Design: Multicohort retrospective study with risk-adjusted mixed-effects modelling
Population: 15,732 adults meeting CDC Adult Sepsis Event and Sepsis-3 criteria for septic shock in a large US health system, 2012-2024
Primary outcome: Cumulative incidence of refractory septic shock and risk-adjusted hospital mortality
Effect: 21.8% met criteria (218 per 1,000, 95% CI 211-224); hospital mortality 64.4%; risk-adjusted odds of death 4.87 times higher (4.46-5.31)

Consensus clinical criteria for refractory septic shock were agreed recently but had never been applied to a population to see how many patients they capture. This study operationalised them as a noradrenaline-equivalent dose above 0.5 micrograms per kilogram per minute occurring together with a lactate above 2 mmol/L, and applied them retrospectively to 15,732 adults with septic shock in a large US health system between 2012 and 2024.

Refractory septic shock occurred in 3,423 patients — 21.8%, or 218 per 1,000 septic shock admissions (95% CI 211-224). Hospital mortality in that group was 64.4%. After adjustment for age, sex, comorbidity burden and non-cardiovascular SOFA score, the odds of hospital death were 4.87 times higher (4.46-5.31) than in patients with septic shock who did not meet the criteria.

The value here is not prognostic novelty — everyone knows a patient on high-dose noradrenaline with a rising lactate is in trouble. It is that a definition now exists which is computable from data every unit already records, applies to a fifth of septic shock patients, and identifies a group with a mortality high enough to justify both trial recruitment and an explicit conversation with the family. That last use is the one available tonight.

Note the operational definition: noradrenaline equivalent above 0.5 mcg/kg/min with lactate above 2 mmol/L, concurrently
Use it to trigger a senior review and an explicit goals-of-care discussion, not a specific therapy
Recognise that no treatment is attached to the label — it identifies, it does not direct
Expect roughly one in five of your septic shock admissions to meet it
Record vasopressor dose in noradrenaline equivalents so the criterion is computable locally

Why it matters

It converts a clinical impression of 'this one is not turning around' into a defined, auditable state that can anchor a conversation and a trial.

The statistics, in plain English

An adjusted odds ratio of 4.87 with a narrow interval reflects the very large sample rather than a subtle finding; with 64.4% mortality, the odds ratio overstates the relative risk, which is the usual distortion when outcomes are common. This is retrospective and observational: the criteria mark severity, and nothing here shows that acting on them changes anything. Patients were drawn from one US health system, so the 21.8% incidence depends on that system's case mix and its thresholds for admission and vasopressor use.