Content uploaded by Veronique M Boscart
Author content
All content in this area was uploaded by Veronique M Boscart on Oct 18, 2017
Content may be subject to copyright.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
j
ourna
l
ho
mepage:
www.ijmijournal.com
The
feasibility
of
an
automated
monitoring
system
to
improve
nurses’
hand
hygiene
Alexander
I.
Levchenkoa,∗,
Veronique
M.
Boscarta,b,
Geoffrey
R.
Ferniea,c
aToronto
Rehabilitation
Institute,
iDAPT
–
Intelligent
Design
for
Adaptation,
Participation
and
Technology,
Ontario,
Canada
bL.
Bloomberg
Faculty
of
Nursing,
University
of
Toronto,
Ontario,
Canada
cDepartment
of
Surgery,
University
of
Toronto,
Ontario,
Canada
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
16
August
2010
Received
in
revised
form
16
February
2011
Accepted
22
April
2011
Keywords:
Monitoring
Hand
hygiene
Embedded
system
Information
system
a
b
s
t
r
a
c
t
Background:
Inadequate
hand
hygiene
(HH)
by
healthcare
staff
results
in
increased
rates
of
hospital
acquired
infections
in
healthcare
institutions,
considerable
waste
of
resources,
and
negative
economic
impact
for
the
healthcare
system.
Toronto
Rehabilitation
Institute
has
developed
an
automated
HH
monitoring
system
that
detects
HH
opportunities,
generates
HH
reminding
signals
when
it
is
necessary
and
enables
hospital
management
to
monitor
individual
and
aggregated
HH
performance
on
ongoing
basis.
Objective:
To
demonstrate
that
HH
improvement
is
feasible
with
the
proposed
technical
solution
and
that
technology
is
acceptable
by
potential
users.
Methods:
The
technology
was
installed
in
four
rooms
on
a
nursing
unit
of
a
larger
complex
continuous
care
hospital.
The
rooms
were
selected
to
make
it
possible
to
automatically
fol-
low
the
same
nurses
for
the
duration
of
their
entire
shift.
Eleven
nurses
were
provided
with
the
wearable
electronic
HH
monitors
as
well
as
with
the
instrumented
personal
wearable
alcohol
gel
dispensers.
Stationary
gel
dispensers
installed
in
the
unit
were
also
instrumented
with
technology.
Results:
Over
145
h
of
testing
the
system
automatically
recorded
a
total
of
1438
events
of
entering
and
leaving
monitored
rooms
and
indicated
an
average
of
6.42
HH
actions
per
hour.
The
baseline
observational
study
indicated
4.2
HH
actions
per
hour.
Approximately
half
of
the
HH
actions
recorded
by
the
system
were
performed
using
personal
wearable
alcohol
gel
dispensers.
Conclusion:
The
results
obtained
when
testing
the
embedded
HH
monitoring
system
demon-
strated
the
feasibility
of
HH
improvement
and
proved
that
proposed
solution
merits
a
larger
and
longer
clinical
trial
to
measure
the
degree
of
improvement
and
the
sustainability
of
that
improvement.
©
2011
Elsevier
Ireland
Ltd.
All
rights
reserved.
1.
Introduction
Hospital
acquired
infections
are
a
significant
threat
to
the
health
of
patients,
yet
health
care
staff’s
hand
wash
compli-
∗Corresponding
author
at:
Toronto
Rehabilitation
Institute,
550
University
Avenue,
12th
floor,
Toronto,
Ontario
M5G
2A2,
Canada.
Tel.:
+1
416
597
3422x7951.
E-mail
address:
Levchenko.Alexander@torontorehab.on.ca
(A.I.
Levchenko).
ance
is
difficult
to
achieve
and
maintain
[1].
The
failure
of
health
care
staff
to
perform
the
simple
procedure
of
disin-
fecting
hands
can
lead
to
serious
consequences
for
patients,
including
an
increased
rate
of
hospital-acquired
infections
[2],
morbidity
and
mortality
[3],
and
a
growing
financial
impact
1386-5056/$
–
see
front
matter
©
2011
Elsevier
Ireland
Ltd.
All
rights
reserved.
doi:10.1016/j.ijmedinf.2011.04.002
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
597
on
the
health
care
system
[4].
Health
care
staff
cite
several
reasons
for
inadequate
HH,
such
as
a
lack
of
time,
no
access
to
alcohol
gel
or
soap
dispensers
or
a
sink,
or
simply
forget-
fulness
[5].
In
order
to
comply
with
the
Ontario
Ministry
of
Health
and
Long
Term
Care
(MOHLTC)
guidelines
health
care
workers
should
perform
hand
hygiene
(HH)
every
time
before
they
enter
a
patient
environment
or
come
in
contact
with
a
patient
or
patient
environment
and
every
time
when
they
leave
a
patient
environment.
To
address
the
HH
compliance
problem,
researchers
at
Toronto
Rehabilitation
Institute
have
developed
an
electronic
HH
monitoring
system
that
automatically
detects
and
records
HH
opportunities
and
HH
actions
and
offers
the
possibility
not
only
to
monitor
and
report
HH
performance,
but
also
reminds
health
care
workers
to
perform
HH
when
it
is
necessary.
