In response to the COVID-19 pandemic, hospitals around the globe restricted inpatient bedside family presence1,2,3 to limit viral spread, preserve personal protective equipment, and protect patients and clinicians.2 Most North American pediatric hospitals initially restricted presence to one caregiver,3,4,5,6 which addressed some ethical issues from caregiver prohibition during the severe acute respiratory syndrome (SARS) epidemic of 2002–2004.7,8 Nonetheless, restricted family presence (RFP) policies represent a significant deviation from the standard of family-centred care (FCC) embraced by most North American children’s hospitals and pediatric intensive care units (PICUs).9

Pediatric intensive care units are a high-stress environment where the majority of pediatric hospital deaths occur.10 Parents, fearing for the life of their child,11 are at risk of trauma12 and in need of care and support from the health care team.13 In PICUs using FCC models, clinicians facilitate family presence and work alongside family members to optimize care of PICU patients.14,15 These humanistic values and interactions improve work satisfaction and decrease clinician burnout.16,17 In this setting, where clinicians value family centredness, RFP policies may result in moral distress.18,19

The experience of restrictions in adult critical care during the COVID-19 pandemic has been explored.20 Within the PICU, one qualitative study found that care provision during the COVID-19 pandemic added stress for clinicians, though it did not focus on the influence of the family presence restrictions.21 To inform future policy and practice, the impact of these policies in a PICU context must be explored to assess the proportionality of the response. This study aimed to explore the impact of RFP policies and practices on PICU clinicians. Our primary objective was to assess the degree of associated moral distress. Secondary objectives were to assess other distress-related mental health impacts associated with restrictions and explore clinician opinions about and experience with RFP policy design and implementation including recommendations for the future.

Methods

Study design and ethical considerations

We conducted a self-administered, web-based, anonymous, cross-sectional survey that adhered to the Consensus-based Checklist for Reporting Of Survey Studies (CROSS, see Electronic Supplementary Material [ESM] eAppendix 1).22 This study was approved by the Research Ethics Board at the University of Alberta (Edmonton, AB, Canada; Study ID: PRO 00102535). A letter of information preceded the survey; participation constituted consent to collect and publish data.

Setting and sample

Through nonprobability voluntary response sampling, we invited any clinician who worked in Canadian PICUs between March and June 2020 to participate in either French or English. Clinicians were defined as any PICU professional who worked with patients and their families including, but not limited to, intensivists, nurses, trainees, respiratory therapists (RRTs), social workers, child life specialists, pharmacists, physiotherapists, and unit aides. On 2 October 2020, we emailed the participation invitation to 19 physician leads and 17 operational managers of the 19 administratively distinct Canadian PICUs (within 17 hospitals),6 and requested forwarding to all PICU staff. The invitation included an untraceable web link and printable posters with quick response codes. Three reminder e-mails were sent to PICU leadership at one-month intervals. This recruitment strategy made it impossible to estimate the number to whom the survey was sent and the response rate. Respondents were not prevented from accessing the survey more than once.

Data sources

We designed a questionnaire (ESM eAppendix 2) that addressed five domains: clinician demographics; baseline and pandemic-related family presence and FCC practices; experience and opinion of RFP policy and practice; moral distress; and impact on the clinicians.

The questionnaire was developed by multiprofessional PICU clinicians, administrators, patient partners, clinician researchers, and an epidemiologist following the methodology described by Burns et al.23 Items were generated using: 1) existing literature on family presence, FCC, and moral distress24 and 2) team discussion of personal and professional experiences of clinicians and patient partners.

We used three tools to assess clinician distress associated with RFP policies: the Moral Distress Thermometer (MDT);25 the Impact of Event Scale (IES);26 and the change in perceived stress using a five-point scale (from 1 [significantly decreased] to 5 [significantly increased]). The MDT is a validated single-item visual analog scale that provides a definition and asks participants to rate their moral distress from 0 (no distress) to 10 (worst distress possible).25 The IES26 is a reliable and valid27 15-item scale that assesses intrusive and avoidance responses to traumatic or stressful events. The IES asks respondents to think about and describe a specific situation or event while answering how often they experienced the given symptoms in the last week (not at all = 0, rarely = 1, sometimes = 3, often = 5). The IES may be interpreted as: 0–19 low; 20–29 moderate; and ≥ 30 high responses,28 with a suggested cut-off of 26 for a clinically relevant reaction.29

We also used the PICU-Family Presence Index (PICU–FPI) tool. This is a reliable and valid 20-item tool developed by our team to assess clinician perception of their PICU’s approach to family presence and participation.30 Participants were asked whether each statement applies to their PICU (yes/no) with each statement assigned a score of − 1 (nonfamily centred) or 1 (family centred). Scores range from − 8 to 12 (see ESM eAppendix 3 for tool development and testing).

