Publication

• Title: Effect of Treatment With Low Doses of Hydrocortisone and Fludrocortisone on Mortality in Patients With Septic Shock
• Acronym: Not applicable
• Year: 2002
• Journal published in: JAMA
• Citation: Annane D, Sébille V, Charpentier C, Bollaert P-E, François B, Korach J-M, et al. Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock. JAMA. 2002;288(7):862-871.

Context & Rationale

• Background

  • Earlier high-dose corticosteroid strategies in sepsis had not improved survival and raised safety concerns.
  • Septic shock was conceptualised as a state of “relative adrenal insufficiency” in a subset of patients, potentially driving catecholamine-refractory vasoplegia.
  • Physiologic and small clinical studies suggested “stress-dose” hydrocortisone could improve vascular responsiveness and shorten shock duration.
  • Mineralocorticoid dysfunction (and potential benefit of fludrocortisone) was biologically plausible but clinically uncertain.
• Research Question/Hypothesis

  • In adults with early, catecholamine-dependent septic shock, does 7-day “replacement” therapy with hydrocortisone plus fludrocortisone (vs placebo) improve 28-day survival distribution in patients with impaired adrenal responsiveness to corticotropin stimulation?
  • Secondary: does the regimen accelerate haemodynamic recovery (vasopressor withdrawal), and does benefit vary by corticotropin response status (responder vs nonresponder)?
• Why This Matters

  • Septic shock carried high mortality with few interventions showing clear patient-important benefit.
  • A reproducible strategy to reverse shock and improve survival would be practice-changing and would shape future trial design (including biomarker-stratified treatment effects).
  • Clarifying the role (or futility) of adrenal testing for “targeted” steroid therapy had major implications for bedside decision-making and guideline recommendations.

Design & Methods

• Research Question: In adults with early catecholamine-dependent septic shock, does 7-day therapy with hydrocortisone plus fludrocortisone (vs placebo), initiated after a corticotropin test, improve 28-day survival distribution in corticotropin nonresponders?
• Study Type: Multicentre, randomised, double-blind, placebo-controlled, parallel-group trial in 19 adult ICUs (France); investigator-initiated; randomisation stratified by centre (block size 4); enrolment April 1995 to March 1999.
• Population:
  • Setting: Adult ICUs; all enrolled patients required mechanical ventilation at inclusion.
  • Inclusion (septic shock definition): Suspected/proven infection plus systemic response and early shock with hypotension and vasopressor dependence, meeting all of:
  • Suspected infection.
  • Temperature >38.3 °C or <35.6 °C.
  • Heart rate >90/min.
  • Systolic arterial pressure <90 mm Hg despite adequate fluid resuscitation.
  • Need for vasopressor support for ≥1 hour: dopamine ≥15 µg/kg/min and/or epinephrine or norepinephrine >0.2 µg/kg/min; plus (after protocol amendment) arterial lactate >2 mmol/L.
  • Mechanical ventilation.
  • Timing: Randomisation within 3 hours of shock onset (protocol later amended to allow up to 8 hours).
  • Key exclusions: Corticosteroids within prior 6 months; etomidate within 6 hours prior to inclusion; recent ketoconazole, rifampin, phenytoin, or phenobarbital use; other exclusions as per trial protocol (Not reported in detail in the published manuscript).
  • Adrenal responsiveness strata (measured pre-randomisation): 250 µg corticotropin (tetracosactrin) stimulation test; “nonresponse” defined as increase in serum cortisol ≤9 µg/dL.
• Intervention:
  • Hydrocortisone 50 mg IV every 6 hours (total 200 mg/day) for 7 days.
  • Fludrocortisone 50 µg enterally once daily for 7 days.
  • Both components accompanied by matched placebos (IV and enteral) to preserve blinding.
• Comparison:
  • Matched IV placebo every 6 hours plus matched enteral placebo once daily for 7 days.
  • All other care (antibiotics, fluids, vasopressors, ventilation and supportive therapies) provided per treating teams; rescue open-label corticosteroid use not reported.
• Blinding: Double-blind (patients, clinicians, investigators, and outcome assessors); study drugs and placebos indistinguishable and dispensed in sequentially numbered boxes.
• Statistics: Power calculation: a total of 270 patients planned to detect a 20% absolute difference in 28-day mortality among corticotropin nonresponders with 90% power using a one-sided 5% significance level (assumptions included 95% mortality in placebo nonresponders and 40% nonresponder frequency); two interim analyses planned using O’Brien–Fleming stopping boundaries; primary analysis used time-to-event methods (Cox proportional hazards) with pre-specified subgrouping by corticotropin response; analyses were primarily intention-to-treat with a single post-randomisation consent withdrawal excluded.
• Follow-Up Period: 28 days (primary); additional follow-up to ICU discharge, hospital discharge, and 1 year.

