5

Journal of Hospital Infection 95 (2017) 344e351 Available online at www.sciencedirect.com

Journal of Hospital Infection journal homepage: www.elsevierhealth.com/journals/jhin

Evaluation of hand hygiene compliance and associated factors with a radio-frequency-identification-based real-time continuous automated monitoring system J-C. Dufour a, b, *, P. Reynier a, b, c, S. Boudjema c, d, A. Soto Aladro c, d, R. Giorgi a, b, P. Brouqui c, d a

Aix Marseille Univ, INSERM, IRD, SESSTIM, Sciences Economiques and Sociales de la Sante´ et Traitement de l’Information Me´dicale, Marseille, France b Assistance Publique Hoˆpitaux de Marseille, Service Biostatistique et Technologies de l’Information et de la Communication, Hoˆpital de la Timone, Marseille, France c Institut Hospitalo Universitaire Mediterrane´e Infection, Marseille, France d Aix-Marseille Universite´, Unite´ de Recherche sur les Maladies Infectieuses et Tropicales Emergentes, UM63, CNRS 7278, IRD 198, INSERM 1095, Marseille, France

A R T I C L E

I N F O

Article history: Received 23 December 2016 Accepted 1 February 2017 Available online 6 February 2017 Keywords: Hand hygiene Electronic monitoring Radiofrequency identification Continuous monitoring

S U M M A R Y

Background: Hand hygiene is a major means for preventing healthcare-associated infections. One critical point in understanding poor compliance is the lack of relevant markers used to monitor practices systematically. Methods: This study analysed hand hygiene compliance and associated factors with a radiofrequency-identification-based real-time continuous automated monitoring system in an infectious disease ward with 17 single bedrooms. Healthcare workers (HCWs) were tracked while performing routine care over 171 days. A multi-level multi-variate logistics model was used for data analysis. The main outcome measures were hand disinfection before entering the bedroom (outside use) and before entering the patient care zone, defined as the zone surrounding the patient’s bed (inside/bedside use). Variables analysed included HCWs’ characteristics and behaviour, patients, room layouts, path chains and duration of HCWs’ paths. Findings: In total, 4629 paths with initial hand hygiene opportunities when entering the patient care zone were selected, of which 763 (16.5%), 285 (6.1%) and 3581 (77.4%) were associated with outside use, inside/bedside use and no use, respectively. Hand hygiene is caregiver-dependent. The shorter the duration of the HCW’s path, the worse the bedside hand hygiene. Bedside hand hygiene is improved when one or two extra HCWs are present in the room. Interpretation: Hand hygiene compliance at the bedside, as analysed using the continuous monitoring system, depended upon the HCW’s occupation and personal behaviour, number of HCWs, time spent in the room and (potentially) dispenser location. Meal tray distribution was a possible factor in the case of failure to disinfect hands. ª 2017 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Address: Faculte ´ de Me ´decine de la Timone, Aix-Marseille Universite ´, 27 Boulevard Jean Moulin, 13005 Marseille, France. Tel.: þ33 (0)4 91 32 46 00. E-mail address: [email protected] (J-C. Dufour). http://dx.doi.org/10.1016/j.jhin.2017.02.002 0195-6701/ª 2017 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

