Distress and the Long-Stay Pediatric Intensive Care Unit Admission: A Longitudinal Study of Parents and the Medical Team

Abstract

Prolonged critical illness in children has emotional consequences for both parents and providers. In this observational cohort study, we longitudinally surveyed anxiety and depression in parents and moral distress in pediatric intensive care unit (PICU) providers (attendings, fellows, and bedside registered nurses) and explored their trajectories and relationships. Anxiety/depression and provider moral distress were measured using the Hospital Anxiety and Depression Scale and the Moral Distress Thermometer, respectively. The relationships of parental and provider distress were evaluated using Spearman's correlations, and their trajectories and potentially associated variables were explored using quadratic random slope and intercept models. Predetermined associated factors included demographic and clinical factors, including parent psychosocial risk and intubation status. We found parental anxiety and depression decreased over their child's admission, and parental psychosocial risk was significantly associated with anxiety (coefficient = 4.43, p  < 0.001). Clinicians in different roles had different mean levels and trajectories of moral distress, with fellows reporting greater distress early in admissions and nurses later in admissions. Parental anxiety/depression and provider distress were significantly, though moderately, correlated. We conclude that anxiety and depression in parents of children with prolonged PICU admissions and the moral distress of their clinicians correlate and vary over time and by provider role.

Introduction

Children with serious chronic conditions are living longer lives, and many can only do so because of the care they receive in the pediatric intensive care unit (PICU). These admissions can be prolonged ¹ and accompanied by distress, not only for the child but for the parents and clinicians as well. Cross-sectional studies show that parents of critically ill children experience anxiety and depression both at the beginning and after the PICU stay. ^{2 3 4 5} Parents of children with prolonged PICU admissions may have greater parental anxiety and depression ⁶ and certainly experience the downsides of rotating clinicians (e.g., changing physicians and nurses, some of whom are unacquainted with their child, who can have different plans and approaches) which is better suited to care for acute illnesses. ^{7 8 9} Among clinicians caring for these complex patients, emotional and moral distress are well described. ^{10 11 12 13 14} Moral distress is the resultant emotional disturbance when a practitioner wishes to care for a patient in a certain way (i.e., “do the right thing”) but cannot. ¹⁵ Moral distress can stem from varied reasons, including perceived futility, perceived poor patient care, and, in the case of children with prolonged ICU stays, from an inability to meet their needs for continuity of care, appropriate developmental stimulation, and chronic illness management due to the acute-care focus of PICU care. ^{9 11} While we suspect that these emotional outcomes are intertwined and occasionally stem from similar reasons, the distress of parents and clinicians of long-stay PICU patients has never been studied longitudinally or in relation to each other.

Given family-centered pediatric critical care seeks to mitigate both the adverse secondary psychosocial consequences of critical illness in patients and parents and high levels of distress experienced by clinicians, which are associated with compassion fatigue and burnout, ¹⁶ we tracked anxiety and depression in parents with children who had prolonged PICU stays and moral distress in their PICU clinicians. Hypothesizing a positive association between family and provider clinician distress, we sought to elucidate the trajectories of these emotional outcomes over time, in relation to various factors, and to each other. Our hope is to inform future efforts to prevent and reduce distress among parents and PICU clinicians.

Materials and Methods

We conducted a prospective observational cohort study in the medical/surgical, neurologic, and cardiac PICUs of a tertiary academic medical center. Institutional Review Board approval was obtained, as was informed consent from all participants.

Participants

Our target populations were parents of children with prolonged PICU admissions and their corresponding PICU providers. Parents of children of any age were eligible if the child remained in the PICU for 12 days or more, as this duration has been used to define prolonged PICU stay. ¹⁷ Parents also became eligible on day 7 after PICU admission if it was clear to the medical team that their child would not be discharged before day 12. Transfer from the PICU to another ward followed by unplanned readmission to the PICU within 48 hours did not affect eligibility for recruitment or on-going data collection. If two parents of one child wished to participate, we analyzed data from the one who was present in the PICU most frequently, to maximize continuity and completeness of longitudinal data collection. We excluded parents who required English interpretation, expressed difficulty with reading inventories, or could not be reached after multiple attempts. Eligibility required the approval of the medical team to approach.

