Mobilization Safety of Critically Ill Children

We conducted a secondary analysis of the Prevalence of Acute Rehabilitation for Kids in the PICU (PARK-PICU) study, a cross-sectional, 2-day study that included 82 PICUs in the United States (n = 1769 patients). Detailed methods of PARK-PICU are available in the primary manuscript.¹⁴  Data collection was standardized across all study sites. Institutional review board approval was obtained at all sites with waiver of informed consent.

For the PARK-PICU study, all patients were included if they had an LOS ≥72 hours as of 7 am on each of the 2 study days (November 9, 2017, and February 12, 2018). Clinical data for all eligible patients were collected based on their clinical status at 9 am on the study days either via electronic health record or at the patient’s bedside by a site study team member. Data included admission reason, BMI, mechanical ventilation status, sedative infusions and sedation level, delirium status, and use of invasive medical devices. Preadmission functional status was characterized using the pediatric cerebral performance category score, which was obtained from the medical record or in discussion with family and the PICU team.

Mobility and potential safety event data were documented in real time by interprofessional staff involved in mobilization (nurses, physical therapists [PTs], occupational therapists [OTs], respiratory therapists [RTs], or child life) and research team staff on standardized data collection forms, which were delivered to the bedside of all eligible patients by 9 am on both 24-hour study periods. The occurrence of any mobility event was captured, including details of PT-, OT-, nurse-, family-, or other staff-provided mobility and the time of day performed. Mobilization was categorized by the type of in-bed activity (eg, passive range of motion, sitting in bed, or exercises in bed) and out-of-bed activity (eg, standing, mat play, transfers, marching in place, or walking). Twenty-one categories of potential safety events were predefined, including physiologic abnormalities (Table 1), on the basis of published practice recommendations for PICU early mobilization.¹⁸ 

A mobility event was defined as any discrete physical activity provided by a provider or family member. Potential safety event rates in the univariate analysis were defined as the number of potential safety events divided by the total number of mobility events with 95% confidence intervals (CIs). Covariates in our regression model were identified a priori on the basis of clinical relevance and previous literature.¹⁹  A final multilevel, multivariable logistic regression model, with random effects for patients nested within the PICU site, was used to calculate odds ratios (ORs), with 95% CIs, for the odds of a mobilization event having a potential safety event. The covariates included age category (0–2, 3–6, 7–12, 13–18, or >18 years); sex; ethnicity (White, Black, Asian American, Hispanic, or other); PICU LOS as of the study day; unit mobility protocol (yes or no); pediatric overall performance category scale before admission (pediatric cerebral performance category score 1–5); nurse-to-patient ratio (1:1 or 2:1/1:2 or 1:3); reason for admission (medical or surgical); type of respiratory support; delirium status (present, absent, or not screened); highest level of mobility (in-bed, out-of-bed, or walking), and the presence of absence of the following (yes or no): sedative use; vasoactive infusion except for milrinone; central catheter; arterial catheter; indwelling urinary catheter chest tube; nurse participation; family member participation; PT participation; OT participation; or RT participation. A two-tailed P value <.05 was considered statistically significant. Additional details for data analysis are provided in the Supplemental Information. Stata 15 (StataCorp LLC, College Station, TX) was used for all statistical analyses.

Potential safety event data were compiled from 1433 patient-days that included at least 1 mobility event. On these 1433 patient-days, 4658 mobility events occurred, for a median of 2 (interquartile range, 1–4) events per patient-day. There were no differences in frequency of mobility events per patient-day when a potential safety event occurred. Overall, 194 of 4658 mobility events were associated with a potential safety event (4%; 95% CI, 3.6%–4.7%).

The classification of all potential safety events is displayed in Table 1. Among all 194 events, 71 (37%) were classified in >1 category. Most of the potential safety events consisted exclusively of physiologic changes, representing 70% of all potential safety events (135 of 194) and 3% of all mobility events (136 of 4658). Dislodgement of medical equipment, excluding peripheral intravenous catheters and feeding tubes, occurred rarely (15 of 4658; 0.3%). Dislodgment of an endotracheal tube was reported in 2 of 1249 (0.16%) of mobility events during passive range of motion and proning. One tracheostomy out of 864 (0.12%) was dislodged during a patient’s attempt to stand with an OT. Chest tubes and arterial lines were displaced in 3 of 404 (0.7%) and 2 of 970 (0.2%) mobility events with these devices, respectively.

