Risk Factors for Mortality in Hospitalized Coronavirus Disease 2019 Patients

To the Editor:

We read with interest the article published in CHEST (July 2020) by Chen and coworkers,¹ about mortality risk factors in hospitalized coronavirus disease 2019 (COVID-19) patients. In our opinion, the article deserves some attention. First, the same cohort with similar objectives has simultaneously appeared in another journal (JAMA Internal Medicine).² The most relevant difference between the two studies was the variable analyzed, mortality vs a composite variable (death, ICU admission, or mechanical ventilation). Both articles develop a multicomponent score, with a different statistical approach (logistic regression vs multivariate Cox regression). This may explain the different variables selected. In the companion study, 10 predictive variables were included, of which only four were maintained in the current analysis, and two new variables were included. We believe that because the two articles were published simultaneously, the inclusion of different variables to predict evolution in the same cohort merits discussion to avoid reader confusion.

Second, and more relevantly, the current model cannot be applied without an external validation in other populations. External validation is essential in all multicomponent prognostic scores, but in this case it is mandatory, because population and evolution differ greatly from what is reported in other areas of the world, and even other Chinese hospitals on the same dates. This suggests that in most cases the hospital admission criteria in this cohort seem to be related more to epidemiological reasons than clinical disease severity.³ The mortality reported was clearly lower than that observed in European and American cohorts in which it reaches percentages of 10% to 25%. Of note, the mortality reported in the same cohort in Hubei was 7.3%, and outside Hubei it is 0.3%, whereas in three other cohorts of 828 patients hospitalized in Wuhan, mortality on February 7, 2020 was 18.6%, 19.2%, and 16%, respectively.^{4 , 5} For comparative purposes, in our hospital (a 500-bed tertiary hospital in Spain), 723 patients were hospitalized for COVID-19 between February 5 and May 30, 2020. Of these, 29% developed a critical illness, and 17.4% died during admission.

Obviously, with these data, it seems that hospitalization criteria in this cohort may have contributed to containing the spread of the virus, but this strategy was not feasible in other areas where the health system was close to collapsing. More importantly, in our opinion, these differences preclude direct application of the proposed model without a previous external validation in different populations.