Phase 3: paediatric early rehabilitation and mobilisation during intensive care feasibility study – part 1 study feasibility

Introduction

Following the development of the PERMIT ERM intervention manual in Phase 2 we designed and delivered a pilot feasibility study which would allow both the assessment of the implementation of the intervention in the NHS setting and also the trial feasibility of screening, recruitment and enrolment of CYP into a study that would deliver our manualised ERM intervention. In this chapter, we present the quantitative results of the feasibility of delivering the ERM intervention to CYP. In Chapter 6 we described in detail the process evaluation of the implementation, feasibility and delivery of the ERM manual within three NHS PICUs.

Study management

The work package was co-led by BRS, JMen and FK. The study management group provided input into protocol, ethics application design and data interpretation. Sati Sahota was trial co-ordinator. JYT designed, adapted and managed REDCAP database. Statistical analysis performed by BRS, RF and James Martin.

Aim

To explore the feasibility and acceptability of implementing the intervention manual in three PICUs.

Objectives

• Determine enrolment, recruitment and delivery of the PERMIT intervention.
• Monitor the safety of the PERMIT intervention and related AEs.
• Establish whether clinically important outcomes can be measured following delivery of the PERMIT intervention.

Methods

Design

A non-randomised unblinded intervention feasibility study with embedded process evaluation.

Intervention

The PERMIT intervention, as described in Chapter 4, was defined as a PICU-wide, healthcare professional-delivered intervention, aiming to promote opportunities for the delivery of ERM. The intervention included strategies to develop an organisational environment that supported the delivery of ERM, as well as ERM activities that could be tailored for each individual patient. Each PICU received the PERMIT intervention manual (see Table 12). The manual defined two key steps: preparing for the PERMIT intervention; and recruiting patients and delivering the ERM intervention. Within these two steps there were six key phases (Figure 14).

FIGURE 14 Phase 3 feasibility study overview. (Figure not reproduced here; see the original article.)

Step 1 - PICU preparation: to prepare the PICU for implementing the PERMIT intervention, a group of PERMIT champions (lead multidisciplinary healthcare professionals and managers) was formed (Phase 1 ‘build the team’). The team then led a rapid self-assessment of their PICU’s readiness to implement ERM (Phase 2 ‘take stock’). Relevant stakeholders were then brought together to participate in local discussions about ERM (Phase 3 ‘get buy-in’). The local PERMIT team reviewed, adapted and tailored the PERMIT intervention manual to suit the unique circumstances of their PICU and planned how best to incorporate ERM into local work routines (Phase 4, ‘get ready’).

Step 2 - Patient delivery: once the PICU self-assessed themselves as ready to start to deliver the PERMIT intervention to CYP, the site then progressed into Step 2. At this point, the local PERMIT team started to recruit patients to the trial, receiving ERM as defined by the manual (Phase 5 ‘make it work’). The study team reviewed the implementation process and worked with the local clinical teams to support ongoing education and training, adjust elements of the process as needed and plan for sustaining the programme beyond the end of the study (Phase 6 ‘keep it going’).

Setting

The study is based in PICUs within three NHS organisations: Birmingham Women and Children’s NHS Foundation Trust (Birmingham), King’s College Hospital NHS Foundation Trust (London) and University Hospital Southampton NHS Foundation Trust (Southampton). The PICUs were selected because of their diversity in (1) clinical expertise and the patient populations they serve (e.g. two PICUs were both cardiac and specialist PICUs, one non-cardiac but a large neurocritical care and liver transplantation programme), (2) the types of resources and staff groups they have available (e.g. allied health professionals, play specialists, rehabilitation equipment) and (3) ERM experience to date of implementing ERM within usual care (e.g. two units with minimal experience of implementing ERM and one with nationally recognised expertise in ERM).

Sampling and recruitment

Since this study was focused on feasibility and acceptability, an a priori sample size estimation was not required. Instead, our sample size was sufficient for capturing data across the diverse patient groups receiving ERM and the multidisciplinary healthcare professionals involved in implementing ERM.