Key Results

This trial was not stopped early. Two interim analyses were planned (after 114 and 220 inclusions) with O’Brien–Fleming boundaries; recruitment continued to 300 randomised (299 analysed).

• Benefit was concentrated in pre-specified corticotropin nonresponders: 28-day mortality 53% vs 63% (Adjusted OR 0.54; 95% CI 0.31 to 0.97; P=0.04), alongside faster vasopressor withdrawal (median 7 vs 10 days; HR 1.91; 95% CI 1.29 to 2.84; P=0.001).
• Among corticotropin responders, there was no evidence of survival benefit (28-day mortality 61% vs 53%; Adjusted OR 0.97; 95% CI 0.32 to 2.99; P=0.96), and proportional hazards assumptions were not supported for responders (log-rank P=0.81 for survival distribution).
• Predefined steroid-related adverse events were similar (21% vs 22%; P=0.89), with no signal for excess superinfection (12% vs 13%) or gastrointestinal bleeding (7% vs 5%).

Internal Validity

• Randomisation and Allocation: Central randomisation list (blocks of 4), stratified by centre; sequentially numbered drug boxes preserved allocation concealment.
• Drop out / exclusions: 300 randomised; 1 withdrew consent post-randomisation (excluded); 299 analysed; 1 patient died before receiving study drug but remained included in analyses.
• Performance/Detection Bias: Double-blind with matched IV and enteral placebos; primary outcomes objective (mortality; time to vasopressor withdrawal).
• Protocol Adherence: Study drug administration was near-complete: placebo 148/149 received study drug; steroid arm 150/150 received study drug (as reported); 7-day fixed regimen minimised exposure variability.
• Baseline Characteristics: Broadly similar severity and haemodynamic profile at inclusion (e.g., SAPS II 57 ± 18 vs 60 ± 18; LOD 7.8 ± 3.1 vs 7.9 ± 2.7; lactate 4.3 ± 3.4 vs 4.6 ± 3.0).
• Heterogeneity: Multicentre (19 ICUs) with a clinically heterogeneous septic shock cohort; the pre-specified effect modification by corticotropin response was central to the design rather than a post hoc subgroup.
• Timing: Early enrolment by design (≤3 hours from shock onset, later ≤8 hours); time from shock onset to vasopressor initiation 3.5 ± 2.2 vs 3.7 ± 2.2 hours; time on vasopressor before study drug 4.1 ± 3.0 vs 4.1 ± 2.4 hours.
• Dose: Hydrocortisone 200 mg/day plus fludrocortisone 50 µg/day for 7 days; regimen consistent with “stress-dose” glucocorticoid plus mineralocorticoid replacement.
• Separation of the Variable of Interest: Clear haemodynamic separation demonstrated by vasopressor withdrawal metrics (all patients: median 7 vs 9 days; HR 1.54; 95% CI 1.10 to 2.16; P=0.01; nonresponders: median 7 vs 10 days; HR 1.91; 95% CI 1.29 to 2.84; P=0.001).
• Key Delivery Aspects: Antibiotic adequacy was high and similar (appropriate antibiotic therapy 91% vs 95%); time to appropriate antibiotics 7.1 ± 10.4 vs 6.0 ± 7.3 hours.
• Crossover / adjunctive therapies: Open-label corticosteroid crossover and detailed co-intervention imbalances (e.g., glycaemic control protocols, sedation strategies) were not reported.
• Outcome Assessment: Mortality at 28 days, ICU, hospital, and 1 year provides robust patient-centred endpoints; vasopressor withdrawal is clinically meaningful but can be influenced by local practice (though the effect size was large in nonresponders).
• Statistical Rigor: Interim analysis plan stated (O’Brien–Fleming); primary time-to-event analyses used Cox models; proportional hazards assumption was not supported among responders (log-rank used), which appropriately limits inference for that subgroup.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given concealed randomisation, double blinding, objective endpoints, near-complete follow-up, and clear intervention separation on haemodynamic outcomes; the main threats are the reliance on a biologically defined subgroup (corticotropin nonresponse) and limited reporting of co-interventions and crossover.

