De Jonge

Effects of selective decontamination of digestive tract on mortality and acquisition of resistant bacteria in intensive care: a randomised controlled trial

Evert de Jonge, Lancet 2003; 362: 1011–16

Clinical Question

  • Does the use of Selective Decontamination of the Digestive Tract (SDD) affect ICU and hospital mortality and the acquisition of resistant organisms?

Design

  • Prospective
  • Controlled
  • Randomised
  • Unblinded

Setting

  • Sept 1999 to Dec 2001
  • Single hospital in Amsterdam
  • 2 separate ICUs – 1 acted as intervention arm, the other control

Population

  • Inclusion
    • Adults, aged over 18
    • ICU admission
    • Expected duration of mechanical ventilation of > 48hrs
    • Expected length of stay on ICU > 72hrs
  • Exclusion
    • Previous ICU admission in last 3 months
    • Known hypersensitivity to study medication
    • Pregnancy
    • Perceived imminent death
    • Participation in another investigational study
  • 934 patients randomised. 2 year follow up

Intervention

  • QDS 0.5g oral paste (2% polymixin E, 2% tobramycin and 2% amphotericin B) to buccal cavity
    • Patients with tracheostomies, had paste applied to skin surrounding tracheostomy
  • 100mg polymixin E, 80mg tobramycin and 500mg amphotericin B via gastric tube
    • Patients with blind bowel loops received suppositories containing 42mg amphotericin B, 42mg polymixin E and 64mg tobramycin twice to four times a day
  • QDS 1gm cefotaxime IV for first 4 days
  • SDD continued until discharged from ICU
  • Surveillance cultures from rectal swabs, throat swabs, and sputum were taken at admission and twice weekly during the stay on the ICU
  • If aerobic gram negative bacteria or yeasts were cultured from the sputum on more than one occasion, 80mg nebulized polymyxin E four times daily or 5mg amphotericin B four times daily administered until cultures became negative

Control

  • Standard oral care
  • Rinsing mouth with water QDS
  • Tooth brushing BD
  • Stress ulcer prophylaxis not routinely given

Outcome

  • Primary outcome:
    • Acquired colonisation of gram-negative aerobic bacteria
      • 16% in intervention group vs 26% in control (RR 0.61, P = 0.001)
    • ICU mortality
      • 14.8% in intervention group vs 22.9% in control (RR 0.65, P = 0.002)
    • Hospital mortality
      • 24.2% in intervention group vs 31.2% in control (RR 0.78, P = 0.02)
  • Secondary outcomes:
    • ICU length of stay
      • 6.8 days in intervention group vs 8.5 days in control (p < 0.0001)
    • Antibiotic cost
      • 11% lower in intervention group

Authors’ Conclusions

  • In a setting with low prevalence of vancomycin resistant enterococcus and methicillin-resistant S aureus, SDD can decrease ICU and hospital mortality and colonization with resistant gram-negative aerobic bacteria.

Strengths

  • Randomised
  • Large numbers
  • Address the 2 main concerns over SDD – mortality and antibiotic resistance
  • Intervention protocol acceptable practice

Weaknesses

  • Power calculation: one-sided alpha error – should this be 2 tailed? Intervention could increase OR decrease resistance
  • Trial ended early and did not meet sample size targetted due to the unit moving to a different site
  • 2 year follow up means that effect on antibiotic resistance after this period is unknown
  • Unblinded study
  • Low baseline prevalence of VRE and MRSA in population

The Bottom Line

  • The use of SDD continues to divide opinion amongst ICU clinicians – concerns remain over its beneficial effect on patient mortality and the promotion of antibiotic resistance. The fact that the study population had low prevalence of VRE and MRSA would limit its application. My view is exactly that of the SuDDICU study – I would be prepared to introduce SDD in the context of a clinical trial.

External Links

Metadata

Summary author: your name: @avkwong
Summary date: 5th September 2014
Peer-review editor: @stevemathieu75

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.