The survey was pretested (six multidisciplinary PICU clinicians and two PICU family members) with clinical sensibility testing. The revised survey was then pilot tested for readability and flow (five PICU clinicians). The final instrument included 48 closed and seven open-ended questions. It was offered in English and French, and administered using QualtricsXM (Qualtrics, Provo, UT, USA).

Data analysis

Survey results were cleaned and exported into IBM SPSS for Windows version 25.0 (IBM Corp., Armonk, NY, USA). Respondents who indicated they did not experience RFP were excluded; their baseline data were compared descriptively with those included in the full analysis. For missing responses due to skip patterns, we used the actual number of responses as the denominator. Attrition was assessed using linear regression analysis. Nominal variables were reported as percentages; ordinal or skewed continuous data as median and interquartile range [IQR]; and normally distributed continuous variables as mean and standard deviation (SD) or, when compared between two groups, as mean and 95% confidence interval (CI). Visual inspection of histograms was used to assess the normal distribution of the data. We used independent samples t test to compare differences in means between two groups, and the independent samples Kruskal–Wallis test to compare medians. One-way ANOVA was used to compare means between more than two groups, with a Tukey Honestly Significant Difference correction applied to multiple comparisons. Univariate regression analysis assessed the correlations of categorical baseline variables with degree of moral distress, perceived stress, and with the IES. All variables with a bivariate association at the P < 0.10 level were subsequently included in the multivariable stepwise linear regression model, with a minimum of ten subjects per independent variable for adequate statistical power in the analysis. Multicollinearity was assessed using the Variance Inflation Factor and Tolerance statistic. A P value < 0.05 was considered statistically significant.

Free text answers to open-ended questions were analyzed using the General Inductive Approach described by Thomas.31 One member of the research team with significant qualitative analytic experience who is not a clinician (M. R.) read all responses, generated a coding framework, applied the framework adding codes as needed, then grouped the codes into categories. These were assessed and verified by another team member with qualitative analytic experience (J. R. F.).

Results

Respondent demographics

Three hundred and eighty-eight responses were received from 17 (90%) of 19 Canadian PICUs between 6 October 2020 and 9 February 2021. Of those who responded, 368/388 (95%, 17 PICUs) indicated that they experienced RFP policies and were included in the full analysis. There were no significant baseline differences for respondents not included in the whole analysis (ESM eAppendix 4). Respondents were predominantly female (333/368, 91%), English-speaking (338/368, 92%), and registered nurses (240/368, 65%) (Table 1). Only free-text questions were skipped. For non free-text questions, response attrition was 21.7% and fit a linear model (y =  − 5.8x + 366 [R2 = 0.90]).

Table 1 Respondent demographics
Full size table

Mental health impacts

Moral distress

The mean (SD) MDT rating that respondents associated with RFP policies and practices was 4.5 (2.4) out of a possible 10 (n = 307). The bivariate relationship between each of the measures of distress (moral distress, IES, and perceived change in stress) and respondent demographic, impact, and experience variable are outlined in Table 2 and Fig. 1. The results of the multivariable stepwise linear regression models for moral distress, IES, and perceived change in stress are reported in Table 3.

Table 2 Associations between demographic and dichotomous variables and measures of clinician distress
Full size table
Fig. 1
figure 1

Categorical responses and associations with measures of distress

Full size image
Table 3 Stepwise multivariable linear regression models
Full size table

Mean moral distress was higher in females (4.6; 95% CI, 4.3 to 4.8 vs 3.7; 95% CI, 2.9 to 4.6; P = 0.04). When adjusted for multiple comparisons, moral distress differed by profession only between “other professionals” and RRTs (mean, 5.9; 95% CI, 4.8 to 7.0 vs 3.7; 95% CI, 3.0 to 4.4, respectively; P = 0.01), and did not differ significantly by years of experience. In the multivariable analysis, increased moral distress was associated with perceptions of differential impact of RFP on some families, increased stress for families, and increased workload. Less moral stress was associated with the beliefs that RFP 1) decreased workload, 2) was congruent with PICU values, 3) meant the hospital and PICU valued clinician health, and 4) helped families cope well with restrictions.