2. Related
work
There
have
been
few
attempts
to
include
technology
as
an
intervention
to
improve
HH.
Kinsella
et
al.
[6]
used
electronic
devices
to
time
stamp
activations
of
gel
and
soap
dispensers.
This
technology
did
not
automatically
detect
HH
opportuni-
ties
and
was
used
as
a
part
of
a
HH
quality
improvement
initiative.
Venkatesh
et
al.
[7]
and
Swoboda
et
al.
[8]
evalu-
ated
an
electronic
prompting
system
that
used
light
beams
and
motion
detectors
at
the
threshold
of
the
room
in
a
sur-
gical
unit.
The
system
provided
audible
and
visual
reminding
to
perform
HH
if
it
had
not
been
completed
before
entering
or
after
exiting
the
room.
The
system
was
installed
in
one
room,
worked
only
with
a
stationary
dispenser
and
did
not
record
individual
HH
performance
data.
More
complex
approaches
have
included
real
time
location
tracking
systems.
There
have
been
a
number
of
patents
on
using
radio
frequency
identifi-
cation
methods
(RFID)
[13],
short
range
radio
frequency
(RF)
[10],
infrared
(IR)
[11],
or
ultrasound
[9,12]
based
technologies
and
wearable
electronic
devices
to
implement
some
sort
of
location
tracking
to
detect
HH
opportunities,
provide
HH
alerts
and,
in
some
implementations,
record
HH
performance
data,
usually
by
a
central
base
unit
controlling
the
whole
system.
Unfortunately,
there
are
no
publications,
to
our
knowledge,
on
the
implementation
of
these
technologies
in
actual
clini-
cal
settings
and
no
publications
describing
even
the
results
of
pilot
clinical
trials.
The
distributed
embedded
HH
monitoring
system
devel-
oped
at
Toronto
Rehabilitation
Institute
does
not
track
location
of
the
staff
continuously
throughout
the
hospital
but
imple-
ments
monitoring
functions
in
the
areas
essential
for
HH
by
detecting
a
presence
of
a
caregiver
within
monitored
zones,
relating
this
information
with
HH
actions
performed
by
a
care-
giver
and
generating
HH
reminding
signals
when
necessary.
This
approach
reduces
the
cost
of
implementation,
simplifies
installation
and
helps
to
avoid
potential
ethical
and
privacy
issues
associated
with
continuous
location
tracking
of
indi-
viduals.
3.
Hand
hygiene
monitoring
system
We
previously
reported
[14]
on
details
of
implementation
and
the
advantages
of
the
developed
HH
monitoring
system
as
well
as
initial
findings
on
acceptability
of
automated
monitoring
by
healthcare
staff
[15].
This
paper
describes
improvements
that
have
been
made
to
the
technology
and
the
results
obtained
when
testing
the
system
in
a
complex
continuous
care
setting.
In
the
embedded
system
developed
at
Toronto
Rehabili-
tation
Institute,
all
HH
monitoring,
data
recording
and
HH
prompting
functions
are
performed
by
the
personal
wear-
able
electronic
monitors.
The
monitors
are
badge-size
devices
including
a
microcontroller,
IR
detector
and
RF
transceiver.
They
communicate
with
the
sets
of
IR
emitters,
mounted
on
the
ceiling
to
define
the
monitored
areas
(zones)
where
HH
compliance
is
critical
(individual
patient
environments,
room
entrances,
shared
bathrooms,
dirty
utility
rooms,
etc.).
IR
emitters
transmit
the
codes
which
are
unique
for
each
zone
and
include
the
zone
identification
number,
zone
type,
loca-
tion
and
other
information
required
for
monitoring
purposes.
Usage
of
IR
communication
links
with
regulated
intensity
allows
the
monitored
areas
to
be
defined
with
high
accuracy
[14]
and
makes
performance
of
the
system
independent
on
structural
characteristics
of
the
environment.
IR
communication
links
are
also
used
to
integrate
wall-
mounted
dispensers
into
the
system.
Two
collimated
IR
emitters
are
installed
above
each
dispenser
and
transmit
for
a
few
seconds
after
dispenser
activation.
Personal
wearable
dispensers
are
instrumented
with
a
microcontroller
and
RF
transceiver
so
that
the
dispenser
activation
is
transmitted
to
corresponding
wearable
electronic
monitor.
The
monitors
process
all
HH
actions
performed
by
a
care-
giver
with
the
personal
wearable
or
wall-mounted
dispensers.
The
record
includes
time
of
activation,
dispenser
identifica-
tion
code,
and
type
of
dispenser.
The
monitors
also
process
and
record
exact
times
when
a
caregiver
enters
or
leaves
mon-
itored
areas
along
with
identification
codes
of
these
areas
and
HH
status
of
a
caregiver.
Using
this
information
and
HH
activ-
ity
data
the
personal
wearable
monitors
make
a
decision
on
when
it
is
necessary
to
generate
HH
reminding
signals.
The
HH
reminding
signal
can
be
implemented
as
an
audible
prompt,
a
vibration,
or
a
combination
of
both.
