Journal of Hospital Infection 83 (2013) 276e283

Available online at www.sciencedirect.com

Journal of Hospital Infection

journal homepage: www.elsevierhealth.com/journals/jhin

Predictors of hand hygiene compliance in the era

of alcohol-based hand rinse
G. Lebovic a, N. Siddiqui b, M.P. Muller a, b, c, *
a
Applied Health Research Center, Keenan Research Centre, Li Ka Shing Knowledge Institute, St Michael’s Hospital,
Toronto, Canada
b
Infection Prevention and Control Service, St Michael’s Hospital, Toronto, Canada
c
Department of Medicine, St Michael’s Hospital, Toronto, Canada

A R T I C L E I N F O S U M M A R Y

Article history: Background: Predictors of hand-hygiene compliance have not been re-evaluated in the
Received 23 May 2012 alcohol-based hand rinse (ABHR) era.
Accepted 7 January 2013 Aim: To re-evaluate predictors of hand-hygiene compliance in the era of ABHR.
Available online 14 February Methods: Hand-hygiene compliance was monitored at a Canadian teaching hospital for
2013 a period of two years using direct observation. Standardized definitions of compliance
were used and potential predictors of compliance were recorded. A generalized linear
Keywords: mixed model was developed to evaluate the impact of predictors of hand-hygiene
Alcohol-based hand rinse compliance while correcting for clustering.
Hand hygiene Findings: We observed 7364 opportunities for hand hygiene among 3487 healthcare
Hand-hygiene compliance workers. Hand-hygiene compliance was 45% and did not vary over time. Predictors of
improved compliance on multivariate analysis included the indication for hand hygiene
with higher compliance seen after body fluid exposure (odds ratio: 4.7; 95% confidence
interval: 3.7e6.1) and after patient contact (3.9; 3.5e4.4) compared with hand hygiene
prior to patient contact. Glove use was associated with higher compliance (1.3; 1.1e1.4).
A professional designation other than nurse or physician was associated with lower
compliance (0.72; 0.61e0.86). The number of hand hygiene opportunities per hour was not
associated with lower compliance. Higher ward level use of ABHR (vs use of soap/water)
was associated with better compliance (P ¼ 0.035).
Conclusions: In the ABHR era a higher frequency of hand-hygiene opportunities is no
longer the primary barrier to achieving optimal hand-hygiene compliance. However,
heterogeneous use of ABHR by ward may still provide a target for improvement.
ª 2013 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Address: Infection Prevention and Control,
St Michael’s Hospital, 30 Bond Street, Room 4-179, Cardinal Carter
Wing North, Toronto, Ontario, Canada M5B 1W8. Tel.: þ1 416 864 5568;
fax: þ1 416 864 5310.
E-mail address: mullerm@smh.ca (M.P. Muller).
Introduction
In 2002 it was estimated that healthcare-associated infec-
tions (HCAIs) affected 1.7 million patients in the USA and
resulted in 99,000 deaths.1 The performance of hand hygiene
by healthcare workers is critical to preventing HCAI and is
mandated in international and national infection control
guidelines.2,3 Despite this, hand-hygiene compliance among
healthcare workers remains poor.4

0195-6701/$ e see front matter ª 2013 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jhin.2013.01.001
 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
Several observational studies have identified risk factors
for non-compliance; in particular it has been suggested that
a higher number of hand-hygiene opportunities per hour
correlates directly with lower hand-hygiene compliance.3,5e7
This result suggests that efforts to improve hand hygiene
should include interventions designed to reduce the time
required to perform hand hygiene. Such strategies focus on
the use of alcohol-based hand rinse (ABHR) located at or
near the point of care rather than soap and water, as this
reduces the time required to perform hand hygiene sub-
stantially.2,3,8
In recent years our hospital, as well as many others, has
moved away from soap and water and towards ABHR as the
primary mode of hand hygiene and has implemented multi-
modal strategies to promote hand hygiene. We hypothesized
that these changes may have reduced the impact of
previously identified predictors of poor hand-hygiene
compliance.
To address this, we evaluated observational data on hand-
hygiene compliance collected at our institution over a two-
year period.
100
277
Methods
Study design, population and setting
The study is a longitudinal observational study of hand-
hygiene compliance conducted at St Michael’s Hospital,
a 450-bed teaching hospital in Toronto, Canada. Data were
collected by direct observation for two years from 1 October
2008 to 30 September 2010 from all inpatient units and from
the emergency department and haemodialysis unit.
Prior to the study period, ABHR dispensers were already
located in most patient rooms and our hospital provided
infection control training to all new staff that included
a session on appropriate hand hygiene. In the year prior to the
study onset, a hand-hygiene improvement team was estab-
lished, administrative support for hand-hygiene improvement
was obtained, a single promotional poster campaign was
launched and sporadic hand-hygiene audits were conducted.
During the study period, stepwise implementation of
a multimodal hand-hygiene improvement intervention occurred
(Figure 1). The intervention included the hiring of a hand-

