Introduction

In line with the global response to the emergence of the novel coronavirus 2019 (COVID-19), the Ontario Provincial Government declared COVID to be a Public Health Emergency on March 17, 2020 [1]. Predictably, the implementation of a number of public health measures to reduce the spread of infection greatly impacted the Ontario healthcare system [2–6], which, in turn, greatly impacted patient treatment options and experiences [7–9]. Emergency departments (EDs), in particular, saw a substantial decrease in the number of patients accessing emergency services, largely due to public health restrictions and fears of COVID exposure in hospital settings [4, 10–12]. This hesitancy to access emergency services had a disproportionate effect on already vulnerable and underserved patient populations during the pandemic, namely those suffering from mental health and/or substance use (MH/SU)-related disorders, for whom emergency services were a primary source of assistance for some prior to the COVID pandemic [13–15]. Of particular concern, research into the impact of COVID on emergency service provision has shown a significant reduction in both psychiatric care and substance use treatment resources during the pandemic [9, 16–18] despite the notable increase in MH/SU-related concerns during this time [19–26].

In light of the observed reduction in both emergency and psychiatric service access due to the COVID pandemic, one aspect of patient treatment that deserves further exploration is whether the pandemic also impacted the use of restraint and seclusion (R/S) interventions amongst patients during service provision in emergency care facilities. The impact of COVID on the use of R/S interventions during treatment is of concern for a number of reasons, including the fact that COVID was an unprecedented event that affected service provision at every level of the healthcare system [2–6]. Given the fact that the frequency of R/S intervention use is considered an indicator of the quality of care [27–30], due to the very real potential for physical (e.g., asphyxiation, fractures, muscle atrophy and contracture, etc.) and psychological injury (e.g., agitation, delirium, PTSD, etc.) [31, 32], understanding how R/S use was impacted by the pandemic offers a useful measure of the broader impact of COVID on health care system and patient experiences.

Existing research on COVID-related changes in R/S intervention use is in its early stages, with reported changes in R/S use varying substantially depending on the patient population and hospital setting. For example, a study from a tertiary mental health hospital in Ontario, Canada [33] noted a substantial reduction (49–100%) in the use of mechanical restraints in adolescent, forensic, and geriatric patient programs after the implementation of COVID-related restrictions. However, they only sampled data for the first 6 weeks after the COVID onset. Similarly, a high-needs psychiatric ward in Dublin, Ireland [34] reported substantial reductions in the use of R/S interventions after voluntarily reducing ward occupancy to facilitate social distancing and other infection prevention strategies. Specifically, reducing bed availability by only 5% resulted in the number of R/S events decreasing by more than 50% over nine months following the onset of the COVID pandemic. On the other hand, research from a mental health hospital in Malaga, Spain, found no change in the likelihood of mechanical restraint [35].

Most consistently, though, researchers have reported increased use of R/S interventions after the onset of the COVID pandemic. For example, a substantial increase in the incidence rate of physical restraint use was observed in geriatric dementia patients presenting to an acute care hospital in Japan, reaching a 50% likelihood of a restraint event approximately four months after the enactment of nationwide public health measures [36]. Similarly, orders for physical restraint interventions among older adults during acute care hospitalization in Ontario, Canada, were found to have increased in prevalence by 2.4% [37], while children and youth presenting with mental health concerns to a pediatric hospital in Michigan, USA were found to be almost four times more likely to experience a restraint event after the onset of the COVID pandemic [38]. A retrospective study on youth and adolescent psychiatric hospitalizations in Israel also noted a small increase in R/S intervention use despite a significant decrease in the number of admissions after the onset of COVID [39]. Similarly, a study of adult psychiatric admissions in Germany found that the number R/S events increased by almost 25% for patients who were voluntarily admitted, and 13% among those who were involuntarily admitted during the pandemic [40]. Finally, a study from the Department of Psychiatry at the Geneva University Hospital in Switzerland reported a slight increase in the number of seclusion events after COVID onset, despite a decrease in both voluntary and involuntary hospital admissions [41].

Materials and methods

Sample

Data for this study were obtained from the NACRS database of emergency care service provision across Canada on November 10, 2021. The raw dataset consisted of administrative and clinical data from 559,817 separate ED visits in Ontario during the fiscal years 2019 (N = 298,734) and 2020 (N = 261,083). These data are anonymized by CIHI prior to dissemination and do not contain information by which an individual accessing ED services could be personally identified.

