Critical Care

Where the Patient Cannot Consent and the Trial Still Has to Run.

Regulatory strategy for the ICU — where sepsis has defeated forty trials and the endpoint is twenty-eight-day mortality.

The consent problem

The ICU Is the Only Place Where Research Runs Without Permission.

A patient in septic shock or cardiac arrest cannot consent, has no time for a surrogate to be found, and needs the intervention in minutes. Regulation's answer is exception from informed consent — a narrow provision requiring community consultation, public disclosure, an independent monitor, and a life-threatening situation with no satisfactory alternative. It is the most ethically loaded machinery in clinical research.

It is also slow, public, and easy to get wrong: the community consultation is a genuine engagement obligation, not a press release. We build EFIC programs that hold up to the scrutiny they invite, because a trial that mishandles consent is not a trial with a protocol problem — it is a trial that should not have enrolled anyone.

A masked patient in a clinic being monitored by an electronic device
Consent by exception

Community consultation, public disclosure, an independent monitor. EFIC is machinery, and it is slow by design.

Forty trials, no drug

Sepsis Is Medicine's Most Expensive Unsolved Regulatory Problem.

More than forty phase 3 sepsis trials have failed. The pharmacology was often plausible; the trial design was the recurring defect — and each failure taught the same lesson.

A syndrome, not a disease

The population is the problem

Sepsis is a clinical definition covering wildly different biology — infection source, organism, host response, and organ trajectory all vary. Enrolling all of it dilutes any real effect to nothing.

28-day mortality

An endpoint that hides benefit

The conventional primary is binary, late, and dominated by competing causes of death. A drug can meaningfully help organ function and still not move it.

Timing is everything

Hours, not days

The therapeutic window may be measured in hours from onset, while enrolment machinery takes longer — so the drug is often given after the biology it targets has moved on.

Enrichment is the answer

Biology, not consensus criteria

Predictive enrichment on biomarkers or host-response phenotypes narrows the population to patients who can respond — and is where the field's credible hope now sits.

The FDA is not the obstacle here. The agency has been consistently open to enrichment strategies, adaptive designs, and alternative endpoints in sepsis; what it will not accept is a broad syndromic population and a hope that mortality moves. The programs worth funding define a biologically coherent population first and let the endpoint follow from it.

Six ICU worlds

Critical Care Regulates Drugs, Machines, and Algorithms Together.

The ICU is device-dense, drug-dense, and increasingly algorithm-dense — and the sickest patients in the building are the test population for all of it.

Enrichment

Sepsis & Septic Shock

Where forty failed trials made predictive enrichment and biologically coherent populations the only credible path.

ARDS · ventilation

Respiratory Failure

Ventilators, ECMO, and the drugs that have never beaten lung-protective ventilation — a field where the standard of care is very hard to improve.

Haemodynamics

Monitoring & Devices

Continuous monitoring, closed-loop systems, and the alarm-fatigue and human-factors questions the FDA now treats as safety issues.

Prediction

Clinical Decision Support

Sepsis prediction algorithms — where the CDS exemption's boundaries, real-world performance, and alert burden all get tested at once.

CRRT

Acute Kidney Injury

Continuous renal replacement in the ICU, and a drug field with no approved therapy and no accepted trial design.

Emergency use

Pandemic Readiness

EUAs, pre-EUA packages, and the hard lesson that authorisation is not approval and the data still has to arrive.

A detailed view of an oxygen outlet on a medical wall panel, used in healthcare facilities
A loan against evidence

EUA moves a product to the bedside in weeks. The confirmatory data still has to arrive — now, and in public.

What the pandemic taught

Emergency Use Authorisation Is Not a Shortcut. It Is a Debt.

COVID-19 ran the largest live experiment in emergency regulation ever conducted. EUAs moved products to bedsides in weeks on a may-be-effective standard — and then revocations, narrowed indications, and the collapse of several high-profile authorisations demonstrated what the standard actually meant. Authorisation was a loan against evidence, and the evidence came due.

The durable lesson for critical care sponsors is that an EUA obligates you to the confirmatory program you would have run anyway, on a harder schedule and in public. We plan emergency pathways with the full approval already designed behind them, because the authorisation is the easy part.

The hardest population in medicine

Everything That Makes the ICU Urgent Makes It Hard to Study.

Unconscious patients, hours-long windows, competing mortality, and a syndrome definition instead of a disease.

40+

phase 3 sepsis trials have failed. Almost none failed for lack of a plausible mechanism — they failed on population and endpoint.

EFIC

Exception from informed consent requires community consultation, public disclosure, and an independent monitor. It is slow, and it is the only way.

28 days

Mortality remains the conventional primary endpoint — binary, late, and capable of hiding a genuine benefit entirely.

Where critical care programs stall

Six Failure Modes We Are Brought In to Prevent.

The ICU punishes optimism about populations more than any other setting.

1

The syndrome as a population

Enrolling consensus-criteria sepsis without enrichment, and diluting a real effect in a biologically incoherent cohort.

2

EFIC underestimated

Treating community consultation as a formality, and meeting an IRB — or a journalist — who does not agree.

3

Mortality as the only endpoint

Building a program around 28-day mortality when the drug's benefit is organ function, and having nothing to show for it.

4

Enrolment window missed

A protocol whose screening and randomisation machinery takes longer than the therapeutic window it is aiming at.

5

EUA without a confirmatory plan

Taking emergency authorisation with no funded, enrolling confirmatory trial, then facing revocation in public.

6

CDS exemption misread

Assuming a sepsis prediction algorithm falls outside device regulation when its output and automation bias put it squarely inside.

People who have run trials in the ICU

Critical Care Regulatory Leadership for the Hardest Setting in Medicine.

Our critical care leads have designed enriched sepsis programs, run EFIC trials through community consultation, and taken ICU devices and EUAs through the agency.

A nurse in blue scrubs examines a medical monitor displaying vital signs in a hospital setting

"Sepsis has not failed for lack of good molecules. It has failed because we kept enrolling a definition instead of a disease. Fix the population and the endpoint stops being impossible."

The discipline we bring across sepsis, respiratory failure, ICU devices, and emergency pathways.

Predictive enrichment strategy EFIC & community consultation Sepsis & ARDS trial design Ventilator & monitoring devices EUA & pre-EUA packages Clinical decision support

Planning an ICU Trial? Define the Population Before the Endpoint.

Bring senior critical care regulatory leadership in while enrichment, consent, and the endpoint are still design decisions.

Senior-led. Embedded in your team. No junior hand-offs.