Publication

• Title: Hydrocortisone in Severe Community-Acquired Pneumonia
• Acronym: CAPE COD
• Year: 2023
• Journal published in: The New England Journal of Medicine
• Citation: Dequin P-F, Meziani F, Quenot J-P, et al. Hydrocortisone in severe community-acquired pneumonia. N Engl J Med. 2023;388(21):1931-1941.

Context & Rationale

• Background

  • Severe community-acquired pneumonia (CAP) can trigger intense pulmonary and systemic inflammation, leading to impaired gas exchange, sepsis, organ failure, and death.
  • Even in high-income settings, mortality is substantial, particularly among patients requiring any form of mechanical ventilation (reported in the CAPE COD background as reaching ~30% in such patients).
  • Earlier randomised trials of systemic glucocorticoids in CAP suggested improved clinical stability and shorter length of stay, but mortality effects were inconsistent and evidence quality varied.
  • A definitive, ICU-focused, blinded trial was needed to determine whether early, protocolised glucocorticoid therapy improves survival in severe CAP while clarifying safety trade-offs (e.g., infection, gastrointestinal bleeding, hyperglycaemia).
• Research Question/Hypothesis

  • In adults treated in ICU (or intermediate care) for severe CAP, does early intravenous hydrocortisone (with a predefined taper/stop algorithm) reduce all-cause mortality at 28 days compared with placebo, alongside standard antibiotics and supportive care?
• Why This Matters

  • Hydrocortisone is inexpensive and widely available; a mortality benefit would have immediate global relevance.
  • Preventing progression to invasive ventilation and shock could plausibly reduce downstream complications, ICU length of stay, and resource use.
  • High-quality evidence is central to resolving long-standing uncertainty and informing guideline recommendations on corticosteroids in severe CAP.

Design & Methods

• Research Question: Whether early, protocolised hydrocortisone improves 28-day survival in ICU-treated severe CAP versus placebo.
• Study Type: Multicentre, randomised, double-blind, placebo-controlled superiority trial conducted in 31 French centres (ICU / intermediate care), within the CRICS-TriGGERSep network.
• Population:
  • Setting: ICU (or intermediate care) patients with severe CAP; enrolment October 28, 2015 to March 11, 2020.
  • Core inclusion: Adult (≥18 years) with community-acquired pneumonia requiring ICU/intermediate care, after ≥1 antibiotic dose since hospital admission, and able to start study infusion within 24 hours after the first severity criterion.
  • Severe CAP definition (≥1 criterion): PSI >130; OR invasive/non-invasive mechanical ventilation (with PEEP ≥5 cm H₂O); OR high-flow oxygen therapy with FiO₂ ≥0.50 and PaO₂:FiO₂ <300; OR oxygen via partial rebreathing mask meeting protocol PaO₂ thresholds (intended to approximate PaO₂:FiO₂ <300).
  • Key exclusions (high-level): Septic shock at inclusion; influenza infection at inclusion; chronic/required systemic corticosteroids for another indication; important immunosuppression; pregnancy/breastfeeding; limitations of life-sustaining treatment (e.g., decision to withhold intubation).
  • Registration: ClinicalTrials.gov NCT02517489; EudraCT 2015-001239-19.
• Intervention:
  • Drug and delivery: Hydrocortisone, continuous intravenous infusion.
  • Initial dose: 200 mg/day for the first 4 days, started within 24 hours after the first severity criterion.
  • Adaptive duration/taper algorithm: If predefined clinical improvement at day 4, taper to 100 mg/day (days 5–6) then 50 mg/day (days 7–8) then stop; if not improved, continue 200 mg/day (days 1–7) then taper to 100 mg/day (days 8–11) and 50 mg/day (days 12–14) then stop.
  • Stopping rule: Study drug discontinued at ICU discharge (i.e., full planned taper not obligatory if discharged earlier).
• Comparison:
  • Placebo: Continuous intravenous infusion using an identical schedule (including the same taper/stop algorithm and ICU-discharge discontinuation).
  • Co-interventions: Standard care (antibiotics and supportive care) at clinician discretion; protocol prohibited systemic corticosteroids outside the trial regimen unless clinically mandated (and such use was captured as premature stopping reason).
• Blinding: Double-blind (participants, treating clinicians, and investigators); placebo infusion designed to maintain masking despite an adaptive duration/taper scheme.
• Statistics: Planned sample size 1,200 to detect a 7% absolute reduction in 28-day mortality (from 27% to 20%) with 80% power at a two-sided 5% significance level (group sequential design with interim monitoring); primary analysis was modified intention-to-treat (excluding patients with withdrawn consent).
• Follow-Up Period: Primary endpoint at day 28; key follow-up through day 90 (including survival and health-related quality of life measures).