Children and young people

We aimed to recruit 30 CYP (10 per unit) to receive ERM. All CYP were screened daily against the following eligibility criteria: aged 0–<16 years at the point of admission, admitted to PICU for any reason and likely to remain in the PICU on day 3 post admission. We excluded individuals where, for any reason, the local clinical team did not feel it was appropriate to include them in the study and we recorded the reasons. The majority of CYP were unable to provide informed consent or participate in an assent process (e.g. because of their very young age, levels of sedation medication). We therefore relied primarily on consent from parents/legal guardians (see below). However, where possible and appropriate, we attempted to take steps to involve CYP in decision-making and inform them about the study to the fullest level of their understanding.

Parents/legal guardians

To gain consent for delivering ERM, we approached parents/legal guardians of eligible CYP on day 1 or 2 of admission. As a minimum, they could choose to consent only to their child receiving ERM and to the accompanying data collection (i.e. clinical data about their child and intervention data about the delivery of ERM). This placed no direct burden of data collection on the parents/legal guardians themselves, offering option of participation, without any additional demands during the PICU admission. If parents declined the study then the child received ‘standard care’ in accordance with the local PICU policy. In addition, parents/legal guardians could choose to consent to completing outcome measures at two different time points – during their child’s stay in intensive care and around the time of discharge or shortly afterwards. They could agree to this when they were first approached on day 1 or 2 of admission or at a later stage during their child’s PICU stay. We also invited parents/legal guardians to take part in qualitative interviews (see Chapter 6: Process evaluation).

Data collection

We collected data about the CYP receiving ERM, the delivery of ERM to patients and the health-related outcomes of relevance to ERM. In this chapter, we report CYP clinical data, PERMIT intervention delivery data and parent/guardian outcome measurement data (Table 14). Chapter 6: Process evaluation contains details about the data collection of debrief conversations with PICU champions, interviews and survey with health professionals and interviews with parents/legal guardians.

Children and young people receiving ERM: currently, all CYP admitted to PICU have clinical data routinely recorded via the Paediatric Intensive Care Audit Network (PICANet).67 We identified PERMIT study participants on the PICANet web-based system and conducted a data download of a pseudo-anonymised comprehensive data set pertaining to each individual CYP enrolled into the PERMIT study.

The data download included the PICANet minimum data set for each individual participant. This included demographic and socioeconomic data (participant’s date of birth, sex, ethnicity); pre-PICU health status (past medical history including underlying conditions and comorbidities); acute illness data [PIM3 (model of PIM that assesses the risk of mortality among children admitted to a PICU)]; PICU admission and discharge diagnoses; comorbidities; operations and invasive procedures performed; type of admission; PICU and hospital LOS, duration of MV, high-frequency oscillatory ventilation, ECMO, renal replacement therapy and vasopressor/inotropic support; sedative medications and days of exposure.

We also collected additional clinical data. The majority of data were collected daily from time of enrolment into the study until PICU discharge (see Table 14) and included:

• Level of organ dysfunction: PELOD-2 score.
• Level of physical activity: Children’s Chelsea Critical Care Physical Assessment Tool (cCPAx).
• Skin integrity: Braden QD was completed to assess skin integrity and risk of pressure damage/injury.
• Presence of delirium: Cornell Assessment of Pediatric Delirium (CAPD).
• Level of sedation: The COMFORT behaviour scale (COMFORT‐B scale) sedation assessment score.
• POPC and PCPC (these were collected at admission to reflect the child’s pre-admission status and PICU discharge only).

Data were recorded on individual patient CRFs and transferred for analysis into the REDCap database for the PERMIT study.

Paediatric early rehabilitation and mobilisation during intensive care intervention data: the Local PERMIT Research Team conducted daily screening of CYP on PICU, to calculate the number of patients fulfilling inclusion eligibility and number of patients approached, consented/declined for the study.