Once a parent of a long-stay patient was enrolled, the attendants (supervising PICU physicians), fellows (PICU physicians in training), and bedside nurses caring for that patient were likewise recruited. Each week the respective patient remained in the PICU, the team members in those roles were verbally consented to participate. Because of the nature of how PICU providers work (e.g., in shifts, a week at a time), individuals in each of these roles changed over the course of a patient's admission. Provider participants could care for more than one child in the study, resulting in them providing data on their moral distress for multiple patients.

Data and Measures

Demographic data were collected from parent participants on enrollment, including age, education level, religion, and their self-reported level of religiosity/spirituality on a 1 to 10 scale. To better understand their psychosocial risk factors, such as family problems and social support, the Psychosocial Assessment Tool 2.0 (PAT) ¹⁸ was administered to each family on enrollment. This 15-item validated parent/caregiver screening tool assesses demographic information, behavior changes in children, and distress of children and families. It has been used to assess baseline psychosocial risk in families with children in a variety of clinical settings, including pediatric acute care. ¹⁹ The PAT has a maximum score of 7, and stratifies households into “universal risk” (score < 1), “targeted risk” (score 1–2), and “treatment/intervention” (score >2) groups based on the Psychosocial Preventative Health Model. ²⁰ Results were reviewed within 48 hours. If a parent had responses on the inventory concerning for potential for self-injurious behavior, a designated psychiatrist was on call for immediate evaluation.

Upon enrollment and weekly thereafter until PICU discharge, clinical patient data from the prior 7 days were collected, including intubation status, use of extracorporeal membrane oxygenation (ECMO), or ventricular assist device (VAD) support, and the need for vasoactive medications by calculating the patient's highest vasoactive-inotropic score (VIS) over the previous 7 days. ²¹

Weekly until PICU discharge, parents were asked to complete the Hospital Anxiety and Depression Scale (HADS). ²² This 42-point screening tool (21 possible points for anxiety and 21 for depression) demonstrates good internal consistency, sensitivity and specificity, and concurrent validity. ²³ The HADS has been studied in a variety of medical contexts, including in parents of critically ill children. ^{12 24} Higher scores indicate higher levels of anxiety and depression, with scores 8 to 10 considered borderline, and scores ≥ 11 considered positive. These numerical results were also reviewed within 48 hours.

Concurrently on a weekly basis, the PICU team respondents of each child in the study was asked to complete the Moral Distress Thermometer (MDT). The MDT is a single-item scale that asks the respondent already familiar with the concept of moral distress to rank their current distress from 0 to 10 (0 = none, 2 = mild, 4 = uncomfortable, 6 = distressing, 8 = intense, and 10 = worst possible). This “thermometer” was designed to measure fluctuations in moral distress over time. The MDT shows good convergent validity and concurrent validity in clinicians who have left or considering leaving their positions. ²⁵ The definition of moral distress was provided to PICU clinicians, and then these clinicians were instructed to mark their distress in relation to the particular child and that child's situation. Despite multiple team members having the same role for study patients, only the clinician in each role caring for the patient that day was approached. If the PICU clinicians were too busy with clinical work to fill out the survey at the time of initial approach, their responses were accepted by the end of the week. The response rate was tracked. If the PICU clinician was involved in the care of more than one study patient, he or she completed an MDT for each study patient. Fig. 1 demonstrates the longitudinal study design.

Fig. 1.

Schema of longitudinal study protocol. Parents and ICU team members were recruited to study 7–12 days into a patient's admission. On a weekly basis, subjects were asked to provide responses for inventories as listed. Clinical variables, including intubation status, extracorporeal membrane oxygenation (ECMO) status, ventricular assist device (VAD) status, and vasoactive-inotropic score (VIS) were collected. HADS, Hospital Anxiety and Depression Scale; ICU, intensive care unit; PAT, Psychosocial Assessment Tool; MDT, Moral Distress Thermometer.