Introduction Healthcare-associated infection (HCAI) affects approximately five million people each year in Europe, at an estimated cost of V13e24 billion, and with an attributable mortality of 50,000e135,000 cases.1 In the USA, it has been estimated that there are 1.7 million cases of HCAI per year, with 99,000 deaths and an economic impact of approximately US$6.5 billion. Worldwide, estimated HCAI prevalence rates vary from 5e10% in all hospital admissions to 20e30% in patients admitted to intensive care units. The literature suggests that as many as 55e70% of HCAIs are preventable with current evidence-based strategies.2 Hand hygiene is key to protecting patients against HCAI and colonization with multi-drug-resistant micro-organisms.3 Hand cleaning with an alcohol-based hand rub (AHR) is a simple and undemanding procedure that only requires a few seconds and is highly effective.4 Although the relative levels of risk of different care activities, and how to best define key moments for hand hygiene action, are still debated among infection control experts, the World Health Organization’s (WHO) ‘My Five Moments For Hand Hygiene’ are now generally accepted as the key times for effective hand hygiene.5 These five moments highlight the importance of two key actions: hand disinfection before contact with the patient, and hand disinfection before leaving a contaminated healthcare zone. There is no standard for measuring compliance with hand hygiene practices; however, direct observation of compliance with hand hygiene recommendations is the method used in most studies.6 Compliance rates observed among healthcare workers (HCWs) have been regarded by public health authorities as unacceptably poor. A recent systematic review of 96 empirical studies reported that the median compliance rate was 40%, varying from <20% to >80%.7 Direct observational surveys suffer from several limitations; they are time consuming and costly, they do not allow continuous monitoring, and they only provide information on a small sample of all hand hygiene opportunities. More importantly, staff members change their behaviour when they know that they are being observed, especially if the observer is standing next to the HCW (the ‘Hawthorne effect’).8 Self-reporting has been used to asses hand hygiene in some studies. This method saves resources, but the validity of self-reported data has been challenged and must be assessed further.9 Video cameras have been used to monitor hand hygiene, limiting direct human observational bias, but this method is time consuming, costly and, like other methods, does not allow for real-time reporting.10 One other method of indirect measurement is consumption of AHR; while not subject to sampling bias, this does not capture the appropriate denominator (when hand hygiene is indicated).9 To address these gaps, various automated monitoring systems such as electronic hand rub dispensers and radio-frequency-identification (RFID)-based systems have been developed. However, none of these are capable of recording hand hygiene opportunities as defined by WHO’s ‘My Five Moments’, and none have been evaluated comparatively with a reference method for accuracy, sensitivity and specificity.11,12 The system used in this study, based on the ‘iCode RFID 15693’ tag technology (Ex NXP), uses a very low electromagnetic energy (13.56 MHz) antenna and tags with no internal power source. It is patented under N FR 12/60453 and called ‘MedihandTrace’ (MHT). The system aims to monitor hand hygiene

345

before entering the patient’s bedroom, before approaching the zone around the bedside (the patient care zone), and after exiting the patient care zone and the bedroom.13 It has been evaluated previously against video recordings, and found to have sensitivity of 95.65%, specificity of 100% and accuracy of 99.02%. The aim of this study was to assess hand hygiene compliance and associated factors before entering the bedroom (outside use) and before entering the patient care zone (inside/bedside use) using an automatic continuous monitoring system.

Materials and methods Study population and data capture This study took place in the Infectious Diseases Ward of the University Hospital in Marseille, France. This is a 17 single-room hospital unit where staff are highly trained in infection prevention and control; annual consumption of AHR solution on the unit is 7.48 times the European mean.13,14 The MHT system (https://www.youtube.com/watch?v¼d1Oa7vNT_iQ) was tested in seven patient rooms in the unit and was used to record the hand hygiene compliance of HCWs in these rooms.15,16 Bedrooms were divided into two care zones: (1) the entire bedroom, and (2) the patient care zone, surrounding the patient’s bed; this was delineated by antennae (Figure A, see online supplementary material). HCWs were expected to follow WHO’s ‘My Five Moments For Hand Hygiene’.5 In order to simplify and adapt the monitoring system, only two initial hand hygiene opportunities (IHHOs) were considered and defined in this study: (1) hand hygiene before entering the bedroom with the AHR dispenser in the corridor (outside use); and (2) hand hygiene before first entering the patient care zone with the AHR dispenser within the room (inside/bedside use); observations were made regardless of the care given to the patient. HCWs should also disinfect their hands each time they leave the bedroom or the patient care zone; these data were not reported in this study. Enrolled HCWs were informed of the study objectives and gave their signed consent. This study was approved by the institutional review board (N 2016-018). All HCWs in the care unit participated in the study.

