Phase 1c: observational study

Introduction

This chapter describes the observational study to ascertain current ERM practices, as well as barriers and facilitators to ERM delivery, within the PICU setting. Following the scoping review and survey (see Chapters 1 and 2), we were interested in the concept of early ERM occurring by day 3 and over the following 7 days of PICU admission, and ERM to include the broad category of any rehabilitation or mobilisation, including both mobility and non-mobility activities. We directly observed current ERM practices within UK PICUs, identify patients who do and do not receive ERM and describe variation between PICUs and factors associated with ERM practices.

Study management

The work package was led by BRS. The study management group provided input into protocol and ethics design. JYT was study co-ordinator and piloted and developed the Research Electronic Data Capture (REDCap) database. Statistical analysis was undertaken by JYT, BRS and James Martin. Data interpretation was by BRS, JYT, JMan and JMen with input from all study management group.

Objectives

• Observe and describe current ERM practice, including barriers and facilitators, in UK PICUs.
• Assess the capability of UK PICUs to deliver ERM.
• Establish and model how many/which CYP may be suitable for ERM in the PICU population using routinely collected data.

Method

Study design

A multicentre prospective observational study.

Target population/setting

All CYP (0 to <16 years) admitted to PICU and remaining on PICU by 9 a.m. on day 3 after PICU admission were eligible to participate. The exclusion criteria included a local decision by PI or treating clinical team not to include patients (e.g. receiving end-of-life care) and parents or guardians who choose to opt out. The broad study inclusion criteria allowed for the observation of all types of patients admitted for PICU care (e.g. planned and unplanned admissions), and all age ranges.

Site and patient selection

The Paediatric Early Rehabilitation and Mobilisation during InTensive care (PERMIT) study was an observational study conducted in 15 UK PICUs across two 21-day periods: (1) 26 November–16 December 2019 and (2) 14 January–3 February 2020. PICUs were identified from the PERMIT survey (see Chapter 2). The PICUs selected were of varying sizes (n = 6 large: >800 admissions/year, n = 5 medium: 500–800 admissions/year, n = 3 small: <500 admissions/year) and reported ERM activity of differing levels in the survey.

This study was conducted in two separate time periods to maximise efficiency, overcome recruitment hurdles and meet the target. Ten sites recruited and collected data during period 1, with a further five sites in period 2. Patients were observed on study day 1–14 with a further week to complete follow-up (study day 15–21). Individual patient data collection and observations took place for up to 7 days after patients were recruited or until PICU discharge, whichever was sooner.

Recruitment/enrolment

The study protocol, manual of operation, checklists and case report forms (CRFs) provided details on the study procedures. Staff at participating sites received remote training on research conduct before, and ongoing support sessions during, the study period.

Research staff screened patients admitted to PICU for the PERMIT study using a bespoke study screening log. The daily screening process ensured patients becoming eligible (e.g. the day before their third day) were identified. Designated research co-ordinators entered data on ERM activities recorded by clinical staff in clinical notes on the study proformas. Data were transferred to a secure electronic database (REDCap™), with scanned copies uploaded for data validation. Data were pseudo-anonymised at the local site before secure transfer to the PERMIT trials office.

Consent

As the study was observational, Regional Ethics Committee (REC) approved data collection without seeking prior consent from parents/legal representatives. In addition, this avoided unnecessary burden for parents/legal guardians in approaching consent during a very sensitive time. Information about the study was provided to all eligible patients’ families and was displayed within public areas of participating PICUs. This explained the study to parents, family, friends and children who were able to make autonomous decisions. Parents/legal guardians were able to opt the child’s data out of the study at any time and were aware that the future care their child would receive would not be affected.

Study procedure and data collection

Site staff collected demographic data on the third day of admission. Clinical and ERM data were collected from enrolment until discharge at the end of the study period. Data were collected twice, between 09.00 and 10.00 and between 14.00 and 15.00, each day.

Patient-level data

Patient characteristics collected at PICU admission included age (in categories); reason for admission; primary diagnosis; the severity of illness using Pediatric Index of Mortality (PIM3) score, with clinical function being assessed at baseline (pre-PICU state) and admission via the PCPC and POPC, which scores from 1 to 6 (1: normal, 2: mild disability, 3: moderate disability, 4: severe disability, 5: vegetative state or coma and 6: death).63 PICU LOS was also recorded.