Data
recorded
by
the
monitors
is
downloaded
to
a
personal
computer
and
the
system
enables
hospital
management
to
monitor
HH
performance
on
an
ongoing
basis,
with
some
of
the
data
processing
routines
implemented
in
the
firmware
of
the
wearable
monitors.
Performance
reports
include
the
num-
ber
of
HH
opportunities
recorded
by
the
monitors
within
a
selected
period
of
time,
with
the
option
to
classify
the
oppor-
tunities
depending
on
HH
status
of
a
caregiver
at
the
moment
when
an
opportunity
was
detected
(clean,
HH
performed
after
prompt,
ignored
HH
prompt).
The
reports
also
include
HH
actions
performed
with
different
types
of
dispensers,
HH
compliance
rates,
comparison
of
individual
and
aggregated
performance,
variations
of
HH
activity
and
compliance
over
specified
period
of
time.
The
system
does
not
use
any
network
infrastructure
and
the
monitors
work
completely
independently,
with
monitor-
ing
logic
and
performance
characteristics
defined
by
their
firmware.
During
earlier
clinical
tests
with
the
monitored
zones
installed
to
define
individual
patient
areas
within
a
multi-bed
room,
we
found
that
the
moment
when
the
wearable
mon-
itor
stops
detecting
the
signal
from
the
previously
detected
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
598
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
patient
zone
cannot
be
reliably
used
as
an
indicator
that
the
caregiver
has
left
the
zone
and
HH
action
is
needed.
When
performing
certain
procedures
a
caregiver
often
needs
to
step
out
of
a
monitored
zone
for
short
periods
of
time.
Moreover,
the
boundaries
marking
a
patient
environment
may
change
dynamically.
For
example,
when
transferring
a
patient
to
a
wheelchair,
a
caregiver
might
be
outside
of
the
predefined
boundary
of
the
patient
environment,
while
continuing
work-
ing
with
the
same
patient.
A
sequence
of
related
events
needs
to
be
detected
and
processed
by
the
system
in
real
time
to
make
a
more
accurate
decision.
These
events
include
the
exact
times
of
HH
actions,
location
of
a
caregiver
when
a
HH
action
was
performed,
time
of
detection
and
type
of
a
monitored
area,
time
of
detection
and
type
of
the
previously
detected
location.
These
findings
resulted
in
significant
modifications
in
the
system
hardware
and
software.
The
logic
defined
in
system
software
is
no
longer
based
on
the
presence
of
a
caregiver
within
a
certain
area
which
is
continuously
checked,
but
on
processing
a
series
of
events
every
time
when
a
monitored
area
is
detected
or
a
HH
action
is
performed.
For
exam-
ple,
when
the
wearable
monitor
detects
that
a
new
patient
environment
is
entered,
it
records
the
HH
opportunity
and
generates
a
HH
prompting
signal,
if
necessary.
The
monitor
however
does
not
make
the
decision
that
the
caregiver
has
left
the
patient
environment
when
it
stops
detecting
the
sig-
nal
from
the
emitters
defining
that
area.
This
decision
is
made
when
the
monitor
detects
a
new
location.
This
can
be
another
patient
zone
in
multi-bed
room,
a
room
entrance
zone
or
a
zone
installed
in
a
shared
ensuite
bathroom.
This
new
approach
allowed
the
size
of
the
monitored
zones
to
be
significantly
reduced.
In
the
latest
configuration
of
the
system,
the
monitored
areas
are
defined
by
one-dimensional
arrays
of
IR
emitters
marking
entrances
to
patient
rooms,
perimeters
of
individual
patient
environments,
entrance
areas
at
shared
bathrooms
and
dirty
utility
rooms.
The
zone
con-
trollers
are
activated
by
passive
infrared
(PIR)
motion
sensors,
with
their
area
of
sensitivity
being
adjusted
to
initiate
the
transmission
only
when
somebody
crosses
the
boundary
defined
by
the
emitters.
The
controllers
of
monitored
zones
and
arrays
of
emitters
marking
boundaries
are
optionally
pow-
ered
by
a
6
V
DC
power
supply
or
by
five
AA
cell
rechargeable
battery
packs,
with
the
intervals
between
charges
being
up
to
one
month.
The
smaller
size
of
the
monitored
zones
increased
the
requirements
to
the
reaction
time
of
the
wearable
moni-
tors.
A
combination
of
high
noise
immunity
IR
detectors
and
a
specially
developed
communication
protocol
allowed
an
inter-
rupt
driven
operation
to
be
implemented,
to
minimize
the
reaction
time
of
the
monitor.
Before
testing
the
system,
a
baseline
observational
study
was
conducted
at
the
nursing
unit,
where
the
system
was
sub-
sequently
installed,
to
collect
the
data
on
HH
opportunities,
HH
actions
performed
with
different
types
of
dispensers
and
to
estimate
the
percentage
of
HH
opportunities
that
could
be
potentially
captured
by
various
configurations
of
the
monitor-
ing
system.
During
the
observational
study
it
was
determined
[16]
that
even
in
multi-bed
rooms
a
nurse
usually
leaves
the
room
after
finishing
working
with
a
patient
and
enters
the
room
again
before
moving
to
another
patient
zone.