Due to the nature of the present study, we limited our sample to cases with complete data regarding R/S intervention use. The availability of this data is subject to the data collection instructions set out by NACRS [45], which mandate reporting of R/S intervention use data only for ED patients presenting with a main problem diagnosis (see Clinical Variables below) from the Mental and Behavioural Disorders (Chapter V) or Intentional Self-Harm categories of the ICD-10-CA [46, 47], regardless of whether an R/S even occurred. Thus, to ensure complete R/S use data, we restricted our sample to cases in which an individual presented to an Ontario ED with an MH/SU-related primary concern for the present analyses, resulting in a final sample of data for 332,046 individual ED visits (2019: N = 174,686 cases; 2020: N = 157,360 cases; Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sample descriptives for clinical and temporal variables of interest for ED visits in Ontario during the 2019 and 2020 fiscal years.

https://doi.org/10.1371/journal.pone.0302164.t001

Results

COVID-related changes in ED visits and the use of R/S interventions

Before exploring factors associated with COVID-related changes in R/S intervention use, we first wanted to determine that there were, in fact, significant changes in the number of ED visits and/or the number of R/S events that occurred in Ontario EDs after the onset of COVID. The overall number of visits to Ontario EDs for MH/SU presentations decreased by 9.9% (17,326 visits) after the onset of the pandemic (Table 1 and Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Number of ED visits by registration year.

Number of ED visits in Ontario for MH/SU concerns before (2019) and after (2020) the onset of the COVID-19 pandemic. ED = emergency department, MH/SU = mental health/substance use.

https://doi.org/10.1371/journal.pone.0302164.g001

While ED visits decreased in general, the number of visits in which R/S interventions were used increased by 9.5% (833 instances; χ²(1) = 185.52, p < .001) during the first year of COVID (Table 1 and Fig 2). This is further confirmed by binary logistic regression results indicating that patients presenting to Ontario EDs during the first year of COVID were 23% more likely to experience an R/S event than those presenting the year prior (OR = 1.23, p < .001, 95% CI [1.19, 1.27]). This observed increase in the prevalence of R/S intervention use was comparable across both restraint (physical/mechanical/chair; 9.6% increase) and seclusion (9.1% increase) intervention methods (Table 1 and Fig 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Number of ED visits by R/S event occurrence.

Number of ED visits in Ontario before (2019) and after (2020) the onset of the COVID-19 pandemic by R/S event occurrence (bar; right axis) with corresponding percent change from pre-COVID (2019) levels (line; left axis). ED = emergency department, R/S = restraint and seclusion intervention.

https://doi.org/10.1371/journal.pone.0302164.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Number of ED visits by type of R/S event occurrence.

Number of ED visits in Ontario before (2019) and after (2020) the onset of the COVID-19 pandemic by type of R/S event occurred (bar; right axis) with corresponding percent change from pre-COVID (2019) levels (line; left axis). ED = emergency department, R/S = restraint and seclusion intervention.

https://doi.org/10.1371/journal.pone.0302164.g003

Factors associated with COVID-related changes in the use of R/S interventions

Registration month.

There was a significant interaction between the month and the year of registration (Wald χ² (11) = 109.48, p < .001), suggesting that the onset of the COVID pandemic had a substantially different effect on the likelihood of an R/S event occurring depending on the month in which a patient visited an ED. Upon exploration of predictor-level interaction effects, a clear pattern of changes in R/S probability emerged primarily centered on the onset of the COVID pandemic in Ontario (Table 1 and Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Month of ED visit.

Percent of ED visits by registration month in which an R/S event occurred before (2019) and after (2020) the onset of the COVID-19 pandemic (bar; right axis) with corresponding changes R/S probability between 2019 and 2020 (line; left axis). ED = emergency department, R/S = restraint and seclusion intervention.

https://doi.org/10.1371/journal.pone.0302164.g004

Most notably, there was a dramatic increase in the probability of an R/S event at the start of the first wave of the COVID pandemic in Ontario, beginning with a 3.6% increase in April 2020, followed by a decreasing trend in the magnitude of the R/S probability change in May (2.5%), and June (1.4%) of 2020, with R/S probability returning to levels comparable to those before the onset of COVID by July of 2020 (> 1% increase in R/S probability; Table 1 and Fig 4). Another period of increased likelihood of R/S was observed during Ontario’s second wave of the COVID pandemic, albeit to a much lesser degree than was seen during the first wave. Specifically, the likelihood of an R/S event began to increase again beginning in September 2020 (the beginning of the second wave) and plateaued at a 1% increase from pre-COVID levels from November 2020 to February 2021 (Table 1 and Fig 4).

