Hand hygiene is the most pivotal behavior a healthcare worker (HCW) can undertake to prevent the patients in their care from acquiring a healthcare-associated infection (HAI).1 The relationship between improved hand hygiene compliance and decreasing HAIs has been amply demonstrated.1–Reference Grayson, Russo and Cruickshank4 The timing of hand hygiene to interrupt the transmission of microorganisms is guided by the World Health Organization (WHO) ‘Five Moments for Hand Hygiene,’ which also provides a structured framework against which compliance can be measured.1
Direct observation has been deemed the gold standard data collection method, and it is mandated in some settings.1,5 This method has a number of established advantagesReference Gould, Creedon, Jeanes, Drey, Chudleigh and Moralejo6–Reference Gould, Drey and Creedon9; perhaps the most vaunted is the ability of auditors to provide immediate, contextual, feedback, and education to those they are observing and hence improve practice and, ultimately, hand hygiene compliance.Reference Pan, Tien and Hung7 However, direct observation has a number of well-documented disadvantages.Reference Fisher, Seetoh and May-Lin10–Reference McLaws and Kwok14 Many now see the method as a time-consuming and labor-intensive way of collecting small amounts of data with inherent biases that do not reflect typical practice.Reference Hagel, Reischke and Kesselmeier15,Reference Boyce, Cooper and Dolan16 Furthermore, immediate feedback occurs only infrequently and is generally not appreciated by frontline clinicians.Reference Boscart, Fernie, Lee and Jaglal17,Reference Azim, Juergens, Hines and McLaws18
Electronic hand hygiene monitoring systems have been proposed as the possible solution to the problems associated with direct observation because they are able to collect large data sets with little human intervention and minimal biases.Reference Srigley, Lightfoot, Fernie, Gardam and Muller19,Reference Ellingson, Polgreen and Schneider20 However, some caution that electronic monitoring systems are not a panacea.Reference Srigley, Furness, Baker and Gardam21 Chief among the disadvantages of the existing systems is the use of proxy measures of hand hygiene compliance rather than the WHO Five Moments framework.22 These electronic approaches rely on either product consumption or room entry and exit tracking as a surrogate measure of compliance.Reference McLaws and Kwok14,Reference Srigley, Lightfoot, Fernie, Gardam and Muller19,Reference Boyce23,Reference Pires and Pittet24 Although the general presumption is that greater product usage or dispenser activity equates to a higher rate of compliance,Reference Gould, Drey and Creedon9 some studies have in fact failed to demonstrate a convincing link.Reference Cohen, Murray and Jia25 Such systems are also often unable to demonstrate the appropriateness of the hand hygiene episode, particularly when it occurs within the patient room or zone. That is, did hand hygiene occur at the optimally correct point in the sequence of the clinical care episode to prevent microbial transmission?1,Reference Srigley, Furness, Baker and Gardam21,Reference Gould, Chudleigh, Drey and Moralejo26,Reference Storey, FitzGerald and Moore27 Video-based monitoring systems (VMSs) may, however, offer a solution.22,Reference Boyce28 Some studies have suggested that such approaches are technically capable of monitoring all 5 momentsReference Haac, Rock and Harris29–Reference Diller, Kelly, Blackhurst, Steed, Boeker and McElveen31; however, no system with this primary aim has been demonstrated thus far.
With these parameters in mind, we developed and tested a VMS to audit hand hygiene compliance of healthcare workers according to the WHO Five Moments framework. The key focus was on capturing actual hand hygiene behaviors for analysis according to the WHO Five Moments criteria rather than proxy measures.
Methods
Setting
This pragmatic quasi-experimental observation trial was conducted at the New South Wales Biocontainment Centre, a purpose-built facility for case management and high-level isolation of high-consequence infectious diseases at Westmead Hospital, New South Wales, Australia.
Ethical considerations
Ethics approval for the study was granted by the Human Research Ethics Committee of the Western Sydney Local Health District and Research Governance Committee at Westmead Hospital, NSW, Australia. This study was also considered low risk and adhered to the National Statement on Ethical Conduct in Human Research 2007 (2018 update).32 There were no variations to the approvals as granted, and no participants withdrew.