External Validity

• Population Representativeness: Enrolled patients were very sick (catecholamine-dependent shock with universal mechanical ventilation), representing a high-acuity ICU septic shock phenotype rather than the full spectrum of septic shock.
• Important exclusions: Recent steroids and recent etomidate exposure were excluded, limiting direct applicability to settings where etomidate is commonly used for intubation in septic shock.
• Era effects: Trial conduct (1995–1999) predates modern sepsis bundles, contemporary ventilation practices, and changes in vasopressor strategies; absolute event rates may not translate directly.
• Intervention feasibility: Hydrocortisone bolus dosing is feasible; enteral fludrocortisone depends on gastrointestinal absorption and access (but is simple and inexpensive).
• Applicability across systems: Likely most applicable to resource-variable ICUs treating severe septic shock with persistent vasopressor requirement; less directly applicable to milder shock or earlier emergency department populations.

Conclusion on External Validity: Generalisability is moderate: the regimen is pragmatic, but the cohort was unusually severe and the trial pre-dates contemporary sepsis care, so translation should focus on patients with ongoing vasopressor-dependent shock rather than all-comers.

Strengths & Limitations

• Strengths: Rigorous double-blind randomised design; early intervention in clearly defined catecholamine-dependent shock; pre-specified biologically defined subgroup; objective primary endpoint (time-to-death distribution); clinically meaningful haemodynamic endpoint with large separation; near-complete follow-up to 1 year.
• Limitations: Primary inference depends on corticotropin nonresponse phenotype with a higher-than-anticipated prevalence (≈76–77%); sample size assumptions (including a one-sided design and very high expected mortality) may limit interpretability; responder subgroup small and statistically unstable; co-interventions and crossover/open-label steroid use not fully reported; sepsis care has evolved since the late 1990s.

Interpretation & Why It Matters

• Clinical signal

  In a very sick, early vasopressor-dependent septic shock cohort, hydrocortisone plus fludrocortisone improved haemodynamic recovery and survival distribution in corticotropin nonresponders, without an excess of predefined steroid-related complications.
• Mechanistic framing

  The trial operationalised “relative adrenal insufficiency” as a treatment-effect modifier, shaping subsequent decades of debate about whether biomarkers (corticotropin response) should direct steroid therapy.
• Modern practice relevance

  Later trials and guidelines moved practice towards using intravenous hydrocortisone for persistent vasopressor-dependent shock without routine corticotropin testing; the incremental role of fludrocortisone remains debated but is supported by later RCT evidence.

Controversies & Subsequent Evidence

• The accompanying editorial emphasised both the importance of the survival signal and the uncertainty around generalisability, the corticotropin test as a gatekeeper for treatment, and whether mineralocorticoid supplementation was a necessary component of benefit.¹
• Correspondence challenged aspects of interpretation, including the reliance on corticotropin-defined “relative adrenal insufficiency” and the implications of withholding corticosteroids in patients suspected to have adrenal dysfunction; these debates presaged later moves away from using the corticotropin test to direct therapy.²
• Subsequent large RCTs evaluating hydrocortisone without routine mineralocorticoid co-administration produced mixed mortality findings while consistently showing faster shock resolution; importantly, the corticotropin response did not reliably identify a subgroup with differential benefit in later work (e.g., CORTICUS).³
• ADRENAL (hydrocortisone 200 mg/day by infusion) did not reduce 90-day mortality but shortened time to shock resolution and ICU discharge, reinforcing haemodynamic benefit without definitive survival improvement in an international cohort.⁴
• APROCCHSS (hydrocortisone plus fludrocortisone) reported improved 90-day survival and more days alive and free of vasopressors, renewing interest in mineralocorticoid supplementation and highlighting that regimen choice (hydrocortisone alone vs combined therapy) may matter in severe shock phenotypes.⁵
• Modern international guidelines generally support intravenous corticosteroids for adults with septic shock and an ongoing vasopressor requirement (conditional/weak recommendations), and do not recommend routine corticotropin testing to guide therapy.⁶⁷⁸
• Systematic reviews/meta-analyses incorporating post-2018 trials generally find that corticosteroids reduce time to shock reversal and may modestly reduce mortality, with residual heterogeneity across dosing strategies and the addition of fludrocortisone.⁹