Key Results

This trial was stopped early. Enrolment was halted after the second planned interim analysis (800 patients randomised), following a Data and Safety Monitoring Board recommendation; the supplementary material documents that the interim p value (0.0055) did not meet the prespecified stopping threshold and that practical/ethical considerations (including prolonged pandemic-related suspension) contributed to the decision.

• Mortality and deterioration: Hydrocortisone was associated with lower 28-day mortality and lower cumulative incidence of subsequent endotracheal intubation and vasopressor initiation.
• Resource-related outcomes: ICU and hospital discharge alive by day 28 occurred more frequently in the hydrocortisone group.
• Harms: Hospital-acquired infection and gastro-intestinal bleeding were similar; insulin requirements were higher with hydrocortisone.
• Prespecified subgroup (mechanical ventilation at baseline): Mortality day 28 was 19/178 (10.7%) vs 25/175 (14.3%) (difference −3.6 percentage points; 95% CI −9.6 to 2.3) among mechanically ventilated patients, and 6/222 (2.7%) vs 22/220 (10.0%) (difference −7.3 percentage points; 95% CI −12.6 to −2.0) among those not mechanically ventilated.

Internal Validity

• Randomisation and allocation:
  • Centralised randomisation with stratification by centre and by use of mechanical ventilation at baseline.
  • Allocation concealment and double-blinding plausibly minimised selection and performance bias.
• Drop out / exclusions:
  • 800 patients randomised; 795 analysed in the modified intention-to-treat population (400 hydrocortisone; 395 placebo).
  • Post-randomisation exclusion was limited to consent withdrawal (5 patients), reducing risk of attrition bias for the primary endpoint.
  • Day 90 mortality denominators were 388 and 389, indicating incomplete longer-term ascertainment; reasons for missingness were not reported in the main outcomes table.
• Performance / detection bias:
  • Blinding was feasible and clinically important given the potential for clinician behaviour to change with perceived steroid exposure.
  • Primary and key secondary endpoints were largely objective (mortality, intubation, vasopressor initiation, discharge alive).
• Protocol adherence:
  • Time from ICU admission to trial agent initiation: median 15.3 hours (hydrocortisone) vs 14.6 hours (placebo).
  • Actual duration of experimental treatment: median 5 days (IQR 3–8) vs 6 days (IQR 3–8).
  • Premature stopping of experimental treatment occurred in 318/400 (79.5%) vs 298/395 (75.4%), most commonly due to ICU discharge alive (261 vs 220) or death (24 vs 42).
• Baseline characteristics and severity:
  • Groups were well balanced for major prognostic variables (e.g., median age 63 vs 63; SAPS II 37 vs 36; SOFA 6 vs 6; PSI 121 vs 121).
  • Baseline respiratory support was comparable (mechanical ventilation 178/400 [44.5%] vs 175/395 [44.3%]).
  • Baseline vasopressor use (10.3% vs 12.9%) indicates some haemodynamic support at inclusion despite exclusion of septic shock, underscoring definitional nuances in “shock” versus vasopressor exposure.
• Timing:
  • Study drug initiation was designed to be early (within 24 hours of the first severity criterion), targeting the hypothesised window of inflammatory amplification.
  • Hospital-to-ICU time was short (median 5.5 vs 5.2 hours), supporting enrolment early in the ICU trajectory.
• Dose:
  • Initial hydrocortisone dose was fixed (200 mg/day), with protocolised extension or taper based on day 4 response.
  • Real-world exposure was shorter than the maximal planned 14-day course for many patients due to ICU discharge, making this a pragmatic test of an “early course” strategy.
• Separation of the variable of interest:
  • Open-label corticosteroid therapy (protocol deviation): 21/400 (5.3%) vs 30/395 (7.7%).
  • Insulin requirement (signal of biological separation): median daily insulin dose 35.5 vs 20.5 IU/day among patients receiving insulin therapy (median difference 8.7; 95% CI 4.0 to 13.8).
• Outcome assessment:
  • Mortality is robust; intubation and vasopressor endpoints are clinically anchored but can reflect local thresholds, making blinding and multicentre design important.
  • Hospital-acquired infection and gastro-intestinal bleeding were reported without central adjudication in the main report; this may reduce sensitivity for small between-group differences.
• Statistical rigour:
  • Group sequential design with interim analyses was prespecified; the trial stopped early after interim review.
  • Secondary endpoints were presented with confidence intervals not adjusted for multiplicity, limiting inferential strength for individual secondary comparisons.

