interventions - Legeforeningen

Effects of interventions
antibiotic use in hospitals
Trial lecture for the PhD dissertation
September 25th, 2014
Jon Birger Haug
University of Oslo
Lett omarbeidet for nettundervisning NFIM, 18.12.2014
Disposition of the lecture
Scope of the lecture
Antibiotic stewardship
Intervention studies –
design and methods
The effects of interventions:
current knowledge
Are intervention effects sustainable?
Factors modifying the effects of interventions
Novel concepts in interventions
A global view
Scope of the lecture
• Only interventions in acute care hospitals are
discussed. To be excluded are:
 Rehabilitation centres
 Long term care facilities
• Outcome effect ("antibiotic use") may also include:
 Microbial outcomes / antibiotic resistance incidence
 Degree of adherence, e.g. to antibiotic guidelines
 Patient outcomes, and the effects on costs
The effect of vaccinations to reduce antibiotic use is less
relevant in a hospital setting and will not be discussed
Antibiotic Stewardship
"Stewardship ….. an ethic that embodies the
responsible planning and management of resources"
IDSA & SHEA Recommendations, 2007
Dellit et al, CID 2007;44
Antibiotic Stewardship program (ASP):
Core interventions
• Prospective audit with intervention and feedback
• Formulary restriction and preauthorization
Supplementary activities / interventions
Guidelines and clinical pathways
Antimicrobial cycling
Antibiotic order forms
Combination therapy
• Streamlining of therapy
• Dose optimization
• Parenteral to oral
Essential ASP elements
(not to be confused with intervention measures)
• An established antibiotic stewardship team, optimally
including specialist(s) of infectious diseases, microbiology,
and clinical pharmacy
• Timely and relevant service from the microbiology
laboratory and the hospital pharmacy
• Computer-based technology (health-care information
databases and surveillance systems)
• Regular reporting of the hospital's antibiotic use
• Regular reporting of antibiotic resistance patterns
• Reporting of "alert microorganisms" (multi-drug resistant)
To organize for optimal interventions….
Local pre-intervention
• Ensure the support from hospital administrators
• Secure the approval from key members of the
medical staff
• Coordinate activities with infection control
• Coordinate also with other hospital units for
patient safety
Identify local areas of major deficits
Prioritize targets
The Pareto Principle, or "Law of the Vital Few"
80 % of outcomes results from
only 20% of the potential causes
• Identification of these causes is important!
• Identification may be achieved by discussions,
patient chart reviews, and surveillance reports
Intervention studies: design and methods
• Interventions should be planned
Unplanned interventions (e.g, acute responses to an
outbreak) may be seriously biased because of
"regression to the mean" – which denotes the tendency
for extreme conditions to return to the normal
• Multicentre intervention studies are needed to
support an optimal applicability of results
• More studies should incorporate end-points related
to patient survival and cost/benefits of interventions
Randomized controlled trials (RCTs)
• The concept is well-known and the
"gold standard" in research
• Less used in interventions studies of antibiotic use
• RCTs are resource demanding (manpower, money)
• Randomization is often difficult, and subject to biases
(especially one-centre studies)
• " Cluster randomization" is the preferred method:
- hospitals are randomized, not wards within one hospital
- possible to control for the "contamination" bias
Example: study of antibiotic treatment duration with
a simple RCT design
• Clearly defined clinical condition:
→ «Pulmonary infiltrates in the ICU»
• Highly relevant for the appropriate use of antibiotics:
«Is a shorter antibiotic treatment course sufficient?»
Inclusion based on a
clinical score (0 – 10)
Compare shortcourse of antibiotics
with standard
treatment duration
Interrupted time-series analysis (ITS)
An ITS is particularly useful when a randomized trial
is not feasible or unethical
• Step 1: construct a time series of rates for your
particular improvement focus (antibiotic use)
• Step 2: test statistically for a change in the
outcome rates in the time periods before and
after the implementation
The analyses should involve several data points before
and after intervention (ideally, 24 monthly rates)
ITS: design and interpretation
Example: ITS analysis with antibiotic use outcome
• New policy for the appropriate use of "Alert
• Concurrent, patient-specific feedback by clinical
Ansari et al, JAC 2003
Other analytic methods used in intervention studies
Controlled clinical trials
Study of one or more intervention groups
compared to one or more control groups
(without randomization)
Controlled before / after studies (CBAs)
Prospective evaluation of outcomes in one
population, before and after intervention(s)
Observational studies are usually not included in reviews
of intervention effects!