Measurements A HCW’s path is a succession of numbered, time-stamped signals recorded by the MHT system, representing locations of a HCW in a patient’s room and his/her use of an AHR dispenser (Figure A, see online supplementary material). A path starts with signal n 1 (or n 2 if the outside dispenser is not used) and ends with signal n 7 (or n 6 if the outside dispenser is not used). Hand hygiene compliance was calculated as the ratio of AHR used outside (signal n 1) or inside (signal n 3) over the total used from first accessing the patient care zone (IHHO) (signal n 4). Paths with IHHOs detected by the MHT system were captured from 18th September 2013 to 9th March 2014. As the MHT system was inoperative for two days, the study period covered 171 days. According to the selection criteria, the following paths were excluded from data analysis: (1) paths performed in a room without patients (determined by the fact that no patient was registered for the room on the Hospital

346

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

Information System at the date/time of the path); (2) paths without contact with the patient care zone (determined by the MHT system as no signal n 4); (3) paths performed in a room under multi-drug-resistant bacteria/Clostridium spp. isolation measures (determined by the unit register); and (4) paths performed by three HCWs (the physician in charge of the care unit and two research nurses) directly involved in the setup of the MHT system and this study. Variables considered for the analyses concerned HCWs, patients, architectural features of the room, HCW flow and temporal characteristics of paths (Table I). Some were variables already known to be associated with hand hygiene, and others were hypothetical untested variables.

Statistical analysis Paths followed by HCWs in specific rooms with specific patients (HCWs, rooms and patients all had their own characteristics) suggested a possible hierarchical data structure. Thus, three hierarchical structure hypotheses were tested: (1) HCW level; (2) room level; and (3) patient level. Furthermore, AHR dispenser use could be significantly different between HCWs or/and rooms or/and patients. These variables are therefore referred to as contextual variables. In order to determine factors affecting the probability of AHR dispenser use, two analyses of dichotomous outcomes were undertaken: in the first analysis, the main outcome compared ‘inside/bedside AHR use’ with ‘no AHR use’ (neither inside nor outside), and in the second analysis, the main outcome compared ‘outside AHR use’ with ‘no AHR use’. For this, two binary variables were considered: ‘inside/bedside use’ and ‘outside use’, with ‘no AHR use’ paths corresponding to the reference group. Hierarchical logistic models were used to take into account the impact of the contextual variables on AHR dispenser use (see Box A, online supplementary material for details of the statistical modelling and the statistical modelling strategy).17 The hierarchical model approach allowed the authors to estimate the distribution of variance in AHR dispenser use between two analysis levels: the path and the contextual variables. The intraclass correlation coefficient (ICC) was used to measure the percentage of variance in AHR dispenser use attributable to contextual variables. The ICC was estimated based on assumptions for a binary variable,18 with the variance attributable to the contextual variable effect as the numerator and the total variance as the denominator. Estimates were obtained using a maximum-likelihood procedure in R Version 3.0.3 with library (lme4) and glmer procedures. All statistical analyses were considered to be significant with P < 0.05 using the two-sided test.

Results Over the study period (171 days), 4629 IHHOs were selected according to the inclusion/exclusion criteria (Figure B, see online supplementary material): 285 ‘inside/bedside AHR use’ before entering the patient care zone (6.1%), 763 ‘outside AHR use’ before entering the patient’s bedroom (16.5%), and 3581 ‘no AHR use’ (77.4%) (Table II). Paths were performed by 42 HCWs (23 medical doctors, eight residents, 12 medical students, three senior doctors, six nurses, nine assistant nurses

and four housekeepers) caring for 132 patients in the seven rooms equipped with the MHT system. Overall, 3849 IHHOs (performed by 39 HCWs) were analysed for ‘inside/bedside use’ and 4316 IHHOs (performed by 39 HCWs) were analysed for ‘outside use’. Hierarchical data structure testing showed that only the HCW effect can be used in the two analyses (Table III). The final model retained eight variables for ‘inside/bedside use’ analysis and seven variables for ‘outside use’ analysis. Several variables were associated with better hand hygiene when entering the patient care zone and using the inside/ bedside AHR dispenser (Table IV). The most significant was the professional occupational group, with physicians associated with a greater likelihood of AHR use (compared with this group, the nurses group had an adjusted odds ratio of 0.08, 95% confidence interval 0.03e0.20). The longer the time between the preceding and current path, or the longer the path duration, the greater the likelihood of AHR use. When a HCW was working with one or two colleagues, hand hygiene was significantly improved (more than 2.54 fold) (the presence of three other HCWs made no significant difference). Hand hygiene compliance was better when leaving the patient care zone than when entering. The temporal characteristics showed that autumn and winter (OctobereJanuary) were associated with a greater likelihood of better hand hygiene. Finally, hand hygiene compliance was better when the patient’s length of stay exceeded seven days, but the meal distribution period decreased hand hygiene compliance significantly. Several variables were associated with better hand hygiene when entering the patient’s bedroom and using the outside AHR dispenser (Table V). Professional occupation was also found to be associated with AHR use before entering the patient’s bedroom. Similarly, path duration and time elapsed between the preceding and current path, as well as AHR use before leaving the patient care zone, were associated with better hand hygiene. While December was also found to be associated with better hand hygiene, there was no significant association with the other months of the year. Monday was not a good day for hand hygiene in this analysis. The worst hand hygiene occurred when the AHR inside/bedside dispenser was hidden by the door. For the ‘inside/bedside AHR use’ analysis, the model fitted both specific characteristics of paths and HCWs, and had an ICC value of 0.15, representing a 76% decrease in interindividual variance compared with the null model. In the case of the ‘outside AHR use’ analysis, the fitted model had an ICC of 0.12, representing a 72% decrease in interindividual variance compared with the null model.