Twenty-eight potential safety events were classified as “other” and included physiologic changes (32%, 9 of 28) that did not meet predefined criteria; provider-perceived risk of endotracheal tube or central venous line dislodgement (14%, 4 of 28); coughing and emesis in a nonintubated patient (14%, 4 of 28); and neurologic or behavior changes (18%, 5 of 28), including increased intracranial pressure, increased activity on EEG, and pain. One patient with known hip dysplasia had a hip dislocation. One patient with a known arrhythmia had increased episodes of arrhythmia during 2 mobility events.

The rate of potential safety events did not differ based on patient age or sex. Compared with rates in White patients (4.4%, 101 of 2311; 95% CI, 3.6–5.3), potential safety events occurred at a lower rate in Hispanic patients (2.2%, 19 of 851; 95% CI, 1.4–3.5) and at a higher rate in patients with ethnicity listed as “other” (8.1%, 28 of 347; 95% CI, 5.6–11.4). Potential safety event rates showed no differences based on BMI, preadmission function, ambulatory status, or LOS (Table 2 ;Supplemental Table 4).

We found no differences in the potential safety event rate based on unit type. Large PICUs, defined as having 20 beds or more, had fewer potential safety events (3.7%, 125 of 3392; 95% CI, 3.1–4.3) than did smaller PICUs (5.5%, 69 of 1266; 95% CI, 4.3–6.8). Units that had a mobility protocol had a higher potential safety event rate (6.4%, 81 of 1259; 95% CI, 5.2–7.9) than did units without one (3.3%, 113 of 3399; 95% CI, 2.8–4). Potential safety event rates were lower when the nurse-to-patient staffing ratio was 1:2 or 1:3 (3.1%, 87 of 2809; 95% CI, 2.5–3.8) than when it was 1:1 or 2:1 (5.7%, 105 of 1832; 95% CI, 4.7–6.9; Supplemental Table 5).

Patients who received no respiratory support had a low potential safety event rate of 2.1% (21 of 999; 95% CI, 1.3–3.2). Rates were higher in patients who were mechanically ventilated via an endotracheal tube (7.3%, 91 of 1249; 95% CI, 6–8.9) and in those receiving noninvasive mechanical ventilation (5.5%, 25 of 456; 95% CI, 3.7–8.0). The rate of potential safety events increased when patients were receiving any sedative or analgesic infusion (5.9%, 111 of 1867; 95% CI, 4.9–7.1). The potential safety event rates of patients who screened positive for delirium (7.8%, 32 of 412; 95% CI, 5.6–10.9) and for those who were not screened (4.7%, 138 of 2965; 95% CI, 3.9–5.5) were higher than those for patients who screened negative (1.9%, 24 of 1280; 95% CI, 1.3–2.8).

Potential safety events occurred at higher rates during mobilizations that included vasoactive infusion (5.5%, 28 of 512; 95% CI, 3.8–7.8), an arterial line (6.4%, 62 of 970; 95%, CI 5–8.1), central line (5%, 153 of 3041; 95% CI, 4.3–5.9), urinary catheter (7.4%, 38 of 514; 95% CI, 5.4–10), or chest tube (6.9%, 28 of 404; 95% CI, 4.8–9.9). Extracorporeal membrane oxygenation cannulas, hemodialysis catheters, and intracranial pressure monitors were not associated with higher rates of potential safety events (Table 3).

A median of 1 staff member (interquartile range, 1–2) was involved in all mobility events, regardless of whether a potential safety event occurred. The potential safety event rate did not change in mobilizations that included a nurse, PT, OT, or family member. Potential safety events occurred at higher rates when an RT participated in the mobilization (12.8%, 19 of 148; 95% CI, 8.3–19.3; Fig 1).

The rate of potential safety events increased when the highest level of mobility achieved during a mobilization was passive movement or range-of-motion activities (5.8%, 74 of 1282; 95% CI, 4.6–7.2) or sitting or exercising in bed (7.5%, 28 of 375; 95% CI, 5.2–10.6). However, the rate of potential safety events was lower when the highest level of mobility achieved during a mobilization was being held by a parent or nurse (3.3%, 46 of 1409; 95% CI, 2.4–4.3; Fig 1). Compared with in-bed mobilizations (6.1%, 108 of 1768; 95% CI, 5.1–7.3), out-of-bed mobilizations were less likely to include a potential safety event (2.8%, 70 of 2454; 95% CI, 2.3–3.6).

