Coronastream: the four C's and a little controversy
Posted on June 3, 2020 by Dr Tim Inglis
In this special blog series, medical microbiologists led by Dr Tim Inglis summarise some of the research that will be essential to inform COVID-19 countermeasures. Find out more about the project in Dr Inglis' Editorial 'Logic in the time of coronavirus', published in the Journal of Medical Microbiology.
In its first month, Coronastream featured sixteen papers, one per week in each of the four Principia aetiologica categories, as explained in the paper that launched the blog, 'Logic in the time of coronavirus'. All previous Coronastream blogs can be seen on the JMM Coronastream blog listing page.
A little controversy
There is a bit more clarity emerging among all the noise and action surrounding the SARS-CoV-2 pandemic. As a consequence, we are starting to see commentary on more general pandemic issues:
The author highlights a worrying development in COVID-19 publications, and refers to work done by the University of Edinburgh on research integrity in over 2000 publications they have reviewed so far. Only 14% of these contain original research, of which 72% is observational. As only a few of these are peer reviewed or pre-registered, there is little control against bias. The article notes a rush to publish, preoccupation with citation index and a breakdown of the review process under the current weight of submissions.
The authors of this provocative article point out that by the beginning of May this year, over 90% of deaths attributed to COVID-19 were in the world's richest countries; yet a one-size-fits-all approach to countermeasures has been applied irrespective of demographic context. Cash and Patel note that there are substantial inequities in access to diagnostic, therapeutic and preventive technology in low-and-middle-income countries, and predict that the universal approach currently being applied to the pandemic is likely to increase inequity in countries that suffer a much higher burden of mortality and morbidity from other diseases.
Congruence: artificial intelligence applied to early COVID-19 diagnosis
Artificial intelligence (AI) gives us a powerful set of tools to analyse, classify and visualise complex data sets. In this research, Mei and colleagues assessed an AI approach to identifying patients with COVID-19 from a combination of clinical features, basic laboratory and CT scan results.
They used a combination of machine learning techniques: convolutional neural network (CNN), support vector machine (SVM), random forest and multilayer perceptron (MLP), and neural network modelling.
When applied to 905 patients, there were 135 patients that tested positive for COVID-19 of whom 33 differed between the AI algorithm and the radiologist's assessment. The algorithm was correct in 23 cases and the radiologist in 10. This study provides a guide to development of AI methods for diagnosis of COVID-19 before changes are apparent on the CT scan.
Consistency: comparing COVID-19 predictive models
Prediction models for diagnosis and prognosis of COVID-19 infection: systematic review and critical appraisal
This paper looked at 27 studies describing 31 COVID-19 prediction models and concluded that the models “are poorly reported, at high risk of bias, and their reported performance is probably optimistic”.
They had different objectives: predicting hospital admission for pneumonia (3), detecting COVID-19 infection (18) or predicting prognosis (10). Only one of these used data from outside China.
The most reported predictors were:
- Presence of COVID-19: age, body temperature, and signs and symptoms
- Severe prognosis in COVID-19: age, sex, from CT scan features, C-reactive protein, lactic dehydrogenase and lymphocyte count
Cumulative dissonance: COVID-19 coagulopathy
In this synthesis of recent publications on coagulopathy in COVID-19, the authors stress the need to monitor D-dimers and fibrinogen levels and use thromboembolism prophylaxis for all patients admitted to hospital with COVID-19 in view of the high prevalence of thrombosis. SARS-CoV-2 infection induces microangiopathy, local thrombus formation and a systemic coagulation defect that results in large vessel thrombosis. The precise cause of this complex coagulopathy is unclear but is thought to involve host defences. Four major contributors to thrombus formation are identified in this paper: the infection-induced cytokine storm, suppression of fibrinolysis, platelet activation and inflammatory endothelial damage.
Countermeasures: vaccine trial
As introduction of an effective COVID-19 vaccine is seen as the best way to bring the pandemic to an end, any progress towards a vaccine attracts a lot of attention. This report of a safety and immunogenicity trial of an adenovirus vectored vaccine marks a step towards a working COVID-19 vaccine, though it is an open-label, non-randomised trial.
The vaccine candidate uses adenovirus Ad5 that expresses SARS-CoV-2 spike glycoprotein. This dose-escalation Phase 1 trial administered one of three intramuscular vaccine doses to healthy adults in a single centre, looked at adverse events in the week following and assessed safety over the four weeks after vaccination. Neutralising antibody and T-cell responses were measured.
A range of adverse responses were reported after vaccination and were commoner in the high vaccine dose group. These included fever, fatigue, dyspnoea, muscle pain, and joint pain. Antibodies increased at day 14 and peaked at day 28, while the specific T-cell response peaked at 14 days.
How to draw the coronavirus