Discussion Measuring hand disinfection in a healthcare setting is a true challenge. To limit observational bias, a number of innovative technologies have been used, but none of them (with the exception of video recording) are able to measure WHO’s ‘My Five Moments’. However, further research of methods to measure compliance with hand hygiene opportunities precisely is required. As MHT is unable to identify the HCW’s task, it is unable to measure M1 and M2 hand disinfection precisely, which are risk-related tasks for microbe transmission.19 However, MHT is capable of tracing HCW hand disinfection before

Table I Variables considered for the analyses Set

HCW characteristics

Architectural features of the room

HCW flow characteristics

Temporal characteristics of path

Sex Age Professionals’ occupational group Job tenure in the service at the time of inclusion in the study Laterality in hand use Sex Age Duration of hospital stay Visibility of the AHR dispenser inside the room (when the door is open) Localization (relatively to the front door) of the AHR dispenser inside the room Room surface Distance from the AHR dispenser inside the room to the head of the bed HCW total number of paths over the whole study HCW number of paths in the day Path durationa Time elapsed between the preceding and the current path of the HCWa Preceding path made in the same room Number of entries/exits within the patient care zone during the HCW’s path Time spent within the patient care zone during the HCW’s path Number of other HCWs detected in the room during the path Occupational groups of the other HCWs detected Rank of the other HCWs detected concerning their hand disinfection before patient contact AHR dispenser use before leaving the bedroom Period of the day Day of the week Month of the year

Order of the path since the beginning of the HCW’s shift Path done during meal times

Modalities or units

Male/female Years Physician/nurse/nurse assistant/housekeeper Days Left/right Male/female Years Less than seven days/more than seven days Not visible/partially visible/totally visible Left/right Square centimetres Centimetres Number Number Short (3e19 s)/medium (20 s to 3 min 12 s)/long (>3 min 12 s) Short (<40 s)/medium (41 s to 5 min 33 s)/long (>5 min 34 s) Yes/no Number Seconds Number Physician/nurse/nurse assistant/housekeeper Number Yes/no Morning/afternoon/evening Monday/other days Type A: SeptembereFebruaryeMarch/ OctobereNovembereDecember/January Type B: December/other months Number Yes/no

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

Patient characteristics

Variables considered

HCW, healthcare worker; AHR, alcohol-based hand rub. a Threshold was chosen according to the distribution of data. The label ‘Short’ is the first quartile; the label ‘Medium’ is the mean; and the label ‘Long’ is the third quartile.

347

348

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

Table II Hand hygiene compliance of healthcare workers (HCWs) Hand hygiene compliance

No. of No. of paths HCWs (%)

Hand hygiene before entering the patient care zone using inside/bedside AHR dispenser Hand hygiene before entering the patient’s bedroom using outside AHR dispenser No AHR use Total

35

285 (6.1)

40

763 (16.5)

41 42

3581 (77.4) 4629 (100)

AHR, alcohol-based hand rub.