Some-
times,
after
leaving
a
patient
zone,
a
nurse
goes
to
a
shared
ensuite
bathroom,
and
in
a
few
cases
directly
to
another
patient
in
the
same
room.
The
results
of
the
observational
study
indicated
that
monitoring
the
entrances
of
the
patient
rooms
and
shared
bathrooms,
in
addition
to
individual
patient
zones,
could
improve
the
accuracy
of
detection
of
HH
oppor-
tunities,
with
more
than
two
thirds
of
them
captured
by
the
monitored
zones
installed
at
patient
room
entrances.
4. Methods
The
objective
of
this
study
was
to
demonstrate
the
feasibil-
ity
of
HH
improvement
with
the
proposed
technical
solution
and
to
prepare
for
a
larger
scale
clinical
trials.
The
system
was
installed
in
four
rooms
(two
four-bed
and
two
two-bed
rooms)
on
a
nursing
unit
of
a
larger
complex
continuous
care
hospi-
tal,
with
three
rooms
instrumented
with
entrance
zones
only
and
one
two-bed
room
instrumented
with
both
entrance
and
individual
patient
zones.
The
instrumented
rooms
were
selected
on
one
side
of
a
nursing
unit
so
as
to
capture
all
nursing
staff
who
delivered
care
to
the
patients
in
these
rooms.
A
typical
nursing
shift
assignment
involved
the
delivery
of
nursing
care
to
3
or
4
patients
in
the
instrumented
rooms,
thereby
making
it
possi-
ble
to
follow
the
same
nurses
almost
for
the
duration
of
their
entire
shift.
The
sample
consisted
of
a
convenience
selection
of
those
nurses
who
delivered
direct
care
to
the
patients
in
the
instrumented
rooms.
In
order
to
be
included,
nurses
had
to
be
employed
full-time
or
part-time
on
the
unit
and
had
to
be
will-
ing
to
receive
an
educational
session
on
the
use
of
the
system.
Eleven
nurses
consented
to
participate
in
the
study.
Eight
of
the
participants
worked
full-time.
All
patients
in
the
instru-
mented
rooms
required
nursing
care
consisting
of
supportive
or
full
care
for
all
activities
of
daily
living
(feeding,
bathing,
dressing,
and
grooming).
Entrance
zones
in
the
system
included
two
stages
so
that
they
could
differentiate
between
the
event
of
entering
and
exiting
the
room,
with
one
of
the
stages
installed
in
the
hall-
way
and
the
other
installed
inside
the
patient
room.
Depending
on
the
size
of
the
room
entrance,
one
or
two
motion
sensors
per
room
were
installed
to
activate
the
zones.
All
zones
were
battery-powered
and
for
both
stages
duration
of
transmission
was
set
at
20
s
after
activation.
At
the
nursing
unit
where
the
study
was
conducted,
wall-
mounted
gel
dispensers
were
installed
only
in
the
hallways
near
the
doorways
of
the
patient
rooms.
All
wall-mounted
dispensers
used
by
the
nursing
team
participating
in
the
study
were
instrumented
with
external
controllers
[14]
so
the
HH
actions
performed
with
these
dis-
pensers
could
be
recorded
and
processed
by
the
system.
Each
participant
also
had
a
personal
wearable
alcohol
gel
dispenser
communicating
with
a
personal
wearable
electronic
monitor,
so
HH
actions
performed
with
the
wearable
dispensers
were
also
recorded
and
processed.
A
state-machine
diagram
illustrating
the
concept
described
above
and
the
logic
of
operation
of
the
wearable
monitors
in
this
study
is
shown
on
Fig.
1.
Most
of
the
time
over
the
dura-
tion
of
the
shift
the
monitor
is
in
the
states
S1
or
S3
depending
on
the
location
of
a
caregiver.
Any
HH
action
changes
the
HH
status
flag
to
“clean”
and
switches
the
monitor
to
the
state
S2
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
599
S1
S4 S2
S5 S3
Internal st
age of
entrance zone
is detected
HH ex
p
ir
y
time tex
p
1ela
p
se
d
Dispenser activation
Internal st
age of
entrance zone
is detected
Prompti
ng time
inter
val
ela
psed
External stage of entrance zone is detected
Dispenser
acti
vati
on
Dispenser activation
External stag
e of
entrance zone
is detect
ed
S6
Prompti
ng time
inter
val
ela
psed
HH ex
p
ir
y
time tex
p
2ela
p
se
d
Fig.
1
–
Generalized
state-machine
diagram
for
the
wearable
electronic
monitor.
S1
–
a
caregiver
is
outside
of
the
patient
room;
S2
–
a
caregiver
is
outside
the
patient
room,
HH
status
set
to
“clean”;
S3
–
a
caregiver
is
inside
the
patient
room;
S4
–
a
caregiver
is
inside
the
patient
room,
HH
status
set
to
“clean”;
S5
–a
caregiver
is
inside
the
patient
room,
HH
prompt
is
generated;
S6
–
a
caregiver
is
outside
of
the
patient
room,
HH
prompt
is
generated;
texp1–
HH
expiry
time
outside
of
the
room;
texp2–HH
expiry
time
inside
the
room.
or
S4.
