Contact precautions in single-bed or multiple-bed rooms for patients with extended-spectrum Beta-lactamase-producing enterobacteriaceae in Dutch hospitals

Kluytmans-van den Bergh. Lancet Infectious Diseases 2019; Published online 23rd August 2019. doi: 10.1016/S1473-3099(19)30262-2.

Clinical Question

  • In non-ICU, non-haematology patients that have been identified as carriers of extended-spectrum beta-lactamase-producing enterobacteriaceae (ESBLs), is a strategy of contact precautions in a multiple-bed room compared to contact precautions plus single-bed room isolation non-inferior for preventing transmission?


  • Antibiotic resistance is a growing problem worldwide
  • Enterobacteriaceae, which include the ‘coliform’ group of bacteria such as E.coli and Klebsiella, frequently produce beta-lactamase proteins that increase their resistance to antibiotic therapy (ESBLs)
  • Hospital policies usually result in carriers of ESBL-producing bacteria being isolated from other patients to prevent the spread of the resistant strain of bacteria. This can be difficult to deliver in hospitals where multiple-bed rooms are more common and isolation facilities are limited
  • The effectiveness of isolation has not been adequately tested


  • Pragmatic, cluster-randomised, crossover design
    • Hospitals were the unit of randomisation
    • Individual patients were the unit of analysis
    • Hospitals were randomised 1:1 ratio to one of two potential study sequences
      • Single-bed room strategy –> Multiple-bed room strategy
      • Multiple-bed room strategy –> Single-bed room strategy
    • Study period were preceded by a 2-month washout period to limit carryover between study periods
    • Random allocation sequence was computer generated and stratified by hospital status (University vs Non-university)
    • Allocation was concealed in opaque envelopes until the last hospital was randomised
    • Patients and staff were aware of allocation but laboratory staff remained blinded to allocation
  • Non-inferiority statistical design
    • 241 patients were required per study arm
    • Expected incidence of 10% for transmission of ESBL-producing Enterobacteriaceae from index patients to ward mates in the single-bed room strategy
    • Non-inferiority margin of 10% for the risk difference between strategies
    • Power 90% and one-sided α 5%
    • Non-adherence 15% (single-bed) and 5% (multiple-bed)
  • Per-protocol analysis and intention-to-treat analysis
  • Written, informed consent was not required for inclusion, but consent was required for data collection and microbial screening
    • Index patients were those included with an ESBL-producing Enterobacteriaceae who were randomised to either a strategy of single-bed room isolation or a strategy of remaining in a multi-bed ward area
    • Ward mates were those included that had been cohabiting within the same hospital ward with an index patient, and were followed-up for evidence of transmission during their stay


  • Medical and surgical wards (non-icu and non-haematology) at 16 hospitals in the Netherlands
    • Only 13 hospitals completed the study
      • Two withdrew due to loss of equipoise after a large-scale outbreak at a different Dutch hospital
      • One withdrew due to structure redesign and were unable to offer single-bed room strategy
  • April 2011 to February 2014


  • Inclusion:
    • Index patients were consecutively screened and considered for eligibility on the day their routine clinical screening returned with a positive ESBL result
    • Ward mates were included if they had been cohabiting a ward with an index patient
  • Exclusion:
    • Index patients were excluded if they were younger than 18 years, had other indications for strict isolation, were discharged prior to their routine clinical screening result or if no result was available by day 7
    • Ward mates were excluded if they were younger than 18 years
  • 1652 index patients were screened, 830 were eligible (most being excluded due to discharge from hospital prior to screening results), 693 were enrolled, 616 were available for complete analysis
  • 12875 ward mates were screened in relation to the 616 index patients, 9527 were enrolled and 9368 were available for complete analysis
  • Characteristics between strategies were similar, except for the number of unprotected ward days, which were greater in the single-bed room strategy group


  • Multiple-bed room strategy
    • Contact precautions were instituted for all episodes of care of the index patient
    • The index patient remained in a multiple occupancy ward area (two to six patients) as the preferred strategy
    • Index patients could either have private or shared bathroom and toilet facilities
    • Ward mates remained in their existing bed spaces and were screened for ESBL-producing Enterobacteriaceae at day 7


