Scholarlink

Background <p>Identifying individuals hospitalized for SARS-CoV-2 infection at increased risk of death is crucial for clinical decision making. Analyses must consider simultaneously the multitude of biomarkers across several domains and how these biomarker profiles change over time.</p> Methods <p>This electronic health records-based study included individuals hospitalized at a Massachusetts General Brigham hospital for at least 24&#xa0;h within 5&#xa0;days prior and 30&#xa0;days after diagnosis of COVID-19. K-means clustering was used to identify profiles among 20 eligible biomarkers and proportional hazards models were used to model 30-day mortality at hospitalization and 7&#xa0;days after hospitalization (i.e., landmark models).</p> Results <p>Twelve thousand, nine hundred forty-two individuals were included, among whom 1<i>,</i>198 died within 30&#xa0;days. Six states were identified, characterized by the following abnormalities: (1) normal/reference, (2) hematologic, (3) inflammatory and hematological, (4) metabolic, (5) kidney, hematologic, and metabolic, and (6) cardio-thrombotic, liver, and metabolic. Risk of death within 30&#xa0;days was higher in States 3, 4, 5, and 6 (adjusted hazard ratios ranging from 3.6 to 7.8) compared to individuals in State 1 at hospitalization. Landmark model findings were similar.</p> Conclusions <p>Distinct sub-phenotypes based on biomarker profiles were identified among patients hospitalized with SARS-CoV-2 infection, and certain phenotypes are associated with greater risk of 30-day mortality.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Biomarker states and risk of death among individuals hospitalized with SARS-CoV-2 infection

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月24日

查看全文
  • Tanayott Thaweethai,
  • Caitlin A. Selvaggi,
  • Ta-Chou Ng,
  • David Cheng,
  • Tingyi Cao,
  • Lori B. Chibnik,
  • Daniel J. Shinnick,
  • Andrea S. Foulkes

摘要

Background

Identifying individuals hospitalized for SARS-CoV-2 infection at increased risk of death is crucial for clinical decision making. Analyses must consider simultaneously the multitude of biomarkers across several domains and how these biomarker profiles change over time.

Methods

This electronic health records-based study included individuals hospitalized at a Massachusetts General Brigham hospital for at least 24 h within 5 days prior and 30 days after diagnosis of COVID-19. K-means clustering was used to identify profiles among 20 eligible biomarkers and proportional hazards models were used to model 30-day mortality at hospitalization and 7 days after hospitalization (i.e., landmark models).

Results

Twelve thousand, nine hundred forty-two individuals were included, among whom 1,198 died within 30 days. Six states were identified, characterized by the following abnormalities: (1) normal/reference, (2) hematologic, (3) inflammatory and hematological, (4) metabolic, (5) kidney, hematologic, and metabolic, and (6) cardio-thrombotic, liver, and metabolic. Risk of death within 30 days was higher in States 3, 4, 5, and 6 (adjusted hazard ratios ranging from 3.6 to 7.8) compared to individuals in State 1 at hospitalization. Landmark model findings were similar.

Conclusions

Distinct sub-phenotypes based on biomarker profiles were identified among patients hospitalized with SARS-CoV-2 infection, and certain phenotypes are associated with greater risk of 30-day mortality.