Clinical data

Data collected during PICU stay included healthcare interventions; requirement for MV; sedation and level of consciousness; presence of delirium; critical care interventions; indicators of physiological status; and individual patient PICU resource use. We calculated PELOD score (PELOD-2),64 which is a measure to describe the severity of organ dysfunction/illness in critically ill CYP, daily at 9 a.m.

Observed early rehabilitation and mobilisation active interaction

Clinical staff performing ERM activities were instructed to record the planned and delivered ERM activity duration in medical records. A research nurse or co-ordinator used these data to complete the bespoke active interaction CRF, which was submitted to the PERMIT study office. CRFs were completed hourly between 9 a.m. and 5 p.m. by the local site research nurse. A retrospective review of clinical case records of ERM activities that occurred overnight was carried out, with CRFs completed accordingly. Overnight ERM interventions were defined as the time from the end of the observed active interaction period 17.01 until 08.59 before the start of the next period.

We defined a priori ERM as (1) any ERM activity, (2) any mobility ERM activity and (3) mobility activity out of bed, informed by scoping review and survey (see full breakdown of ERM activities, ERM group and level of ERM detailed in Appendix 3, Table 40). We excluded chest PT, tracheal tube suctioning and routine nursing ‘cares’ (such as mouth and eye cleaning).

Patients were defined as receiving any ERM intervention if any ERM was recorded on study day. Details on interventions such as type of ERM activity (mobility, non-mobility, out-of-bed, passive, active and psychological) and safety events (such as changes in heart rate, oxygen saturation, removal of tubes or falls) were recorded.

Primary outcome

• the delivery of any ERM activity on day 3 post admission (study day 1).

Secondary outcomes

• the delivery of any ERM activity on days 4–10 post admission (study day 1–7);
• the delivery of ERM involving any mobility activity and out-of-bed mobility ERM on days 3–10;
• the number, type and duration (e.g. dose) of ERM delivered on each day;
• predictive factors related to the delivery of ERM on day 3 post admission.

Data analysis

We reviewed data for errors: missing data, duplicated records and outliers. Extreme values were set to missing if they were deemed impossible, based on their validity range. Continuous variables were reported as mean and SD or median and IQR based on data distribution. Categorical variables were described in numbers and/or percentages.

The prevalence and scope of ERM were described as the proportion of patients provided with any ‘active interaction’ of any ERM on day 3 post admission. Proportions of eligible patients receiving ERM interventions were analysed for each day. Rate ratios of patients receiving an ERM intervention during the study period were analysed using a Poisson regression model.

Cumulative prevalence for each day in PICU after day 3 up to day 10 post admission was calculated. Cumulative proportion of patients receiving ERM interventions as per prespecified categories during the study period was described graphically using Kaplan–Meier estimation and event rate plots.

We undertook further analysis to understand potential predictive factors associated with ERM and the incidence of ERM. We performed multivariable logistic regression to evaluate predictive factors of ERM provided on day 3. Factors of interest were established following the PERMIT survey and expert group consensus. These included age; baseline PCPC score; unplanned versus emergency admission; ventilation status; requirement of vasoactive infusion, sedative infusion, or neuromuscular blocking drugs; presence of urinary catheter, CVC or arterial line; family member present or participating in ERM activity; and presence of PICU protocol.

We calculated level of mobility activity using a modified progression score previously described in the EU-PACK (European Prevalence of Acute Rehabilitation for Kids in the PICU) study42 – Level 1: passive ROM, 2: sitting and exercise in bed, 3: sitting edge of bed, 4: held by parent or nurse (cuddle), 5: transfer to chair, 6: mat play, 7: standing, 8: walking in room/PICU, 9: walking out of PICU (see Appendix 3, Table 40). Further post hoc categorisation of ERM activities as enrichment, passive and active activities was also applied.

Adverse events rates were calculated per ERM activity. For zero rate observed AEs, the upper 95% CI are presented using Hanley’s formula.65

Patient and public involvement and engagement

For an overview of the approach to patient and public involvement and engagement (PPIE) adopted throughout the study, please see Chapter 8. With this element of the study, we explored the potential consent model, especially the acceptability of an ‘opt out’ approach to consent. There was a study recruiting at the time on PICU with an ‘opt out’ approach – the Sedation AND Weaning In Children trial (SANDWICH) trial.66 At the time, the SANDWICH trial had recruited over 700 patients at Birmingham Children’s Hospital (BCH), with only three families choosing to opt out of their child’s data being collected. It was therefore regarded as a highly successful approach to consenting.

