The best strategy for managing fever in patients with septic shock remains unknown. In a pilot study, the investigators showed that fever control at normothermia allowed a better control of shock and evolution of organ failures. In this second trial the investigators will conduct a multicentre, open-label, randomized controlled, superiority trial in which two strategies will be compared:

1. Respect of fever
2. Fever control at normothermia using external cooling The primary end point will be d-60 mortality.

Sepsis is a common syndrome responsible for multiple organ failure. Septic shock, defined as sepsis with hyperlactatemia and cardiovascular failure requiring vasopressor infusion despite adequate fluid resuscitation has an extremely high mortality rate. Fever is a frequent disease process during sepsis. Fever increases oxygen consumption and can worsen imbalance between oxygen supply and oxygen requirements. Fever increases inflammation but reduces viral and bacterial growth. The beneficial effects of active fever control on inflammation have been mainly shown in a context of lung injury. Pneumonia represents the first cause of septic shock in developed countries.

In a pilot study (SEPSISCOOL I), we showed that fever treatment using external cooling significantly increased the resolution of shock, improved organ functions and decreased d-14 mortality. Although reduced, hospital mortality was not significantly different. This study was underpowered to allow conclusion on mortality. A more pronounced beneficial effect was observed among the most severely ill patients with elevated serum lactate level.

Fever treatment is commonly applied in septic patients but its impact on survival remains undetermined.

The main objective of the study is to compare two strategies of fever management in febrile (body temperature > 38.3°C) septic shock patients requiring invasive mechanical ventilation and sedation. These patients will be randomly allocated in two arms:

1. Fever respect
2. Fever control by external cooling to obtain normothermia during 48 hours

A covariate-adaptive randomization will be used to ensure the comparability of the two groups at each stage of the study. We will use an adaptive multistage population-enrichment design with a pre-specified subgroup of patients with ARDS identified at randomization.

An independent Safety and Data Monitoring Committee will review data on serious adverse events. The decision of study stop for potential harmful effect of one strategy will be let at the entire responsibility of the committee.

One interim analysis will be performed by independent observers after enrolment of half of the population. The assumption that fever treatment is more effective in patients with ARDS will be confirmed or not. According to pre-defined rules based on the conditional power calculated in the two subgroups, the trial will be stopped for futility, continued as planned or continued by enrolling only patients with ARDS.

Inclusion Criteria:

• Documented or suspected infection either communautary or hospital acquired
• Septic shock defined by the need for vasopressor and lactate>2 mmol/l despite adequate fluid resuscitation (sepsis-3 definition)
• Patients under invasive mechanical ventilation
• Body core temperature>38.3°C
• Intravenous sedation or opioids
• Ongoing antimicrobial treatment and/or intervention for infection source control
• Attending physician confirms clinical equipoise without substantial risk if the patient participates in the trial
• Informed consent of next of kin/other designated person before inclusion or procedure for inclusion in emergency situation

Exclusion Criteria:

• Cardiac arrest within previous 7 days
• Acute brain injury within previous 7 days
• Extensive burns or epidermal necrolysis
• <18 years old
• Body core temperature >41°C
• Under legal guardianship
• No affiliation with the French health-care system
• Pregnancy
• Participation in another interventional study with mortality as the primary endpoint
• An investigator's decision not to resuscitate
• Patient already recruited in the trial

Strictly anonymized data will be available upon justified request from the coordinating investigator and the study sponsor

After publication upon justified request from the coordinating investigator and the study sponsor