The effects of interventions:
current knowledge
General remarks on current scientific evidence
• A majority of studies have methodological flaws!
• Effect evaluation is often made difficult by
considerable heterogeneity of studies
• Low external validity (applicability) of results from
carefully monitored studies (e.g. RCTs) is a general
aspect to be considered in "real life" situations…
Cochrane Collaboration (Davey et al)
"Interventions to improve
antibiotic prescribing practices
for hospital inpatients"
Issue published 2005: studies from 1980 up to November 2003
Issue publisher 2013: studies from 1980 up to December 2006
Selected for review were:
Randomized clinical trials
Interrupted time series studies
Controlled clinical trials
Controlled before-and after studies
Cochrane 2013: Studies overview
Type of studies:
• 89 studies, 95 interventions reported
• 56 studies (63%) used interrupted timeseries analysis
• 25 studies (28%) were randomized
controlled trials, of which 5 were cluster
Summary of the main findings
Main comparisons
Effect size difference at
1 month post-intervention
Quality of
Appropriate prescribing of
antibiotics (40 ITS studies)
32% (95% CI 2–61%)
Microbial outcomes
53% (95% CI 31–75%)
Mortality risk 0.92% (N.S.)
Diff. length of stay (N.S.)
Frequent readmissions
Very low
Very low
(14 ITS studies)
Patient outcomes
(11 cRCT, RCT and CCT studies )
Patient outcomes
(improve prescribing for
pneumonia – 3 CBA, 1 RCT)
Mortality risk 0.89%
(CI 0.82 - 0.97)
Main categories of interventions
Persuasive interventions: use of e.g. education,
feedback and reminders to change prescribers
Restrictive interventions: restriction of the freedom
of prescribers to select some antibiotics
A majority of the 89 Cochrane studies were "multifaceted" – that is, more than one type intervention
was used, often with a mix of persuasive and
restrictive components.
Persuasive intervention that are effective
Type of intervention
Methods used
Educational material,
- Teach or otherwise disseminate
knowledge of best practices
- Implement updated guidelines
Educational outreach
("Academic detailing")
"Interactive education" by an expert, or
one-to-one discussion with the prescriber
Audit with feedback
Prospective audit with feedback to
prescribing physician in case of
inappropriate use
Manual or electronic advice at point-ofcare; e.g. to check indication,
microbiology results, parenteral to oral
Restrictive interventions that are effective
Type of intervention
Methods used
Expert approval
Prescribing of certain antibiotic agents
needs to be approved by an infectious
disease specialist
Compulsory order
When prescribing an antibiotic agent, a
form has to be filled out stating e.g. the
indication for use
Removal of drug choice
Certain antimicrobials are removed from
the hospital's formulary
Review prescriptions
and make change
Prescriptions are reviewed by an expert
and inappropriate use is corrected
without further discussions
Structural interventions that are effective
Type of intervention
Rapid microbiology
New inflammatory
marker – Procalcitonin
Use of computerized
desicion support systems
Methods used
- Reduced time to pathogen detection
(whole genome sequencing)
- Faster susceptibility results by
dectecting resistance markers in
- MRSA, VRE and ESBL screening tests
- Decision aid to discontinue antibiotic
in sepsis, respiratory tract infections
(No substitute for clinical judgement)
(Not a test for primary diagnosis)
"Antibiotic stop orders"
Context-sensitive guideline advice
Effect sizes of intervention categories
• Persuasive interventions: Average median effect
across all study types: 3.5% – 42.5%
• Restrictive interventions: Average median effect
across all study types: 34.7% – 40.5%
• Importantly, restrictive interventions work faster
than persuasive interventions and should be used
when the need is urgent
• This difference between restrictive and persuasive
interventions diminishes over time ( ≥ 6 months)
Newer intervention studies, "post-Cochrane" (2007 – 2014)
exp Anti-Bacterial Agents/ - 538561
Medline search, Sept. 15th 2014
exp Antibiotic Prophylaxis/ - 8965
1 or 2 543022 or exp Hospitals/
- 971084
3 and 4 - 29715
antibiotic us*.mp. 7121
stewardship*.mp. - 1903
antibiotic stewardship*.mp. – 406
antimicrobial stewardship*.mp. - 631
behavioral - 2170
behavioral interventi*.mp - 4337
exp Guideline Adherence/ - 21449
Intervention or Intervention Studies/ 13017
6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 - 49028
5 and 14 - 2146
limit 14 to (English language and humans and yr="2007 -Current") - 1059
1059 studies evaluated on the basis of title and abstracts:
92 studies described interventions for appropriate antibiotic use in hospitals
Are intervention effects sustainable?
