Risk of COVID-19 related admissions in cancer patients in a UK metropolitan region

Purpose: Few studies have investigated the susceptibility of cancer patients to COVID-19. We aim to quantify the risk of hospitalization in active cancer patients and use a machine learning algorithm (MLA) and traditional statistics in predicting clinical outcomes and mortality. Methods: A single UK centre retrospective cohort study was conducted (Rec20/EE/0139;IRAS ID28233). Data on total hospital admissions between March 2018 and June 2020, all active cancer diagnoses between March2019 and June2020 and clinical parameters of all patients with positive COVID-19 admissions between March2020 and June2020 were collected. 30 and 90-day post-COVID- 19 survival was determined. The denominator was the total and cancer population of the Dudley (UK). Logistic regression analyses were performed with SPSS 22.0. R-Studio software was used to determine the association between cancer status, COVID-19 and 90-day survival against variables in a MLA. Results: The estimated infection rate of COVID-19 was 87/22729 (0.4%) in the cancer patients and 526/426658 (0.1%) in the non-cancer population (Odds ratio: 3.105;95% CI: 2.474-3.897;P < 0.001). The median age was 77 years and male to female ratio 1:6. Multivariate analysis showed increases in age (OR1.039[95%CI1.020-1.057], P < 0.001),urea (OR1.005[95%CI1.002-1.007], P < 0.001) and CRP (OR1.065[95%CI1.016-1.116], P < 0.008) is associated with greater 30-day mortality. The MLA model examined the relative contribution of predictive variables for 90-day survival;with transplant patients, age, male gender and diabetes mellitus status being predictors of greater mortality. Conclusions: Active cancer diagnosis has a 3-fold increase in the risk of hospitalization with COVID-19. Increased age, urea, and CRP predict mortality. MLA complements traditional statistical analysis in identifying predictive variables.