entering two targeted zones (the patient’s bedroom, using the AHR dispenser outside; and/or the patient care zone, using the inside/bedside AHR dispenser). One may consider that to prevent patient contamination, hands should be disinfected before entering the patient care zone regardless of whether or not the HCW carries out task M1 or M2. The aim of hand disinfection in M3, M4 and M5 is to avoid carrying pathogens out of the contaminated room by disinfecting hands after leaving the infected care zone (using the AHR dispenser inside and/or outside). However, this study focused on hand disinfection prior to potential patient contact. It is known that hand hygiene varies significantly among HCWs within the same care unit, institution or country.20 Compliance with hand hygiene is an individual behaviour (ICC ¼ 0.43 in this study), thus complicating the comparison of studies between care units, hospitals and countries where the recruited individuals are heterogeneous.7 While this may not have an overly large impact on observational studies recording a low number of events per person, it becomes a major bias in studies such as this that record very high numbers of events per HCW. This justifies the use of multi-level analysis to attenuate bias due to the HCW effect.18 Moreover, this finding emphasizes the fact that hygiene training and education should be targeted at poorly compliant individuals. Depending upon the risk assessment of possible transmission (intensive care unit,

surgery ward, elderly care unit, etc.), it might be considered that individuals failing to comply with hand hygiene should be moved to other less risky situations. This is particularly important in the case of peripatetic HCWs who have greater potential to be super spreaders.21 Risk factors for poor hand hygiene compliance have been determined in several studies, but most were observational or self-reported. To the authors’ knowledge, no studies to date have evaluated factors related to hand hygiene by using automated continuous monitoring systems. Among HCW characteristics, male sex and doctor status have been reported several times as factors in low compliance.19,22 In the present study, physicians systematically displayed better hand disinfection than other HCWs, in agreement with a previous study in the same care unit.14 This result likely reflects the fact that the study is ‘care unit’ dependent, and should be extrapolated with caution. Another explanation would be that nurses are more susceptible to the Hawthorne effect than doctors, reflecting their better score in observational studies. Use of an AHR dispenser at the bedside/inside the room just before entering the patient care zone is dependent on the patient’s length of stay. The longer the patient stays, the higher the compliance of HCWs with hand disinfection, irrespective of patient age or sex. This new information may suggest that disease severity rather than age may enhance hand disinfection compliance.23 Bedside hand hygiene is improved when one or two other HCWs are present in the single hospital room. This positive influence is not associated with the social position of the co-worker, suggesting that the awareness of (possibly) being watched by peers enhances hand disinfection inside a patient’s room independently of the social position of the peers. This behaviour may be interpreted as a sort of normative social influence effect,24 rather than the Hawthorne effect commonly reported in observational studies.25 Hand disinfection as a whole seems to be an individual attitude, as suggested by the fact that use of the AHR dispenser before entering the patient care zone was associated with its use after contact.26 Similar to Dedrick et al.,27 this study found that the shorter the path duration, the worse the compliance with hand hygiene at the bedside, indicating that bedside hand hygiene is dependent upon the time spent in the patient’s room. Some

Table III Null multi-level models Null multi-level models AHR inside use Ref ¼ no hand disinfection

Fixed effect Individual adjusted OR Random effect Interindividual variance ICC Statistical model AIC BIC -2logLik Deviance

AHR outside use Ref ¼ no hand disinfection

OR (95% CI)

P

OR (95% CI)

P

0.11 (0.1e0.2)

<0.0001

0.31 (0.20e0.47)

<0.0001

2.458 0.43

e e

1.658 0.34

e e

1618.6 1631.1 807.3 1614.6

e e e e

3497.1 3509.9 -1746.6 3493.1

e e e e

AHR, alcohol-based hand rub; ICC, intraclass correlation coefficient; OR, odds ratio; CI, confidence interval.

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

349

Table IV Variables associated with hand hygiene before entering the patient’s care zone AHR inside/bedside use (N ¼ 3849 initial hand hygiene opportunities) Ref ¼ no AHR use

Multi-level multi-variate logistic analysis (N ¼ 39 HCWs) N (%)

Null model ORa (95% CI)