Sadly --- NO!
For example….
A classical intervention, using updated antibiotic
guideline dissemination and "academic detailing" in
two paediatric wards of a St. Petersburg hospital.
Significant reduction in total
antibiotic use in the intervention
ward, but not in the control ward
In the follow-up period, both
wards had the same level of use,
similar to the baseline condition
Factors that modify
the effects of interventions
Barriers to intervention effects
• Lack of infectious
diseases personnel
• Lack of financial
• Inadequate health-information systems
• Resistance from hospital administrators
• Opposition from prescribing physicians
• Physicians' lack of knowledge; cultural
factors; "irrational behaviour”
Beneficial to the effect of interventions
• Local strategies tailored to the needs!
• Avoid the "ceiling effect": to intervene on
already optimal areas
• National initiatives
 An "Antimicrobial self-assessment toolkit" for
acute hospitals (UK – 2009)
 A national consensus statements on quality
indicators for antimicrobial prescribing
(Germany – 2014)
• 99 indicators were suggested in a questionnaire to
professionals for detailed ratings (1-9) of relevance and
practicability - 67 were approved
• "Efforts to collect data" and "Implementation barriers"
were often given suboptimal scores
• In a consensus workshop, 21 structure and 21 process of
care indicators were finally selected
How to proceed?
• The 42 quality indicators will be piloted and
undergo feasibility studies in German hospitals
• The indicators would appear to prove valid in
similar health-care settings, e.g. in Scandinavia
Novel concepts in interventions
Innovative approaches are increasingly being
sought to enhance the effect and sustainability of
stewardship efforts
Rapidly improving old, as well as new "tools"
• Electronic health records with antimicrobial
stewardship modules & integrated clinical decision
• Web- or smartphone "app"-based prescriber aids
• Social marketing & behaviour science theories
Computer-based decision support and health records
Already described in the "Annals of Internal Medicine" in 1996:
Computer-based antibiotic stewardship aids:
• Have large potentials to facilitate, improve and prolong the
effect of intervention activities
• Unintentional effects must be considered, especially when
applying restrictive prescribing measures
• Observed increase in number of studies after 2006
Electronic support system with a
sustained intervention effect
28 antibiotics were restricted (given temporal computer-based approval) but for a
duration that varied, based on the indication for use and the prescribed antibiotic
Behaviour sciences
• A potential is recognized for behaviour sciences
to enhance antibiotic stewardship measures
• To date, no study has described the
effect of singular behavioural change
interventions on antibiotic use in
• Enhancement potential:
 Recognize "key drivers" for prescribing behaviour
and identify incentives to alter behaviour
 Target intervention according to physicians
"behaviour profiles"
A global view
• Antibiotic resistance is a world-wide problem, low-income
countries need international support (WHO)
• Large populations and deficient hospital structures calls for
alternative antibiotic stewardship measures
Important to consider: Treatment options in low-income countries are more often
limited by the unavailability of antibiotics than by antibiotic-resistant pathogens….
Intelligent use of smartphones and
computer technology may be one approach …
… being widely used by even by poor patients
- and also their doctors
Standard units per 1 000 000 population
Some national sale trends for carbapenem antibiotics *
* "Antibiotic resistance—the need for global solutions" - The Lancet (2013)
• For optimal effects of interventions, an Antibiotic Stewardship Program
is essential and should be tailored to your hospital's need
• An array of interventions have proven effective without compromising
patient safety, in the size of 30% for antibiotic use and 50% for
microbial outcomes
• Up until 2006, intervention studies have had low quality; RCTs are
often biased and ITS analysis is the preferred method
• A generally low sustainability of intervention effects is problematic
• To find intervention strategies which will work in low-income countries
is a global challenge
• Computer support and behaviour change strategies are promising
novel approaches to enhance the effects of interventions

similar documents