Critical care Topic

Start the steroid?

Six bedside calls on sepsis, septic shock and severe pneumonia. Two decades of trials have alternated between a survival benefit and no difference; the finding that repeats is faster shock reversal.

Reviewed October 2026 · verify against current guidelines

Steroids · 1 of 9

Start the steroid?

Sepsis · septic shock · severe pneumonia. Make your call on each — give or don't — before reading on. The answers are at the end.

Your call

Septic shock, norepinephrine 0.3 µg/kg/min for 6 h
Sepsis, lactate 3, no vasopressors
Severe pneumonia on high-flow oxygen, no shock
Influenza pneumonia, intubated
Adding fludrocortisone to hydrocortisone
Methylprednisolone 1 g pulse for shock

Steroids · 2 of 9

It restores the pressor response.

In septic shock, the cortisol response is often inadequate for the degree of illness. Hydrocortisone makes vessels respond to catecholamines again. It does not treat the infection.

Where it acts

Septic shock (inflammation) → low cortisol for the illness → vessels ignore norepinephrine. Hydrocortisone acts on that last step: it restores adrenergic receptor response, reduces nitric oxide vasodilation and dampens the cytokine response.

It does	It doesn't
Shorten time on vasopressors	Prevent shock in sepsis without it
Shorten mechanical ventilation in some trials	Reliably change survival — the question this series covers

The two drugs

Hydrocortisone	IV, 200 mg a day as bolus or infusion. Glucocorticoid with some mineralocorticoid effect.
Fludrocortisone	Enteral only, 50 µg a day. Pure mineralocorticoid; absorption in shock varies widely.

Why testing cortisol fell away — Annane 2002 found benefit in corticotropin nonresponders; CORTICUS found the test did not predict who benefited. Later trials skipped it: ADRENAL and APROCCHSS enrolled on vasopressor need alone.

Steroids · 3 of 9

The pendulum keeps swinging.

Each large trial since 2002 has pushed practice one way, and the next has pushed it back. Mortality results alternate; faster shock reversal is the one finding that repeats.

Major trials, mortality result

Year	Trial	Mortality	Population and finding
2002	Annane	Lower	300, shock. Hydrocortisone and fludrocortisone; benefit in nonresponders to corticotropin.
2008	CORTICUS	No difference	499, shock. Faster reversal in those who recovered; more superinfection.
2016	HYPRESS	No difference	380, sepsis without shock. Did not prevent shock.
2018	ADRENAL	No difference	3,800, shock on ventilation. Hydrocortisone infusion.
2018	APROCCHSS	Lower	1,241, severe shock. Hydrocortisone and fludrocortisone.
2023	CAPE COD	Lower	800, severe pneumonia without shock.
2025	REMAP-CAP	No difference	658, severe pneumonia. Stopped for futility.

What the guidelines say

SSC 2026	Suggests IV corticosteroids in septic shock; conditional, low certainty. Based on 45 RCTs, 9,543 patients.
SCCM 2024	Suggests corticosteroids in septic shock; conditional, low certainty.

Why the trials disagree — different patients: control mortality ranged from about 10% to over 60%, depending on who was enrolled. Different regimens: bolus or infusion, with or without fludrocortisone, 4 to 8 days.

Steroids · 4 of 9

2018: one year, two answers.

Two trials in septic shock, published in the same issue. One found no survival benefit; the other found 6 fewer deaths per 100 patients.

90-day mortality

	ADRENAL · 3,800	APROCCHSS · 1,241
Steroid vs placebo	27.9% vs 28.8%	43.0% vs 49.1%
Effect	OR 0.95 (0.82–1.10). Faster shock resolution, shorter ventilation, fewer transfusions.	RR 0.88 (0.78–0.99). More vasopressor-free days; more hyperglycemia.

How the trials differed

	ADRENAL	APROCCHSS
Entry	Any vasopressor for 4 h, ventilated	Norepinephrine ≥ 0.25 µg/kg/min for 6 h
Regimen	200 mg/day infusion	50 mg every 6 h, plus fludrocortisone 50 µg
Duration	7 days or ICU discharge	7 days, no taper

The leading explanations — severity: placebo mortality was 49% in APROCCHSS and 29% in ADRENAL; sicker patients may have more to gain. Fludrocortisone: only APROCCHSS gave it, and no large trial has tested it head to head.

Bolus or infusion — the trials also differed here; neither design tested which delivery is better.

Steroids · 5 of 9

Septic shock: faster off pressors.

Pooled across trials, steroids reliably speed shock reversal. Any mortality benefit is small, and the pooled confidence interval crosses no effect.

Short-term mortality, steroid against placebo — risk or odds ratio, 95% CI

ADRENAL	0.95
APROCCHSS	0.88
45 trials pooled	0.92

Shock reversal, day 7	RR 1.29 — high certainty, in the SSC 2026 evidence review.
Dose ceiling in pooled data	260 mg a day of hydrocortisone. No added benefit above it.

What the guidelines say

SCCM 2024	Recommends against high dose, short courses: over 400 mg/day for less than 3 days.
SSC 2026	No separate recommendation; pooled RR 1.09 in sepsis without shock.

How to read a 0.92 — for it: a cheap drug, a consistent direction across trials, and a benefit that patients notice: less time on pressors. Against it: the upper limit is 1.01, and the largest trial alone shows no difference.

Steroids · 6 of 9

Fludrocortisone: add it or not?

The trials with a survival benefit both gave fludrocortisone. No large trial has compared hydrocortisone with and without it.