After adjusting for patient and site, the following variables had the strongest positive association with potential safety events (Fig 2; Supplemental Table 6): positive delirium screen (adjusted OR [aOR], 5.86; 95% CI, 2.17–15.86) or no delirium screen (versus negative screen, aOR, 3.98; 95% CI, 1.82–8.72), RT presence during mobilization (aOR, 4.11; 95% CI, 1.85–9.11), mobility protocol (aOR, 2.71; 95% CI, 1.52–4.81), and family presence (aOR, 2.08; 95% CI, 1.28–3.39). Variables that had the strongest negative association with a potential safety event included Hispanic ethnicity (aOR, 0.41; 95% CI, 0.18–0.94), large PICU size (aOR, 0.51; 95% CI, 0.29–0.95), and out-of-bed mobilization as the highest level of mobility (versus in-bed, aOR 0.41; 95% CI, 0.24–2.72).

Our analysis of real-time data from 4658 mobility events across 82 US PICUs (one-third of all PICU beds)²⁰  revealed that PICU mobilization is safe, with an overall potential safety event rate of 4%. Most potential safety events were transient physiologic changes, and dislodgement of medical equipment was rare, with no cardiac arrests or falls. Younger children were not at higher risk of potential safety events. Importantly, no PICU intervention including respiratory support or vasoactive infusions were associated with increased potential safety events in our multivariable analysis. However, children who screened positive for delirium or who were not screened were at increased risk for potential safety events. These findings provide important insights that can be used to guide patient selection for PICU rehabilitation interventions, as well as considerations for staff education around early mobility.

A strength of our study is the utilization of a standardized definition of potential safety events during PICU mobilizations. Our multicenter finding of a 4% event rate is consistent with previous PICU reports of potential safety events during mobilizations, which ranged from 1% to 6%.^14–17  Importantly, adverse events are common in the PICU during standard care, even without mobilization activities.^21–23  Similar to the adult literature on ICU mobilizations, we found that most potential safety events were transient physiologic changes.^12,13  When mobilizing critically ill children, providers are most concerned about dislodgement of endotracheal tubes and central lines,¹⁰  but these events were rare. The rate of endotracheal tube dislodgement in our study was 1.6 per 1000 mobility events. The rate of unplanned extubation is low during both mobility events and routine PICU care.^24,25  No central lines were dislodged during this study, a finding comparable to that in the literature^16,17  and a rate less than that associated with routine PICU care.²⁶  Importantly, no cardiorespiratory arrests occurred in the 4658 mobilizations. Overall, the low frequency of potential safety events is reassuring, though it may indicate that exercise intensity during mobility events is often inadequate. No data exist on the appropriate dose, duration, or frequency of PICU rehabilitation.

Despite PICU providers’ concerns and perceived barriers to mobilizing infants,¹⁰  we found that younger age was not associated with an increase in potential safety events. These findings are consistent with the neonatal ICU literature for preterm and low-birth weight infants,¹⁸  and are reassuring given that two-thirds of patients admitted to the PICU for >72 hours are <3 years old. Yet, PICU patients under 3 years of age are mobilized less frequently than are older children.^14,17  Thus, our data support the premise that mobilizing young infants is safe and that additional education should be aimed at improving provider comfort in this age group.

Our finding that the 10% of children who screened positive for delirium had twice the potential safety event rate is very important. The majority of these potential safety events were transient physiologic changes (69%), which consisted of predominantly of respiratory rate and heart rate changes. Furthermore, the fact that 64% of patients who were not screened for delirium also had an elevated potential safety event rate suggests that a high proportion of patients had potentially unrecognized delirium. These findings are consistent with previous literature showing that 25% of critically ill children experience delirium²⁷  and that PICU patients are not sufficiently screened for delirium.²⁸  Implementation of an early mobility program has been shown to decrease rates of delirium in the PICU.⁵  Our findings emphasize the need to consistently screen patients for delirium and carefully consider mobilization of delirious patients. If delirium recognition, prevention, and mitigation efforts are part of unit practice, mobilization safety will be optimized and nondelirious patients can undergo rehabilitation to prevent subsequent delirium.

We found that no PICU intervention was associated with increased rates of potential safety events in our multivariable analysis, including respiratory support. These findings differ from reports in the adult literature that patients with endotracheal tubes experience an increased adverse event rate during ICU rehabilitation.¹²  Our findings also contrast a large, single-center study where mechanically ventilated children receiving routine care had a high adverse event rate of 3.6 per 100 ventilator days.²⁹  Therefore, our findings should reassure providers that mobilizing mechanically ventilated children is safe. Additionally, our results add to the existing literature on appropriate patient selection for PICU mobilization¹⁸  and can potentially expand the clinical characteristics deemed acceptable for team-based PICU rehabilitation.