Conclusion on Internal Validity: Internal validity is moderate-to-strong: randomisation, concealment, and blinding support causal inference for objective endpoints, and contamination was limited; however, early stopping (outside the prespecified interim threshold) and incomplete 90-day ascertainment introduce uncertainty regarding the stability and magnitude of the estimated effect.

External Validity

• Population representativeness:
  • Represents ICU-treated severe CAP in a high-resource system, including both mechanically ventilated patients and those requiring high-flow oxygen therapy.
  • Key exclusions (notably septic shock at inclusion, influenza at inclusion, and important immunosuppression) restrict applicability to common “real-world” severe CAP phenotypes.
  • Protocol-required early initiation (within 24 hours of the first severity criterion) may be challenging where ICU access or early severity recognition is delayed.
• Applicability:
  • The regimen (continuous infusion hydrocortisone with response-adapted taper) is feasible in most ICUs, but requires reliable infusion delivery and glycaemic monitoring capacity.
  • Generalisation is most defensible for severe CAP patients similar to the trial population (early ICU presentation, no septic shock at inclusion, and no contraindication to corticosteroids).
  • External validity is uncertain for viral pneumonias and for severe CAP complicated by septic shock at presentation, where other corticosteroid evidence and sepsis guidance may dominate decision-making.

Conclusion on External Validity: Generalisability is moderate, strongest for severe CAP patients admitted early to ICU without septic shock at baseline; applicability is limited for shock-at-presentation, viral pneumonia, and substantially immunosuppressed populations.

Strengths & Limitations

• Strengths:
  • Double-blind, randomised design across 31 centres, reducing bias and enhancing internal validity.
  • Early initiation strategy aligned with hypothesised biology (attenuating early inflammatory escalation).
  • Clinically important, mostly objective outcomes (mortality, intubation, vasopressors, discharge alive).
  • Protocolised dosing and taper algorithm supports reproducibility and mitigates ad hoc steroid exposure.
• Limitations:
  • Stopped early after interim review, with a decision-making rationale that may influence precision and the stability of effect estimates.
  • Observed placebo mortality (11.9%) was substantially lower than assumed in planning (27%), raising questions about baseline risk, selection, and transportability to higher-risk cohorts.
  • Exclusion of septic shock and influenza at inclusion narrows applicability to common ICU CAP populations.
  • Infection and bleeding outcomes were not centrally adjudicated in the main report, limiting detection of subtle safety signals.

Interpretation & Why It Matters

• Clinical signal

  Hydrocortisone was associated with lower 28-day mortality (6.2% vs 11.9%) and less progression to invasive ventilatory and haemodynamic support, suggesting a meaningful effect on early organ-failure trajectory.
• Mechanistic plausibility

  The pattern of benefit (reduced intubation and vasopressor initiation) is consistent with attenuation of inflammatory lung injury and systemic inflammatory vasoplegia, rather than solely late “rescue” effects.
• Practical implementation

  Adoption requires early identification of severe CAP, prompt initiation (within 24 hours of severity), avoidance in viral pneumonias where harm is plausible, and structured monitoring for hyperglycaemia (higher insulin exposure was observed).