A/P
90

E/F/L
80
70
60
% compliance
50
40
30
20
10
0

Q4/08 Q1/09 Q2/09 Q3/09 Q4/09 Q1/10 Q2/10 Q3/10

Quarter

Figure 1. Quarterly hand-hygiene compliance, October 2008 to September 2010, showing date of initiation of components of the
multimodal hand-hygiene programme. A: initiation of hospital-wide hand-hygiene audits; P: initiation of hand-hygiene poster campaigns;
E: initiation of hospital-wide staff hand-hygiene education (ended Q2/10); F: initiation of quarterly compliance feedback to senior
administration; L: initiation of hospital-wide programme to ensure correct placement of hand sanitizer. (Univariate P-value for the
association between quarter and compliance, adjusted for random effects ¼ 0.42.)
278 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
hygiene coordinator, routine hospital-wide hand-hygiene audits using cubic splines was performed to evaluate the relationship
with feedback to our medical and administrative leadership, an between ward level use of ABHR (vs soap and water and
education programme for all healthcare workers, an awareness compliance).
campaign involving regular poster campaigns promoting hand
hygiene, and an ongoing collaboration with engineering, en-
Results
vironmental services and clinical units to ensure that all rooms
had appropriately located and functional ABHR dispensers.
During the two-year study period 3487 healthcare workers
and 7364 hand-hygiene opportunities were observed in 1180
Methodology of hand-hygiene auditing and data audit sessions or 393 h of observation. A mean of 148 audit
collection sessions were conducted per quarter, rising from 29 in the first
quarter to a maximum of 354 in the seventh quarter. Of the
The auditing methodology was developed by the Ontario 1180 sessions, 3% (35) were conducted in duplicate (i.e. with
Ministry of Health and Long Term Care (MOHLTC). Indications two observers) and the inter-observer agreement on whether
for hand hygiene were defined using the ‘four moments of hand hand hygiene was performed was high (kappa ¼ 97%). The
hygiene’ that represent a modification of the World Health mean number of hand-hygiene opportunities per hour of
Organization’s ‘five moments of hand hygiene’.3 The four observation was 19.3 (SD: 12.2) and varied by ward from 9
moments are defined as follows. Moment I: hand hygiene per- (palliative care) to 24 [cardiovascular intensive care unit
formed before contact with the patient/patient’s environment; (ICU)]. The median number of healthcare workers observed per
moment II: hand hygiene performed before an aseptic proce- session was 3 (range: 0e13). Of the healthcare workers
dure; moment III: hand hygiene performed after a body fluid observed, 64% (2237) were nurses, 19% (649) were physicians,
exposure; moment IV: hand hygiene performed after contact and the remaining 17% (601) were distributed among 14
with a patient/patient’s environment. Trained auditors directly different professional categories. Nurses accounted for 67% of
observed healthcare workers’ hand hygiene in 20 min observa- observed hand-hygiene opportunities, physicians for 15% and
tion sessions on weekdays throughout the study period and other healthcare workers for 18%. Approximately 7% (82/1180)
results were recorded on a standardized data collection tool. of sessions involved patients in isolation and 32% of observed
Data were checked for completeness and entered into a data- opportunities involved healthcare workers wearing gloves.
base (Microsoft Access 2002, Microsoft, Redmond, CA, USA). Glove use was more frequent when patients were in isolation
Sessions with two audits were conducted intermittently to (47% vs 31%, P < 0.05).
ensure agreement between auditors, especially new auditors.
Table I
Statistical analysis Factors associated with hand hygiene compliance
Category Variable Compliance OR (95% CI)
All statistical analysis was performed using R Statistical
Software (R version 2.13.0). Inter-observer agreement was Indication Moment I 31% (879/2868) 1.00
measured using Cohen’s kappa. A generalized linear mixed for hand Moment II 34% (128/376) 1.18 (0.90e1.53)
model (multilevel model) was fitted to the data to control for hygiene Moment III 61% (251/410) 5.07 (3.94e6.50)**
nesting of repeated measurements.9 Repeated measurements Moment IV 56% (1927/3419) 3.81 (3.40e4.30)**
within the same healthcare worker, within the same session Mixed 37% (107/291) 1.50 (1.10e2.03)*
and within the same ward were considered as random effects Isolation Patient 44% (296/673) 1.02 (0.79e1.32)
within this model. Random effects are used in repeated status isolated
measures modelling to account for correlation arising within No isolation 45% (2996/6691) 1.00