Once parents/legal guardians (and CYP where appropriate) had consented, the patients received the PERMIT intervention until PICU discharge. The Local PERMIT Research Team used the PERMIT CRF for each individual CYP to collect the following data about the PERMIT intervention at ward rounds, ERM intervention sessions and through discussion with local clinicians:

• patient acuity level;
• prescribed ERM activity level and specific ERM activities;
• delivered ERM activity levels and specific ERM activities (including timing, duration, number and type of staff or family member assisting);
• reasons for deviating from the prescription, where relevant;
• use of pause and reassess criteria;
• safety and AEs;
• any intervention/manual tailoring proposed or undertaken within the site.

Outcome measurement tools: the Local PERMIT Research Team distributed outcome measurement tools (questionnaires) to parents/legal guardians at two time points: during their child’s PICU admission and at the point of PICU discharge or within 30 days after PICU discharge, whichever occurred first. The outcome measurement tools were administered initially as paper questionnaires then by additional methods at parents’/legal guardians’ preference. The information was entered into the PERMIT study REDCap database.

During PICU admission: parents/legal guardians provided a retrospective report based on their child’s pre-admission status (2 weeks before) by completing a QoL and a fatigue measure (1 and 2) and an assessment of the family pre-admission status (3) from the Pediatric Quality of Life Inventory (PedsQL)™ family of questionnaires:

1. PedsQL™ Infant Scales Version 4.0 – Acute (Aged: 1–23 months) OR PedsQL™ Generic Core Scales Version 4.0 – Acute (Aged: 2 years+);
2. PedsQL™ Multi-dimensional Fatigue Scale Version 3.0 – Acute;
3. The PedsQL™ Family Impact Module Version 2.0 (parent report).

At point of PICU discharge or within 30 days post admission to PICU: parents/legal guardians completed questionnaires based on their child’s current health status (1–3) and the family status (4–7):

1. PedsQL™ Infant Scales Version 4.0 – Acute (Aged: 1–23 months) OR PedsQL™ Generic Core Scales Version 4.0 – Acute (aged: 2 years+);
2. PedsQL™ Multi-dimensional Fatigue Scale Version 3.0 – Acute;
3. the PedsQL™ Family Impact Module Version 2.0 (parent assessment);
4. parent Stressor Scale;
5. the EMpowerment of PArents in The Intensive Care – 30 Item Version (EMPATHIC-30);
6. Patient Health Questionnaire-4 (PHQ-4).

Feasibility and acceptability of the questionnaires were captured through monitoring completion rates and feedback from the parent/carer interviews. This is reported in Chapter 6: Process evaluation.

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.

Descriptive statistics were used to summarise the PERMIT intervention data, including: success of implementation, recruitment, proportion of consented patients triaged from the acuity table, proportion of consented patients allocated an ERM intervention from the manual appropriate to their acuity level, proportion of consented patients receiving the prescribed ERM, number of AEs and proportion of patients experiencing AEs. Results are presented in text and tables with a narrative summary of findings. Stata v16 was used for all statistical analyses.

Patient and public involvement and engagement

For an overview of the approach to PPIE adopted throughout the study, please see Chapter 8. PPIE in this phase of the study focused on exploring the appropriate model of consent, refining participant-facing documents and establishing the overall acceptability of the study. We had planned to speak to parents with direct experience of involvement in research studies, but the research suspension imposed in response to directives to prioritise COVID-related research made this challenging.100 We spoke to parents of four children (aged 1 month to 16 years) who had experienced one or more PICU admission(s) at BCH and Kings College London. One parent had experience of their child being recruited to a number of non-PICU research studies.

The concept of a staged approach to consent to help reduce the burden about deciding on all aspects of the study at once was liked by the parents (Table 15). The challenge of speaking to families so early on in the child’s PICU stay was recognised. One parent described the overwhelming emotions and the importance of reducing the demands being placed upon them:

There’s also the challenge of emotional engagement. When * first came in, there was an overwhelming, primeval pain. This developed into a rollercoaster of positive and negative emotions which over-rided my ability to absorb information ….

