Discussion

Most studies on EMU patient safety practices across the world have shown that these practices were often heterogeneous between centers. ^{Reference Kobulashvili, Höfler and Dobesberger1,Reference Baheti, Rathore and Bansal3,Reference DeVries-Rizzo7–Reference Buelow, Privitera, Levisohn and Barkley11} Differences in EMU structure, patient monitoring, staffing, use of standardized protocols, and general safety measures against injuries all contribute to this heterogeneity. Safety is paramount in the EMU, especially given that patients are often weaned off their medication to precipitate seizures. With adverse events occurring in approximately 7% of EMU admissions, ^{Reference Sauro, Wiebe, Macrodimitris, Wiebe, Lukmanji and Jetté5} it has become increasingly evident that clear and implementable safety standards could help homogenize EMU practices and improve overall patient safety. Consequently, the publication of the 2021 ILAE/IFCN clinical practice recommendations, although not only focused on patient safety, was both warranted and important. ^{Reference Tatum, Mani and Jin2} In this cross-sectional study, we surveyed Canadian EMUs to (a) collect information on EMU safety practices in a post-COVID era and (b) compare Canadian practices to those in the 2021 ILAE/IFCN guidelines. In doing so, we highlight the heterogeneous nature of safety practices in Canadian EMUs and reinforce the need to improve patient safety by implementing the latest recommendations.

The 2021 ILAE/IFCN guidelines were written in such a way so that certain recommendations could be directly extracted, whereas others could be inferred from the text and the tables. ^{Reference Tatum, Mani and Jin2} These recommendations were as follows: (a) the maximal LTVEM technologist-to-patient ratio is one to two, (b) the maximal LTVEM nurse-to-patient ratio is one to four, (c) patient companions during the night are recommended for children and adults with cognitive or behavioral challenges, (d) a dedicated hospital area should be used for LTVEM, (e) there should be a designated EMU, (f) ECG monitoring is recommended, whereas oximetry, extraoculography, and polygraphy are optional, (g) automated algorithms for seizure detection may be used, (h) informed consent should be obtained before LTVEM, and (i) a written, standardized protocol may be used to manage and test patients during seizures.

Concerning staffing, our study emphasized much variability in personnel ratios across Canadian EMUs. These personnel ratios varied according to the time of day, during weekends, during holidays, and for intracranial EEG patients. The ILAE/IFCN recommended a maximum technologist-to-patient ratio of one to two. Although most EMUs respected this ratio during the day, there was most often no technologist supervision in the evening, in the night, during weekends, and during holidays. However, it remains unclear both in the general literature as well as in the 2021 ILAE/IFCN guidelines if 24/7 technologist monitoring is even desirable. Hence, the one to two technologist-to-patient ratio recommended by the ILAE/IFCN may not have necessarily been constructed to be applied to all hours of the day. ^{Reference Tatum, Mani and Jin2} The American Society of Electroneurodiagnostic Technologists underlined the role of EEG technologists in patient observation, maintaining of recording integrity (and following of protocols), identification of EEG patterns and activity, and emergency interventions, which reduce risks of undesirable outcomes. ^{13,Reference Thomas, Curtis, Adams, Brown and Lau14} Another study, prior to the 2021 ILAE/IFCN guidelines, recommended a technologist-to-patient ratio of one to four. ^{Reference Thomas, Curtis, Adams, Brown and Lau14} A 2014 revision of practices in EMUs speculated that the lack of continuous and appropriate supervision by both nurses and technicians led to an increase in adverse events. ^{Reference Rheims and Ryvlin15} Findings from our survey add to the body of available data on variability in technologist-to-patient ratios, focusing on differences based on the time of day or year. Nevertheless, it is important to note that prior surveys often did not describe technologist availability in terms of technologist-to-patient ratios. The 2016 survey in Canada, for instance, did not include ratios, but instead noted that 67% of adult EMUs provided 24/7 on-call technician support (56% for pediatric) and that 8% had weekend daytime on-call support (33% for pediatric). ^{Reference DeVries-Rizzo7} This method of presenting data on technologist availability may be more realistic than presenting technologist-to-patient ratios based on the time of day or year, since continuous monitoring by technologists 24/7 would be difficult to implement and is probably rare.