While
the
monitor
is
in
one
of
these
states
a
caregiver
can
enter/exit
monitored
areas
with
no
HH
prompt
generated.
When
the
HH
action
expires
the
monitor
returns
to
the
state
S1
or
S3.
If
a
monitored
room
is
entered
(or
exited)
while
the
monitor
is
in
one
of
these
states,
the
device
starts
generat-
ing
a
HH
prompting
signal
and
switches
to
the
state
S5
(S6).
If
a
caregiver
responds
by
performing
a
HH
action
the
mon-
itor
immediately
stops
generating
the
HH
reminder,
changes
its
HH
status
to
“clean”,
records
that
HH
was
performed
after
prompting
and
switches
to
the
state
S4
or
S2.
If
no
HH
action
is
performed
within
the
duration
of
the
prompting
signal
(20
s
in
this
study),
the
monitor
records
that
a
caregiver
decided
to
ignore
it,
stops
generating
the
HH
reminder
and
switches
back
to
the
state
S1
or
S3
depending
on
the
location
of
a
care-
giver.
For
simplicity
the
diagram
does
not
illustrate
certain
details
of
the
logic
applicable
to
individual
patient
zones
in
a
multi-bed
environment.
The
main
difference
for
a
multi-
bed
room
is
that
a
HH
opportunity
associated
with
leaving
an
individual
patient
environment
is
processed
only
when
the
monitor
detects
the
location
which
is
different
from
the
pre-
viously
detected
individual
patient
zone.
For
example,
if
after
finishing
working
with
a
patient
a
caregiver
leaves
the
room,
the
monitor
detects
the
external
stage
of
the
room
entrance
zone,
makes
the
decision
that
the
patient
environment
has
been
exited,
and
switches
to
the
state
S2
or
S6
depending
on
whether
or
not
a
HH
action
was
performed
before
leav-
ing
the
room.
If
a
caregiver
goes
directly
from
one
patient
bed
to
another,
HH
opportunities
associated
with
leaving
and
entering
a
patient
environment
essentially
merge,
the
monitor
switches
to
state
S5
reminding
about
the
need
for
a
HH
action
if
it
had
not
been
preformed
before
the
new
patient
zone
was
detected.
Expiry
time
constants
texp1and
texp2were
determined
based
on
time-stamped
data
collected
during
the
baseline
obser-
vational
study,
with
HH
expiry
time
outside
of
the
patient
room
(texp1)
set
to
60
s
and
HH
expiry
time
inside
the
patient
room
(texp2)
set
to
10
s
for
this
testing.
An
audible
signal
with
the
duration
of
20
s
was
used
in
this
study
as
the
HH
reminder.
A
baseline
observational
study
conducted
before
testing
the
system
was
organized
to
cover
different
time
slots
over
the
duration
of
the
shift
for
each
participant
and
totally
6
h
over
the
shift
were
observed
for
each
participant.
Three
methods
of
HH
were
recorded
in
the
observational
study:
wall-mounted
gel
dispensers,
personal
wearable
gel
dispensers
and
sink
and
soap
dispensers
installed
in
the
patient
bathrooms.
The
testing
of
the
electronic
system
started
five
days
after
completion
of
the
baseline
observational
study.
5.
Results
A
total
of
1438
events
of
entering
and
leaving
monitored
rooms
were
recorded
by
the
system
over
145
h
of
testing.
In
387
instances
out
of
1438
(Fig.
2A),
the
HH
status
of
a
care-
giver
was
recorded
as
clean,
indicating
that
a
HH
action
had
already
been
performed
within
the
previous
60
s
before
enter-
ing
a
patient
room,
or
within
the
previous
10
s
before
leaving
a
patient
room.
In
467
instances
(Fig.
2B)
the
HH
status
of
a
caregiver
was
not
set
to
clean
at
the
moment
of
entering
or
leaving
a
monitored
room,
but
a
HH
action
was
performed
within
the
duration
of
the
prompting
signal
generated
by
the
wearable
monitor.
Group
C
on
Fig.
2
includes
255
events
when
a
caregiver
did
not
perform
HH
but
left
a
patient
room
for
a
short
period
of
time
(less
than
the
duration
of
prompting
signal,
set
to
20
s
in
this
study)
and
returned
to
the
same
room,
or
when
a
caregiver
spent
less
than
20
s
in
the
room.
Group
D
on
Fig.
2
includes
HH
opportunities
that
were
clas-
sified
as
“ignored”.
This
group
is
of
particular
interest
and
it
is
further
divided
into
three
categories
(Fig.
3)
based
on
the
events
recorded
by
the
system
after
an
ignored
HH
prompt.
In
83
instances
(D.1)
the
next
recorded
event
was
still
a
HH
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
600
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
0
100
200
300
400
500
DCBA
Fig.
2
–
HH
opportunities
detected
by
the
system:
A
–
HH
status
set
to
clean;
B
–HH
performed
after
prompting
signal;
C
–
a
caregiver
entered
(left)
a
monitored
room
for
a
period
of
time
less
than
the
duration
of
prompting
signal,
no
HH
performed;
D
–
other
ignored
HH
prompts.
action,
with
almost
a
half
of
them
performed
in
less
than
a
minute
after
the
ignored
HH
prompt.
