Critical illness is characterized by a catabolic, proinflammatory state. Anabolic agents, such as testosterone, have therefore been proposed as therapeutic targets. Our objectives were to assess the effects of testosterone in critically ill populations on patient-important outcomes and identify design limitations to inform future studies. We searched for randomized control trials (RCTs) through Medline, Embase, and EBM Reviews databases from inception through February 24, 2026, including English language articles enrolling adults (≥18 years) admitted to ICU where anabolic androgen therapies (AAT) were compared with placebo or standard of care. Studies had to report at least one of: mortality, ICU and hospital lengths of stay, or duration of mechanical ventilation. We extracted data independently using a standardized data extraction tool, and feedback was received from all co-authors to ensure agreement. For each outcome, we performed meta-analyses using a random-effects model with inverse variance weighting in RevMan. We used the GRADE approach to assess certainty in pooled estimates of effect. Of 1325 screened articles, we found 4 that fit our inclusion criteria. Together, we judged risk of bias as 'some concerns' in 3 trials and 'high' in the final trial, and ultimately found that the effects of anabolic-androgen therapy on patient-important outcomes uncertain. With the uncertainty of current evidence for the effects of anabolic-androgen therapy in critically ill adults, there is insufficient support for its routine use. Future randomized evidence is needed to determine whether anabolic-androgen therapy improves clinically-important outcomes and better define its safety profile in critically ill adults.

Accumulating evidence suggests that outcomes are shaped not only by age and comorbidities, but also by preexisting biological states. A key determinant is immune resilience-the capacity to restore immune function and organ homeostasis. Aging-or accelerated biological aging-promotes accumulation of senescent cells and the senescence-associated secretory phenotype (SASP), driving chronic low-grade inflammation ("inflammaging"), impairing immune function, and reducing host resilience. SASP may also prime the immune system for exaggerated inflammatory responses in sepsis, contributing to cytokine storm. Notably, hyperinflammatory and immunosuppressive responses often emerge concurrently rather than sequentially. Immune cell death pathways help explain this apparent contradiction, as highlighted in the review featured in this issue of the Journal. These pathways provide a mechanistic bridge between these seemingly opposing states. They are not merely consequences of sepsis but active determinants of immune trajectory. Loss of SCFA- producing microbiota may impair immune regulation and promote dysregulated cell death, sustaining both hyperinflammatory and immunosuppressive states. This framework provides a testable model and supports interventional studies to determine whether restoration of SCFA signaling-such as through early administration of butyrate-producing microbes or tributyrin (a stable butyrate triglyceride)-can favorably alter immune trajectories and improve clinical outcomes.

Sepsis remains a leading cause of death, driven not only by early hyperinflammation but also by a catastrophic collapse of adaptive immunity during the late phase. This failure is orchestrated by distinct regulated cell death (RCD) pathways - apoptosis, pyroptosis, necroptosis and ferroptosis - that differentially deplete T cells, B cells and dendritic cells while shaping the immunological milieu. Apoptosis silently eliminates lymphocytes and promotes immunosuppression; pyroptosis and necroptosis release damage-associated molecular patterns, fueling inflammation that paradoxically destroys adaptive effectors; and ferroptosis, an iron-dependent lipid peroxidation process, selectively targets metabolically active CD8[+] and T helper cells. This review proposes that these pathways do not operate in isolation but converge on a "cell death decision network" centred on caspase-8, receptor-interacting serine/threonine-protein kinase 1(RIPK1), reactive oxygen species (ROS) and mitochondria, whose integration determines lymphocyte fate under septic stress. Understanding this network opens opportunities for precision immunotherapy. Emerging strategies targeting these pathways hold promise, but their success will require phase-specific application, biomarker-guided patient stratification and cell-type-selective delivery. Targeting the quality, as well as the quantity, of cell death may restore adaptive immunity and improve survival in sepsis.

Trauma patients with traumatic intracranial hemorrhage (tICH) who require extracorporeal membrane oxygenation (ECMO) present a difficult therapeutic dilemma. Early anticoagulation may worsen intracranial bleeding, whereas prolonged anticoagulation-free ECMO may increase thrombotic complications. This narrative review examines the time-dependent balance between these competing risks and proposes a conceptual time-dependent risk framework for anticoagulation initiation. A focused search of PubMed, Embase, and the Cochrane Library identified studies on the temporal profile of hemorrhagic progression after tICH, thrombotic outcomes during anticoagulation-free ECMO, and clinical experience with ECMO in trauma patients with intracranial hemorrhage. Hemorrhagic progression after tICH is concentrated within the first 24 to 48 h and, based predominantly on non-ECMO cohorts, decreases substantially after the first week. Anticoagulation-free ECMO is associated with a thrombotic complication rate of approximately 23% over a median support duration of 4.75 days. Despite originating from separate populations, multiple independent lines of evidence converge directionally on declining hemorrhagic risk and rising thrombotic risk over time. These data suggest that the timing of anticoagulation initiation should be guided by the changing balance between bleeding and clotting risk, with multidisciplinary review in selected patients demonstrating radiographic stability. The framework is limited by heterogeneous data and indirect evidence. It is intended to support multidisciplinary bedside discussion, not to define a fixed treatment threshold, and requires prospective validation.