Nurses are the cornerstone of mobility programs in the PICU. In the PARK-PICU study, nurses facilitated most mobilizations (67%),¹⁴  and nurse participation in mobilizations was associated with a low potential safety event rate (4.4%). In critically ill adults, nursing-led mobility initiatives have proven to be safe, feasible, and effective.^30–32  Surprisingly, we found that the presence of a mobility protocol within a PICU increased the rate of potential safety events. This finding may indicate that PICUs with mobility programs provide mobility at increased duration or intensity. Duration of mobilization events was not captured in the PARK-PICU study¹⁴  and no data exist on the appropriate duration or intensity of PICU mobilization. Additionally, all pediatric early mobility protocols include guidance on identifying potential safety events.^6,8,33–35  It is conceivable that providers at institutions with mobility programs had increased sensitivity for identifying transient physiologic changes as potential safety events. Furthermore, we found that large PICUs with >20 beds had lower potential safety event rates, suggesting that higher exposure to critically ill pediatric patients may increase interprofessional staff expertise at mobilizing them.

Finally, we found that out-of-bed mobilizations had a lower potential safety event rate than in-bed mobilizations, consistent with the adult ICU literature.^13,36  This finding likely reflects the fact that in-bed mobilizations are the first mobility activity in a progression. Therefore, if a potential safety event occurs during an in-bed mobilization, a child will likely not advance to an out-of-bed mobilization during the session.¹³  Potential safety events occurred at an increased frequency when an RT was involved in the mobilization. Of the 19 potential safety events that occurred with an RT, 68% occurred during an in-bed mobilization. RTs are more likely to be involved in mobilizing higher-risk patients during their first several mobilizations in the PICU, which may account for this association. Of the mobilizations that RTs participated in, nearly 80% were in mechanically ventilated children. It is possible that the rate of potential safety events would have been higher in those patients had the RT not been involved. In our multivariable analysis, family member involvement in a mobilization increased the odds of a potential safety event. Of the 63 potential safety events that occurred when a family member participated, 78% occurred during out-of-bed mobilizations. It is conceivable that patients require more staff support than anticipated for initial out-of-bed mobilizations.

Our study has multiple strengths and limitations. First, although the parent PARK-PICU study included a large proportion of US PICUs with robust data, the design does not establish causality.¹⁴  Second, although potential safety events were predefined, unrecognized potential safety events or incomplete reporting might have led to an underestimate of event rates. Third, knowledge of the study by the clinical staff may have increased documentation of potential safety events and thereby overestimated potential safety event rates. However, potential safety events were prospectively recorded on both days; thus, we believe reasonable means were used to ensure that most potential safety events were appropriately captured. Fourth, data were not collected on the clinical significance of physiologic changes or, specifically, whether a medical intervention was required in response to the potential safety event. Fifth, illness severity was not captured utilizing a standardized measure, such as the Pediatric Risk of Mortality III score; therefore, we were not able to control for illness severity in our multivariable regression. Proxies of illness severity, such as RT presence and nurse-patient ratio may be confounded, though PICU interventions were accounted for in the model. Sixth, data captured regarding patient ethnicity were gathered from the electronic health record, which is frequently inaccurate³⁷  and only included limited ethnic groups, making conclusions regarding ethnicity difficult to interpret.

In US PICUs, mobilization has a strong safety profile and rarely leads to dislodgement of medical equipment; however, delirium is associated with increased potential safety events. No PICU interventions were associated with increased safety events. Interprofessional PICU teams, including physicians, nurses, family members, RTs, PTs, and OTs, should incorporate these encouraging findings into the selection of appropriate mobilizations and mobility protocol development for critically ill children. The reassuring safety profile of PICU rehabilitation overwhelmingly highlights the need to systematically improve provider education and confidence in mobilizing critically ill children.

We thank Ms Claire Levine, MS, ELS, scientific editor, Department of Anesthesiology and Critical Care Medicine, Johns Hopkins University, for her editorial assistance with this manuscript. We also thank the contributions of collaborators from the PARK-PICU Investigators, whose names are provided in full in the Supplemental Information.

aOR

adjusted odds ratio

CI

confidence interval

LOS

length of stay

OR

odds ratio

OT

occupational therapist

PARK-PICU

Prevalence of Acute Rehabilitation for Kids in the PICU

PT

physical therapist

RT

respiratory therapist