Fixed effect ORa e 0.11 (0.1e0.2) HCW characteristics Professionals’ occupational group Physicians 479 (12.4) Nurse 680 (17.7) e Nurse assistant 2215 (57.6) e Housekeeper 475 (12.3) e Patient characteristics Duration of patient hospital stay Less than seven days 1520 (39.5) More than seven days 2329 (60.5) e Temporal characteristics of path Month of the year September, February, March 1140 (29.6) October, November, December 1676 (43.6) e January 1033 (26.8) e Meal times Yes 992 (25.8) No 2857 (74.2) e HCW flow characteristics Path duration Short 209 (5.4) Medium 1584 (41.2) e Long 2056 (53.4) e Time elapsed between the preceding and the current path Short 1088 (28.3) Medium 833 (21.6) e Long 1928 (50.1) e Number of other HCWs detected in the room during the path 0 3291 (85.5) 1 463 (12.0) e 2 74 (1.9) e 3 21 (0.6) e Use of AHR dispenser after patient care zone contact Yes 2707 (70.3) No 1142 (29.7) e Random effect Interindividual variance e 2.458 ICC e 0.43 Proportion of variance explained e e Statistical model AIC e 1618.6 BIC e 1631.1 -2logLik e -807.3 Deviance e 1614.6

Full model p

ORa (95% CI)

P

<0.0001

0.10 (0.02e0.39)

0.0009

e e e

1 0.08 (0.03e0.20) 0.07 (0.03e0.16) 0.13 (0.05e0.35)

<0.0001 <0.0001 <0.0001

e

1 0.73 (0.54e0.98)

0.036

e e

1 0.39 (0.27e0.56) 0.29 (0.18e0.47)

<0.0001 <0.0001

e

1 1.48 (1.01e2.18)

0.043

e e

1 3.43 (1.03e11.45) 5.97 (1.80e19.80)

0.0448 0.0035

e e

1 2.80 (1.74e4.49) 1.69 (1.09e2.62)

<0.0001 0.0191

e e e

1 2.40 (1.69e3.41) 2.54 (1.30e4.95) 2.68 (0.82e8.78)

<0.0001 0.0064 0.103

e

1 0.60 (0.44e0.83)

e e e

0.593 0.15 76%

e e e

e e e e

1462.5 1568.9 -714.3 1428.5

e e e <0.0001

0.00152

HCW, healthcare worker; AHR, alcohol-based hand rub; ICC, intraclass correlation coefficient; ORa, individual adjusted odds ratio; CI, confidence interval.

previously published video footage taken during routine care shows that nurses make numerous entries and exits during nursing time to supply care materials.28 One may suggest that hand hygiene is more frequently performed upon first entry

than in subsequent paths; consequently, re-organization of care by means of a nursing kit may lead to a reduction in the number of entries and exits, and consequently enhance hand hygiene compliance. This hypothesis is currently under

350

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351

Table V Variables associated with hand hygiene before entering the patient’s bedroom AHR outside use (N ¼ 4316 initial hand hygiene opportunities) Ref ¼ no AHR use

Multi-level multi-variate logistic analysis (N ¼ 39 HCWs) N (%)

Null model ORa (95% CI)

Fix effect ORa e 0.31 (0.20e0.47) HCW characteristics Professionals’ occupational group Physicians 587 (13.6) Nurse 808 (18.7) e Nurse assistant 2431 (56.3) e Housekeeper 490 (11.4) e Room characteristics Visibility of AHR dispenser inside the room (when the door is open) Not visible 290 (6.7) Partially visible 809 (18.8) e Totally visible 3217 (74.5) e Temporal characteristics of path Month of the year Other months 677 (15.7) December 3639 (84.3) e Day of the week Monday 704 (16.3) Other days 3612 (83.7) e HCW flow characteristics Path duration Short 265 (6.1) Medium or long 4051 (93.9) e Time elapsed between the preceding and the current path Short 1132 (26.2) Medium 1065 (24.7) e Long 2119 (49.1) e Use of AHR dispenser after patient care zone contact Yes 2927 (67.8) No 1389 (32.2) e Random effect Interindividual variance e 1.658 0.34 ICC e Proportion of variance explained e e Statistical model AIC e 3497.1 BIC e 3509.9 -2logLik e -1746.6 Deviance e 3493.1

Full model P

ORa (95% CI)

P

<0.0001

0.18 (0.08e0.37)

<0.0001

e e e

1 0.19 (0.09e0.40) 0.13 (0.07e0.25) 0.10 (0.04e0.25)

<0.0001 <0.0001 <0.0001

e e

1 1.99 (1.17e3.38) 3.40 (2.09e5.53)