Statistical Analysis

Parents' and patients' demographics, as well as their PAT results and HADS scores, were presented using summary statistics, including means with standard deviations (SDs). We reported the number and proportion of HADS scores that were above the borderline and positive thresholds. ²² For clinician, all results were stratified by clinician group (attending, fellow, and nurse). MDT scores were presented as means with SD. We reported the number and proportion of MDTs that were ≥3 because this threshold has been found to be associated with nurses leaving a position due to moral distress. ²⁵ The relationships between anxiety, depression, and moral distress were examined using Spearman's rank-order correlations analysis.

To model the longitudinal trajectory of the HADS and MDT scores, linear mixed models with patient-level random coefficients were used. In these models, main predictors were quadratic terms of week. ²⁶ To accommodate for intraclass correlations of parents and clinicians (by role), patients were treated as random effects. The STATA “marginsplot” command was used to display the trajectories at weekly intervals with 95% confidence intervals (CIs). Graphic trajectories were limited to 15 weeks because few patients had data beyond that point. Additional models were fitted with clinical and baseline demographic variables to explore their relationship with our longitudinal outcomes. The prespecified independent variables that were chosen were: intubation status, ECMO/VAD status, VIS, mortality, PAT score, whether the patient had a previous ICU admission, and religiosity/spirituality score. Due to our small sample size, these independent variables were explored in separate regressions.

STATA version 16 (StataCorp LLC, College Station, Texas, United States) was used for all statistical analysis and graphs. A p -value of <0.05 was considered statistically significant.

Results

From April through December 2019, a total of 74 long-stay patients were identified. After exclusions and seven parents who were approached and declined participation, 37 (50%) families and children were enrolled and included in analysis ( Fig. 2 ). No parents withdrew early from the study. Parents and clinicians contributed data for 2 to 24 weeks. Partial weeks were not included. Table 1 shows characteristics of patients and parents in the study. Thirty-two (86%) parents were female. Parental age, race, and education background were diverse, although 31 (84%) parents identified as Christian. Of the 37 parents enrolled in the study, 23 (62%) PAT scores fell into the “Universal Risk” (score <1) category, 11 (30%) “Targeted Risk” (score 1–2), and 3 (8%) “Clinical Risk” (>2; mean = 1.0, SD = 0.6). One parent had a response concerning for suicidal ideation that resulted in immediate psychiatric consultation. Thirty (81%) families reported that their child had a previous PICU admission. The demographics of the 37 study patients are also found in Table 1 . The most common admission diagnosis categories were cardiac and respiratory. Two patients died during the study.

Fig. 2.

Flowchart demonstrating subject recruitment. PICU, pediatric intensive care unit.

Table 1. Characteristics of parents and patients.

Characteristic n (%)	n  = 37 (100%)
Female parent	32 (86)
Parent age (y)
20–29	8 (22)
30–39	13 (35)
40–49	10 (27)
50–59	5 (14)
60–69	1 (3)
Parent race/ethnicity
White/Caucasian	16 (43)
African American	4 (11)
Asian/Indian/Pacific Islander	3 (8)
Hispanic or Latino/a	10 (27)
Other/mixed	3 (8)
Unspecified/unknown	1 (3)
Parent highest level education
Some high school	2 (5)
High school degree	7 (19)
Some college	5 (13)
Associate's degree	7 (19)
Bachelor's degree	9 (24)
Graduate degree	7 (19)
Parent religion
Christian	31 (84)
Other	4 (11)
Unspecified/unknown	2 (5)
Religiosity/spirituality level (1–10 scale) Median(IQR)	7 (6–8)
Female patient	15 (41)
Patient age
0–6 months	6 (16)
7 months-–1 year	8 (22)
2–6 years	5 (14)
7–12 years	6 (16)
13–18 years	10 (27)
≥18 years	2 (5)
Acute primary physiologic derangement
Cardiac	16 (43)
Respiratory	10 (27)
Pulmonary hypertension	3 (8)
Neurologic	1 (3)
Infectious	1 (3)
Orthopaedic	1 (3)
Gastrointestinal	5 (14)
Previous ICU admission	30 (81)

Abbreviations: ICU, intensive care unit; IQR, interquartile range.

