JMIR Publications

ABSTRACT

Background:

The COVID-19 pandemic has led to an unprecedented global healthcare challenge for both medical institutions and researchers. Recognizing different COVID-19 subphenotypes and their severity characterization may assist clinicians during the clinical course, research efforts, surveillance system, and the allocation of limited resources.

Objective:

We aimed to discover age-gender unbiased COVID-19 patient subphenotypes based on easily available phenotypical data before admission, such as pre-existing comorbidities, lifestyle habits and demographic features, to study the potential early severity stratification capabilities of the discovered subgroups through characterizing their severity patterns, including prognostic, ICU and morbimortality outcomes.

Methods:

We used the Mexican Government COVID-19 open data, including 778,692 SARS-CoV-2 population-based patient-level data as of September 2020. We apply a meta-clustering technique that consists of a two-stage clustering approach combining dimensionality reduction –i.e., principal components analysis and multiple correspondence analysis– and hierarchical clustering using Ward’s minimum variance method with Euclidean squared distance.

Results:

56 clusters from independent age-gender clustering analyses supported 11 clinically distinguishable meta-clusters (MCs). MCs 1-3 showed high recovery rates (90.27-95.22%), including healthy patients of all ages; children with comorbidities alongside priority in medical resources; and young obese smokers. MCs 4-5 showed moderate recovery rates (81.3-82.81%): patients with hypertension or diabetes of all ages; and obese patients with pneumonia, hypertension and diabetes. MCs 6-11 showed low recovery rates (53.96-66.94%): immunosuppressed patients with high comorbidity rate; chronic kidney disease patients with poor survival length and recovery; elderly smokers with chronic obstructive pulmonary disease; severe diabetic elderly with hypertension; and oldest obese smokers with chronic obstructive pulmonary disease and mild cardiovascular disease. Group outcomes conformed to the recent literature on dedicated age-gender groups. Age is not necessarily linked to mortality but physiological age, which also considers the combination of unhealthy habits and comorbidities. Centenarians are prone to better outcomes due to a low prevalence of pre-existing conditions. Mexican states and several types of clinical institution showed relevant heterogeneity regarding severity, implying a crucial socioeconomic-inequality and healthcare resource level inequality, important for consideration in further studies.

Conclusions:

The proposed two-stage cluster analysis methodology produced clinically coherent models with discriminative characterization and explainability over age and gender. These results can potentially help in the clinical patient understanding and their stratification towards automated early triage prior to further tests and laboratory results are available, and even locations where additional tests are not available or help decide resource allocation among vulnerable subgroups such as prioritize vaccination among subgroups that are more vulnerable when they get infected during the pandemic.