Data collection
This study included 2 participant groups. Group 1 comprised 5 volunteer HCWs, 3 registered nurses, and 2 infectious diseases registrars at Westmead Hospital who participated in the scenarios relative to their professional designation. Group 2 comprised 4 volunteer local doctoral students who were simulated patients for the HCW participants. Data collection occurred over 2 consecutive days. We conducted 3 separate simulation scenarios, each of which was undertaken on day 1 and then subsequently repeated on day 2. The HCW participants differed from day 1 to day 2, with only 1 HCW taking part on both days. The scenarios ranged from 22 to 49 minutes in duration, with 100 minutes recorded on day 1 and 106 minutes on day 2.
Scenarios were developed by the research team that included practicing clinicians. The objective of the scenarios was to provide realistic representations of clinical care to be captured by the VMS and to encourage the participants to relate to the actor patients in a realistic way. To promote authentic HCW– patient-actor interactions, the scenarios had high fidelity; they incorporated prosthetics, symptom triggers, and cues, as well as medical equipment and consumables. We provided numerous prompts for the potential performance of hand hygiene and the use of personal protective equipment (PPE) to explore whether compliance with such practices could be audited from recorded footage. Despite the detail in the scenarios, however, they were not heavily scripted, leaving participants to make decisions and manage their patients as they deemed appropriate. Thus, the simulations evolved in unexpected ways.
The simulations were recorded by fixed cameras (Fig. 1) present in the quarantine (Q class) rooms of the New South Wales Biocontainment Centre where the study took place. The Q class suite contains 8 cameras in 3 rooms, 2 in the donning area, 3 in the patient room, and 3 in doffing area, with the feed for each camera being displayed as a multiview arrangement. This system records continuously, with footage stored on a secure server for export and review. Individual images can be enlarged and reviewed using this modality. The system utilizes Milestone XProtect SmartClient_viewer software (Milestone Systems, Lake Oswego, OR) for footage review. This system provides up to 9 camera views at one time, which facilitates tracking from space to space, for example, from donning PPE to patient room and thence to the doffing area (Fig. 2). The system also allowed full-screen focus on a single camera view as well as the ability to zoom, freeze, fast forward, and rewind.
Fig. 1. Camera system in place in the NSW Biocontainment Unit; placemt of cameras, close up image of camera and sample of camera field of view.
Fig. 2. Single and multi-screen views of footage reacorded during trial in simulation.
Fig. 3. Hand hygiene compliance according to day, moment and HCW as audited from recorded footage.
Data analysis
The recorded footage was reviewed by the researcher McKay, who is a current Gold Standard Auditor using the WHO Five Moments definitions of what constituted a hand hygiene moment and utilizing the tools developed by the National Hand Hygiene Initiative (NHHI) for direct observational auditing (Supplementary Material online).
The metrics included the compliance with hand hygiene practice according to moment and healthcare worker designation and the total time taken to audit each scenario. In addition, auditor reflections regarding the process of auditing from the recorded footage were also noted.
Results
In total, 206 minutes (3.4 hours) of footage were recorded over the 2-day trial. This footage was recorded by each of the 8 fixed-camera feeds, which could be viewed either collectively as an 8-view split screen or individually as a full-screen view.
Auditing of the 6 simulations was undertaken by researcher McKay and was verified by the other researchers on the study team (Fig. 3). Viewing the 8 camera feeds simultaneously cast to a large screen was determined to be the easiest method. This method allowed participants to be tracked from camera to camera as they moved from room to room, and it captured hand hygiene opportunities and/or moments that occurred outside the patient zone in the donning and doffing areas. The footage was viewed at different playback speeds, paused, or rewound as needed. The system also had the capacity to enhance and zoom into individual views so that practice could be examined more closely.
The recorded simulation footage on day 1 totaled 100 minutes, and auditing time from the footage required 49.5 minutes (Table 1) due to the ability to fast forward the recording. In total, 42 moments were collected, with a compliance rate of 88% overall (range, 70%–96% depending on scenario) (Table 2). However, 15 episodes of unnecessary hand hygiene were also observed. Compared to the total duration of the scenario, this method represented a savings of >50 minutes in auditing time with a collection rate of ∼70 seconds per moment. In terms of cost based on the auditor wage of $50 AUD ($36 USD) per hour, the overall cost of auditing from VMS footage was $41.25 AUD ($30 USD) or ∼$0.98 AUD ($0.71 USD) per moment.
Table 1. Results of Auditing From Footage: Total Time, Auditing Time and Compliance Rate
a Simulation 1: Basic physical assessment and history taking in a patient with a fever.
b Simulation 2: Management of a returned traveller with an infected leg wound.
c Simulation 3: Care of a patient with suspected tuberculosis.