Category
D.2
includes
the
situations
similar
to
those
in
group
C,
but
when
a
care-
giver
entered
or
left
the
room
for
a
period
of
time
longer
than
the
duration
of
the
ignored
HH
prompt.
Finally,
81
instances
in
group
D.3
include
the
situations
when
a
caregiver
left
a
mon-
itored
room,
ignored
the
HH
prompt,
and
the
next
recorded
event
was
entering
a
different
patient
room.
The
system
recorded
a
total
of
931
HH
actions,
with
429
HH
actions
performed
with
wall-mounted
gel
dispensers
and
502
performed
with
personal
wearable
gel
dispensers.
6.
Feedback
from
nurses
Ten
nurses
participated
in
the
structured
individual
interview
upon
completion
of
the
testing.
0
100
200
300
400
500
D.3D.2D.1
Fig.
3
–
Events
recorded
by
the
system
after
ignored
HH
prompts
(group
D);
D1
–HH
action
performed
after
more
than
20
s
following
the
prompt,
D2
–
a
caregiver
leaves
the
room
(spent
more
than
20
s
inside
the
room)
or
returns
to
the
same
room
(room
was
left
for
more
than
20
s);
D3
–a
caregiver
goes
to
a
different
room.
Responses
from
participants
in
relation
to
parameters
of
HH
monitoring
system
are
summarized
in
Table
1.
When
discussing
the
prompting
functions
of
the
device,
staff
found
HH
prompts
acceptable
when
entering
patient
rooms
if
they
had
not
disinfected
their
hands.
Eight
partici-
pants
did
not
support
prompting
when
they
had
left
a
patient
room
for
a
short
time
and
returned
to
the
same
patient.
They
felt
that
picking
up
linens
from
a
cart
parked
in
the
hallway
or
retrieving
a
wheelchair
at
the
other
end
of
the
unit
should
not
warrant
a
prompt
from
the
device
to
‘rewash’
their
hands.
At
those
times
they
chose
to
ignore
the
prompt.
Interestingly,
these
participants
also
mentioned
that
in
these
circumstances
it
is
technically
impossible
to
perform
HH
as
they
are
wearing
gloves
and
would
not
remove
their
gloves
when
leaving
the
room
to
pick
up
an
item
for
the
patient.
One
participant
stated:
‘I
am
back
in
a
second
to
finish
off
with
that
patient’.
The
duration
of
the
prompting
signal
was
deemed
accept-
able,
and
participants
also
liked
the
idea
that
the
prompting
signal
was
not
continuous
(generated
until
HH
is
performed),
but
was
only
generated
for
a
short
period
of
time.
As
stated
by
one
nurse:
“Obviously
I
did
not
wash
my
hands
for
what-
ever
reason,
so
don’t
keep
beeping
at
me”.
On
the
other
hand,
most
participants
indicated
that
they
would
prefer
a
vibration
signal
rather
than
audible
prompt.
Reasons
for
this
included
that
patients
are
often
surrounded
by
equipment
that
produces
similar
beeping
signals
(i.e.,
feeding
pumps,
PCA
pumps,
etc.)
and
the
anonymity
that
comes
with
a
vibration
signal
rather
than
a
signal
that
other
staff
and
patients
can
hear.
None
of
the
participants
expressed
concerns
about
being
monitored
or
watched.
These
findings
confirmed
previous
research
about
the
acceptability
of
automated
HH
monitoring
[15].
In
relation
to
the
physical
form
of
the
devices,
nursing
staff
indicated
that
the
size
and
weight
of
the
wearable
compo-
nents
were
acceptable
and
that
devices
did
not
interfere
with
their
regular
care
routines.
Staff
did
not
feel
that
wearing
these
devices
negatively
affected
how
patients
and
colleagues
per-
ceived
them.
7.
Discussion
Our
purpose
at
this
stage
was
not
to
determine
the
precise
effectiveness
of
the
system.
This
trial
was
certainly
too
small
and
of
short
duration
to
make
a
reliable
estimate
of
the
impact
of
the
system
on
HH
and
the
sustainability
of
the
improved
performance.
However,
this
study
does
indicate
promising
potential
for
the
system
that
justifies
the
larger
trial.
The
data
recorded
by
the
system
indicated
an
average
of
6.42
HH
actions
per
hour,
that
is
about
53%
greater
than
the
rates
(4.2
HH
per
hour)
obtained
in
the
baseline
observational
study.
This
improvement
may
have
been
underestimated
because
the
presence
of
an
observer
is
likely
to
have
resulted
in
increased
HH
during
the
baseline
study
(Hawthorne
effect).
In
addition
it
should
be
noted
that
25%
of
HH
actions
during
the
baseline
study
were
performed
with
the
sink
and
soap
dispenser
in
a
patient
bathroom.
At
the
time
of
this
study
these
soap
dispensers
were
not
integrated
into
the
system
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
601
Table
1
–
Responses
from
nurses
participating
in
interviews.