clusters and measure natural heterogeneity due to unmeasured Profession Nurse 46% (2258/4914) 1.00
factors.9 Additional predictors including professional group, Physician 44% (485/1108) 0.89 (0.76e1.05)
indication for hand hygiene, glove use, isolation status, ward Other 41% (549/1342) 0.77 (0.66e0.90)**
type and number of hand-hygiene opportunities per hour were Glove use Gloves used 47% (111/2386) 1.08 (0.96e1.22)
included in the multivariate model as fixed-effects. To calcu- No gloves 44% (2181/4978) 1.00
late univariate odds ratios for the fixed-effect predictors, each Ward type Medical 45% (1331/2946) 1.00
predictor was incorporated individually into a model adjusting Surgical 45% (792/1749) 1.09 (0.73e1.61)
for all of the random effects. Additionally, variability of the Intensive care 48% (820/1724) 1.26 (0.83e1.91)
random effects variables was expressed by evaluating the odds Other/ 37% (349/945) 0.72 (0.46e1.10)
of compliance at the 75th percentile of the distribution versus outpatient
the 25th percentile for each variable. Activity Quartile
To evaluate the impact of workload or busyness on compli- indexa (range)
ance, a surrogate measure or ‘activity index’ was defined as Q1 (3e11) 49% (344/702) 1.00
the number of opportunities per session (session level analysis) Q2 (12e18) 44% (805/1828) 0.78 (0.63e0.98)
or the number of opportunities per session per ward (ward level Q3 (19e27) 45% (889/1971) 0.84 (0.66e1.06)
analysis) expressed as opportunities per hour. A moving linear Q4 (>27) 44% (1254/2863) 0.76 (0.60e0.95)
regression smoother (loess method) was used to visually eval- Total 45% (3292/7364)
uate the relationship between activity level and compliance OR, odds ratio; CI, confidence interval.
and tests of significance were performed using a univariate *P < 0.05, **P < 0.005.
model adjusting for random effects.10,11 Univariate analysis a
Number of hand-hygiene opportunities per hour.
 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
Overall hand-hygiene compliance was 45% and remained
stable throughout the study period (Figure 1). The most
common indications for hand hygiene were hand hygiene after
contact with patients or their environment (46% of observa-
tions) and hand hygiene before contact with patients or their
environment (39% of observations). Hand hygiene was most
frequently performed using ABHR rather than soap and water
(74% vs 26%) but this practice varied by both ward and
profession.
In our univariate analysis, the only significant predictors of
improved compliance were indication for hand hygiene after
patient/patient environment contact or body fluid contact (as
compared to hand hygiene indicated prior to patient/patient
environment contact) while the only significant predictor of
reduced compliance was a professional designation as ‘non-
nurse, non-physician’ (Table I). Physician status, glove use, and
isolation status were not associated with reduced compliance.
Ward type (i.e. ICU vs medical vs surgical vs other) was not
predictive, and compliance varied among the 21 individual
wards observed, from 31% to 65%, and the best-performing
wards were relatively consistent throughout the study period.
We also evaluated the impact of ‘activity level’ on compli-
ance, both at the session and at the ward level. No significant
association was found between compliance and session
(P ¼ 0.13) or ward ‘activity level’ (Table I; Figures 2 and 3).
100
90
80
70
60
% compliance
50
40
30
20
10
0
10 15
Hand-hygiene opportunities per hour
Figure 2. Hand-hygiene compliance and hand-hygiene opportunities per hour at the ward level.
279
Finally, the impact of ward-level uptake of ABHR on
compliance was evaluated. Use of ABHR varied widely by ward,
from a high of 98% on a medical unit to 38% in our haemodialysis
unit. A trend towards higher compliance on units with more
frequent use of ABHR (vs soap/water) was apparent and was
statistically significant (P ¼ 0.035) (Figure 4). The neonatal ICU
was a high-compliance, low-ABHR use outlier. This finding is
explained by a strong tradition of excellent infection control
practices on that unit, combined with a reluctance to adopt
ABHR because of concerns that soap and water may be more
appropriate for their patient population due to the frequent
potential for contamination with body fluids. There was also
some variation in the use of ABHR by profession, with higher
use seen in physicians (82%) and non-nurse, non-physician staff
(80%) compared with nurses (71%).
In the multivariate model the predictors of improved
compliance were hand hygiene performed after body fluid
exposure (odds ratio: 4.7; 95% confidence interval: 3.7e6.1)