Our survey showed that most Canadian EMUs seemed to follow the ILAE/IFCN recommendations for the optimal maximum nurse-to-patient ratio of one to four. Almost all EMUs had nurses who supervised only up to four EMU patients, whether it be in the day, in the evening, in the night, on the weekend, or on holidays. Nurses supervising intracranial EEG patients mostly only supervised one patient. Nevertheless, a substantial amount of missing data was present in the nurse-to-patient ratio section, with many EMU administrators responding that they did not have this information. Although missing data affect the validity of our findings on nurse-to-patient ratios, the fact that many EMU medical directors did not know the nurse-to-patient ratio in itself represents an interesting yet worrisome finding. In addition, as many EMUs have nurses who were not dedicated to EMU patients, the nurse-to-patient ratios reported by some centers may not reflect the true number of patients supervised by one nurse, since one nurse could also supervise non-EMU patients. The importance of the nurse-to-patient ratio of one to four has been evaluated in prior studies, in which it is supported that the number of supervising nurses has an impact on both patient supervision and safe patient care in the EMU. ^{Reference Spanaki, McCloskey and Remedio16–Reference Harris18} In fact, it was found that nurses who had to supervise a higher number of patients were less able to quickly attend to seizures and perform dedicated supervision. ^{Reference DeVries-Rizzo7} As for intracranial EEGs, higher risks of complications and hemorrhage brought the need for a lower nurse-to-patient ratio. ^{Reference Tatum, Mani and Jin2,Reference Hedegärd, Bjellvi, Edelvik, Rydenhag, Flink and Malmgren19} Our survey mirrors previous findings on staff-to-patient ratios in the 2016 Canadian survey: most EMUs employed a nurse-to-patient ratio of one to four or less and modified the ratio for invasive monitoring. No changes in this ratio were reported for daytime, nighttime, days of the week, and pediatric versus adult units in the 2016 survey. ^{Reference DeVries-Rizzo7} The question of optimal nurse-to-patient ratio and technologist-to-patient ratio is extremely relevant in Canada today, given the difficulty many centers face in recruiting new EMU staff members. It is therefore encouraging to see that nurse-to-patient ratio remained more or less unchanged when compared with the 2016 survey.

In our survey, most adult EMUs recommended caregivers for adults with cognitive or behavioral challenges, and most pediatric EMUs recommended caregivers for all children. The 2021 ILAE/IFCN publication recommended caregivers for the aforementioned scenarios, ^{Reference Tatum, Mani and Jin2} although one may instinctively support the presence of caregivers for all patients, since caregivers can lessen reaction time and precipitate the examination of patients by staff members by pushing call buttons at seizure onset. ^{Reference Dericioğlu, Albakir and Saygi20}

Our study showed variability in unit structure across Canadian EMUs. Indeed, most centers reported having a dedicated hospital area for LTVEM, and most (albeit a lesser proportion) of the participating EMUs practiced monitoring from a designated unit with dedicated epilepsy beds. The importance of dedicated hospital areas and designated EMUs was highlighted in the ILAE/IFCN recommendations. ^{Reference Tatum, Mani and Jin2} A study done in the UK to characterize the ideal EMU briefly mentioned the role of unit dedication in patient comfort et monitoring effectiveness. In fact, it determined that proximity and availability of oxygen and suction equipment contributed to environment safety by reducing staff response time. The proximity of other surrounding rooms necessary to epilepsy monitoring (e.g., central monitoring or nursing stations) increased both comfort and security as well. ^{Reference Scott, Fish and Allen21} Another study on patient care in pediatric EMUs suggested that installing effective alarm systems and measures against patient injuries (e.g., rails, shower seats, recessed sinks) as EMUs were established would provide better adaptation and a more secure environment. ^{Reference Perkins and Buchhalter22} The installation of all these pieces of equipment practically requires that the EMU be standalone, with its own reserved beds. Our findings support the notion that unfortunately not all EMUs operate within dedicated areas with dedicated beds, as was also shown in the 2016 Canadian survey. ^{Reference DeVries-Rizzo7}

ECG and oximetry monitoring were used variably across Canadian EMUs. Most EMUs used continuous heart monitoring but did not employ alarms. On the other hand, most EMUs did not use continuous oximetry monitoring. The ILAE/IFCN guidelines, however, do recommend ECG monitoring based on previous studies and suggestions. ^{Reference Tatum, Mani and Jin2,Reference Hamandi, Beniczky and Diehl9} Serious cardiac abnormalities, including cardiac arrest, can occur during the peri-ictal time frame (postictal arrhythmia being of higher importance), highlighting the important role of heart rate monitoring in the EMU. ^{Reference Baumgartner, Pirker, Levin and Chauvel23} A 2016 survey on the EPILEPSY network recorded that 96% of EMUs monitored heart rate, although only 17% of European EMUs in 2015 used alarm systems. ^{Reference Baumgartner, Pirker, Levin and Chauvel23} Thus, Canadian EMUs seem to employ less ECG monitoring than their European counterparts (65% vs 96%), although percentages for the use of alarms for cardiac anomalies remain somewhat similar (31% vs 17%). Although the use of oximetry monitoring is described as optional in the 2021 ILAE/IFCN guidelines, the fact that most EMUs did not use continuous oximetry monitoring is still concerning, especially since postictal respiratory depression is thought to play a role in SUDEP.^{Reference Thijs, Ryvlin and Surges24}

