Clustering patients with COVID-19 according to respiratory support requirements, and its impact on short- and long-term outcome (RECOVID study)

Pulmonology. 2025 Dec 31;31(1):2442175. doi: 10.1080/25310429.2024.2442175. Epub 2025 Jan 3.

Abstract

Introduction: The Spanish Society of Pulmonology and Thoracic Surgery created a registry for hospitalised patients with COVID-19 and the different types of respiratory support used (RECOVID). Objectives. To describe the profile of hospitalised patients with COVID-19, comorbidities, respiratory support treatments and setting. In addition, we aimed to identify varying profiles of patients according to outcomes and the complexity of respiratory support needed.

Methods: Multicentre, observational study in 49 Spanish hospitals. A protocol collected demographic data, comorbidities, respiratory support, treatment setting and 1-year follow-up. Patients were described using either frequency and percentages or median and interquartile range, as appropriate. A cluster analysis made it possible to identify different types of profile among the patients.

Results: In total, 2148 of 2454 hospitalised patients (87.5%) received care in the conventional ward, whilst 126 in IRCU and 180 in ICU. In IRCU, 30% required high-flow nasal oxygen whilst 25%, non-invasive mechanical ventilation and 17%, mechanical ventilation. Four clusters of patients were identified. Two clusters were more likely to require IRCU/ICU admission, although primarily Cluster 2: Cluster (C) 1 consisted of patients without comorbidities and C2, those with comorbidities. Both presented higher inflammatory levels and lower lymphocyte count and SpO2/FiO2; however, C2 showed worse values. Two different clusters identified patients requiring less complex respiratory support. C3 presented higher comorbidities and elevated lymphocyte count, SpO2/FiO2 and low C-reactive protein (CRP). C4 included those without comorbidities except for arterial hypertension, lymphopenia and an intermediate CRP. In-hospital mortality and subsequent 1-year mortality were greater for C2 (28.6% and 7.1%) and C1 (11.1%, 8.3%) than for C4 (3.3%, 1.8%) and C3 (0%, 0%).

Conclusions: The cluster analysis identified four clinical phenotypes requiring distinct types of respiratory support, with great differences present per characteristics and outcomes.

Keywords: COVID-19; NIMV; RECOVID; SARS-CoV-2; SEPAR.

Publication types

  • Observational Study
  • Multicenter Study

MeSH terms

  • Aged
  • COVID-19* / epidemiology
  • COVID-19* / mortality
  • COVID-19* / therapy
  • Cluster Analysis
  • Comorbidity
  • Female
  • Hospitalization / statistics & numerical data
  • Humans
  • Intensive Care Units / statistics & numerical data
  • Male
  • Middle Aged
  • Registries
  • Respiration, Artificial* / statistics & numerical data
  • SARS-CoV-2
  • Spain / epidemiology
  • Treatment Outcome