and hand hygiene performed after patient/patient environ-
ment contact (3.9; 3.5e4.4) compared with hand hygiene
performed before patient/patient environment contact. Glove
use was a weak predictor of improved compliance (1.3;
1.1e1.4) whereas a professional group other than nurse or
physician was associated with lower compliance (0.72;
0.61e0.86).
20 25
280 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
100
90
80
70
60
% compliance
50
40
30
20
10
0
0 10 20 30
Hand-hygiene opportunities per hour
Figure 3. Hand-hygiene compliance and hand-hygiene opportunities per hour at the session level.
In this analysis, clustering occurred at the healthcare
worker, session, and ward levels. The standard deviation for
the random effect was smallest for the ward level (SD: 0.37)
compared with the level of the healthcare worker (SD: 0.76)
and the session level (SD: 0.76). This can further be understood
by examining the odds ratio for comparison between high-
performing (75th percentile) and low-performing (25th
percentile) levels of wards, healthcare workers and sessions
which were 1.65, 2.78, and 2.77 respectively.
Discussion
In this study, trained observers observed 7364 hand-hygiene
opportunities and 3487 healthcare workers, making this one of
the largest observational studies of the predictors of hand
hygiene compliance to date. Whereas other large studies have
addressed this issue, our study is important given the changes that
have occurred in the related fields of hand hygiene, infection
control, and patient safety over the past 10 years.3e5,7 Changes
over this time-period include the recognition that multimodal
hand-hygiene improvement strategies may effectively improve
hand-hygiene compliance, release of comprehensive World
Health Organization (WHO) hand-hygiene guidelines and the
widespread adoption of alcohol-based hand rub in North Amer-
ican hospitals.3,12 Additionally, mandatory public reporting of
40 50 60 70 80
hospital hand-hygiene compliance was implemented in Ontario in
2009.13 Therefore, this study provides an opportunity to re-
evaluate the epidemiology of hand hygiene after a decade of
intensive efforts to improve its profile and performance.
One of the surprising findings of our study was the consis-
tently low level of compliance observed, despite the imple-
mentation of a multimodal hand hygiene campaign during the
study period that included audit and feedback, education,
regular awareness (i.e. poster) campaigns, improved access to
ABHR at or near the point of care and administrative support.
Over the two-year study period, compliance was 45% and was
stable on a quarterly basis.
In retrospect, we believe that these results should not be
interpreted as indicative that multimodal strategies for hand-
hygiene promotion are not effective. Rather, the success of
multimodal interventions cannot be predicted based solely on
the type of intervention utilized (e.g. audit and feedback,
education) independent of an understanding of the specific
details of how such interventions were implemented. A careful
evaluation of each component of our intervention identified
potential weaknesses that have since been addressed (e.g.
feedback was initially provided only to upper levels of
management and never reached front-line workers; turnout at
educational sessions was poor on many units and in-services are
now supplemented with a mandatory online learning module).
 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
100
90
80
70
60
% compliance
50
40
30
20
10
0
30 40 50
% of hand hygiene performed using ABHR
Figure 4. Hand-hygiene compliance and use of alcohol-based hand rinse (ABHR) vs soap and water at the ward level (univariate
P ¼ 0.035).
Another intriguing finding in our study was the absence of an
association between ‘activity level’ (the mean number of
opportunities identified during a session or on a specific ward)
and compliance. Previous studies identified a strong inverse
relationship between ‘activity level’ and compliance. These
studies postulated that busier healthcare workers are less
likely to perform hand hygiene and that the implementation of
ABHR at the point of care would improve compliance by
reducing the time required for hand hygiene.3e7,12 The failure
to observe this association in our study likely occurred because
ABHR is already widely available at our facility. We did,
however, note an association between ward use of ABHR and
compliance; wards using more ABHR and less soap tended to
have higher compliance. For two units the reasons for low ABHR
uptake were apparent (i.e. poor location of dispensers in the
haemodialysis unit, concern that body fluid exposure in the
neonatal ICU requires washing with soap and water). A useful