As for seizure detection systems, the 2021 ILAE/IFCN recommendations stated that automated algorithms could be implemented as complementary aid for expert assessment. ^{Reference Tatum, Mani and Jin2} Some authors have shown that automated seizure detection software may improve seizure recognition and hasten intervention from nursing staff, positively contributing to patient safety. ^{Reference Kamitaki, Yum and Lee25,Reference Rommens, Geertsema, Jansen Holleboom, Cox and Visser26} A 2015 European survey of monitoring practices found that only 8% of EMUs used automatic detection methods for seizure-related movements and 15% for ictal EEG abnormalities. ^{Reference Rubboli, Beniczky and Claus8} Similarly, a study on the E-EPILEPSY network practices showed that 19% of EMUs used automatic seizure and spike detection software. ^{Reference Kobulashvili, Höfler and Dobesberger1} On the other hand, a survey of American EMUs found that a 82% of surveyed EMUs used a seizure detection software. ^{Reference Buelow, Privitera, Levisohn and Barkley11} In our survey, 23% of EMUs reported using a real-time seizure detection software, all of which were EEG-based. Our findings further highlight the heterogeneity of seizure detection system use across Canadian centers as compared to the USA and Europe.

The 2021 ILAE/IFCN guidelines stipulated that informed consent should be obtained before LTVEM and that written standardized protocols should determine appropriate practices (for managing and testing during seizures). On the matter of consent, patients should be properly informed of the purpose, procedures, and risks of LTVEM, and they should ideally formally consent with their signature. ^{Reference Baumgartner, Pirker, Levin and Chauvel23} The 2021 ILAE/IFCN article did not go into the details on what should be written on the consent form, however. On the matter of protocols, it has been suggested that standardized protocols for managing patients could be beneficial especially in settings in which staff rotations were frequent, but insufficient evidence exists to back these supposed benefits. ^{Reference Tatum, Mani and Jin2,Reference Rheims and Ryvlin15} Our survey showed that most Canadian EMUs did not require patients to fill out consent forms prior to admission, although most EMUs had protocols for managing falls, rescue medications, status epilepticus, and seizure clusters.

Finally, we investigated whether EMU patients were continuously monitored, either by video or physical presence, throughout their stay. Continuous monitoring in this sense was not explicitly recommended in the 2021 ILAE/IFCN guidelines but seemed important to evaluate given previous recommendations on the subject. A 2017 study of UK EMUs recommended that continuous supervision should be performed by direct observation of the patient, along with video monitoring and nurse alarms. Indeed, continuous patient observation maximizes staff interventions during seizures. ^{Reference Hamandi, Beniczky and Diehl9} A 2016 study showed that 81% of E-EPILSPY network EMUs employed continuous observation during regular working hours and 63% outside of working hours. ^{Reference Kobulashvili, Höfler and Dobesberger1} We showed in our survey that although almost every EMU had a video monitoring station available all day and night, most EMUs did not have staff dedicated to watch the live feed continuously throughout the day and night. In these EMUs, most did not provide continuous physical monitoring of patients (e.g., staff member observing patient from outside their room). As such, a substantial proportion of EMUs did not provide continuous monitoring of patients, and this phenomenon seemed to be worse at night. It is important to note, however, that most of these EMUs were pediatric EMUs, which may not have necessitated continuous monitoring of patients by staff members because caretakers were required to be onsite. Our survey was not designed to appropriately assess whether a caretaker was present at all times by bedside. The proportion of EMUs without continuous patient monitoring may therefore be inflated by these pediatric EMUs. Nevertheless, once again, although continuous monitoring of patients was not required in the 2021 ILAE/IFCN guidelines, we believe Canadian EMUs should look into improving this area, as there has been evidence that it may improve patient safety and, at least intuitively, early detection of a patient’s seizure may lead to quicker interventions. Our group has shown in a previous publication, for instance, that patients who were being continuously monitored by specialized EMU staff during the day benefited from quicker interventions than patients who were being intermittently monitored by less specialized nursing staff during the night. ^{Reference Li, Toffa and Bou Assi27} This publication did not show, however, that continuous monitoring by specialized EMU staff resulted in lower rates of adverse events.