0.011 <0.0001

e

1 1.56 (1.22e2.00)

0.00037

e

1 1.42 (1.10e1.84)

0.0079

e

1 0.60 (0.42e0.85)

0.0038

e e

1 6.20 (4.64e8.29) 1.99 (1.50e2.65)

<0.0001 <0.0001

e

1 0.53 (0.44e0.65)

<0.0001

e e e

0.458 0.12 72%

e e e e

3176.7 3259.6 -1575.4 3150.7

e e e e e e <0.0001

AHR, alcohol-based hand rub; ICC, intraclass correlation coefficient; OR, individual adjusted odds ratio; CI, confidence interval.

evaluation in the authors’ care unit. As shown here, hand disinfection compliance is much lower during meal times. It is understandable that hand disinfection before entering the patient care zone is somewhat difficult when HCWs are carrying the meal tray. Many questions remain unanswered. While this study suggests that the location of the dispenser may play a role in hand hygiene compliance, it is not known why this study found better hand disinfection before entering the bedroom when the bedside/inside AHR dispenser is visible. Similarly, there is no explanation for the observed difference in hand disinfection compliance during the year or over the course of

the week, while others have reported that hand disinfection was better during the weekend.29 Finally, only 22.6% of HCWs disinfected their hands before entering the patient care zone, either using the outside AHR dispenser or using the bedside/ inside AHR dispenser. This should be considered a very poor result; at the same time, the same care unit declared AHR consumption of 7.48 times the European mean. This indicates that there is a great discrepancy between the two markers, and consumption of AHR may overestimate hand disinfection.13 Solutions to enhance hand disinfection compliance, such as reminders, have been reported several times in the literature,

J-C. Dufour et al. / Journal of Hospital Infection 95 (2017) 344e351 but long-term follow-up has not been reported to date. MHT meets most requirements for an automated hand hygiene monitoring system as demanded by WHO: continuously recording a huge range of HCW paths, capturing hand disinfection opportunities, minimizing the Hawthorne effect, providing real-time feedback of hand hygiene compliance, and requiring very little time-consuming and technical expertise. In light of these results, and in addition to reminders, hand hygiene compliance would benefit from specific training of the least compliant HCWs, care re-organization including nursing kits and working in pairs, modified location of bedside dispensers, and automated continuous hand hygiene monitoring systems to evaluate the effectiveness of planned interventions.

Acknowledgements The authors wish to thank all the HCWs who took part in the study for their active participation and confidence in the researchers. Conflict of interest statement Prof. Philippe Brouqui funded MedihandTrace SAS, a startup devoted to research and development of innovative tools to tackle infectious agent transmission problems in health care. The other authors declare no conflicts of interest. Funding sources A*MIDEX project (ANR-11-IDEX-0001-02), APFR 2011 and FEDER 2007-2013 N 42171 MedihandTrace, Oseo and Region PACA, and Foundation ‘Mediterrane ´e infection’.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jhin.2017.02.002.

References 1. Pittet D, Allegranzi B, Boyce J. The World Health Organization Guidelines on Hand Hygiene in Health Care and their consensus recommendations. Infect Control Hosp Epidemiol 2009;30:611e622. 2. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ. Estimating the proportion of healthcare-associated infections that are reasonably preventable and the related mortality and costs. Infect Control Hosp Epidemiol 2011;32:101e114. 3. Widmer AF, Conzelmann M, Tomic M, Frei R, Stranden AM. Introducing alcohol-based hand rub for hand hygiene: the critical need for training. Infect Control Hosp Epidemiol 2007;28:50e54. 4. Pittet D, Boyce JM. Hand hygiene and patient care: pursuing the Semmelweis legacy. Lancet Infect Dis 2001;1(Suppl. 1):9e20. 5. Sax H, Allegranzi B, Uckay I, Larson E, Boyce J, Pittet D. ‘My five moments for hand hygiene’: a user-centred design approach to understand, train, monitor and report hand hygiene. J Hosp Infect 2007;67:9e21. 6. World Health Organization. WHO guidelines on hand hygiene in health care: first global patient safety challenge: clean care is safer care. Geneva: WHO; 2009. 7. 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.