Table 2. Results of Audit of Recorded Footage, According to Day and Moment According to the WHO Five Moments Criteria
Footage recorded on day 2 was 106 minutes in duration. Auditing of the scenarios from the footage took 69.5 minutes, for a collection rate of 60 seconds per moment. In total, 69 moments were collected; they achieved a compliance rate of 88% (range, 82%–93%) (Table 2). Notably, 23 episodes of unnecessary hand hygiene occurred, including several episodes of the application of hand rub to gloves. The overall auditing cost for the scenarios was ∼$58 AUD ($42 USD) or ∼$0.84 AUD ($0.60 USD) per moment.
The footage analysis provides a breakdown according to moment and HCW classification, which follows current NHHI reporting requirements. These results clearly illustrate the capacity of a VMS to collect data consistent with the WHO Five Moments criteria and in a format suitable for submission to the NHHI.
Discussion
Existing research has suggested the utility of electronic systems for the collection of hand hygiene compliance data. However, because this research is largely limited to the use of proxy measures such as product consumption and/or room entry and exit, these systems are of concern.Reference Srigley, Furness, Baker and Gardam21,Reference Sahud, Bhanot, Narasimhan and Malka33 Stewardson and PittetReference Stewardson and Pittet34 warned of the importance of monitoring behaviors and actions that actually “correspond conceptually with patient safety” (p. 1030). They noted that although monitoring and promoting hand hygiene prior to door opening may be simple, convenient, and indeed technologically possible, it may not be either useful or desirable in terms of best practice for patient safety. Thus, rather than defining compliance as adherence to practices which best decrease the risk of microbial transmission, there could be a risk of changing the definition of what constitutes hand hygiene compliance to fit with the monitoring capacity of the technology.
Although this study utilized a trial in simulation with volunteer participants, rather than actual clinical care episodes to record footage, this does not negate the validity of the results. The high-fidelity nature of the scenarios provided a realistic simulacrum of HCW–patient interaction and demonstrates the utility of using video monitoring to capture and assess hand hygiene compliance according to the 5 moments framework. The use of the multiple camera array meant that the entire hand hygiene journey of the participant HCWs could be captured. Subsequent review of the footage demonstrated the ability to clearly distinguish all 5 of the WHO Five Moments as they occurred and according to the WHO Five Moments criteria rather than by proxy measures. Proxy measures have been reported in the literature, notably the research using room-entry and -exit measures via video camera by Armellino et alReference Armellino, Hussain and Schilling35 and research using patient-zone entry coupled with dispenser activation by Al Salman et al.Reference Al Salman, Hani, de Marcellis-Warin and Isa36
The time, and consequently financial, savings demonstrated in our results confer additional advantage to the approach. We determined that auditing from the recorded footage for both days represented a saving of 86 minutes. When compared to time taken if direct observational auditing was employed, this equated to a saving of 84 seconds per moment. Furthermore, this calculation assumed that a human auditor would be able to capture all moments with the same certainty as was able to be achieved from the footage. We suspect that a higher yield of moments was achieved from the footage due to the capacity to freeze, slow, and rewind the footage, in addition to the lack of distraction during auditing, which took place away from the clinical setting. Auditing using recorded footage would cost $99.17 AUD ($ 71.59 USD) or $0.89 AUD ($0.64 USD) per moment for the 6 simulation scenarios, whereas direct observational auditing would have cost $171.67 AUD ($123.92 USD) or $1.55 AUD ($1.12 USD) per moment. These figures are lower than the cost per moment of $2.24 AUD ($1.62 USD) suggested by Azim and McLaws.Reference Azim and McLaws37 With data from the NHHI in Australia showing the collection of >1.8 million moments per year,38 the potential financial benefits could be considerable. The cost and time savings of auditing practice from footage has been noted in the literature. Hu et alReference Hu, Peyre and Arriaga39 estimated a 50%–80% time savings in data review due to the ability to fast-forward footage when appropriate. This report supports the enhanced efficiency of video-monitoring technologies. Recorded footage may also be reviewed at a more convenient time compared to direct observation data gathering, which typically occurs during the day shift on weekdays. The rationale for selecting this time is that the clinical areas are the busiest and larger amounts of hand hygiene will be being performed; hence, data gathering is most efficient. However, this practice may not only foster selection bias but may create conflict for local, ward-based auditors with dual clinical and auditing roles.Reference Haas and Larson8,Reference Boyce23,Reference Daniels40
Another useful aspect of collecting data using video monitoring technology is the potential to collect less biased results. The practices of all HCWs who come within the view of the camera lens are captured, hence eliminating or at the very least significantly reducing selection bias. In addition, observer bias may be reduced, interauditor reliability issues may be solved by auditing from footage because multiple auditors can review the same episode and compare results.Reference Woolrych, Zecevic and Sixsmith41 Footage can be manipulated to ensure certainty (ie, by rewinding, slowing motion, and freezing frames) and to facilitate discussion, review, and clarification of difficult or ambiguous situations. In addition, observation bias or the Hawthorne effect is reduced by camera versus direct observation; a camera creates less behavioral change than a human observer. Several studies involving video recording of participants in clinical or simulated clinical contexts have shown the reduction or elimination of the Hawthorne effect in the presence of cameras.Reference Clack, Scotoni, Wolfensberger and Sax30,Reference Hu, Peyre and Arriaga39,Reference Filho, Marra and Magnus42 More accurate hand hygiene data would allow the identification of practice deficits and open the possibilities of targeted change strategies to improve patient safety and quality of care.
