fold risk (95% CI, 2.6-5.8, P < .001) of mortality. The first wave cohort had more males (76.7% vs 64%; P = .014), higher 180-day survival (71.1% vs 53.3%; P = .003), more venous thrombosis alone (46.4% vs 29.2%; P = .02), and lower circuit thrombosis (9.2% vs 28.1%; P < .001). The second wave cohort received more steroids (121/150 [80.6%] vs 86/159 [54.1%]; P < .0001) and tocilizumab (20/150 [13.3%] vs 4/159 [2.5%]; P = .005). CONCLUSION: MB and thrombosis are frequent complications in patients on VV-ECMO and significantly increase mortality. Arterial thrombosis alone or circuit thrombosis alone increased mortality, while venous thrombosis alone had no effect. MB during ECMO support increased mortality by 3.9-fold.

Scaling up a National Extracorporeal Membrane Oxygenation Referral Service for adult patients in acute severe respiratory failure at the time of a pandemic.

The National ECMO Service for patients in acute severe respiratory failure in England responded to the challenge of the coronavirus pandemic by implementing a central electronic referral system within days. Prior to this, each ECMO centre managed independently around 20 ECMO referrals per month. Early during the pandemic, we recognised the need for a referral system to co-ordinate the anticipated increased number of referrals. We implemented rapidly a referral system with universal access across England. This allowed the five National ECMO centres to manage over 1000 referrals in the first seven weeks of the pandemic. Key benefits of the new system included facilitated communication and collaboration between centres; data on demand; and capacity shared in real-time. We believe this was instrumental in allowing us to continue to provide for the whole country, respond at scale, and facilitate our collaborative work as a multidisciplinary team.