Back

Clinical and historical features associated with severe COVID-19 infection: a systematic review

Pigoga, J. L.; Friedman, A.; Broccoli, M.; Hirner, S.; Naidoo, A. V.; Singh, S.; Werner, K.; Wallis, L. A.

2020-04-27 infectious diseases
10.1101/2020.04.23.20076653 medRxiv
Show abstract

BackgroundThere is an urgent need for rapid assessment methods to guide pathways of care for COVID-19 patients, as frontline providers need to make challenging decisions surrounding rationing of resources. This study aimed to evaluate existing literature for factors associated with COVID-19 illness severity. MethodsA systematic review identified all studies published between 1/12/19 and 19/4/20 that used primary data and inferential statistics to assess associations between the outcome of interest - disease severity - and historical or clinical variables. PubMed, Scopus, Web of Science, and the WHO Database of Publications on Coronavirus Disease were searched. Data were independently extracted and cross-checked independently by two reviewers using PRISMA guidelines, after which they were descriptively analysed. Quality and risk of bias in available evidence were assessed using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) framework. This review was registered with PROSPERO, registration number CRD42020178098. ResultsOf the 6202 relevant articles found, 63 were eligible for inclusion; these studies analysed data from 17648 COVID-19 patients. The majority (n=57, 90{middle dot}5%) were from China and nearly all (n=51, 90{middle dot}5%) focussed on admitted adult patients. Patients had a median age of 52{middle dot}5 years and 52{middle dot}8% were male. The predictors most frequently associated with COVID-19 disease severity were age, absolute lymphocyte count, hypertension, lactate dehydrogenase (LDH), C-reactive protein (CRP), and history of any pre-existing medical condition. ConclusionThis study identified multiple variables likely to be predictive of severe COVID-19 illness. Due to the novelty of SARS-CoV-2 infection, there is currently no severity prediction tool designed to, or validated for, COVID-19 illness severity. Findings may inform such a tool that can offer guidance on clinical treatment and disposition, and ultimately reduce morbidity and mortality due to the pandemic.

Matching journals

The top 10 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.