Controversies & Subsequent Evidence

• Accompanying editorial perspective¹
  • Framed CAPE COD as a pivotal test of an early, ICU-targeted steroid strategy and highlighted the potential for hydrocortisone to be a pragmatic adjunct in carefully selected severe CAP.
  • Emphasised that trial eligibility and exclusions define the clinical space where the evidence is most coherent (early severe CAP requiring ICU-level respiratory support, without influenza or septic shock at inclusion).
• Correspondence: interpretive challenges and phenotype questions²
  • Shock evolution and open-label steroids: One letter noted that while septic shock was excluded at inclusion, haemodynamic deterioration occurred more often in placebo, and argued that prohibition of open-label glucocorticoids could have amplified between-group differences if shock developed after randomisation.
  • Competing phenotype (COPD/asthma overlap): Another letter highlighted that ~30% of placebo patients had COPD/asthma and raised concern that withholding steroids could disadvantage those with concurrent obstructive airway exacerbation.
  • Author reply (post hoc analysis): Trialists reported post hoc day-28 mortality differences in patients with COPD/asthma (6/102 [5.9%] vs 14/112 [12.5%]; difference −6.6 percentage points; 95% CI −14.3 to 1.1) and without COPD/asthma (19/298 [6.4%] vs 33/283 [11.7%]; difference −5.3 percentage points; 95% CI reported in the reply), supporting a broadly consistent direction of effect.
• How does this align with other severe CAP steroid trials?
  • The correspondence and editorial contrasted CAPE COD with a similarly sized ICU trial of low-dose methylprednisolone in severe CAP that did not demonstrate a mortality benefit (Intensive Care Med 2022;48:1009–1023).³
  • Across the steroid-in-CAP literature, differences in corticosteroid choice (hydrocortisone vs methylprednisolone), dosing strategy (continuous infusion vs intermittent), permitted rescue steroids, and patient phenotype (shock, COPD overlap, viral exclusions) plausibly contribute to heterogeneous results.
• Guidelines and practice uptake
  • European joint guideline recommendations for severe CAP management (ERS/ESICM/ESCMID/ALAT) provide a structured framework for severity assessment and ICU management; corticosteroid decisions must be integrated with phenotype (e.g., shock, viral pneumonia) and co-morbidity considerations.⁴
  • The 2024 SCCM focused update addresses corticosteroid use across sepsis, ARDS, and community-acquired pneumonia, synthesising post-2018 evidence to guide contemporary practice decisions in CAP populations relevant to critical care.⁵

Summary

• CAPE COD was a multicentre, double-blind RCT in 31 French ICUs/intermediate care units testing early hydrocortisone (200 mg/day with protocolised taper) versus placebo in severe CAP.
• Hydrocortisone reduced 28-day mortality (6.2% vs 11.9%; absolute difference −5.6 percentage points; P=0.006) and reduced subsequent intubation and vasopressor initiation among those not receiving these supports at baseline.
• ICU and hospital discharge alive by day 28 were more frequent with hydrocortisone; ventilator-/vasopressor-free day medians were similar, reflecting the complexities of “free days” metrics when mortality differs.
• Major safety signals were not observed for hospital-acquired infection or gastro-intestinal bleeding, but insulin requirements were higher (median 35.5 vs 20.5 IU/day among insulin-treated patients).
• Interpretation must account for early stopping and for phenotype boundaries (notably exclusion of septic shock and influenza at inclusion), which shape both internal and external validity.

Overall Takeaway

CAPE COD is a landmark ICU-focused, double-blind trial showing that early, protocolised hydrocortisone can reduce 28-day mortality and key deterioration endpoints in selected patients with severe community-acquired pneumonia. Its impact is strongest in defining a feasible, broadly implementable steroid strategy with an apparently acceptable short-term safety profile, while also clarifying that patient phenotype (shock, viral pneumonia, COPD overlap) and early-trial stopping considerations remain central to how the results should be applied.

Bibliography

• 1Metlay JP, Waterer GW. Time to treat severe community-acquired pneumonia with steroids? N Engl J Med. 2023;388(21):2001-2002.
• 2Friedrich JO, Gouvêa Bogossian E, van Geffen WH, Zijlstra GJ, Dequin PF, Meziani F, et al. Hydrocortisone in severe community-acquired pneumonia. N Engl J Med. 2023;389(7):671-673.
• 3Meduri GU, Shih MC, Bridges L, et al. Low-dose methylprednisolone treatment in critically ill patients with severe community-acquired pneumonia. Intensive Care Med. 2022;48:1009-1023.
• 4Torres A, Martin-Loeches I, Niederman MS, et al. Guidelines for the management of severe community-acquired pneumonia: recommendations from the European Respiratory Society, European Society of Intensive Care Medicine, European Society of Clinical Microbiology and Infectious Diseases, and Asociación Latinoamericana del Tórax. Intensive Care Med. 2023;49:615-632.
• 5Chaudhuri 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.