Parents' mean anxiety score was 9.2 (SD = 4.6), and mean depression score was 7.3 (SD = 3.7). Eighty (67%) anxiety scores were ≥8, and 41 (34%) were ≥11 (consistent with anxiety). Fifty-seven (48%) depression scores were ≥8 (borderline), and 21 (18%) were ≥11 (consistent with depression). Parent anxiety and depression scores were highly correlated ( r _s  = 0.72, p  < 0.001; Table 2 ).

Table 2. Correlation matrix of parent anxiety and depression and PICU team moral distress.

Characteristic n (%)	Parent anxiety	Parent depression	Attending moral distress	Fellow moral distress
Parent anxiety
Parent depression	0.72 c
Attending moral distress	0.2	0.26 a
Fellow moral distress	0.35 b	0.25	0.32 b
Nurse moral distress	0.29 a	0.36 b	0.03	0.13

Abbreviation: PICU, pediatric intensive care unit.

^ap  < 0.05.

^bp  < 0.01.

^cp  < 0.001.

Regression modeling demonstrated that trajectories of parental emotional reactions changed longitudinally ( Fig. 3 ), with both anxiety and depression decreasing throughout the admission. For the first 4 weeks of the PICU stay, projected parental anxiety remained above the “borderline” threshold, but subsequently decreased below it. Parental depression remained below the “borderline” threshold throughout.

Fig. 3.

Regression modeling of longitudinal Hospital Anxiety and Depression Scale scores of parents with children in the PICU Shading represents 95% confidence intervals. Short-dash line, threshold for “borderline” value; Long-dash line, threshold for “positive” value. PICU, pediatric intensive care unit.

Thirty-eight percent of MDT responses were from attendings (response rate = 86%), 33% from fellows (response rate = 74%), and 28% from nurses (response rate = 63%). Mean moral distress scores were 1.0 (SD = 1.4) for attendings, 3.2 (SD = 2.3) for fellows, and 2.4 (SD = 2.2) for nurses. Fourteen (11%) of attending MDT scores, 61 (54%) of fellow scores, and 35 (37%) of registered nurse (RN) scores were ≥3.

Parent anxiety and depression were positively correlated with attending, fellow, and nurse moral distress. Spearman's correlation revealed parent anxiety was significantly correlated with fellow and nurse moral distress ( r _s  = 0.35, p  = 0.007 and r _s  = 0.29, p  = 0.025, respectively). Parent depression was significantly correlated with attending moral distress and nurse moral distress ( r _s  = 0.26, p  = 0.047 and r _s  = 0.36, p  = 0.005, respectively). Additionally, attending and fellow moral distress significantly correlated ( r _s  = 0.32, p  = 0.012), but neither attending nor fellow moral distress significantly correlated with nursing moral distress ( Supplementary Table S1 ; available in the online version).

Trajectories of clinician moral distress visually varied by clinician group and some modestly changed over time ( Fig. 4 ). Compared with attendings and nurses, fellow moral distress was relatively higher initially (in the 3–4 range) but then downtrended below nurses' as admissions prolonged. Nurse moral distress fluctuated within a more narrow range. Attending moral distress consistently remained the lowest. Trajectories of moral distress and parent anxiety and depression are simultaneously represented in Supplementary Fig. S1 (available in the online version), with parent anxiety and depression and fellow moral distress decreasing throughout the ICU stay and nursing and attending moral distress remaining relatively stable.

Fig. 4.

Regression modeling of longitudinal PICU team moral distress in relation to specific long-stay patients by provider role. Shading represents 95% confidence intervals. PICU, pediatric intensive care unit.