For the SANDWICH trial, parents were given a leaflet outlining that data collection was taking place on data routinely collected as part of ‘normal care’ which goes to the national PICU audit.67 They were told that the SANDWICH team would have access to some of this information. In addition, the research team would also collect information from the child’s medical notes and charts. All the data were anonymised and there were no interventions (at the individual level), just data collection. Parents were not asked for their informed consent. They received the Participant Information Sheet (PIS). If they did not want to take part, they then spoke to the clinical staff, who informed the research staff, and this was documented as an ‘opt out’.

We spoke to families who had received the information to ask how they had experienced the process of approach for the SANDWICH study. In addition, we approached families who were participating in other research studies known to the PPIE lead (JMen) and whose children had recently been discharged from hospital. We spoke to six parents of four children (aged 0.3–6 years) who had experienced one or more PICU admission(s) and had experience of their child being recruited to research. We also spoke to three young people (aged 17–20) naive to PICU to participate as PPIE participants. Different models of consent were discussed and where there was no intervention then an opt-out model of consent was universally popular (Table 3). We also discussed the PISs and poster to inform parents about the observational study with them and they suggested a number of changes to the language, graphics and layout of them.

Results

Eligibility and enrolment

A total of 169 patients were enrolled into the study from 15 PICUs, with each PICU enrolling a median (IQR) of 10.0 (9.5–15.3) patients.

During the 14-day enrolment period, the median census on each PICUs was 10.5 patients (IQR 7.0–17.0; range 3–26). Overall, there was a median (IQR) of 1 patient (0–2) eligible per PICU per study day of whom 1 patient (0–1) was enrolled into the study per PICU per study day. This identified 203/2447 (8.7%) patients within each PICU eligible, of whom 158/203 (77.8%) were enrolled over the 14-day enrolment period (enrolment data were missing from 1 unit which recruited 11 patients). Ineligible patients had either not reached day 3 or had already reached day 4 or greater on day of screening. Table 4 shows PICU patient census, eligibility and enrolment proportion. During days 8–14, some sites reported not enrolling as they had already reached their target of 10 patients. Enrolment rate in study days 1–7 was 98/108 [90.7% (95% CI 83.6% to 95.4%)].

Demographics

Of the 169 patients, 59.2% were male; median age was 4.5 months (IQR 1.1–37.9). The majority (81%) were <4 years with 62.7% <1 year. Only 48 (28.4%) were ambulatory prior to PICU admission (key demographics at admission are in Table 5).

The most common admission diagnosis was bronchiolitis in 55 (32.5%). Eighty-four (49.7%) were admitted from another hospital, requiring retrieval into PICU. There were 127 (75.1%) emergency admissions, 150 (89.3%) required invasive ventilation at admission, 5 (3%) extracorporeal membrane oxygenation (ECMO) and 60 (35.5%) required cubicle isolation. Prior to admission, 66% had a PCPC score of 1 or 2 and 50% a POPC score 1 or 2, indicating a high proportion with moderate to severe disability pre-PICU. Admission predicted probability of mortality, as measured by PIM3, was median 1.2% (0.5–4.4%) and 68 (40.2%) children were enrolled in a PICU with an existing ERM protocol (as reported in the PERMIT survey – Chapter 2).

Patient clinical status on day 3 of admission (study day 1)

Between PICU admission and study day 1, 12 (7.1%) had surgery, 1 had a cardiac arrest. Most (119; 70%) remained ventilated by an ETT, 40 (23.7%) required vasoactive infusions and 3 remained on ECMO (Table 6).

Sedative medications were used in 108 (63.9%); 102 on opiates, 40 on benzodiazepines and 26 (15.3%) were on neuromuscular blocking drugs. In patients who could be assessed and on sedative drugs, comfort B sedation score was median (IQR) 12 (11–14), and 12 (12–15) if not receiving sedation. No patient was screened for delirium in any PICU. PELOD2 severity of illness median score was 4 (IQR 2–6). Central venous access was used in 103 (60.9%), arterial access in 72 (42.6%), and 103 (60.9%) had a urinary catheter. Pressure ulcers were reported in 6 (3.6%) patients.

Early rehabilitation and mobilisation prevalence

On the first day of PERMIT study (day 3 post-PICU admission) overall 162/169 (95.9%) received at least one ERM activity. A mobility ERM activity was delivered to 147/169 (87.0%) of which an out-of-bed mobility ERM was delivered to 64/169 patients (37.9%).

