Impact of hand hygiene on healthcare associated infections:
Four key studies
Josh FreemanDepartment of Clinical Microbiology
ADHB
• Before-after intervention study (Quasi-experimental)– “Before” - 1994– “After” - 1995-1998
• Large teaching hospital (University of Geneva Hospitals)• Standardisation of outcome measures
– auditing of hand hygiene compliance (5 moments)– nosocomial infections (NNISS definitions) measured by annual
prevalence surveys– MRSA attack rate – new hospital acquired cases per 100
admissions
HH compliance
HH compliance improved overall from 47.6% at baseline in 1994 to 66.2% in December 1997 (p<0.001)
Impact on nosocomial infections and MRSA attack rate
Between 1994 and 1998, MRSA infections decreased from 2.16 /10000 PD to 0.93 / 10000 PD (p<0.001)
MRSA bacteraemia decreased from 0.74 to 0.24 / 10000 PD (p<0.001)
Evidence that HH affects healthcare-associated infections: Strengths and Weaknesses
Strengths• Large hospital with large
sample size• Temporal associaton
between improved hand hygiene practice and reduced nosocomial infections and MRSA attack rate
• Extended time frame post intervention
Weaknesses• Other interventions targeting
MRSA carried out simultaneously
• Few data points for nosocomial infection rates (particularly pre-intervention)
• “Nosocomial infections” may be subject to classification bias despite standardised definitions
• Before-after intervention study – Before: Jan 1999-May 2001– After: May 2001- April 2004
• Five wards at Austin Health, Melbourne• Intervention
– HH programme – Mupirocin / triclosan for MRSA colonised patients on admission
• Outcomes – Standardised HH compliance auditing (“5 moments”)– Standardised definitions
• MRSA bacteraemia• MRSA clinical isolates• ESBL-E. coli and K. pneumoniae clinical isolates
HH compliance pre and post intervention
Overall – 21% pre-intervention to 42% post intervention
MRSA rates pre and post intervention
Slope<0; p<0.001
Slope<0;p=0.003
ESBL rates pre and post intervention
Evidence that HH affects healthcare-associated infections: Strengths and Weaknesses
Strengths• Substantial and significant
association between program onset, improved HH compliance and improved trends in MRSA bacteraemia; MRSA clinical isolates; and ESBL clinical isolates
• Large number of data points pre and post intervention
Weaknesses• Intervention included
decolonisation for MRSA, therefore difficult to estimate relative impact of HH on MRSA rates
• Before-after study (Quasi-experimental study)– Standardised process and outcome measures
• auditing of HH compliance 4 monthly (5 moments)• Standardised definitions of MRSA bacteraemia / clinical isolates
• Pilot study– 6 Victorian healthcare institutions over 24 month period
• Statewide study – 75 Victorian hospitals over 12 month period– Rolled out in two stages: Stage 1 – March 2006-April 2007 and stage 2
July 2006-June 2007
HH Compliance: Pilot program
MRSA bacteraemia rates: Pilot program
MRSA clinical isolates: Pilot program
HH compliance: Statewide rollout
MRSA bacteraemia rates: Statewide rollout
MRSA clinical isolates: Statewide rollout
Evidence that HH affects healthcare-associated infections: Strengths and Weaknesses
Strengths• Large, multicentre study• Improved HH practice temporally
associated with significant reductions in MRSA bacteraemia in both the pilot and statewide studies
• Large number of data points before and after the intervention
• MRSA bacteraemia–less vulnerable to classification bias than many endpoints (“hard” endpoint)
Weaknesses• Quasi-experimental, non-
randomised study with historical controls– therefore intrinsically vulnerable to confounding
• Concurrent MRSA-specific measures not documented
• MRSA clinical isolates started to decrease prior to commencing the program (raising possibility that factors other than the HH program may have been driving change)
24 month Outcomes from the Australian National Hand Hygiene Initiative (MJA -in press)
• National HH initiative (Hand Hygiene Australia)• Quasi-experimental study (2009-2010)
– “Before” - Jan 2007-Dec 2008– “After” - Jan 2009-Dec 2010
• Nationally standardised – Auditing of HH compliance (“5 moments”)– MRSA bacteraemia– Hospital-onset SA/MRSA bacteraemia
HH compliance by state: 2009-2010 (post intervention)
Overall 43.6% at baseline to 67.8% post intervention
National MRSA bacteraemia rates pre and post intervention
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Apr-07 Nov-07 Jun-08 Dec-08 Jul-09 Jan-10 Aug-10 Feb-11
Rate
of
MRS
A b
acte
raem
ia p
er 1
0,00
0 PD
's
Month
Pre NHHI
Post NHHI Implementation
Slope<0;p=0.008
Hospital-onset S. aureus bacteraemia rates post intervention
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Rate
of
S.au
reus
bac
tera
emia
s pe
r 10
,00
PD's
Month
MRSA /10,000
MSSA/10,000
SAB/10,000
Linear (MRSA /10,000)
Linear (MSSA/10,000)
Linear (SAB/10,000)
Evidence that HH affects healthcare-associated infections: Strengths and Weaknesses
Strengths• First nationwide before-after
intervention study for HH programme with outcome data
• Large multi centre study• Utilises a “hard” endpoint
(MRSA bacteraemia)• Strong temporal association
with improved compliance and reduced MRSA bacteraemia
• Large number of data points pre and post intervention
Weaknesses• Vulnerable to confounding
(like all quasi-experimental studies)
• No reduction in “hospital-onset” S. aureus bacteraemia or “hospital-onset” MRSA bacteraemia
HH compliance infection: Association versus causality
Bradford Hill Criterion Supports improved HH compliance as a means to reduce infection?
Association is strong? YES – Statistically significant association between HH and infection rates
Association is seen consistently? YES – Consistent association in four well designed studies
Cause precedes effect? YES – Improvements in hand hygiene have preceded / coincided with reduced infections
Biological gradient (dose-response between cause and effect)?
YES - Inverse correlation between HH compliance rates and rates of infection
Biologically plausible? YES – Hands of healthcare workers known to be frequently contaminated with potential pathogens including MRSA
Coherence (compatible with existing knowledge)?
YES- A causal relationship would be consistent with accepted models of pathogenesis of healthcare-associated infections
Subject to experiment? NO – experimental studies not feasible / ethical
Alternate explanations for association?
YES – Difficult to rationalise temporal association between HH compliance and infectious endpoints based on alternative explanations