Table 1.

Potential factors contributing to the lack of “bench-to-bedside” translational success of preclinical sepsis research

Domain	Preclinical model	Human condition
Construct validity
Patient population	Historically male, however more female animals in recent years Healthy Young/juvenile Limited environmental exposure Genetically homogeneous	Female and male Medical comorbidities Old and young Environmental stressors Genetically diverse
Site of infection	Non-bacterial surrogates (e.g. endotoxin) Polymicrobial abdominal/enteric Gram-negative Pneumonia (rare) Fungi/protozoa (rare) Virus (very rare)	Soft tissue Gram-positive Abdominal Gram-negative, including biliary Pneumonia (common) Virus (common) Fungi/protozoa
Intercurrent therapy	None Antibiotics (monotherapy) Fluids Anesthesia/analgesia Experimental therapy	Antibiotics (poly-therapy) Fluids Blood products Vasopressors/Inotropes Sedation/analgesia Baseline medication regimen Adjunct therapies (e.g. steroids, heparin)
Outcomes	Non-mortality surrogate Short term Organ failure (often single) Molecular biomarkers (common) Organ histology	Mortality Short and long term ICU/hospital length of stay Validated multi-organ failure score Molecular biomarkers (rare) Organ histology (very rare)
Research methodology
Biostatistics	Lack of sample size calculation	Study powered to detect difference in pre-specified outcome
Reduce bias	Randomization rare Lack of blinding	Randomization Double-blinded
Standardization	Single centre Variations in practice	Multicentre Shared protocol
Reporting	Inconsistent Incomplete Difficult to synthesize	Required Comprehensive Conducive to systematic review

There are knowledge gaps in construct validity and research methodology between preclinical models and the human condition of sepsis