Editor’s Note: Find the latest COVID-19 news and guidance from Medscape Coronavirus Resource Center..
Combining clinical observations on admission, patient demographics, and blood measurements into simple scores predicted possible in-hospital mortality in more than 57,000 predictive observation cohorts of COVID-19.
Dr. Karam Semple
“We were surprised that eight out-of-the-box features and simple mathematical calculations were sufficient to accurately predict mortality risk,” said study author Dr. Calum Semple. Medscape Medical News.
Mortality scores for the Coronavirus Clinical Characterization Consortium (4C) outperformed other clinical predictors, including more complex machine learning calculations. “Comparing this tool to existing tools, we found that they were superior,” said Semple, a professor of child health and developmental medicine at the University of Liverpool, UK.
Research Published online September 9 BMJ..
Triage guidance
The scoring system classifies patients as having a low, medium, high, or very high risk of death based on a total score from 0 to 21. The higher the number, the higher the risk.
Researchers point out that people in low-risk groups may potentially be community-managed. Patients in the intermediate group may be monitored in the ward, while those at high risk of death may be triaged for prompt and aggressive treatment. Patients at high risk may be treated with steroids and transferred to, for example, emergency care.
They called for producing clinically relevant risk stratification scores due to a lack of validated clinical tools to predict mortality risk for people hospitalized with COVID-19, they say.
They developed the tool using data from 35,463 adults admitted to one of 260 British hospitals between February 6 and May 20, 2020. The overall mortality rate was 32%.
Age, gender, number of comorbidities, respiratory rate, peripheral oxygen saturation, Glasgow Coma Scale Score, urea Levels and C-reactive protein concentration are eight factors included in mortality prediction.
Interactive scoring system Available online According to Semple, it’s intended for use by clinicians in British hospitals. “We welcome cooperation with researchers in other countries to test the use of the tool.”
table. 4C Mortality Score Results
| Score range | Percentage of research population | Mortality | |
|---|---|---|---|
| Low | 0-3 | 7.4% | 1.2% |
| Intermediate | 4-8 | 21.9% | 9.9% |
| high | 9-14 | 52.2% | 31.4% |
| Very high | 15+ | 18.6% | 61.5% |
The researchers examined the scores of another 22,361 adults admitted for COVID-19, assessed after May 20. This step “has shown that this tool can guide clinician decisions, including treatment escalation,” they explain.
Semple and colleagues validated the scores against 15 other risk stratification techniques identified in a systematic literature search. “The 4C mortality score was often compared to these existing risk stratification scores in predicting in-hospital mortality,” the researchers write.
Sensitivity analysis showed that the score remained valid across ethnic groups and different geographical cohorts.
The researchers warn that the scoring system is not designed for community use, and may result differently in populations at low risk of death. Further validation is needed to determine if the score applies to young people with COVID-19 and countries outside the UK.
“Practical risk score”
“This study develops and validates a practical risk score for predicting mortality in patients admitted with COVID-19. A mortality prediction tool that accurately stratifies patients into low states, not associated with a medium, high, or very high mortality risk,” said Tim Q. Duong. Medscape Medical News When asked for a comment.
Dr. Tim Duong
The large cohort of patients with COVID-19, the ready availability of clinical variables, and the fact that 4C mortality scores exceeded or were comparable to other risk stratification tools were strengths of the study. Duong, director of MRI research at the Renaissance School, said. Medicine, Stony Brook University, Stony Brook, New York.
Duong was a senior author on July 16, 2020. Investigation This assessed the role of deep learning artificial intelligence in predicting COVID-19 mortality.
“Patients are admitted to hospital at various stages of disease severity, but it can be difficult to predict mortality far downstream, a potential limitation of current studies,” Duong said. In addition, mortality rates may depend on patient load at each hospital, available resources, and treatment plans.
“Still, this research laid the foundation for future prospective research,” he said.
Semple received a grant from the Department of Health and Human Services at the National Institutes of Health Research, and a grant from the University of Liverpool’s UK Medical Research Council for Emerging and Zoonotic Diseases and Health Conservation Research Unit. Duong does not disclose the financial relationship involved.
BMJ.. Published online September 9, 2020. Full text
Follow Damian McNamara on Twitter. @MedReporter.. For news on Medscape Neurology, Facebook And twitter..
Follow Medscape for more information. Facebook, twitter, Instagram,and YouTube..