When respondents were asked to share any additional comments, thoughts, or experiences related to moral distress, 19 of the 97 free-text responses addressed the theme of “end-of-life care.” Conversely, eight participants identified distress related to aggressive or noncompliant family members and four from lax rules or enforcement.

Impact of event scale

The mean (SD) total IES score associated with RFP was 29.7 (10.5) (n = 290), consistent with a moderate degree of distress.28 We identified seven categories of events or experiences that respondents described as impactful (Table 4). The most frequently reported category (39%) was “concern about the impact on family”, followed by “nonfamily-centred end-of-life situations” (29%).

Table 4 Impactful experiences during restricted family presence
Full size table

There was no association between IES score and years of experience, profession overall, or gender. Professionally, the PICU fellow/trainee/clinical associate group (mean, 19.4; 95% CI, 13.3 to 25.4) had significantly lower mean scores than RNs (30.4; 95% CI, 29.0 to 31.8; P < 0.01), RTs (31.9; 95% CI, 28.5 to 35.3; P < 0.01) and other clinicians (32.1; 95% CI, 26.9 to 37.3; P = 0.02), though not than staff MDs (24.6; 95% CI, 20.7 to 28.5; P = 0.71).

General stress

Most respondents (176/317, 56%) indicated an increase/significant increase in stress attributable to RFP; 27% (86/317) indicated no change. Change in stress was not correlated with years of experience, profession, or gender (Table 2).

Examining the relationship between measures of clinician distress, moral distress, and alteration in general stress were strongly correlated (rs = 0.6; P < 0.001).32 The IES scores correlated moderately with moral distress (rs = 0.4; P < 0.001) and weakly with alteration in general stress (rs = 0.2; P < 0.001).

Experience and opinion of restricted family presence policy and practice

The mean (SD) PICU-FPI for the era before the pandemic was 6. 8 (2.4) vs 1.3 (2.8) early in the pandemic, with a mean paired difference of − 5.5 (95% CI, 5.2 to 5.8; P < 0.001) (see ESM eAppendix 4). Responses to experience and opinion questions and their associations with measures of clinician distress are displayed in Fig. 1. Most respondents indicated that they were not consulted prior to implementation. Although most agreed that the hospital and PICU valued their health when implementing the policies (246/331, 74%), 57% (188/332) disagreed that their experience with RFP was mainly positive. Many respondents (198/308, 64%) perceived that RFP policies impacted families differentially (Table 5). While 52% (171/331) of respondents agreed that RFP made their job easier, 37% (121/331) indicated an increased workload (Fig. 2).

Table 5 Groups perceived to have been differentially impacted by restricted family presence policies and practices by bedside PICU health care providers
Full size table
Fig. 2
figure 2

Care elements resulting in an increase in workload. Only respondents who indicated an increase in workload due to restrictions in family presence (n = 121, 36.6%). Total > 100% as respondents checked all that applied.

Full size image

When invited to provide free-text comments on policies, 38 of 164 respondents indicated appreciation for restrictions for reasons beyond infection control; 12 perceived that restrictions should be more strictly enforced, particularly for patients with infectious symptoms.

When asked what future family presence policies should look like, respondents provided answers that fit into three themes: 1) policy priorities (n = 35) including a need to balance competing priorities, maintain FCC, and ensure policy flexibility; 2) policy development (n = 120) that includes stakeholder, and particularly PICU frontline clinician, input and must address the clinician’s opinion that both parents should be present; and 3) policy implementation (n = 84), which must involve clear communication of consistent rules with provisions for enforcement that maintain a therapeutic relationship, additional sources of support for families, and transparent processes for granting rules exceptions. See ESM eAppendix 5 for themes, subthemes, and exemplar quotes.

Discussion

Although Canadian PICUs did not implement the restrictions barring all family presence seen in adult intensive care units (ICUs),6 this study showed that clinicians nonetheless experienced moral distress as a result of RFP policies. Policies and practices were perceived to be implemented without clinician input, to violate PICU values, and to result in decreased family-centredness of care. Although some respondents appreciated having fewer family member interactions, most experienced increased psychological burden associated with denying family access to critically ill children, separating families, and isolating family members from supports. Drawing on their frontline experiences, respondents expressed that future family presence policies should be consistent, developed with stakeholder input, clearly communicated, enable at least two caregivers at once, and provide transparent processes for compassionate rule exceptions.

