ISTH Congress Abstracts

Official abstracts site for the ISTH Congress

MENU 
  • Home
  • Congress Archive
    • ISTH 2020 Congress
  • Resources
  • Search

Identification of a COVID-19 Subpopulation Responsive to Hydroxychloroquine Using Machine Learning: The IDENTIFY Trial

C. Lam, S. Mataraso, A. Lynn-Palevsky, A. Green-Saxena, G. Barnes, J. Hoffman, J. Calvert, E. Pellegrini, R. Das

Dascena, Inc., Oakland, United States

Abstract Number: PB/CO04

Meeting: ISTH 2020 Congress

Theme: Diagnostics and OMICs » Epigenetics, OMICs and Bioinformatics

Background: Hydroxychloroquine has emerged as a controversial treatment for COVID-19. While some studies have suggested a survival benefit for patients prescribed hydroxychloroquine, other studies have suggested an increased risk of mortality and cardiovascular complications including de-novo arrhythmias. Precision medicine based methods to identify a subpopulation of COVID-19 positive patients who are likely to benefit from the use of hydroxychloroquine may help to reduce COVID-19-related mortality while preventing hydroxychloroquine complications.

Aims: To identify a subpopulation of COVID-19 patients for whom treatment with hydroxychloroquine improves survival.

Methods: We performed a pragmatic clinical trial of a machine learning algorithm to identify patients likely to respond positively to hydroxychloroquine. Patients were enrolled from 7 U.S. health centers. Treatment with hydroxychloroquine was not randomized; inverse probability of treatment weighting was used to adjust for baseline confounding factors, including demographic characteristics, medical history, and medication use. The primary endpoint was a composite outcome of ventilation or mortality in the subpopulation identified by the algorithm. The secondary endpoint was in-hospital mortality in the general COVID-19 population. Outcomes were assessed using Fine and Gray Cox Proportional Hazards models for competing risks to account for the competing outcome of hospital discharge.

Results: 201 patients were enrolled. In the subpopulation identified by the algorithm, those treated with hydroxychloroquine were less likely to experience mechanical ventilation or death (adjusted HR 0.29, 95% CI 0.11 – 0.75, p = 0.01) (Figure 1). In the general population, hydroxychloroquine was not associated with decreased risk of ventilation or mortality (adjusted HR 1.59, 95% CI 0.89 – 2.83, p = 0.12) (Figure 2).

Conclusions: It is possible to identify a subpopulation of COVID-19 patients for whom hydroxychloroquine is associated with increased survival. Identification of these patients can reduce COVID-19 mortality as well as morbidity associated with COVID-19 complications, such as pulmonary embolism and cerebral thromboembolism.

[Figure 1. Adjusted survival curves among those identified as suitable for treatment by the algorithm]


[Figure 2. Adjusted survival curves for the whole study population.]

To cite this abstract in AMA style:

Lam C, Mataraso S, Lynn-Palevsky A, Green-Saxena A, Barnes G, Hoffman J, Calvert J, Pellegrini E, Das R. Identification of a COVID-19 Subpopulation Responsive to Hydroxychloroquine Using Machine Learning: The IDENTIFY Trial [abstract]. Res Pract Thromb Haemost. 2020; 4 (Suppl 1). https://abstracts.isth.org/abstract/identification-of-a-covid-19-subpopulation-responsive-to-hydroxychloroquine-using-machine-learning-the-identify-trial/. Accessed March 3, 2021.
Save to PDF

« Back to ISTH 2020 Congress

ISTH Congress Abstracts - https://abstracts.isth.org/abstract/identification-of-a-covid-19-subpopulation-responsive-to-hydroxychloroquine-using-machine-learning-the-identify-trial/

Simple Search

ISTH 2020 Congress site

Visit the official web site for the ISTH 2020 Virtual Congress »

  • Help & Support
  • About Us
  • Cookies & Privacy
  • Wiley Job Network
  • Terms & Conditions
  • Advertisers & Agents
Copyright © 2021 John Wiley & Sons, Inc. All Rights Reserved.
Wiley
This site uses cookies: Find out more.