Hematocrit-Albumin Gap and Sepsis Risk in ICU Patients: A Multicenter Study

A clinical research dataset from a retrospective cross-sectional study investigating the association between the hematocrit-albumin gap (HAG) and sepsis risk. The study used two cohorts: a derivation cohort of 19,134 adults from the MIMIC-IV database and a validation cohort of 630 patients from a hospital in China (2022–2023). The dataset was authored by Fenggao Zhou and last updated in April 2026.

Use Cases

• Modeling the non-linear relationship between hematocrit-albumin gap and sepsis risk based on the described S-shaped association.
• Conducting subgroup analyses for sepsis prediction based on conditions like anemia, liver disease, and metastatic solid tumors mentioned in the results.
• Validating the diagnostic utility of a simple biomarker calculation (HAG) for sepsis in critical care settings.
• Exploring interaction effects between biomarker gaps and patient comorbidities for risk stratification.

Strengths

• Derived from a large derivation cohort of 19,134 adult ICU admissions.
• Includes an independent validation cohort of 630 patients for result verification.
• Study reports specific odds ratios and threshold values (e.g., HAG between -9.162 and 17.27) for sepsis risk.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the actual shared dataset is unknown, which may limit suitability assessment.
• The data is presented in a DOCX file (22.5 KB), suggesting it may be a summary table or manuscript rather than a raw data file.

Provenance

Source

Derivation cohort from the MIMIC-IV database; validation cohort from The First People's Hospital of Yunnan Province, China.

Collection Method

Retrospective cross-sectional analysis using logistic regression and smooth curve fitting.

Time Range

Validation cohort data from 2022–2023.

Freshness

Last updated 2026-04-17 05:21:18; freshness should be verified.

Geography

Multicenter; includes data from a U.S. database (MIMIC-IV) and a hospital in China.