Prediction of hospital mortality among critically ill patients in a single centre in Asia: comparison of artificial neural networks and logistic regression-based model.

INTRODUCTION: This study compared the performance of the artificial neural network (ANN) model with the Acute Physiologic and Chronic Health Evaluation (APACHE) II and IV models for predicting hospital mortality among critically ill patients in Hong Kong. METHODS: This retrospective analysis included all patients admitted to the intensive care unit of Pamela Youde Nethersole Eastern Hospital from January 2010 to December 2019. The ANN model was constructed using parameters identical to the APACHE IV model. Discrimination performance was assessed using area under the receiver operating characteristic curve (AUROC); calibration performance was evaluated using the Brier score and Hosmer-Lemeshow statistic. RESULTS: In total, 14 503 patients were included, with 10% in the validation set and 90% in the ANN model development set. The ANN model (AUROC=0.88, 95% confidence interval [CI]=0.86-0.90, Brier score=0.10; P in Hosmer-Lemeshow test=0.37) outperformed the APACHE II model (AUROC=0.85, 95% CI=0.80-0.85, Brier score=0.14; P<0.001 for both comparisons of AUROCs and Brier scores) but showed performance similar to the APACHE IV model (AUROC=0.87, 95% CI=0.85-0.89, Brier score=0.11; P=0.34 for comparison of AUROCs, and P=0.05 for comparison of Brier scores). The ANN model demonstrated better calibration than the APACHE II and APACHE IV models. CONCLUSION: Our ANN model outperformed the APACHE II model but was similar to the APACHE IV model in terms of predicting hospital mortality in Hong Kong. Artificial neural networks are valuable tools that can enhance real-time prognostic prediction.

Ten-year territory-wide trends in the utilisation and clinical outcomes of extracorporeal membrane oxygenation in Hong Kong.

INTRODUCTION: The utilisation of extracorporeal membrane oxygenation (ECMO) has been rapidly increasing in Hong Kong. This study examined 10-year trends in the utilisation and clinical outcomes of ECMO in Hong Kong. METHODS: We retrospectively reviewed the records of all adult patients receiving ECMO who were admitted to the intensive care units (ICUs) of public hospitals in Hong Kong between 2010 and 2019. Temporal trends across years were assessed using the Mann-Kendall test. Observed hospital mortality was compared with the Acute Physiology and Chronic Health Evaluation (APACHE) IV-predicted mortality. RESULTS: The annual number of patients receiving ECMO increased from 18 to 171 over 10 years. In total, 911 patients received ECMO during the study period: 297 (32.6%) received veno-arterial ECMO, 450 (49.4%) received veno-venous ECMO, and 164 (18.0%) received extracorporeal cardiopulmonary resuscitation. The annual number of patients aged ≥65 years increased from 0 to 47 (27.5%) [P for trend=0.001]. The median (interquartile range) Charlson Comorbidity Index increased from 1 (0-1) to 2 (1-3) [P for trend<0.001] while the median (interquartile range) APACHE IV score increased from 90 (57-112) to 105 (77-137) [P for trend=0.003]. The overall standardised mortality ratio comparing hospital mortality with APACHE IV-predicted mortality was 1.11 (95% confidence interval=1.01-1.22). Hospital and ICU length of stay both significantly decreased (P for trend=0.011 and <0.001, respectively). CONCLUSION: As ECMO utilisation increased in Hong Kong, patients put on ECMO were older, more critically ill, and had more co-morbidities. It is important to combine service expansion with adequate resource allocation and training to maintain quality of care.

Admission triage tool for adult intensive care unit admission in Hong Kong during the COVID-19 outbreak.

Intensive care is expensive, and the numbers of intensive care unit (ICU) beds and trained specialist medical staff able to provide services in Hong Kong are limited. The most recent increase in coronavirus disease 2019 (COVID-19) infections over July to August 2020 resulted in more than 100 new cases per day for a prolonged period. The increased numbers of critically ill patients requiring ICU admission posed a capacity challenge to ICUs across the territory, and it may be reasonably anticipated that should a substantially larger outbreak occur, ICU services will be overwhelmed. Therefore, a transparent and fair prioritisation process for decisions regarding patient ICU admission is urgently required. This triage tool is built on the foundation of the existing guidelines and framework for admission, discharge, and triage that inform routine clinical practice in Hospital Authority ICUs, with the aim of achieving the greatest benefit for the greatest number of patients from the available ICU resources. This COVID-19 Crisis Triage Tool is expected to provide structured guidance to frontline doctors on how to make triage decisions should ICU resources become overwhelmed by patients requiring ICU care, particularly during the current COVID-19 pandemic. The triage tool takes the form of a detailed decision aid algorithm based on a combination of established prognostic scores, and it should increase objectivity and transparency in triage decision making and enhance decision-making consistency between doctors within and across ICUs in Hong Kong. However, it remains an aid rather than a complete substitute for the carefully considered judgement of an experienced intensive care clinician.

Tetramine poisoning.

We report on two patients who presented with status epilepticus due to ingestion of rat poison containing tetramine. Both had eaten the same meal, subsequently presumed to be the source of the poison. Physical examination and investigation were unremarkable and diagnosis was based on patient history. Seizures were ultimately controlled with ketamine, after unsuccessful attempt of benzodiazepine and sodium thiopentone. One week after poisoning, both patients underwent one session of high-volume haemofiltration followed by charcoal haemoperfusion to eliminate the toxin from the body. Plasma tetramine levels then decreased from 0.95 microgram/mL to 0.35 microgram/mL and from 0.53 microgram/mL to 0.40 microgram/mL, respectively.