Heterogeneity and temporal variation in the management of COVID-19: a multinational drug utilization study including 71,921 hospitalized patients from China, South Korea, Spain, and the United States of America

Albert Prats‐Uribe, Lana Yin Hui Lai, Ahmed W-u-r., Heba Alghoul, Osaid Alser, Thamir M. Alshammari, Carlos Areia, William Carter, Paula Casajust, Dalia Dawoud, Asieh Golozar, Jitendra Jonnagaddala, Puja Mehta, G Menchung, Morales, Fredrik Nyberg, Jose Posada, Martina Recalde, Elena Roel, Karishma Shah, Nirja Shah, Lisa M. Schilling, Vignesh Subbian, David Vizcaya, Ashley Williams, Lin Zhang, Yumin Zhang, Hua Zhu, Li Liu, Peter R. Rijnbeek, George Hripcsak, Lane Jce., E Burn, Christian Reich, M A Suchard, Talita Duarte‐Salles, Kristin Kostka, Phil Ryan, Daniel Prieto‐Alhambra

Abstract

Objectives: A plethora of medicines have been repurposed or used as adjunctive therapies for COVID-19. We characterized the utilization of medicines as prescribed in routine practice amongst patients hospitalized for COVID-19 in South Korea, China, Spain, and the USA.

Design: International network cohort study.

Setting: Hospital electronic health records (EHR) from Columbia University Irving Medical Centre, Stanford, Tufts, Premier, Optum EHR, Department of Veterans Affairs (USA), NFHCRD (China), and HM Hospitals (Spain); and nationwide claims from HIRA (South Korea).

Participants: 71,921 patients hospitalized for COVID-19 from January to June 2020.

Main outcome measures: Prescription/dispensation of any medicine on or 30 days after hospital admission date.

Results: 71,921 people were included: 304 from China, 2,089 from Spain, 7,599 from South Korea, and 61,929 from the USA. Common repurposed medicines included hydroxychloroquine (<2% in NFHCRD to 85.4% in HM), azithromycin (4.9% in NFHCRD to 56.5% in HM), lopinavir/ritonavir, and umifenovir. Adjunctive medicines were used with great variability, including enoxaparin, heparin, and dexamethasone. Hydroxychloroquine and azithromycin increased rapidly in use in March-April but declined steeply in May-June.

Conclusions: Multiple medicines were used in the first months of the COVID-19 pandemic, with substantial geographic and temporal variation. Research is needed on the comparative risk and benefit of these treatments.

Introduction: Navigating the "Wild West" of Early COVID-19 Care

In the early months of 2020, clinicians were "flying blind." With no established protocols for a novel virus, hospitals worldwide began repurposing existing medications based on limited in vitro evidence or small, anecdotal reports.

This study sought to quantify the "heterogeneity"—the diversity of treatment approaches—across different countries and over time. By characterizing what was actually prescribed to hospitalized patients in the US, China, Spain, and South Korea, the study provides a baseline for understanding how medical practice evolves during a global health crisis and highlights the gap between experimental repurposing and evidence-based medicine.

Methodology

The researchers conducted a multinational network cohort study using the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). This allowed each participating site to run the same analytical code locally and share only aggregated results, preserving patient privacy while enabling large-scale analysis.

The study classified drugs into two categories:

  1. Repurposed medicines: Drugs with existing indications believed to have antiviral properties (e.g., hydroxychloroquine, lopinavir/ritonavir).
  2. Adjuvant therapies: Drugs used to manage complications like pneumonia or cytokine storms (e.g., corticosteroids, anticoagulants).

Patient Characteristics and Datasets

The study analyzed 10 massive databases encompassing 71,921 hospitalized patients (including 11,512 in intensive care) between January and June 2020. These cohorts represented a diverse range of demographics and clinical backgrounds across four countries.

Demographic and clinical characteristics of hospitalized COVID-19 patients across international datasets.

As shown in the data, there was significant variation in age distribution and comorbidities across the different regions. For instance, the US Veterans Affairs (VA) cohort was predominantly male (93-94%), while other cohorts like HIRA (South Korea) and HM (Spain) showed different sex and age distributions. Comorbidities like hypertension, diabetes, and heart disease were prevalent across most hospitalized populations, though their recorded frequency varied by dataset.

  • The Evidence Pivot: The most striking finding was the temporal shift. Hydroxychloroquine and azithromycin use skyrocketed in March and April but crashed in May and June. This suggests the global medical community reacted with remarkable speed to negative results from major trials like RECOVERY and SOLIDARITY.
  • Geographic Silos: Treatment was highly localized. For example, Umifenovir was used in 78.3% of patients in China but was virtually non-existent in the other three countries. Similarly, Lopinavir/ritonavir saw high use in Spain (56.5%) and South Korea (34.9%) but remained below 3% in the US.
  • Anticoagulation as a Standard: The high prevalence of heparin and enoxaparin across the US and Spain reflects the early clinical realization that COVID-19 was a pro-thrombotic (clot-inducing) disease, leading to aggressive use of blood thinners even before formal guidelines were solidified.
  • Steroid Hesitancy: Despite their eventual status as a gold standard for severe COVID-19, systemic corticosteroids were used sparingly in the early months (January–April), only gaining traction later in the study period.

Summary of Contributions

This study demonstrates the extensive variability in drug repurposing and utilization during the initial phase of the pandemic. It documents how emerging clinical data rapidly impacted international prescribing habits.

Summary of existing knowledge and the novel contributions of this study.

Limitations and Future Work

The authors acknowledge several key limitations, including potential misclassification in EHR/claims data and the lack of data on Remdesivir, which became a cornerstone of care after the study period ended in June 2020. Crucially, this study is descriptive; it tracks what was given, but does not measure effectiveness. The authors emphasize that the next step is conducting comparative risk and benefit studies to determine which of these diverse approaches actually improved patient survival.

Summary updated

BibTeX
@misc{pratsuribe2020heterogeneity,
  author = {Albert Prats‐Uribe and Lana Yin Hui Lai and Ahmed W-U-R. and Heba Alghoul and Osaid Alser and Thamir M. Alshammari and Carlos Areia and William Carter and Paula Casajust and Dalia Dawoud and Asieh Golozar and Jitendra Jonnagaddala and Puja Mehta and G Menchung and Morales and Fredrik Nyberg and Jose Posada and Martina Recalde and Elena Roel and Karishma Shah and Nirja Shah and Lisa M. Schilling and Vignesh Subbian and David Vizcaya and Ashley Williams and Lin Zhang and Yumin Zhang and Hua Zhu and Li Liu and Peter R. Rijnbeek and George Hripcsak and Lane Jce. and E Burn and Christian Reich and M A Suchard and Talita Duarte‐Salles and Kristin Kostka and Phil Ryan and Daniel Prieto‐Alhambra},
  title = {{Heterogeneity and temporal variation in the management of COVID-19: a multinational drug utilization study including 71,921 hospitalized patients from China, South Korea, Spain, and the United States of America}},
  year = {2020},
  doi = {10.1101/2020.09.15.20195545},
  howpublished = {medRxiv},
}
Figure