Simple Blood Ratio Predicts Death Risk in Sepsis-Linked Lung Failure Patients

A routine pair of blood tests that costs pennies and takes minutes to perform may help intensive care physicians identify which patients with sepsis-associated acute respiratory distress syndrome are most likely to die during their hospital stay, according to a retrospective analysis published in BMC Infectious Diseases. The study, led by Yuqian Niu and colleagues at the Affiliated Hospital of Guizhou Medical University in China, examined the serum creatinine-to-albumin ratio, or CAR, in more than two thousand critically ill patients and found that those with the highest ratios faced a dramatically elevated risk of death within thirty days of intensive care unit admission.

Sepsis-associated ARDS represents one of the most lethal syndromes encountered in modern critical care. When overwhelming infection triggers the body’s immune defenses into a destructive cascade, the delicate air sacs of the lungs flood with inflammatory fluid, impairing oxygen exchange and often forcing patients onto mechanical ventilators. Mortality in this population remains stubbornly high despite decades of research into targeted therapies, and clinicians have long sought reliable, early markers that can separate patients who might respond to aggressive treatment from those whose trajectory is likely to be fatal. The new analysis suggests that a simple arithmetic combination of two ubiquitous laboratory values could contribute to that triage.

The research team drew its data from the Medical Information Mart for Intensive Care IV, widely known as MIMIC-IV, a vast, anonymized repository of intensive care records maintained jointly by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. The database, approved for research use by the institutional review boards of both institutions with a waiver of individual informed consent, contains detailed clinical information on tens of thousands of ICU admissions. From this resource, the investigators identified 2,093 patients who met diagnostic criteria for sepsis-associated ARDS and for whom both serum creatinine and albumin measurements were available at admission.

The headline finding was stark. Thirty-day all-cause mortality across the cohort stood at 32.25 percent, meaning nearly one in three patients died within a month. When the researchers divided patients into four groups based on their creatinine-to-albumin ratio, the Kaplan-Meier survival curves, a standard statistical tool that tracks survival over time, separated cleanly. Patients in the highest quartile, designated Q4, experienced significantly worse thirty-day survival than those in the other three groups, with a log-rank test yielding a probability value below 0.001, indicating that the difference was very unlikely to have arisen by chance.

To quantify the association more precisely, the team employed Cox proportional hazards modeling, a regression technique that estimates how a given factor changes the instantaneous risk of an event such as death while accounting for other variables. In the fully adjusted model, each one-unit increase in the creatinine-to-albumin ratio was associated with a 50.7 percent higher hazard of death, with a hazard ratio of 1.507 and a 95 percent confidence interval spanning 1.361 to 1.67. When CAR was treated as a categorical variable, patients in the top quartile faced a 45.1 percent higher risk of death compared with those in the lowest quartile, corresponding to a hazard ratio of 1.451 with a confidence interval of 1.108 to 1.90 and a p-value of 0.007.

Importantly, the researchers did not assume that the relationship between CAR and mortality was a simple straight line. They applied restricted cubic spline regression, a flexible method that allows the shape of the association to bend and curve rather than forcing it into a linear mold. The spline analysis revealed a genuinely nonlinear relationship between the ratio and all-cause mortality, suggesting that the risk conferred by rising CAR may accelerate or plateau at different points along its range. This kind of dose-response characterization matters for clinicians, because it hints at where meaningful thresholds for clinical concern might lie rather than implying that every incremental change carries identical weight.

Skeptics of retrospective database studies often raise a legitimate concern: could the observed association simply reflect confounding by other factors? The authors addressed this on multiple fronts. They assessed multicollinearity among the variables in their models to ensure that overlapping predictors were not distorting the estimates, and they performed subgroup analyses across clinically relevant patient strata, accompanied by interaction tests, to determine whether the CAR-mortality link held consistently across different types of patients. The robustness of the results across these analyses strengthens the case that the ratio captures prognostic information beyond what is already embedded in conventional severity scores and comorbidity measures such as heart failure and diabetes.

The biological plausibility of the finding rests on the two components of the ratio itself. Serum creatinine, a waste product of muscle metabolism cleared by the kidneys, rises when renal function deteriorates, a frequent and ominous complication of septic shock. Albumin, the most abundant protein in blood plasma, falls during severe inflammation as blood vessels become leaky and the liver shifts its synthetic priorities toward acute-phase reactants. Low albumin on its own is a well-established marker of poor prognosis in critical illness, reflecting both the intensity of the inflammatory response and the patient’s nutritional reserve. By dividing creatinine by albumin, the ratio effectively combines two independent danger signals into a single number: it climbs when the kidneys are failing and when inflammation is ravaging vascular integrity, making it a compact index of multi-system derangement.

What makes the result potentially transformative is its accessibility. Unlike novel biomarkers that require specialized assays, proprietary platforms, or lengthy turnaround times, creatinine and albumin are measured routinely in virtually every hospital laboratory on Earth, often within an hour of admission. A prognostic indicator built from tests that are already ordered for nearly every ICU patient could be integrated into early warning systems, triage protocols, and clinical trial stratification without any additional cost or infrastructure. The authors suggest that CAR may aid early risk stratification in sepsis-associated ARDS, potentially helping clinicians decide which patients warrant the most aggressive interventions, earlier nephrology consultation, or expedited discussions about goals of care.

As with any retrospective single-database analysis, caveats remain. The study population derives from a single academic medical center in the United States, and the findings will need validation in external cohorts before CAR can be recommended for routine clinical decision-making. The observational design establishes association, not causation, and the ratio cannot by itself determine whether elevated values mark a modifiable process or simply reflect irreversible injury. Nevertheless, in a field where mortality has proven resistant to decades of therapeutic advances, the identification of a cheap, immediately available marker that flags the highest-risk patients within hours of admission offers a genuinely practical advance, and it underscores a recurring lesson in critical care research: sometimes the most powerful prognostic tools are not new molecules but new ways of reading the numbers clinicians already have.

Subject of Research: Association between serum creatinine-to-albumin ratio and all-cause mortality in sepsis-associated ARDS patients in the intensive care unit

Article Title: Association of serum creatinine-to-albumin ratio with all-cause mortality in patients with sepsis-associated ARDS in the intensive care unit: a retrospective analysis based on the MIMIC-IV database

Article References: Niu, Y., Cao, Y., Zhong, D., Chen, Z., Shao, D., Dai, Y., Liu, Y., Li, K., & Zhang, X. (2026). Association of serum creatinine-to-albumin ratio with all-cause mortality in patients with sepsis-associated ARDS in the intensive care unit: a retrospective analysis based on the MIMIC-IV database. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14549-z

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14549-z

Keywords: sepsis, ARDS, creatinine-to-albumin ratio, all-cause mortality, MIMIC-IV, intensive care unit, prognostic biomarker, Cox proportional hazards, restricted cubic spline, retrospective study, critical care, risk stratification