Main problem diagnosis.

The type of SU/MH concern (main problem diagnosis) a patient presented with significantly interacted with patient registration year (Wald χ² (9) = 30.35, p < .001), suggesting that the onset of the COVID pandemic impacted the probability of experiencing an R/S event differently depending on a patient’s main problem diagnosis. Further exploration revealed that there was a general increase in the probability of an R/S event across all diagnostic categories from 2019 to 2020 (range = 0.7% - 21.3%), and it is only the magnitude of this increase that varied across diagnostic categories (Table 1 and Fig 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Main problem diagnosis.

Percent of ED visits by main problem diagnosis in which an R/S event occurred before (2019) and after (2020) the onset of the COVID-19 pandemic (bar; right axis) with corresponding changes R/S probability between 2019 and 2020 (line; left axis). ED = emergency department, R/S = restraint and seclusion intervention.

https://doi.org/10.1371/journal.pone.0302164.g005

Patients with a diagnosis of dementia saw the greatest increase of 21.3% in the likelihood of experiencing an R/S event after COVID onset, followed by patients diagnosed with a psychotic disorder or polysubstance abuse, who both showed an increase of 17.7% in their likelihood of experiencing an R/S event (Table 1 and Fig 5). Individuals diagnosed with a personality disorder saw a moderate 13.3% increase in R/S probability after COVID onset, while an increase of approximately 10% in R/S probability was found for patients diagnosed with cannabinoid abuse, a mood disorder, or other mental health disorder not otherwise specified. Finally, patients diagnosed with alcohol abuse (0.7% increase), opioid abuse (5.4% increase), or an anxiety disorder (3.4% increase) saw the smallest increases in the probability of experiencing an R/S event from pre-COVID levels.

Discussion

The present study aimed to examine the association between temporal and clinical factors and changes in the use of R/S interventions in Ontario EDs during the COVID pandemic as an indicator of broader changes in the quality of care and patient experiences during this time. To this end, we first wanted to establish whether the onset of COVID affected ED access and the use of R/S interventions during service provision in Ontario EDs. We found that there were approximately 17,000 fewer visits to Ontario EDs for MH/SU concerns after the onset of the pandemic, a decrease of almost 10% from pre-COVID levels. Notably, despite there being fewer ED visits during the first year of the pandemic, the number of visits where an R/S event occurred increased by almost 10%, which translated to 833 more visits in which the patient was restrained or secluded during service provision, which is consistent with previous research exploring COVID-related changes in R/S intervention use.

Further underscoring the fact that this observed increase in R/S intervention use was related to the pandemic onset, the greatest increase in R/S risk was seen for patients who accessed Ontario EDs during the first few months of the first wave of the pandemic (April–June 2020). The pandemic was officially declared a public health emergency in Ontario on March 17, 2020, suggesting that the observed spike in the probability of R/S intervention use aligns with this declaration and the resulting impact on the public and healthcare system. We suggest that the combination of staffing resource shortages, and strict infection prevention protocols likely prevented staff from effectively employing traditional de-escalation methods and less coercive R/S alternatives, such as 1-to-1 patient supervision. Consistent with this interpretation, we also found that the prevalence of R/S use returned to comparable levels by the end of the first wave of infection, suggesting that, as resources and staffing stabilized, the healthcare system was able to adapt and mitigate factors associated with the temporary increase in the use of R/S interventions.