A recent systematic review addressing RFP impacts identified ten adult and no pediatric ICU studies.33 The authors described similar clinician impacts to those herein: moral distress, distress around end-of-life situations, and positive responses to limitation of occupational COVID-19 exposure. Unique to pediatrics, PICU clinicians witnessed trauma in the lone parent coping with their child’s critical illness and felt that family-centred values were threatened. Although other PICU-based studies have examined general sources of COVID-19-related clinician stress,21 the present study examined the effects of RFP policies specifically and its findings raise red flags about the potential for traumatic impact within the interprofessional clinical team.

Moral distress occurs when an individual feels they know the ethically correct course of action, but feel powerless to enact it.34 Participants experienced a moderate degree of moral distress, similar to studies of PICU nurses and physicians facing challenging clinical scenarios like treatment futility, perceived powerlessness, and inadequate team communication during baseline operations.35 Multiple organizational factors may affect the risk for moral distress such as relationships with management and care processes.36 Thus, it is relevant that participants who felt heard by management and those who perceived congruence with PICU values experienced less moral distress. Notably, nurses usually experience higher moral distress than staff physicians during routine operations,37 but it did not differ significantly in this setting. As decisions related to RFP were made by administrators outside the PICU,6 both nurses and physicians were placed in the same powerless situation of “moral hazard,” congruent with adult ICU studies,38 with attendant risks of moral injury, depersonalization, and abandonment of the profession.18,39

Most respondents reported an IES score > 26, a cited indication for psychological referral.29 For comparison, the mean (SD) for individuals experiencing cancer has been reported as 31.5 (17.8)40 and median [IQR] for otolaryngology physicians in high COVID-19 prevalence regions as 17.0 [5.0–28.0].41 The most commonly cited impactful events were those counter to key elements of FCC9,13 and PICU values. Pediatric intensive care unit trainees/fellows/clinical associates/nurse practitioners had lower mean IES scores than other professions except staff physicians. This is consistent with studies comparing distress for physicians and nurses during an epidemic-pandemic42 and may have been influenced by decisional capacity and ability to leave the bedside.43 Nevertheless, predictions of human reactions are complex and identified factors will only explain a portion of the clinician distress.

Restrictions did not significantly increase most clinicians’ workload as it did for their adult counterparts.44 Increases were experienced mostly to negotiate and advocate for policy exceptions for end-of-life care2 and beyond, with resulting suggestions to formalize an equitable exceptions process for future policy. We postulate that the absolute PICU-FPI change from prepandemic to pandemic did not correlate with the degree of moral distress, stress, or traumatic stress because individuals experienced stress from limitations to presence and family centredness irrespective of the degree of change. We hypothesize that the weak correlation between IES scores and perceived change in stress was because IES examines current symptoms of avoidance and intrusion related to one event while our stress measure asked participants to reflect on change in stress from RFP in its entirety. It is possible that distress may have waned at the time of survey completion.

The high mean scores reported for mental wellbeing indices are worrisome. Provisions are needed to support clinicians in managing the psychological burden associated with implementing policies and practices that are counter to clinician-identified PICU values.39 Involving frontline clinicians in future policy development and implementation is feasible and was desired by respondents, and may be an essential preventative action to ensure a stable and healthy workforce.45

Our results are limited by the inability to estimate the size of the interprofessional PICU workforce, with a resulting unknown response rate. There is a risk of response bias, whereby respondents who had the most impactful experiences were the most likely to reply, and of recall bias with responses impacted by recent events or by attenuation of memories and emotions over time or clinicians’ overall mental health during the pandemic. Our survey showed significant attrition, which may be related to its length and requests for free-text responses. As respondents had unrestricted access to the survey, it is possible that some answered the survey more than once, and that respondents who opened the survey multiple times without completing it may have inflated attrition. Although restrictions to presence were implemented globally,3 the cultural and social contexts of our findings may not be generalizable outside North America. Finally, we sought perceptions and opinions from early in the pandemic when restrictions were the most severe; accordingly, these results do not represent policy evolution throughout the pandemic.

Study strengths include the use of validated tools and rigorous survey development methodology with a multidisciplinary team that included patient partners and hospital leadership. The sample had robust geographic representation including most Canadian PICUs and units operating in French and English, and respondents showed both negative and affirmative policy impressions.

Conclusions

Restricted family presence policies in Canadian PICUs during the COVID-19 pandemic increased stress and resulted in mental health impacts on clinicians, including moderate levels of moral distress and trauma-associated distress. Restrictions to family presence place the wellbeing and functioning of the multidisciplinary team at risk. Ensuring policies are consistent and developed with frontline clinician input may decrease negative effects on clinicians.