Summary

• Multicentre, double-blind RCT in 19 French ICUs (1995–1999) randomising 300 adults with early catecholamine-dependent septic shock requiring mechanical ventilation to 7 days of hydrocortisone (200 mg/day) plus fludrocortisone (50 µg/day) vs placebo.
• Primary endpoint (28-day survival distribution) was pre-specified in corticotropin nonresponders and improved with steroids (HR 0.67; 95% CI 0.47 to 0.95; P=0.02), with 28-day mortality 53% vs 63% (Adjusted OR 0.54; 95% CI 0.31 to 0.97; P=0.04).
• Haemodynamic recovery was faster, especially in nonresponders (median vasopressor withdrawal 7 vs 10 days; HR 1.91; 95% CI 1.29 to 2.84; P=0.001).
• No survival benefit was demonstrated in corticotropin responders, and responder subgroup analyses were statistically unstable (non-proportional hazards; log-rank used).
• Predefined steroid-related adverse events were similar between groups (21% vs 22%; P=0.89) with no excess superinfection or GI bleeding reported.

Overall Takeaway

The Annane 2002 trial is “landmark” because it reintroduced physiologic-dose corticosteroids into modern septic shock care, linking a clinically meaningful haemodynamic effect to a survival signal in a pre-specified biologically defined subgroup. Subsequent RCTs refined (and partly challenged) its mechanistic framing—particularly the value of corticotropin testing—while preserving the core message that corticosteroids can accelerate shock resolution, and may improve survival depending on population severity and regimen.

Bibliography

• Abraham E, Evans T. Corticosteroids and septic shock. JAMA. 2002;288(7):886-887.
• Opal SM. Corticosteroids for patients with septic shock. JAMA. 2003;289(1):41-42.
• Sprung CL, Annane D, Keh D, Moreno R, Singer M, Freivogel K, et al. Hydrocortisone therapy for patients with septic shock. N Engl J Med. 2008;358(2):111-124.
• Venkatesh B, Finfer S, Cohen J, Rajbhandari D, Arabi Y, Bellomo R, et al. Adjunctive glucocorticoid therapy in patients with septic shock. N Engl J Med. 2018;378(9):797-808.
• Annane D, Renault A, Brun-Buisson C, Megarbane B, Quenot JP, Siami S, et al. Hydrocortisone plus fludrocortisone for adults with septic shock. N Engl J Med. 2018;378(9):809-818.
• Evans L, Rhodes A, Alhazzani W, et al. Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock 2021. Crit Care Med. 2021;49(11):e1063-e1143.
• Pastores SM, Annane D, Rochwerg B, et al. Guidelines for the diagnosis and management of critical illness-related corticosteroid insufficiency (CIRCI) in critically ill patients (Part II). Crit Care Med. 2017;45(12):2078-2088.
• Chaudhuri D, Nei AM, Rochwerg B, et al. Guidelines on use of corticosteroids in sepsis, acute respiratory distress syndrome, and community-acquired pneumonia: 2024 focused update. Crit Care Med. 2024;52(5):e219-e233.
• Rochwerg B, Oczkowski SJW, Siemieniuk RAC, Agoritsas T, Belley-Cote E, D’Aragon F, et al. Corticosteroids in sepsis: an updated systematic review and meta-analysis. Crit Care Med. 2018;46(9):1411-1420.