This month’s winning publication was “Development and validation of the ISARIC 4C Deterioration model for adults hospitalised with COVID-19: a prospective cohort study”, co-authored by HDR UK member(s) Annemarie B Docherty, Prof. Aziz Sheikh and Prof. Cathie Sudlow.
The team have developed a model based on 11 characteristics to predict whether adults who have been hospitalised for COVID-19 will end up in one of the three states: 1) needing ventilator support 2) critical care or 3) death. This work is valuable because, it informs clinician of the level of risk for each patient at the point of hospitalisation for COVID-19 giving clinicians the opportunity to be able to intervene and prevent progression to more adverse outcomes. The method utilized is informed by routinely collected patient data. Read on for more technical details of this work, provided by Fatemeh Torabi on behalf of the our Early Career Research Committee.
Gupta et al.’s used multivariable logistic regression to develop a prognostic model which predicts risk of clinical deterioration at the point of hospitalisation for COVID-19. The study was conducted across nations, data from 260 hospitals in England, Scotland, and Wales for 74,944 hospitalised adults was used to develop and validate the model. Through backward elimination, 11 predictors (age, sex, nosocomial infection, Glasgow coma scale score, peripheral oxygen saturation at admission, breathing room air or oxygen therapy, respiratory rate, urea concentration, C-reactive protein concentration, lymphocyte count, and presence of radiographic chest infiltrates) were remained, in the final proposed model. The data in this study was divided into 9 NHS regions, of which 8 regions were used as training data for model development and internal-external cross validation. The ninth region used later for out-sample validation, independent of the training set.
Out of all articles reviewed this month, Gupta and Harrison et al best fulfilled the core criteria of the HDR UK ethos for collaboration: a UK-wide multi institutional research, scale: large training and test data across nations, openness and transparency: all stages of model and validation was covered in-text and as supplement. Authors mentioned that they’ve used R as analytical software, ECC members command this effort and would also encourage making the models development code syntax available to the research community. The high potential of impacts to patients and public was clear to the committee from the body of work as well as authors’ acknowledgement of the used of data provided by patients; and the impact these can have on informed clinical decisions and therapeutic interventions. The ECC members felt that, not only should Gupta et al. be commended for this scientific contribution, but also for the way they have formed a UK-wide collaboration to draw a large sample size from across the nations and their effort in obtaining a diverse sample to make the findings generalizable.
Our Early Career Committee would like to congratulate and commend Rishi K Gupta and the authorship team for their contribution to our vision of uniting the UK’s health data to enable discoveries that improve people’s lives.
BHF Data Science Centre
Improving the public’s cardiovascular health through the power of large-scale data and advanced analytics across the UK.
Early Career Committee
Meet our Early Career Committee (ECC) who each month, select an Open Access Publication of the Month.