FLUDRESS 2024 · 153 · 28-day mortality by dose

Fludrocortisone	None	50 µg	100 µg	200 µg
28-day mortality	24%	17%	11%	11%

Phase 2, open label, stopped early. Not powered for mortality. Time to shock resolution: 3 days in every group.

Pooled, 7 studies · 90,756 patients

In-hospital mortality 40.8% with the combination vs 42.8% in controls (OR 0.86). Nearly all patients come from one observational US database study.

What the guidelines say

SSC 2026	No recommendation. 63% of the panel never add it; 10% usually or always do.
SCCM 2024	Recommends corticosteroids without naming a molecule or combination.

The two positions — add it: it was in both trials that cut mortality, costs little, and observational data lean in its favor. Skip it: 200 mg of hydrocortisone already has mineralocorticoid effect, and enteral absorption in shock is unreliable.

Steroids · 7 of 9

Severe pneumonia: back in doubt.

CAPE COD halved deaths and guidelines moved to recommend steroids. Then REMAP-CAP stopped its hydrocortisone arm for futility, with a nonsignificant excess of deaths on steroid.

Mortality, steroid against control

Trial	Population	Direction
CAPE COD · 2023	800, no shock, 28 d	Lower on steroid — about 6% vs 12%
ESCAPe · 2022	586, methylpred, 60 d	Slightly lower on steroid — about 16% vs 18%
REMAP-CAP · 2025	658, half in shock, 90 d	Higher on steroid — about 15% vs 10%

What the guidelines say

SCCM 2024	Recommends corticosteroids in severe bacterial CAP; strong, moderate certainty.
ATS 2025	Suggests corticosteroids in severe CAP, not due to influenza; conditional, low quality.

Why REMAP-CAP may differ — different patients: CAPE COD excluded shock and influenza; REMAP-CAP had almost half in shock. A weaker design: open label, 536 vs 122 after a randomization error, and 23% of controls got steroids anyway.

Influenza — excluded from CAPE COD. Observational data link steroids to higher mortality; certainty is low.

Steroids · 8 of 9

What it can cost.

At 200 mg a day for a week, the common harms are metabolic. The serious ones depend on the infection and the dose.

Harms

Hyperglycemia	RR 1.19 in pooled trials. More frequent in APROCCHSS; more insulin use in CAPE COD.
Hypernatremia	RR 1.64 in pooled trials, moderate certainty. The largest relative increase.
Weakness	Neuromuscular weakness: RR 1.13 (0.48–2.65), very low certainty. Effect uncertain.
Superinfection	More new sepsis episodes in CORTICUS. Not seen in ADRENAL or APROCCHSS.
Influenza	Observational studies link steroids to higher mortality (OR 3.90). Low certainty, but no trial shows benefit.

Dose limits in the guidelines

SCCM 2024	Strong recommendation against over 400 mg/day hydrocortisone equivalent for under 3 days.
SSC 2026	No added benefit above 260 mg/day. 90% of the panel give 200 mg, mostly as intermittent doses.

Taper or stop: unsettled — APROCCHSS stopped abruptly after 7 days, without tapering. The SSC 2026 panel is split: 61% stop without a taper; 39% taper. No trial compares the two.

Steroids · 9 of 9

The answers, and the dose.

Steroids fit vasopressor-dependent shock. Outside it, the case is weaker or absent.

Slide 1 answers

Give	Shock on norepinephrine 0.3 — APROCCHSS entry level; faster shock reversal.
Don't	Sepsis without vasopressors — HYPRESS: did not prevent shock.
Give	Severe pneumonia, no shock — CAPE COD, SCCM 2024, ATS 2025; REMAP-CAP dissents.
Don't	Influenza pneumonia — untested in trials, harm signal in observational data.
Consider	Adding fludrocortisone — in both positive trials, never tested head to head.
Don't	Methylprednisolone 1 g pulse — SCCM strong recommendation against.

Trial regimens, adults

Septic shock	Hydrocortisone 50 mg IV every 6 h, or 200 mg/day infusion, for 7 days.
If adding	Fludrocortisone 50 µg enterally once a day, for 7 days.
Severe CAP	CAPE COD: hydrocortisone 200 mg/day infusion for 4 or 8 days, then taper.

The deciding questions — on vasopressors after fluids? If yes, the septic shock evidence applies. If no, steroids have trial support only in severe bacterial pneumonia. Could this be influenza? If so, the pneumonia evidence does not apply.

Educational reference only — not medical advice. Verify against local protocol.

Sources

Verify against current guidelines and local protocol before acting.

Annane 2002 — hydrocortisone and fludrocortisone in septic shock (n = 300).
CORTICUS 2008 — hydrocortisone in septic shock (n = 499).
HYPRESS 2016 — hydrocortisone in sepsis without shock (n = 380).
ADRENAL 2018 — hydrocortisone infusion in septic shock on ventilation (n = 3,800).
APROCCHSS 2018 — hydrocortisone plus fludrocortisone in severe septic shock (n = 1,241).
ESCAPe 2022 — methylprednisolone in severe pneumonia (n = 586).
CAPE COD 2023 — hydrocortisone in severe community-acquired pneumonia without shock (n = 800).
FLUDRESS 2024 — fludrocortisone dose in septic shock, phase 2 (n = 153).
REMAP-CAP 2025 — hydrocortisone in severe pneumonia (n = 658).
Surviving Sepsis Campaign guidelines 2026.
SCCM corticosteroid guideline 2024.
ATS community-acquired pneumonia guideline 2025.

Downloads

Every card behind this topic, as published — view full size or download.

Carousel9 slides