Lymphocyte Counts Modestly Distinguish Modeled Corticosteroid Benefit
Baseline blood-cell markers were associated with model-estimated corticosteroid benefit in septic shock, but modest discrimination and observational evidence do not establish a reliable treatment-selection rule.
Lower starting counts of lymphocytes, a type of immune cell, modestly distinguished septic-shock cases for which a statistical model estimated lower intensive-care mortality with corticosteroids. This does not establish which patients actually benefit from these anti-inflammatory drugs.
Evidence
The study of blood markers in septic shock, published September 29, 2026, analyzed records from the Medical Information Mart for Intensive Care IV database. It included 5,533 stays involving adults with septic shock—severe infection-related circulatory failure—who spent at least 24 hours in intensive care. Corticosteroids were given in 1,009 stays, or 18.23%.
The researchers used a retrospective target trial emulation: they structured a treatment comparison using existing observational records rather than random assignment. Their adjustment method, augmented inverse probability weighting, combines statistical weighting with outcome modeling to address measured differences between treated and untreated groups. The primary outcome was death in intensive care.
Lower baseline lymphocyte count had a discrimination score of 0.65 for identifying model-estimated corticosteroid benefit. Higher platelet-to-lymphocyte ratio scored 0.63, and higher neutrophil-to-lymphocyte ratio scored 0.62. These ratios compare counts of platelets, which help blood clot, or neutrophils, another immune-cell type, with lymphocyte counts.
The discrimination measure—area under the receiver operating characteristic curve—describes how well a marker separates the model’s estimated-benefit categories. A score of 0.5 indicates chance-level discrimination; 1 indicates perfect discrimination.
The percentage predicted to benefit increased with higher neutrophil-to-lymphocyte ratios, using cutoffs of 3.93, 6.56 and 12.33.
Analysis
Interpretation: these blood-cell measurements may help describe variation in estimated treatment effects within the study’s model. Their performance was modest, however. A discrimination score of 0.65 is neither a 65% probability of treatment benefit nor a measure of mortality reduction.
This is a single-study finding, not independent confirmation of a marker-guided treatment strategy. The ratio cutoffs describe the analysis; they are not established treatment thresholds.
Limitations
The corticosteroid-treated group was substantially sicker, with higher organ-failure scores and greater requirements for drugs that support blood pressure. Statistical adjustment addresses measured differences but cannot exclude residual or unmeasured confounding—other factors influencing both treatment receipt and mortality.
The retrospective, nonrandomized design therefore cannot establish definitive causal treatment effects. The available results do not include confidence intervals or absolute mortality differences, limiting assessment of precision and benefit size. The numerical differences between marker scores do not establish that one marker reliably outperforms another.
What to watch
Independent testing could assess whether these markers retain their modest discrimination in another septic-shock population. Confidence in treatment-directed use would increase if a randomized study found different corticosteroid effects on mortality across prespecified lymphocyte-count groups, with absolute mortality differences and confidence intervals reported.