351

8. Haessler S. The Hawthorne effect in measurements of hand hygiene compliance: a definite problem, but also an opportunity. BMJ Qual Saf 2014;23:965e967. 9. Haas JP, Larson EL. Measurement of compliance with hand hygiene. J Hosp Infect 2007;66:6e14. 10. Armellino D, Hussain E, Schilling ME, et al. Using high-technology to enforce low-technology safety measures: the use of third-party remote video auditing and real-time feedback in healthcare. Clin Infect Dis Off Publ Infect Dis Soc Am 2012;54:1e7. 11. Cheng VCC, Tai JWM, Ho SKY, et al. Introduction of an electronic monitoring system for monitoring compliance with Moments 1 and 4 of the WHO ‘My 5 Moments for Hand Hygiene’ methodology. BMC Infect Dis 2011;11:151. 12. Marra AR, Edmond MB. New technologies to monitor healthcare worker hand hygiene. Clin Microbiol Infect Off Publ Eur Soc Clin Microbiol Infect Dis 2014;20:29e33. 13. Brouqui P, Aladro AS. Alcohol hand rub consumption objectives in European hospitals need to be revisited. Clin Microbiol Infect 2016;22:577. 14. Barrau K, Rovery C, Drancourt M, Brouqui P. Hand antisepsis: evaluation of a sprayer system for alcohol distribution. Infect Control Hosp Epidemiol 2003;24:180e183. 15. Boudjema S, Dufour JC, Aladro AS, Desquerre I, Brouqui P. MediHandTrace: a tool for measuring and understanding hand hygiene adherence. Clin Microbiol Infect Off Publ Eur Soc Clin Microbiol Infect Dis 2014;20:22e28. 16. Dufour J-C, Reynier P, Soto Aladro A, Brouqui P. Input of innovative technology for surveillance and improvement of hand hygiene: the Medihandtrace contribution to hand disinfection monitoring and intervention. Clin Microbiol Open Access 2015;4:216. 17. Guo G, Zhao H. Multilevel Modeling for Binary Data. Annu Rev Sociol 2000;26:441e462. 18. Snijders TA, Bosker R. Multilevel analysis: an introduction to basic and advanced multilevel modeling. Los Angeles CA: Sage; 2012. 19. Pittet D, Mourouga P, Perneger TV. Compliance with handwashing in a teaching hospital. Infection Control Program. Ann Intern Med 1999;130:126e130. 20. Allegranzi B, Pittet D. Healthcare-associated infection in developing countries: simple solutions to meet complex challenges. Infect Control Hosp Epidemiol 2007;28:1323e1327. 21. Temime L, Opatowski L, Pannet Y, Brun-Buisson C, Boe ¨lle PY, Guillemot D. Peripatetic health-care workers as potential superspreaders. Proc Natl Acad Sci USA 2009;106:18420e18425. 22. Hugonnet S, Perneger TV, Pittet D. Alcohol-based handrub improves compliance with hand hygiene in intensive care units. Arch Intern Med 2002;162:1037e1043. 23. Pittet D, Stephan F, Hugonnet S, Akakpo C, Souweine B, Clergue F. Hand-cleansing during postanesthesia care. Anesthesiology 2003;99:530e535. 24. Lankford MG, Zembower TR, Trick WE, Hacek DM, Noskin GA, Peterson LR. Influence of role models and hospital design on hand hygiene of healthcare workers. Emerg Infect Dis 2003;9:217e223. 25. Srigley JA, Furness CD, Baker GR, Gardam M. Quantification of the Hawthorne effect in hand hygiene compliance monitoring using an electronic monitoring system: a retrospective cohort study. BMJ Qual Saf 2014;23:974e980. 26. Whitby M, McLaws ML, Ross MW. Why healthcare workers don’t wash their hands: a behavioral explanation. Infect Control Hosp Epidemiol 2006;27:484e492. 27. Dedrick RE, Sinkowitz-Cochran RL, Cunningham C, et al. Hand hygiene practices after brief encounters with patients: an important opportunity for prevention. Infect Control Hosp Epidemiol 2007;28:341e345. 28. Boudjema S, Reynier P, Dufour JC, et al. Hand hygiene analyzed by video recording. J Nurs Care 2016;5:1e5. 29. Pittet D. Improving adherence to hand hygiene practice: a multidisciplinary approach. Emerg Infect Dis 2001;7:234e240.