The results of this study demonstrate that video-based auditing of hand hygiene is indeed technically possible. These findings suggest that it may be achieved in a more efficient manner than via traditional human direct observation, a suggestion which was raised by participants in preparatory work.Reference McKay, Ferguson and Shaban43 In addition, resultant data may represent a truer reflection of practice than data produced by direct observation. These results are significantly less biased, they are able to be collected without intrusion, and they have enhanced credibility due to inherent accuracy and completeness of the record. This approach allows for what Broyles et alReference Broyles, Tate and Happ44 have described as nuanced levels of analysis.
The COVID-19 global pandemic has accelerated the development, use and acceptability of various video-based “telehealth” modalities.Reference Vilendrer, Patel and Chadwick45,Reference Ostrowsky, Weil and Olaisen46 Vilendrer et alReference Vilendrer, Patel and Chadwick45 suggest that the optimal set up would be to have the facility for such approaches in every patient room as a way of optimizing communication for patients requiring isolation precautions and preventing pathogen transmission. They go on to suggest the utility of using the technology for other purposes such as “virtual sitters,” the facilitation of visitor interaction and the enhancement of clinician convenience and efficiency.Reference Vilendrer, Patel and Chadwick45 Furthermore, developments in the use of cameras for fall prevention and elder abuse (so called “granny cams”) also have additional possibilities in aged-care settings.Reference Woolrych, Zecevic and Sixsmith41,Reference Ostervang, Vestergaard, Dieperink and Danbjørg47,Reference Berridge, Halpern and Levy48 Video monitoring within the patient zone has the potential to enhance patient safety and quality of care through the accurate identification and subsequent remediation of practice shortfalls.
This study had several limitations. Auditing was conducted by a single member of the research team who is a validated Gold Standard Auditor under the NHHI in Australia. Although this could represent a limitation of the study, the practice is consistent with direct observational auditing under the NHHI framework where results are collected and submitted based on the interpretation of a single individual.
In conclusion, our earlier works explored the question of the methodological and technical considerations for video-based auditing of hand hygiene practiceReference McKay, Ferguson and Shaban43,Reference McKay, Shaban and Ferguson49 and established the basis of the current study. We have demonstrated the technical capacity of a multicamera VMS to record footage of clinical practice via high-fidelity simulation which can subsequently be audited for hand hygiene compliance against the WHO Five Moments criteria rather than proxy measures such as entry and exit.
Furthermore, utilizing this approach may produce output data that are more representative of true, unbiased hand hygiene compliance behavior and provide time and cost savings as well as convenience for auditors. This research is intended as a launching point for the development and refinement of camera systems within the patient zone with which to audit hand hygiene as well as other infection prevention and control behaviors. Another question that needs to be addressed, notwithstanding the technical capacity to record and subsequently audit hand hygiene behaviors, is this: What is the acceptability of the practice to HCW and patients? This factor could be an obstacle despite the benefits of this approach.
Acknowledgments
We thank the volunteers who participated in our study. We acknowledge Dr Cristina Sotomayor-Castillo, Dr Shizar Nahidi and Dr Keren Kaufman-Francis for their support and assistance with this study.
Financial support
This study was supported by a University of Sydney PhD candidate stipend, and the 2020 Seed Grant Fund from the Sydney Institute for Infectious Diseases at the University of Sydney.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
