Risk Factors of Fatal Outcome in Hospitalized Subjects With Coronavirus Disease 2019 From a Nationwide Analysis in China

Abstract

BACKGROUND:The novel coronavirus disease 2019 (COVID-19) has become a global health emergency. The cumulative number of new confirmed cases and deaths are still increasing out of China. Independent predicted factors associated with fatal outcomes remain uncertain.RESEARCH QUESTION: The goal of the current study was to investigate the potential risk factors associated with fatal outcomes from COVID-19 through a multivariate Cox regression analysis and a nomogram model. STUDY DESIGN AND METHODS:A retrospective cohort of 1,590 hospitalized patients with COVID-19 throughout China was established. The prognostic effects of variables, including clinical features and laboratory findings, were analyzed by using Kaplan-Meier methods and a Cox proportional hazards model. A prognostic nomogram was formulated to predict the survival of patients with COVID-19. RESULTS:In this nationwide cohort, nonsurvivors included a higher incidence of elderly people and subjects with coexisting chronic illness, dyspnea, and laboratory abnormalities on admission compared with survivors. Multivariate Cox regression analysis showed that age $ 75 years (hazard ratio [HR], 7.86; 95% CI, 2. 44-25.35), age between 65 and 74 years (HR, 3.43; 95% CI, 1.24-9.5), coronary heart disease (HR, 4.28; 95% CI, 1.14-16.13), cerebrovascular disease (HR, 3.1; 95% CI, 1.07-8.94), dyspnea (HR, 3.96; 95% CI,, procalcitonin level > 0.5 ng/mL (HR, 8.72; 95% CI,, and aspartate aminotransferase level > 40 U/L (HR, 2.2; 95% CI, 1.1-6.73) were independent risk factors associated with fatal outcome. A nomogram was established based on the results of multivariate analysis. The internal bootstrap resampling approach suggested the nomogram has sufficient discriminatory power with a C-index of 0.91 (95% CI, 0.85-0.97). The calibration plots also showed good consistency between the prediction and the observation. INTERPRETATION:The proposed nomogram accurately predicted clinical outcomes of patients with COVID-19 based on individual characteristics. Earlier identification, more intensive surveillance, and appropriate therapy should be considered in patients at high risk.

ANDC: an early warning score to predict mortality risk for patients with Coronavirus Disease 2019

“…In accordance with previous studies on severe acute respiratory syndrome (SARS) [11], Middle East respiratory syndrome (MERS) [12] and COVID-19 [13], older age was also identi ed as a predictor for poor prognosis of patients with COVID-19 in our study. It is hypothesized that immunosenescence and/or underlying comorbidities might deem geriatric patients more vulnerable to developing severe COVID-19 illness.…”

The role of essential organ-based comorbidities in the prognosis of COVID-19 infection patients

“…Previous studies reported non-survivors present a higher proportion of various co-existing chronic illnesses in univariate analysis [4,[10][11][12]. Consistent with previous reports [10,13,14], our study indicated that COVID-19 patients with heart, liver, and kidney comorbidities had a higher risk of related organ injuries and death, compared with those without comorbidities. The cardiac, liver, and kidney complications can occur precipitously at any point during hospitalization and are increasingly being described as late complications that can occur after improvements in a patient's respiratory status.…”

Ideas for how informaticians can get involved with COVID-19 research

“…variables) that ML can better utilize to represent the problem/target outcome, (2) feature selection, applying expert domain knowledge, statistical methods, and/or ML methods to remove 'irrelevant' features from consideration and improve downstream modeling, (3) data harmonization, allowing for the integration of data collected at different sites/institutions, (4) handling different outcomes and related challenges, e.g. binary classification, multi-class, quantitative phenotypes, class imbalance, temporal data, multi-labeled data, censored data, and the use of appropriate evaluation metrics, (5) ML algorithm selection for a given problem can be a challenge in itself, thus strategies to integrate the predictions of multiple machine learners as an ensemble are likely to be important, (6) ML modeling pipeline assembly, including critical considerations such as hyper-parameter optimization, accounting for overfitting, and clinical interpretability of trained models, and (7) considering and accounting for covariates as well as sources of bias in data collection, study design, and application of ML tools in order to avoid drawing conclusions based on spurious correlations.…”

“…The coronavirus disease 2019 (COVID-19) pandemic has had a significant impact on population health and wellbeing. Research efforts are underway to identify vaccines [1], improve testing [2,3], understand transmission [4], develop serologic tests [5], develop therapies [6], predict risk [7], and develop mitigation and prevention strategies [8,9]. Biomedical informatics is central to each of these research efforts and for the delivery of healthcare for COVID-19 patients.…”