Being able to stagger the informed-consent process until a parent was better placed to consider what was being asked of them was endorsed. Other factors that helped were having these messages clearly conveyed in the PIS and ensuring that this process was supported by someone who could provide clarification and answer questions. Parents noted that the material was written at the right level for them; however, concerns reflected the volume of information – ‘It’s like a book!’ – to read and digest. All commented on the value of figures or illustrations to help visualise the sort of interventions that would be involved.

We had also hoped to explore outcome measures and the interview schedules (for consenting parents and declining parents) but this was not possible. All four patients were inpatients on either PICU or had just been discharged to the ward. The parents we worked with were the only parents allowed with their child, so they had a lot of competing demands. It was not felt to be appropriate to provide them with multiple outcome measure questionnaires or interview schedules to review; in fact, this was not allowed in some ward areas due to COVID. The study team therefore drew on our PPIE work from other recent PICU studies that were involved in selecting questionnaires (the OCEANIC study101) and qualitative interviews (BRICC trial102) to shape the decision-making.

The study received Health Research Authority approval (ref 21/SC/0127), and was registered at https://clinicaltrials.gov/ct2/show/NCT04909762.

Results

Site implementation and recruitment to PERMIT study

Overall, three PICUs participated in the PERMIT feasibility study. The PICUs were chosen as a pragmatic sample with variation in average annual admissions, number of overall PICU staff, types of paediatric specialties and stand-alone children’s hospital, co-located adult/paediatric centre.

Sites 1 and 2 started Step 1 of the PERMIT implementation programme in early June 2021. Site 3 was delayed due to contract issues until 2 August 2021 (Figure 15). All three sites (100%) achieved the primary outcome and progressed through phases 1–4 of Step 1 (PICU preparation phase) of the PERMIT manual. Time to complete Step 1 and progress to Step 2; site 1: 12.9 weeks, site 2: 15.0 weeks, site 3: 8.4 weeks. Table 16 reports key feasibility outcomes.

FIGURE 15 Progress of sites through Phase 3 feasibility study. (Figure not reproduced here; see the original article.)

All three sites (100%) recruited their target of 10 patients within Step 2. Site 2 over-recruited (n = 11) as one patient was discharged from PICU a few hours after signing consent and was excluded from reporting below. Full recruitment was completed within 53, 23 and 57 days from the start of Step 2. Overall, 31/35 (89%) of eligible patients were successfully recruited. Only four families declined consent to PERMIT.

Sites completed follow-up to discharge or 30 days (whichever was sooner) of all recruited patients by 91, 32 and 57 days respectively of commencing Step 2.

Of the 31 patients who agreed to consent A (participation in data collection and to receive ERM interventions), 21/31 (68%) also consented to consent B (follow-up assessment tool collection and to be approached about an interview). One patient was discharged from PICU very soon after consent so no data were collected. Therefore, 30 patients were included in ERM assessment data.

Outcome assessment tool completion

Assessment tools were completed at baseline, during daily clinical status assessments and at follow-up (discharge or 30 days after PICU admission) (Table 17).

Baseline assessment tools for PCPC and POPC were recorded in 30/30 (100%) of cases. Additional baseline forms were scored for the families consenting to consent B (n = 20). Of these, 19/20 (95%) completed the Family Impact Scale and PedsQL parental core reports, with 7/7 (100%) of those eligible completing the PedsQL Multi-dimensional Fatigue Scale.

The daily patient assessment scores and organ dysfunction scores (PELOD-2 and cCPAX) were collected in nearly all cases (see Table 17). In more than 91% of available PICU study days the CAPD, Braden QD and Comfort B scores were collected. Missing data were related to unavailability of research staff to calculate scores contemporaneously, indicating that these were not routinely collected clinical scores by bedside staff.

