Risk Score Predicts COVID-19 Mortality, Outperforms Others

Damian McNamara

September 10, 2020

Editor's note: Find the latest COVID-19 news and guidance in Medscape's Coronavirus Resource Center.

Clinical observations, patient demographics, and blood measurements at hospital admission combined in a brief score predicted the likelihood of in-hospital mortality in a prospective, observational cohort of more than 57,000 individuals with COVID-19.

Dr Calum Semple

"We were surprised that eight readily available features and a simple math calculation were sufficient to accurately predict risk of death," study author Calum Semple, PhD, told Medscape Medical News.

The Coronavirus Clinical Characterization Consortium (4C) Mortality Score outperformed other clinical prediction measures as well, including more complex machine learning calculations. "When the tool is compared to existing tools, it was found to outperform them," said Semple, professor of child health and outbreak medicine at the University of Liverpool in the United Kingdom.

The study was published online September 9 in The BMJ.

Triage Guidance

The scoring system classifies patients as having low, intermediate, high, or very high likelihood of death on the basis of a total score from 0 to 21, with higher numbers reflecting greater risk.

People in the low-risk group could potentially be managed in the community, the researchers note. Those in the intermediate group might be monitored on a hospital ward, whereas patients with a high risk for death could be triaged to prompt, aggressive treatment. High-risk patients might receive steroid treatment and be transferred to critical care, for example.

A lack of validated clinical tools to predict mortality risk among people hospitalized with COVID-19 prompted Semple and colleagues to develop a clinically relevant risk stratification score, they say.

They developed the tool using data from 35,463 adults admitted to any of 260 UK hospitals between February 6 and May 20, 2020. The median age of the patients was 73 years, 42% were women, and 76% had at least one comorbidity. The overall mortality rate was 32%.

Age, sex, number of comorbidities, respiratory rate, peripheral oxygen saturation, Glasgow coma scale score, urea level, and C reactive protein concentration are the eight factors included in the mortality predictor.

The interactive scoring system is available online for use by clinicians in UK hospitals, Semple said. "We would welcome collaborations to test the use of the tool with researchers in other countries."

Table. 4C Mortality Score Results

  Score range Proportion of study population Mortality rate
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 validated the score in another 22,361 adults hospitalized with COVID-19 who were evaluated after May 20. This step "showed this tool could guide clinician decisions, including treatment escalation," they explain.

Semple and colleagues also validated their score against 15 other risk stratification methods identified in a systematic literature search. "The 4C Mortality Score compared well against these existing risk stratification scores in predicting in-hospital mortality," the researchers write.

Sensitivity analyses demonstrated that the score remained valid in ethnic minority groups and among different geographic cohorts.

The researchers caution that the scoring system is not designed for use in community settings and might perform differently in populations at lower risk for death. More validation is required to determine whether the score applies to younger people with COVID-19 and in countries other than the United Kingdom.

"Pragmatic Risk Score"

"This study develops and validates a pragmatic risk score to predict mortality in patients admitted to hospital with COVID-19. The mortality score is an easy-to-use and validated prediction tool for in-hospital mortality that accurately stratifies patients as being at low, intermediate, high, or very high risk of death," Tim Q. Duong, PhD, who is not affiliated with the study, told Medscape Medical News when asked to comment.

Dr Tim Duong

The large cohort of COVID-19 patients, the ready availability of clinical variables, and the fact that the 4C Mortality Score outperformed or was comparable to other risk stratification tools were strengths of the study, said Duong, director of MRI research, Renaissance School of Medicine, Stony Brook University, Stony Brook, New York.

Duong was senior author of a July 16, 2020, study that evaluated the role of deep-learning artificial intelligence in predicting COVID-19 mortality.

"Patients come into the hospital at different stages of disease severity, and it may be challenging to predict mortality far downstream," which is a potential limitation of the current study, Duong said. Furthermore, mortality rates may depend on each hospital’s patient load, available resources, and treatment regimens.

"This study nonetheless set the foundation for future prospective studies," he said.

Semple received grants from the Department of Health and Social Care, National Institute for Health Research UK and grants from the Medical Research Council UK and the Health Protection Research Unit in Emerging and Zoonotic Infections at University of Liverpool during the conduct of the study. Duong has disclosed no relevant financial relationships.

BMJ. Published online September 9, 2020. Full text

Follow Damian McNamara on Twitter: @MedReporter. For more Medscape Neurology news, join us on Facebook and Twitter.

For more news, follow Medscape on Facebook, Twitter, Instagram, and YouTube.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.