improvement strategy may be to target education and human
factors, engineering approaches on units where both compli-
ance and uptake of ABHR use is low, focusing on other strate-
gies (e.g. audit with feedback to frontline staff, use of role
models) on units where uptake of ABHR is complete but
compliance is not yet optimal.
Our study failed to identify a gap in performance between
physicians and nurses; historically, compliance has been much
281
60 70 80 90 100
higher in nurses than in physicians.4,7,14,15 We postulate that
the increased use of ABHR (vs soap) by physicians compared
with nurses may account for this finding. The study does not
address the issue of whether the higher reliance on soap and
water on some units was appropriate (e.g. used when hands are
visibly soiled) or inappropriate (e.g. used for routine hand
hygiene due to lack of appropriately placed or maintained
ABHR dispensers). We did, however, note that compliance was
significantly lower in the ‘non-nurse, non-physician’ group.
This may have been a result of our educational strategy which
was specifically tailored to physicians and nurses and may have
failed to capture other groups (e.g. housekeepers, radiology
technicians, chaplains, etc.)
Our study also failed to identify an association between
glove use and lower hand-hygiene compliance. In fact, a weak
association between glove use and higher compliance was
identified in the multivariate model. We suspect the reason for
this difference is that all of our educational material strongly
emphasizes the importance of hand hygiene regardless of
whether gloves are worn.
Because both glove use and isolation status were monitored,
and because infection control policies require glove use for
almost all isolation indications at our facility, our hand-hygiene
audits also provide a surrogate audit on adherence to recom-
mended additional practices. Most glove use was for general
282 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283
patient care given that >30% of observed hand hygiene
opportunities involved glove use (whereas only 7% of observa-
tion sessions involved isolated patients). However, whereas
glove use was more frequently observed for isolated patients
(47% vs 31%), this nevertheless represents a surprisingly low
compliance with required use of personal protective equip-
ment for patients in isolation. Given the cost and difficulty of
implementing a robust hand hygiene audit and feedback pro-
gramme, it is a useful observation that, without substantially
additional workload, data on compliance with other aspects of
appropriate infection control practices can also be evaluated.
Evaluation of the random effects variables (i.e. healthcare
worker, session and ward) demonstrated lower variability
between wards than between healthcare workers or sessions.
This may be related to our measurement methodology, as there
were large numbers of observations for each ward, with
smaller numbers per session or per individual healthcare
worker. Furthermore, compliance at the level of an individual
healthcare worker or session was frequently zero or 100%
(Figure 3).
Our data have several limitations. Hand hygiene data were
measured by direct observation and they overestimate
compliance due to the effect of observation on the healthcare
workers being observed (i.e. the Hawthorne effect).16 However,
direct observation, despite its limitations, remains the gold
standard for monitoring hand-hygiene compliance.3,16 As part of
the study, we did not formally measure inter-observer agree-
ment on a randomly selected subset of observation sessions.
While this is an important limitation, few hand-hygiene studies
report inter-observer agreement, and our results likely under-
estimate agreement as duplicate observation sessions targeted
less experienced observers and/or observers obtaining unusual
results.4,17 The definition of hand-hygiene compliance used
here differs slightly from the US Centers for Disease Control and
Prevention and WHO definitions of hand hygiene.2,3 Specifically,
contact with the patient or the environment are considered part
of the same moment. However, the definitions are identical in
most aspects and are largely consistent with these other defi-
nitions. Because audits were conducted anonymously, esti-
mates of clustering effects at the healthcare worker level only
evaluated healthcare workers with repeated hand-hygiene
opportunities within a session and could not evaluate effects
that clustered across sessions. Also, elements of a multimodal
hand hygiene improvement plan were implemented during the
observation period. However, compliance results were stable