Even more dramatic than the observed temporal effects of COVID was the finding that the risk of experiencing an R/S event doubled for almost all patients who presented to Ontario EDs with primary complaints associated with MH/SU concerns after the onset of COVID, regardless of the specific disorder. This further supports the suggestion that increases in R/S intervention use during COVID were a function not only of the exacerbation of psychiatric and behavioural symptoms specific to certain disorders but also the systemic issues related to health care service provision described above, namely, the availability of relevant resources. In addition to increasing the likelihood that patients with MH/SU-related concerns had to turn to ED services during COVID, the lack of alternative resources may have also led patients to delay seeking treatment, resulting in significantly greater symptom severity and in turn, R/S risk when the patient did seek care. Notably, the probability of R/S use nearly tripled for patients presenting with MH/SU concerns that were already associated with increased R/S risk prior to COVID onset, namely psychosis, personality disorder, dementia, and substance abuse. Arguably, these patients are also more likely to have had difficulty complying with strict infection prevention protocols during service provision due to the cognitive and behavioural symptoms specific to these disorders. This in turn may have contributed to the disproportionate increase in R/S intervention use we observed, as the use of physical restraint or patient seclusion may have been necessary to minimize infection risk in non-compliant patients.

Future directions

This study is just an initial step toward understanding the underlying issues that contributed to the increased use of R/S interventions among patients presenting to EDs with MH/SU-related concerns. Naturally, more research into the factors associated with changes in R/S intervention use due to the COVID pandemic is needed to fully understand how such events impact patient care. A better understanding of the underlying factors driving the differences in the likelihood of an R/S event across different patient populations and treatment settings is of particular importance, not only to identify additional exacerbating factors, but also mitigating factors that effectively reduced R/S risk under similar circumstances. In line with this, the synthesis of the research into the factors associated with increased R/S intervention use is equally important, as it has the potential to be translated into valuable improvements to existing policies and practices related to the use of R/S interventions. Similarly, continued attention to education and training regarding the use of R/S interventions is equally important, with particular emphasis on appropriate use and practical alternatives to coercive methods. For example, in response to the COVID pandemic, there is a clear need to develop strategies to minimize the need to use such methods as a form of infection prevention among vulnerable populations that struggle with treatment compliance.

In particular, as suggested above, the observed increase in R/S intervention use in the present study is likely not only due to an increase in psychiatric symptom severity but also because of the overall strain on the healthcare systems leading to a lack of resources. Therefore, efforts to prevent similar increases in the number of R/S events during future major public health crises should focus on non-patient-related contextual factors that are likely to be impacted, such as access to non-emergent psychological services and substance abuse programming. One of the outcomes of the pandemic was the advent and refinement of telehealth programs that made non-emergent healthcare services much more accessible while minimizing infection risk. This now-existing infrastructure may help mitigate similar situations during future public health crises by allowing easier and earlier access to services specifically intended to address concerns related to MH/SU issues before they require emergency care and the use of R/S intervention methods. Having said this, the public health system cannot hope to rely solely on telehealth alternatives and should develop appropriate emergency response plans to deploy and allocate resources with the knowledge that major public health crises can have a disproportionately negative impact on the experience of already vulnerable patient populations, as well as patients who may be experiencing MH/SU issue for the first time due to the unique stressors associated with major public health crises, such as COVID.

Conclusions

The results of the present study provide evidence that the use of R/S interventions in Ontario EDs increased substantially among patients presenting with MH/SU concerns after the onset of COVID, despite a decrease in the number of ED visits during the same period. Further underscoring the detrimental effect of COVID on the Ontario health care system, the observed increase in R/S intervention use was most prominent during the first few months after the public health emergency declaration in Ontario, when resources were under the most significant strain and uncertainty surrounding the virus was at its highest. Finally, we demonstrated that patients presenting to Ontario EDs with MH/SU-related concerns were at a substantially increased risk of experiencing an R/S event during the first year of the pandemic, regardless of the specific nature of their presenting concern, but that this effect was most pronounced for patients with MH/SU concerns traditionally associated with increased risk of R/S intervention use.

Taken together, the findings of the present strongly support the conclusion that the observed increase in R/S intervention use in Ontario EDs may be a result of the general state of upheaval in the Ontario healthcare system caused by the COVID pandemic. In context with the existing literature, the results of this study have the potential to further our understanding of COVID-related changes in the presentation of MH/SU issues in ED facilities, as well as how the healthcare system responded to these changes during the pandemic. Furthermore, this knowledge has the potential to inform and guide improvements to existing policies and procedures related to the use of R/S interventions to minimize the inherent risk to both patients and staff.

Supporting information