  • Single-bed room strategy
    • Contact precautions were instituted for all episodes of care of the index patient
    • The index patient was moved to a single-bed room for isolation as the preferred strategy
    • A private bathroom and toilet was used solely by the index patient
    • Ward mates remained in their bed spaces and were screened for ESBL-producing Enterobacteriaceae at day 7

Management common to both groups

  • Standard precautions were routinely applied to all patients at all times – this included hand hygiene and personal protective equipment  (gloves and gowns) when there was likely exposure to blood or body fluids
  • Contact precautions were additionally applied to index patients after their results reported positive for ESBL producing Enterobacteriaceae – this included, in addition to standard precautions, the wearing gloves for all contact with a patient, their belongings or their immediate surroundings
  • Unprotected ward stay was the period from taking the routine clinical screening sample until the subsequent positive result that led to the initiation of contact precautions
  • Protected ward stay was the period after the institution of contact precautions until the screening of ward mates (day 7 after enrolment of their associated index patient) or discharge of ward mates


  • Primary outcome: transmission of ESBL-producing Enterobacteriaceae to ward mates (day 7 rectal carriage of a related species to index patient sample) was not statistically significantly different between strategies and the 90% confidence interval fell within the pre-defined threshold to determine non-inferiority
    • Multiple-bed room strategy transmission rate: 7.45%
    • Single-bed room strategy transmission rate: 4.00%
    • Absolute Risk Difference: 3.45% (90% CI -0.25% to 7.15%; P = 0.0815)
  • Secondary outcome:
    • Adherence to allocated strategy:
      • Multiple-bed room strategy: 62% (mostly due to other medical reasons for isolation)
      • Single-bed room strategy: 88% (mostly due to no single room being available)
    • Median length of stay: 11 days in both groups
    • Mortality at 30 days: 4% in both groups
    • Prevalence of ESBL-producing Enterobacteriaceae:
      • Multiple-bed room strategy: 122 per 100,000 bed days
      • Single-bed room strategy: 115 per 100,000 bed days
    • Transmission by timing of institution of contact precautions:
      • Transmission to at least one ward mate if any period of unprotected ward stay: 7%
      • Transmission to at least one ward mate if no period of unprotected ward stay (that is, contact precautions were implemented on admission due to known ESBL carriage or hospital policy): 3%
      • Absolute Risk Difference: 3.4% (95% CI 0.0 to 6.8)

Authors’ Conclusions

  • In this study of non-icu and non-haematology patients, an isolation strategy of contact precautions in a multiple occupancy ward area was non-inferior compared to a strategy of single-bed room isolation for preventing the transmission of ESBL-producing Enterobacteriaceae


  • Very important clinical question, given the increasing demands on healthcare and the increasing prevalence of antibiotic resistant species or bacteria
  • The cluster-randomised, cross-over trial design is pragmatic and efficient
  • The non-inferiority design is appropriate given the clinical question
  • The bio-statistics are appropriate for non-inferiority testing, including per-protocol analysis as the primary outcome
  • The multi-centre design increases generalisability


  • Non-inferiority margin
    • The study authors defined the non-inferiority margin at an absolute risk difference of 10%
    • Given they anticipated a transmission rate of 10% in their control group, this equates to a doubling of the transmission rate before they would have concluded inferiority
    • It could be argued that the 10% non-inferiority margin is too extreme
  • The significant numbers excluded may have introduced bias, but this was unavoidable and attempts were made to minimise this impact, including adjusted analysis and consecutive screening
  • Many clinical samples were not reported within 7 days, resulting in many exclusions
  • This is not necessarily generalisable to Critical Care / Intensive Care wards, where contact precautions are often implemented from the point of admission, and the invasive nature of therapies and the high incidence of diarrhoea may increase transmission

The Bottom Line

  • This study raises an important question and provides some data that may lead to changes in infection control policies in some Intensive Care units
  • However, repetition of this trial in the Intensive Care setting, with careful consideration of the non-inferiority margin, is needed before widespread change is recommended

External Links


Summary author: Duncan Chambler
Summary date: 20 September 2019
Peer-review editor: Adrian Wong

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.