The relationships between predictor variables and emotional outcomes, explored in separate regressions, are shown in Supplementary Table S1 (available in the online version). A higher parental PAT score was significantly associated with higher longitudinal parental anxiety (coefficient = 4.43 [95% CI: 2.76–6.10], p  < 0.001). The association between PAT score and depression did not meet the same level of significance (coefficient = 1.51 [95% CI: −0.02 to 3.04], p  = 0.054). The trajectories of anxiety and depression also varied by PAT scores; parents with higher than “universal” psychosocial risk reported more anxiety and depression for almost all of their child's prolonged stay, compared with parents with universal risk ( Fig. 5 ), although this association was statistically significant only for anxiety and not depression. Depression scores of parents with higher psychosocial risk remained in the “borderline” range. Similarly, higher PAT scores were significantly associated with fellow moral distress (coefficient = 1.32 [95% CI: 0.38–2.24], p  = 0.006). A child being intubated was associated with significantly higher parent anxiety levels (coefficient = 1.23 [95% CI: 0.00–2.48], p  = 0.050), but not with parent depression or clinician moral distress. If the patient died, attendings had significantly higher moral distress (coefficient = 0.83 [95% CI: 0.08–1.58], p  = 0.030) but this was not the case for other clinicians. ECMO or VAD use, VIS, previous ICU admission, acute disease states, or religiosity/spirituality score were not associated with parent anxiety and depression or clinician moral distress.

Fig. 5.

( A ) Regression modeling of longitudinal anxiety scores of parents with children in the PICU, categorized by the Psychosocial Assessment Tool (PAT) score. Orange line represents parents who scored “targeted” or “clinical” risk on the PAT during study enrollment; red line represents parents who scored “universal” risk. Shading represents 95% confidence intervals. Short-dash line, threshold for “borderline” value; long-dash line, threshold for “positive” value. ( B ) Regression modeling of longitudinal depression scores of parents with children in the PICU, categorized by PAT score. Green line represents parents who scored “targeted” or “clinical” risk on the PAT during study enrollment; blue line represents parents who scored “universal” risk. Shading represents 95% confidence intervals. Short-dash line, threshold for “borderline” value; long-dash line, threshold for “positive” value. PICU, pediatric intensive care unit; RN, bedside registered nurse.

Discussion

A consequence of some children having increasingly complex medical needs has been some require prolonged PICU admissions. While many parents of children with critical illness experience anxiety and depression, ² parents of children with prolonged critical illness may be at risk for greater levels of distress, ⁶ as prolonged admissions have higher mortality, morbidity, and complications, ^{17 27 28 29} as well as presumably greater disruptions to families' routines and finances. ³⁰ Additionally, studies show PICU clinicians commonly experience moral distress when caring for children in challenging circumstances, ^{10 31 32} and these emotions may be exacerbated when the child involved has a prolonged critical illness. ^{9 11}

Going beyond previous cross-sectional studies, we longitudinally tracked anxiety and depression in parents of children admitted for prolonged PICU stays (in our cohort, 3–26 weeks) and the moral distress of their PICU clinicians—for the first time—as it relates to specific children and their circumstances. Additionally, we are perhaps the first to explore the relationship of these emotional outcomes to each other and the factors that may affect their trajectories. We found that mean parental anxiety was above the “borderline” threshold. While regression modeling showed parental anxiety and depression decreased over the course of their child's prolonged PICU admission, their anxiety was notably higher during the first several weeks. Considering most patients survive, the majority of parents may grow accustomed to their child being ill in the PICU and adopt an expectation that they will eventually be discharged. We found that parents' baseline psychosocial risk was significantly associated with increased anxiety and depression over the course of their child's PICU stay. In parents with higher psychosocial risk, depression increased throughout the PICU admission. PAT scores of our cohort were seemingly lower than another study of parents of children with serious chronic illnesses, which showed their scores were among the highest of all previously studied pediatric populations, with 55% showing “Targeted” or “Clinical” risk. ³³ Parental anxiety, but not depression, was also significantly higher when their child was intubated. Notably, parent emotional stress may be a sign that provider moral distress is also present (and vice versa), given that they correlate and all parties may recognize that the limitations of modern medicine to cure and of acute-illness-focused PICU care to address chronic needs.