Figure 5 shows the reduction in number of patients remaining in PICU following enrolment to PERMIT and the prevalence for (1) any ERM, (2) mobility ERM and (3) out-of-bed ERM mobility. Of note, by day 7 post PICU admission, half of patients enrolled into PERMIT had been discharged and by day 9, 113 (67%) had been discharged and 2 (1%) had died.

FIGURE 5 Prevalence of ERM for each study day in PICU. (Figure not reproduced here; see the original article.)

The rate of receiving ERM during the PERMIT study period, analysed using a Poisson regression model, did not change across the study period. We did not identify a significant trend with rate ratios of (1) any ERM: 0.98 (95% CI 0.92 to 1.05, p = 0.57), (2) mobility ERM: 0.97 (95% CI 0.93 to 1.01, p = 0.14) and (3) out-of-bed mobility 0.97 (95% CI 0.93 to 1.01; p = 0.14).

Cumulative probability of early rehabilitation and mobilisation

Figure 6a shows that while over 95% of enrolled patients were observed to have received an ERM intervention on day 1 of the study period, mobility ERM and out-of-bed ERM were less frequent (see Figure 6b and c). However, by day 6 of the study, 98% of patients had received a mobility ERM and 80% (see Figure 6b) of patients had received an out-of-bed ERM at some point during their observed period (see Figure 6c).

FIGURE 6 Cumulative probability of (a) any ERM, (b) mobility ERM and (c) out-of-bed ERM. (Figure not reproduced here; see the original article.)

Description of early rehabilitation and mobilisation

In total, 3696 ERM episodes capturing 4978 ERM activities occurred during 729 patient days. On the first study day (day 3 post PICU admission), 169 patients received 977 ERM episodes and 1302 ERM activities [median IQR 7 (4–10) ERM activities per patient].

Analysing the whole study period, positioning (which incorporated the mobility element) (1205/4978; 24%) and non-mobility positioning (1177/4978; 23.6%) were the most frequent activities (Figure 7). Active mobility was less frequent: the majority were active movement (e.g. rolling, active ROM) and sitting up in bed or transfer out of bed to chair or mat. No in-bed cycling was reported throughout the observation period.

FIGURE 7 All ERM activities ranked by active, enrichment and passive categories. (Figure not reproduced here; see the original article.)

Table 5 compares the baseline characteristics of patients receiving any ERM, mobility ERM or out-of-bed ERM and those who did not on the first study day. Patients receiving mobility ERM tended to be older that those who did not; however, the opposite was seen in out-of-bed mobility, where younger patients (especially <3 months) were more likely to receive mobility out of bed (e.g. cuddles). We identified no difference across ethnicity groups or diagnostic admission group in the proportion of patients receiving each category of ERM. We did identify that patients with moderate (PCPC 3) or severe disability (PCPC 4) received less out-of-bed mobility.

Table 6 compares the clinical status of patients with ERM type on the first study day. Patients receiving or not receiving any ERM or mobility ERM were similar in respect to the majority of measured clinical status factors. The only differences were between patients receiving out-of-bed mobility or not. There were more patients receiving invasive ventilation (high-frequency oscillation and conventional ventilation), use of neuromuscular blocking drugs, sedative drugs and with additional lines or tubes (e.g. central venous catheter, arterial line, urinary catheter and chest tubes) in the group not receiving out-of-bed mobility. This was also reflected in the higher PELOD2 organ dysfunction score for patients not receiving out-of-bed ERM [5 (IQR 3–6)] versus those who did receive [2.5 (IQR 0–5)].

Mobility activities

Using the ranking scale described by Ista et al.42 on the first day of study 147/169 (87%) had a mobility ERM activity. The ranked activities are shown in Table 7. Most were either passive ROM or cuddles. Only 29 (17%) involved ranked 5–9 activities (transferring out of bed, mat play, standing or walking).

Staffing and parental involvement in early rehabilitation and mobilisation

We examined staffing and parent data across all ERM episodes (which may have involved a combination of activities). As shown in Figure 8, registered nursing staff were involved with the majority of ERM episodes (3127/3696, 85%). Parents/guardians were present for 1614/3696 (44%) and participated in delivering 1372/3696 (37%). Physiotherapists delivered only a small proportion of ERM episodes (361/3696; 10%).

FIGURE 8 Number of parents, guardians or healthcare professionals involved in ERM episodes. (Figure not reproduced here; see the original article.)

For individual patients, on the first study day, only 57/162 (35%) of patients received at least one ERM episode delivered by a physiotherapist. Across the duration of the study period this increased to 95/162 (59%) of patients having the input of a physiotherapist for at least one ERM activity. While the majority of ERM episodes were delivered by registered nursing staff or parents, without specialist therapy input there is a risk that ERM activity quality may be low (e.g. not targeted at the patient acuity or developmental age).