Parameters
of
the
monitoring
system
Responses
Acceptable
type
of
HH
prompting
signal Audible
signal
2
Vibration
signal 8
Acceptable
duration
of
HH
prompting
signal Continuous
signal
1
Non-continuous
signal
9
Privacy
concerns
related
to
automated
HH
monitoring Indicated
privacy
concerns
0
No
privacy
concerns
10
Physical
form
and
shape
of
the
devices Acceptable
size
and
weight
10
Interference
with
care
routines 0
and
therefore
no
HH
actions
performed
with
this
method
were
processed
by
the
electronics.
A
total
of
853
HH
actions
were
related
to
automatically
detected
HH
opportunities
of
categories
A
and
B
(Fig.
2),
which
is
about
90%
of
the
total
HH
actions
recorded
by
the
system.
Even
if
only
these
HH
actions
are
taken
into
account,
the
average
HH
rate
indicated
by
the
system
is
5.88
per
hour
com-
pared
to
4.2
per
hour
obtained
in
the
observational
study
and
calculated
using
the
total
number
of
HH
actions
performed,
including
those
performed
with
sink
and
soap
dispensers
in
patient
bathrooms,
which
were
not
integrated
with
the
system
in
this
study.
As
it
appears
from
Fig.
4,
more
than
80%
of
HH
opportuni-
ties
with
HH
status
set
to
clean
were
detected
when
a
caregiver
entered
a
patient
room,
while
HH
opportunities
performed
as
a
result
of
prompting
signal
(Fig.
5)
were
mostly
associ-
ated
with
leaving
a
patient
room.
This
finding
opens
up
an
interesting
debate
of
the
nurses’
awareness
and
consequent
actions
to
disinfect
their
hands
when
they
start
a
new
care
task,
yet,
ignoring
the
importance
of
HH
upon
completing
a
task.
Groups
C
and
D
shown
on
Fig.
2
represent
the
events
of
entering
or
leaving
a
patient
room
with
no
HH
action
per-
formed.
Group
C
include
the
situations
when
a
patient
room
was
entered
or
left
for
a
very
short
period
of
time
(less
than
10
s
for
two
thirds
of
instances
in
this
category)
and
a
caregiver
probably
decided
that
HH
was
not
necessary.
0
100
200
300
400
500
A.2A.1
Fig.
4
–
Entering/leaving
monitored
rooms
with
HH
status
set
to
clean
(group
A);
A.1
–
opportunities
associated
with
entering
a
patient
room
and
A.2
– opportunities
associated
with
leaving
a
patient
room.
The
results
demonstrate
that
the
software
employed
by
an
electronic
HH
reminding
and
monitoring
system
must
be
quite
sophisticated.
For
example,
if
a
simple
system
is
used
that
prompts
every
time
a
caregiver
enters
or
leaves
a
patient
room
then
the
impact
of
the
prompts
will
likely
be
lessened
because
in
nearly
one
third
of
these
cases
our
system
showed
that
the
caregiver
had
already
performed
HH
shortly
before
entry
or
exit.
It
follows
that
any
automated
HH
monitoring
system
should
include
hardware
and
software
to
detect
the
performance
of
HH
along
with
location
detection
functions
and
should
be
capable
of
relating
the
events
of
performing
HH
and
entering/exiting
monitored
areas.
In
any
given
clini-
cal
environment
a
detailed
time-stamped
observational
study
has
to
be
conducted
to
determine
time
constants,
logic,
and
locations
of
monitored
areas
prior
to
the
implementation
of
an
automated
HH
monitoring
solution.
Approximately
half
of
the
HH
actions
were
performed
using
wearable
alcohol
gel
dispensers
presumably
because
of
their
constant
availability
and
convenience.
In
order
to
be
most
effective
it
follows
that
a
HH
monitoring
system
should
be
able
to
include
instrumented
wearable
dispensers.
8.
Limitations
Wall-mounted
soap
dispensers,
installed
in
ensuite
bath-
rooms
were
not
integrated
into
the
system
at
the
time
of
this
0
100
200
300
400
500
B.2B.1
Fig.
5
–
Entering/leaving
monitored
rooms
when
HH
was
performed
within
the
duration
of
prompting
signal
(group
B);
B.1
– opportunities
associated
with
entering
a
patient
room
and
B.2
–
opportunities
associated
with
leaving
a
patient
room.
Author's personal copy
Journal
Identification
=
IJB
Article
Identification
=
2751
Date:
July
2,
2011
Time:
3:7
am
602
i
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
m
e
d
i
c
a
l
i
n
f
o
r
m
a
t
i
c
s
8
0
(
2
0
1
1
)
596–603
study.
HH
actions
performed
with
this
type
of
dispensers
were
not
recorded
and
processed
so
the
improvement
in
HH
caused
by
the
system
was
probably
underestimated.
It
is
also
important
to
interpret
the
results
in
light
of
the
clinical
setting,
that
is,
complex
continuing
care,
which
might
warrant
a
different
acuity
and
care
model
then
other
clinical
settings.
9.