From an exploratory perspective, the degree and trajectory of fellow and nurse moral distress, as it related to specific children, seemed to differ from attendings, which was consistently lower. Fellows reported relatively greater moral distress early in patients' admissions but then decreased over time. These differences may be related to years of experience or amount or degree of direct interaction with patients, as nurses and fellows generally provide more bedside care than attendings. Attending moral distress, which was usually low, increased when patients died, perhaps in the setting of more direct patient interaction. A patient's death was not associated with changes in nurses' or fellows' reported moral distress. Finally, parental distress and PICU team moral distress were significantly associated, though they had only a weak to moderate correlation.

These findings can inform interventions that seek to alleviate emotional distress in parental and professional caregivers. For example, efforts to address anxiety and depression in parents of long-stay PICU patients may be more efficacious early in their child's admission and in those who have higher baseline psychosocial risks. Similarly, efforts to address moral distress might be more advantageous in PICU nurses and trainees, as opposed to attendings, which other studies also suggest. ^{10 34} Palliative care may be an already-existing option to support these caregivers (notably, pediatric palliative care subspecialists were not available in our institution at the time of this study). Palliative care seeks to address physical and psychological symptoms and burdens in patients and families, and early palliative care efforts have been shown to reduce parental stress in related circumstances. ³⁵ Similarly, it can help anticipate and address moral and psychological distress in PICU clinicians, through collaborative forums to share, validate, and normalize clinicians' experiences and emotions and mindfulness/meditation training, ^{36 37} even if the patients themselves are not formally receiving palliative care services. Ideally, all PICU providers should be adept in and practice the core competencies of palliative care (i.e., practice “primary” palliative care). ³⁸ Numerous online resources and tools exist to enhance primary ICU palliative care knowledge and skills. ³⁷ Other ancillary services, including social work, therapists (music/art) and chaplaincy may be helpful resources for both families and staff at times of heightened emotions and distress. Such efforts may also be helpful in mitigating familial post–intensive care syndrome ³⁹ (i.e., high levels of anxiety, depression, posttraumatic stress disorders, and/or complicated grief after a loved one's ICU stay) and professional burnout. ⁴⁰

Limitations

There are several limitations to this study. First, because the study was conducted in a single institution and focused on English-speaking/reading participants, its generalizability may be limited. Second, the children represented here heterogeneous in their chronic medical conditions, which may have impacted emotional outcomes in unmeasured ways. Third, our results are at risk for several forms of bias, including self-reporting bias (parents or clinician giving responses they thought the investigators wanted) and attrition bias (patients discharged earlier were likely different from and contributed less data than those discharged later). Similarly, we did not account for the nonindependence of observations, as PICU clinicians could provide moral distress rating for more than one child. Fourth, data were only collected weekly which may have been insufficient to accurately represent participants' distress over the previous week. We only collected data during the day, so the nurses who worked at night were not represented. Fifth, clinicians were stratified by role but were otherwise treated as interchangeable, and thus, we could not account for intraclinician correlations. Similarly, we did not collect clinicians' demographics, so we do know how those factors impact moral distress. Finally, we did not collect any data on excluded families or clinicians in other roles.

Conclusion

Optimal care of critically ill children requires all their caregivers to be emotionally capable of supporting the child throughout the PICU stay, even if days become weeks or months. However, the challenges of a child's prolonged critical illness can compromise their emotional well-being. Hence, PICU clinicians should be mindful that parents' emotional distress is dynamic and is impacted by what is going on inside and outside the PICU, and that provider distress is variably experienced depending on their role. Additional studies are needed to corroborate and expand upon our findings, so that interventions can effectively help vulnerable families and clinicians.