Timing and duration of early rehabilitation and mobilisation

ERM episodes were commenced throughout the 24-hour period of PICU with 1855/3730 (49.7%) occurring between 09.00 and 17.00 (Figure 9). The median duration of ERM episode was 15 (IQR 10–30) minutes and 37% of ERM episodes were of short duration (1–14 minutes) (Figure 10).

FIGURE 9 Histogram of ERM start time. (Figure not reproduced here; see the original article.)

FIGURE 10 Duration of ERM episode (data available n = 3410). (Figure not reproduced here; see the original article.)

As the professional group present at the bedside 24/7, nursing staff delivered 85% of ERM activities and 48% of these activities were commenced outside ‘9 a.m. to 5 p.m. office hours’ and emphasise the need to view ERM as a 24/7 intervention and not just an activity performed during office hours by physiotherapists. Less than 10% of ERM episodes delivered by a physiotherapist were outside 9 a.m. to 5 p.m..

Predictive factors associated with out-of-bed mobility

We created a multivariable logistic regression model to explore predictors of interest for out-of-bed ERM on study day 1 (day 3 of PICU) (Figure 11). Factors associated with a decreased odds of out-of-bed mobility included pre-PICU severe disability (PCPC score 4 vs. 1) [OR 0.09 (95% CI 0.01 to 0.61)], invasive ventilation [OR 0.23 (95% CI 0.06 to 0.83)] and presence of a urinary catheter [OR 0.16 (95% CI 0.16 to 0.42)]. The only statistically significant factor associated with increased odds of out-of-bed mobility was the presence of a parent or guardian and/or their involvement in the ERM activity [OR 13.46 (95% CI 1.05 to 172.7)], although the CI was wide. Of note, a vasoactive infusion was associated with increased out-of-bed mobility [OR 1.85 (95% CI 0.5 to 6.82)] and use of neuromuscular blocking drugs was associated with decreased out-of-bed mobility [OR 0.33 (95% CI 0.08 to 1.4)]; however, both of their CIs crossed 1. We were unable to create models for any ERM or mobility ERM because of the high rate of positive events for both on study day 1.

FIGURE 11 Forest plot of odds of out-of-bed mobility on study day 1 by key factors. (Figure not reproduced here; see the original article.)

Adverse events

There were 106 recorded AEs during 78 separate ERM episodes (Table 8). Overall proportion of ERM activities with an AE was 106/3696 (2.87%; 95% CI 2.35 to 3.45), which equates to 1 in 35 (95% CI 1 in 29 to 1 in 43) ERM activities. The most frequent reported event was desaturation in 38 (1.03%; 1 in 97) of ERM activities and discomfort of patient or tiredness in 18 (0.49%; 1 in 205). There was only one ETT tube dislodged, and seven other tubes dislodged, during the entire study period. Overall ERM delivery was safe, with a very low rate of AEs.

Summary of findings to inform the PERMIT study

• Paediatric intensive care units were able to enrol 90% of eligible patients into the observational study during week 1, using an opt-out consent model.
• All participating UK PICUs delivered some form of ERM to patients. However, the range of ERM delivered is broad, with most of the time and resources delivered to basic patient positioning, family holding of patients and active movement in bed.
• ERM, using its broadest definition, is delivered to nearly all PICU patients on day 3 after admission and throughout their PICU stay. We did not identify a change in the prevalence of ERM delivery across study days. However, mobility and especially out-of-bed mobility was less frequent.
• We identified some barriers and modifiable factors associated with delivery of out-of-bed mobility (e.g. invasive ventilation, presence of urinary catheter, pre-existing severe disability and parental presence and involvement in ERM). Strategies to address these could improve rates of out-of-bed mobility.
• The presence of an existing ERM protocol neither increased nor decreased the odds of receiving out-of-bed mobility.
• Reported safety profile of patients receiving ERM is very good. There is a very low level of incidents, safety events, or physiological effects reported during ERM delivery.
• Important information regarding PICU practice was obtained. Delirium screening is universally absent in participating PICUs. Nursing staff and parents deliver the majority of ERM activities throughout the 24-hour period. Physiotherapists only delivered 10% of ERM, although 59% of patients received at least one PT delivered ERM episode at some point. There was minimal input from other medical, therapy, or support staff reported. Any ERM manual or intervention plan will need to utilise available staff.