Follow-up outcome assessment score completion rate was much lower (12/20; 60% of patients), although these were all fully filled in. The protocol defined follow-up time occurring at point of PICU discharge or within 30 days post admission to PICU with flexibility for research staff to assess feasibility. Time from PICU admission to follow-up assessment completion was a median (IQR) of 31.5 (30–40) days. The length of time from discharge of PICU to follow up was median (IQR) 26 (13–36) days.

Demographics

The median age of the 30 patients recruited was 1.8 years (IQR 0.4–5.4) with around a third (9/30; 30%) <12 months (Table 18). The most common primary admission diagnosis was a respiratory illness (40%) and 8/30 (27%) were admitted following surgery. Prior to admission, 16/30 (53%) had a normal (PCPC 1) cerebral performance score, and 11/30 (37%) a normal (POPC 1) overall performance score with 19/30 (63%) ambulatory. This population was similar to the general population observed in Phase 1c observational study (see Chapter 3).

On day 1 of the study (day 3 PICU stay), the median PELOD 2 score was 4 (IQR 3–6). Nearly all were on assisted ventilatory support; 23/30 (77%) invasive MV and 5/30 (10%) on non-invasive support [bilevel positive airway pressure (BIPAP), continuous positive airway pressure (CPAP) or high flow nasal cannula], 6/30 (20%) were on vasoactive infusions and 2/30 (7%) were on neuromuscular blocking drugs (Table 19).

Prior to the first study day, four patients had screened positive for delirium. On the first study day the median CAPD score was 14.5 (IQR 9–19). The standard cut-off CAPD score indicating delirium is a score of >12. Therefore, a high proportion (16/30; 59%) screened positive for delirium.

Patients’ total LOS on PICU from admission was a median (IQR) of 5 (1–13) days. Following recruitment to PERMIT, the number of study days per study participant was median (IQR) 2 (1–7) days and patients were potentially available to receive ERM on 191 study days. Site 1 had the largest number of ERM study days (n = 108) compared to site 2 (n = 42) and site 3 (n = 41) as a result of a longer LOS for patients at site 1 (Table 20).

Number of early rehabilitation and mobilisation activities prescribed per children and young people following patient acuity screening

Of the 191 available study days, 174 (91%) had a morning ward round observed. During the ward round the patient acuity was recorded on 161/174 (93%) occasions. The median (IQR) acuity level (level 1, most unwell to level 4 least, unwell) for study patients was 3 (3–4); only 23/161 (14%) were level 1 or 2 (i.e. most sick). A corresponding ERM activity level for the patient was documented for 159/174 (91%) and an ERM activity prescribed on 137/174 (79%) occasions. PERMIT local researchers were present at the time of the ward round on over half of occasions (58%), providing a range of levels of support (from 17% ‘lots of support’ to 29% ‘no support’) to the ward round clinical team.

In total, 825 ERM activities were delivered within 328 ERM episodes on 140/191 (73%) of patient study days across the three study sites.

Type of early rehabilitation and mobilisation

On study day 1, 26/30 (87%) patients had an ERM activity prescribed and 22/30 (73%) had an ERM episode delivered. Of the eight patients who did not receive ERM, five had an ERM prescription for an activity. One of the four patients not prescribed ERM received an ERM activity on study day 1. Across the first 7 days of the study (Figure 16), the rate of receiving a mobility ERM was similar to receiving any ERM. The proportion of patients receiving out-of-bed mobility gradually increased across the study period. For illustration, Appendix 28 shows the range of ERM activities prescribed and the corresponding proportion that were delivered on study day 1. All sites prescribed a wide range of ERM with a good conversion rate from prescribed to delivered for passive, enrichment and active mobility ERM activities.

FIGURE 16 Prevalence of ERM delivered on study day 1 by any, mobility and out-of-bed categories. (Figure not reproduced here; see the original article.)

Duration of early rehabilitation and mobilisation

Times were recorded at the start and end of ERM episodes. Recording was variable in the database, with some documented episodes as a single ERM activity, others recorded as multiple ERM activities, or continuous delivery of ERM making interpretation difficult.

