Torres: Effect of Corticosteroids on Treatment Failure Among Hospitalized Patients with Severe Community-Acquired Pneumonia and High Inflammatory Response

Torres. JAMA 2015; 313(7):677-687. doi:10.1001/jama.2015.88

Clinical Question

  • In patients with severe community-acquired pneumonia and high inflammatory response, does the use of corticosteroids compared to placebo improve outcome?


  • Multi-centre
  • Double-blind, randomised-controlled trial
  • Sample size calculation
    • 80% power to detect absolute risk reduction of 20% in treatment failure (assuming placebo group failure rate of 35%)
    • Sample size n = 60
    • Pre-specified interim analysis resulted in no change in sample size target


  • 3 Spanish teaching hospitals
  • June 2004 – February 2012


  • Inclusion
    • Age > 18
    • Clinical symptoms suggestive of community-acquired pneumonia (CAP)
    • New chest radiographic infiltrates
    • Severe CAP as defined by either/or
      • Modified American Thoracic Society criteria for severe CAP
      • Pneumonia Severity Index risk class 4
    • C-reactive Protein (CRP) > 150mg/L
  • Exclusion
    • Prior treatment with systemic corticosteroids
    • Nosocomial pneumonia
    • Reported severe immunosuppression
    • Pre-existing medical condition with a life expectancy of less than 3 months
    • Uncontrolled diabetes mellitus
    • Major gastrointestinal bleed within 3 months
    • Condition requiring acute treatment with > 1mg/kg/day of methylprednisolone equivalent
    • H1N1 influenza A pneumonia
  • 519 screened, 120 patients randomised (61 to intervention, 59 to control)


  • Methylprednisolone
    • 0.5mg/kg per 12 hours of methylprednisolone for 5 days
    • Commenced within 36 hours of hospital admission


  • Identical placebo
    • Placebo injection every 12 hours for 5 days
    • Commenced within 36 hours of hospital admission

Antibiotic therapy according to international guidelines


Table defining primary outcome
  • Primary outcome: less overall (early + late) treatment failure occurred in the methylprednisolone group
    • This was predominantly due to a difference in late treatment failures, and specifically the occurrence of increased pulmonary infiltrates on radiographs
  • Secondary outcomes: there were no significant differences in the duration until stability, length of stay or mortality
Table of results

Authors’ Conclusions

  • Among patients with severe CAP and high initial inflammatory response, the acute use of methylprednisolone compared to placebo decreased treatment failure.


  • Randomised, blinded, multi-centre
  • Patients, investigators and data assessors were blinded
    • It’s likely bedside clinicians were blinded too, although not explicitly stated
  • Targeted severe CAP as defined by recognised definitions
  • Adequate power calculation and observed baseline rate in control group approximated that assumed for the power calculation


  • Long trial duration
    • Other practices may have changed during the trial
    • Authors compared first 3 years against last 4 years and found no differences
    • Unlikely to affect internal validity but generalisability may be reduced
  • Treatment discontinued in 10% of patients in intervention group
  • Choice and duration of antibiotics not specified
  • Unusual composite primary outcome makes it difficult to clinically apply
    • Authors argue that ‘treatment failure’ is a surrogate for resource use and patient mortality
  • Fragility index = 2 (see further reading link below)
    • If 2 additional patients of the 61 in the intervention group had been ‘treatment failures’ then the result would have been non-significant (p > 0.05) and the conclusion may have been drawn differently

The Bottom Line

  • This study raises an interesting question, and the methodology to answer this is reasonable. However, the small numbers and composite, surrogate primary outcome limit the generalisability.
  • The authors are conducting a larger trial to investigate this further
  • Clinical practice should not be changed based upon this trial alone

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.