The most serious adverse event that may occur in the EMU is the death of a patient. Across Canadian centers, three have reported a SUDEP in their EMU. The 2013 MORTEMUS study showed that across 147 EMUs around the world, 16 SUDEP occurred during monitoring, 14 of which were at night. ^{Reference Ryvlin, Nashef and Lhatoo6} This study called for enhanced nocturnal supervision to mitigate the risk of SUDEP. In fact, better general patient supervision, not only at night, may decrease SUDEP risk. ^{Reference Schulz, Bien and May28} Nevertheless, the three SUDEP that occurred in Canadian EMUs all occurred in large adult EMUs (none were reported in pediatric EMUs) that have now reported continuously (day and night) monitoring their patients, with nurse-to-patient ratios within the recommended ratio of one to four. A few explanations as to why these SUDEP occurred notwithstanding these levels of supervision may be postulated. Firstly, we do not know exactly when these SUDEP happened; perhaps they occurred when supervision was not as tight, and EMUs strengthened their supervision in reaction to the SUDEP. Secondly, even if an EMU theoretically boasts a 24/7 patient supervision, in practice, staff shift changes and human error may make it so that there are moments in which continuous supervision is broken. If the staff members that ensure this continuous supervision for some reason do not concentrate their attention on the patient (e.g., they are tending to another patient), adverse events such as SUDEP may still occur. EMUs that attempt to ensure continuous supervision should evaluate on a case-by-case basis whether their theoretically continuous supervision holds in practice. In addition, with changes in staff composition, it may be difficult to consistently offer specialized training to all personnel. Finally, with SUDEP being rare, as time goes on, centers may relax their practices; this in turn may predispose to adverse events.

In sum, while this study did not yield data on EEG technologist availability in Canadian EMUs, we suggest that all EMUs review their practices so that technologists may be as available as possible. Furthermore, we would like to stress the importance of abiding by the ILAE/IFCN-recommended nurse-to-patient ratios of one to four (one to one for intracranial EEG patients). Nurses should ideally be “dedicated” to EMU patients; we have shown in our survey that EMU nurses often supervise non-EMU patients, and this practice may have underestimated nurse-to-patient ratios in certain centers. Seeing as several EMU directors were unaware of their unit’s nurse-to-patient ratios, we suggest that EMU directors regularly and actively communicate with their nursing staff for better continuity of care and for audit purposes. We encourage EMUs to recommend caregivers for adults with cognitive or behavioral challenges and for all children, despite 92% EMUs already doing so. Although EMUs should ideally be standalone with dedicated beds, drastically modifying unit structures is not always achievable or warranted. Rather than focus on EMU structure, changes in staffing and equipment seem more feasible and important. EMUs should seriously consider continuous ECG and oximetry monitoring with alarms. Only a minority of Canadian EMUs employ alarms with their ECG or oximetry monitoring. Seizure detection systems may be helpful, although further research should be conducted on these systems to analyze their impact on patient safety and intervention time. EMUs should consider using informed consent forms before patient admission, a seemingly uncommon practice in Canadian EMUs. When possible, standardized protocols for handling certain emergencies can be implemented across EMUs to promote a homogeneous management of adverse events. Finally, we recommend a continuous monitoring of patients, either physically or by video. EMUs should particularly review how monitoring is done at night.

Our study has some limitations. Firstly, some responders were not aware of all their EMU practices (e.g., nurse-to-patient ratio). As such, this study featured some missing data, which were treated with pairwise deletion. Secondly, the 2021 ILAE/IFCN guidelines were not entirely explicit in all their recommendations. A certain level of extrapolation had to be employed to extract the relevant recommendations, and it is possible that some recommendations may have been missed in the process. Thirdly, we based most of our survey items almost exclusively on the 2021 ILAE/IFCN guidelines. The field of EMU safety is wide-spanning, and a plethora of questions unrelated to these guidelines could have been included in the survey. We chose to use a guidelines-based approach since the adhesion of an EMU to these guidelines seemed to be an efficient and easily comparable indicator of the overall safety of EMUs. We believe this approach also helped maximize participation rate, as a lower number of more curated questions made the survey less time-consuming to answer. Nevertheless, in retrospect, many additional topics could have been interesting to explore, the most important being the following: the types and frequency of adverse events, whether EMU directors knew of the new ILAE/IFCN guidelines, the size and exact makeup of the EMU staff, whether nurses and technologists had proper cardiopulmonary resuscitation training, whether nurses were specially trained in epilepsy, the absolute number of technologists screening the recordings, whether the EMU had epilepsy fellows, which person was usually called for emergencies, whether patients were kept over the weekend or holidays, how often the EMU was closed due to hospital overflow, and how patients were tracked during and/or after a seizure. A more in-depth survey not aimed specifically at assessing if centers adhered to ILAE/IFCN guidelines could fill in these knowledge gaps left behind by our current survey.