Conclusions
The
results
obtained
when
testing
the
embedded
HH
monitor-
ing
system
demonstrated
the
feasibility
of
HH
improvement
and
proved
that
proposed
solution
merits
a
larger
and
longer
clinical
trial
to
measure
the
degree
of
improvement
and
the
sustainability
of
that
improvement.
The
study
demonstrated
that
hardware
and
software
con-
figurations
must
employ
appropriate
time
constants
in
order
to
implement
HH
monitoring
algorithms
which
would
allow
making
appropriate
decisions
on
the
need
for
prompting
HH
actions.
These
algorithms
might
vary
between
clinical
settings
and
need
to
be
determined
prior
to
the
installation
of
technol-
ogy
in
order
to
provide
a
practical
and
credible
HH
monitoring
solution.
Efficient
systems
that
minimize
loss
of
caregiving
time
should
also
include
instrumented
wearable
alcohol
gel
dis-
pensers.
Authors
contributions
1.
Alexander
Levchenko
is
a
Control
Systems
Specialist
with
R&D
Technology
Team
at
Toronto
Rehabilitation
Institute.
He
is
the
lead
developer
of
the
automated
hand
hygiene
monitoring
system.
He
contributed
in
data
acquisition,
interpretation
of
data,
and
writing
the
manuscript.
2.
Veronique
Boscart
is
a
Registered
Nurse
with
R&D
Tech-
nology
Team
at
Toronto
Rehabilitation
Institute.
She
contributed
in
design
of
the
study,
data
interpretation
and
writing
the
manuscript.
3. Geoffrey
Fernie
is
VP
Research
at
Toronto
Rehabilitation
Institute
and
the
leader
of
R&D
Technology
Team.
He
con-
tributed
in
drafting,
revising
and
final
approval
of
the
article.
Conflict
of
interest
The
potential
conflict
of
interest
includes
the
anticipated
commercialization
of
the
new
hand
hygiene
technology
in
the
future.
We
have
applied
for
patent
protection
for
this
intellectual
property.
Royalties
generated
from
the
IP
will
be
distributed
according
to
the
institutional
policy
at
Toronto
Rehabilitation
Institute.
Two
of
the
authors
who
are
cited
as
inventors
of
the
technology
may
benefit
from
royalties
in
the
future
(Dr.
Levchenko,
Dr.
Fernie).
Dr.
Fernie
has
committed
to
donating
his
personal
share
of
royalties
to
Toronto
Reha-
bilitation
Institute.
One
of
the
commercial
partners
holds
the
option
to
license
the
IP
from
this
research.
Summary
points
What
was
known
before
the
study:
•
Despite
various
educational
and
management
inter-
ventions
compliance
with
hand
hygiene
requirements
by
healthcare
staff
remains
inadequate.
•Pilot
clinical
experiments
have
been
conducted
in
sim-
ulated
and
occupied
patient
rooms
to
test
technical
performance
of
the
automated
hand
hygiene
mon-
itoring
system
and
to
collect
initial
feedback
from
caregivers.
What
this
study
has
added:
•
For
the
first
time
nurses
used
the
technology
for
the
duration
of
their
entire
shifts
and
hand
hygiene
per-
formance
data
was
recorded
automatically
with
no
observer
present.
•The
results
of
the
study
demonstrated
potential
of
hand
hygiene
improvement
with
automated
moni-
toring
system
and
confirmed
our
initial
findings
on
acceptability
of
technology.
Acknowledgments
This
research
was
supported
by
Health
Technology
Exchange
(HTX),
Canadian
Institutes
of
Health
Research
(Funding
ref-
erence
#HTX
80047
and
XHT
83447),
Registered
Nurses
Foundation
of
Ontario,
Andrew
J
Hart
Enterprises
Limited,
Toronto
Rehabilitation
Institute
and
the
Ontario
Centers
of
Excellence
(Project
#
TO-CR-1007
4-08).
Toronto
Rehabilitation
Institute
receives
funding
under
the
Provincial
Rehabilitation
Research
Program
from
the
Ministry
of
Health
and
Long-Term
Care
in
Ontario.
The
authors
wish
to
acknowledge
the
contri-
butions
of
other
members
of
the
hand
hygiene
project
team
including:
S.
Pong,
J.
Ibbett,
and
P.
Holliday.
r
e
f
e
r
e
n
c
e
s
[1]
W.
Jarvis,
R.
Walker,
Selected
aspects
of
the
socioeconomic
impact
of
nosocomial
infections:
morbidity,
mortality,
cost,
and
prevention,
Infect.
Control
Hosp.
Epidemiol.
17
(8)
(1996)
552–557.
[2]
P.W.
Stone,
E.
Larson,
L.N.
Kawar,
A
systematic
audit
of
economic
evidence
linking
nosocomial
infections
and
infection
control
interventions:
1990–2000,
Am.
J.
Infect.
Control
30
(3)
(2002)
145–152.
[3]
C.
Zhan,
M.R.
Miller,
Excess
length
of
stay,
charges
and
mortality
attributable
to
medical
injuries
during
hospitalization,
J.
Am.
Med.
Assoc.
290
(14)
(2003)
1868–1874.
[4]
M.H.
Wilcox,
J.
Dave,
The
cost