Appendix 29 shows the distribution of recorded durations of ERM: 36/266 (14%) recorded times were <15 minutes duration and 61/266 (23%) 15–29 minutes. The median duration was 30 (IQR 15–75) minutes, which was longer that the median time observed in the Phase 1c observational study of 15 minutes.

Safety and adverse event rates during an early rehabilitation and mobilisation activity

The ‘pause and assess’ criteria embedded within the ERM manual for safe deliver and re-assessment of patients before and during an ERM activity were used in 21/330 (6%) of ERM episodes across 13/30 (43%) patients (Table 21). The most frequent categories were desaturation limit exceeded (5/21; 24%) and agitation, anxiety and distress (4/21; 19%).

Only 2/328 (0.6%) of ERM activities in two patients were associated with an AE as specified in the study protocol (Table 22). The first was a nasogastric tube dislodged during an ERM episode, but not related to the activity and with mild severity impact on the patient. The second was a blood-pressure change, possibly related to ERM activity, with mild impact on the patient.

Summary of findings to inform the PERMIT study

Overall we successfully and safely introduced a multidisciplinary delivered, PICU wide, early rehabilitation programme using the PERMIT manual to three UK PICUs. All sites implemented the PERMIT programme as described in the PERMIT manual requirement within 8–15 weeks to reach Step 2 to allow patient recruitment.

• A diverse group of PICU patients were recruited on day 3 of PICU admission which represented the wider population of patients identified in Phase 1 observational study.
• All sites successfully recruited the 10-patient target.
• All patients had an acuity level scored and these were repeated on 84% of ward rounds. The acuity level was correctly linked to ERM activity prescription and then subsequently to ERM activity delivered. The level of activity was broadly representative of the acuity level. However, the manual description of progression of activity levels over time was difficult to translate to bedside clinical decision-making and data-recording (e.g. demonstration of increased dose of standing, or mat play).
• All sites modified a local ‘menu’ of ERM activities specific for each acuity level to allow wider choice of ERM prescriptions.
• A large number of potentially clinically relevant patient outcomes were measured and recorded successfully through validated tools. Baseline and daily tools were recorded over 90% by research and bedside clinical staff and the REDCap database.
• Clinical trial data integration with PICANet existing audit network data collection was successful and allowed reduction in trial data collection.
• All patients received ERM activities safely using the pause and assess criteria with only two trial reported AEs and no severe AEs. The rate of physiological deterioration (e.g. desaturation limit exceeded) was lower than the self-reported AEs in the Phase 1 observational study. None adversely affect the patient, as assessed by the bedside clinical team.
• Although the feasibility study did not aim to do a pre-post implementation assessment on ERM delivery, we did identify that ERM episode durations overall were longer, increasing from a median of 15 minutes in Phase 1 to 30 minutes in Phase 3. In addition, a higher proportion of ERM activities were ‘active’ compared to ‘passive’ or ‘enrichment’ and PT delivered ERM increased from 59% of patients in phase 1 to 93% of patients in Phase 3.

A future trial of ERM therefore appears to be feasible; however, there are a number of recommendations (see Tables 27, 28 and 29). Key messages include:

• A future trial would need to plan for at least a 12-week (3-month) implementation training programme to work through the key components of implementing the PERMIT manual.
• The provision of adequate resources for research data entry and research co-ordinator and regular research team presence on ward rounds appear important to retain the quality and accuracy of data collection.
• Automated integration with existing PICU clinical information systems and workflows would allow additional improvement to data collection.
• We also identified a high rate of delirium when research staff completed the CAPD screening. Previously, PICUs did not screen for delirium automatically (0% of the 15 PICUs in Phase 1c observational study). Delirium screening is a key component of the wider ABCDEF intensive care unit (ICU) liberation bundle.103 Training and implementation of delirium screening and further investigation of modifiable factors and potential treatments should remain a priority for NIHR funders. Future PERMIT trials should include delirium screening as a key component of the PERMIT manual.