throughout the study, suggesting that the programme had little
impact and it is unlikely that the programme could have altered
the impact of hand-hygiene predictors without improving
compliance. Finally, the study was conducted at a single facility
and care should be taken in assuming that our results are broadly
generalizable. However, our hospital is a typical teaching
hospital, similar to other hospitals where hand-hygiene epide-
miology has been assessed.2,3,7
In summary, we have re-evaluated the epidemiology of
hand-hygiene compliance at a large teaching hospital in the
ABHR and patient safety era. Although many predictors of
hand-hygiene compliance remain unchanged, such as the
tendency of healthcare workers preferentially to perform hand
hygiene after, rather than before, patient contact, significant
changes were also observed, including relative improvement in
compliance among physicians compared with nurses and the
loss of an association between ‘activity level’ and compliance,
possibly related to the widespread use of ABHR. Whereas this
study failed to demonstrate improvements related to the
implementation of a multimodal hand-hygiene improvement
campaign, we have identified potential targets for future
quality improvement efforts including using strategies to
improve uptake of ABHR on low-ABHR-use wards and
strengthening the components of our hospital-wide multidis-
ciplinary improvement strategy.
Acknowledgements
We would like to acknowledge the hard work and dedication
of all the hand-hygiene auditors who collected the data and
who have contributed to our efforts to improve hand hygiene
compliance through audit and feedback.
Conflict of interest statement
None declared.
Funding sources
None.
References
1. Klevens RM, Edwards JR, Richards Jr CL, et al. Estimating health
care-associated infections and deaths in U.S. hospitals, 2002.
Public Health Rep 2007;122:160e166.
2. Boyce JM, Pittet D. Healthcare Infection Control Practices Advi-
sory Committee, HICPAC/SHEA/APIC/IDSA Hand Hygiene Task
Force. Guideline for Hand Hygiene in Health-Care Settings.
Recommendations of the Healthcare Infection Control Practices
Advisory Committee and the HICPAC/SHEA/APIC/IDSA Hand
Hygiene Task Force. Am J Infect Control 2002;30:S1e46.
3. Pittet D, Allegranzi B, Boyce J, et al. The World Health Organi-
zation guidelines on hand hygiene in health care and their
consensus recommendations. Infect Control Hosp Epidemiol
2009;30:611e622.
4. Erasmus V, Daha TJ, Brug H, et al. Systematic review of studies on
compliance with hand hygiene guidelines in hospital care. Infect
Control Hosp Epidemiol 2010;31:283e294.
5. O’Boyle CA, Henly SJ, Larson E. Understanding adherence to hand
hygiene recommendations: the theory of planned behavior. Am J
Infect Control 2001;29:352e360.
6. Pan A, Mondello P, Posfay-Barbe K, et al. Hand hygiene and glove
use behavior in an Italian hospital. Infect Control Hosp Epidemiol
2007;28:1099e1102.
7. Pittet D, Mourouga P, Perneger TV. Compliance with handwashing
in a teaching hospital. Ann Intern Med 1999;130:126e130.
8. Voss A, Widmer AF. No time for handwashing!? Handwashing versus
alcoholic rub: can we afford 100% compliance? Infect Control Hosp
Epidemiol 1997;18:205e208.
9. Diggle P, Heagerty P, Liang K-Y, et al. Random effects models. In:
Diggle P, Heagerty P, Liang K-Y, Zeger SL, editors. Analysis of
longitudinal data. 2nd ed. Oxford: Oxford University Press; 2002.
169e184.
10. Cleveland WS. Robust locally weighted regression and smoothing
scatterplots. J Am Statist Assoc 1979;74:829e836.
11. Harrell Jr FE. Exploring trends with nonparametric regression. In:
Harrell Jr FE, editor. Regression modeling strategies with appli-
cations to linear models, logistic regression and survival analysis.
1st ed. New York: Springer; 2001. p. 299e326.
12. Pittet D, Hugonnet S, Harbarth S, et al. Effectiveness of a hospital-
wide programme to improve compliance with hand hygiene.
Lancet 2000;356:1307e1312.
13. Muller MP, Detsky AS. Public reporting of hospital hand hygiene
compliance e helpful or harmful? JAMA 2010;304:1116e1117.
 G. Lebovic et al. / Journal of Hospital Infection 83 (2013) 276e283 283
14. Hugonnet S, Perneger TV, Pittet D. Alcohol-based handrub 16. Haas JP, Larson EL. Measurement of compliance with hand
improves compliance with hand hygiene in intensive care units. hygiene. J Hosp Infect 2007;66:6e14.
Arch Intern Med 2002;162:1037e1043. 17. Gould DJ, Chudleigh J, Drey NS, Moralejo D. Measuring hand-
15. Lipsett PA, Swoboda SM. Handwashing compliance depends on washing performance in health service audits and research
professional status. Surg Infect 2001;2:241e245. studies. J Hosp Infect 2007;66:109e115.