RADAR-2 will be a randomised, open-label, allocation concealed, pilot trial of conservative fluid administration and deresuscitation compared with usual care in patients who are critically ill.

The optimal approach to fluid balance in critically ill patients is uncertain. A recent systematic review found low quality evidence in favour of a conservative fluid or deresuscitative approach (active removal of accumulated fluid using diuretics and/or renal replacement therapy) compared with a liberal strategy or usual care. The RADAR-2 pilot randomised trial will compare conservative fluid and deresuscitation with usual care in patients who are mechanically ventilated in an intensive care unit. The main hypothesis is that in critically ill patients, a post-resuscitation fluid strategy comprising conservative fluid administration and active deresuscitation reduces net fluid balance, is safe and improves clinical outcomes.

Inclusion Criteria:

1. Invasive mechanical ventilation
2. Treating ICU doctor expects patient to require treatment in an ICU beyond the next calendar day
3. Between 24 and 48 hours from ICU admission at the time of randomisation

Exclusion Criteria:

1. Age < 16 years
2. Body weight <40kg (measured or estimated)
3. Diabetic ketoacidosis or Hyperosmolar hyperglycaemic state
4. Non-traumatic subarachnoid haemorrhage
5. Acute cardiac failure or cardiogenic shock
6. End-stage renal failure (on dialysis)
7. Known to be pregnant
8. Suspected or proven active diabetes insipidus (DDAVP within 24 hours)
9. Not expected to survive for 72 hours
10. Active 'Do not attempt resuscitation' order
11. Refusal of consent
12. Inability of personal consultee to understand written or verbal information and for whom no interpreter is available
13. Known allergy to one or more of the study drugs
14. Inability to measure fluid balance

Group assignment and early (up to 28 day) data will be locked and unavailable to investigators undertaking 6-month follow up.