Introduction
This review aimed to evaluate early mobilisation and rehabilitation (ERM) within paediatric intensive care as reported within the published literature. We characterised the evidence base using a narrative synthesis approach to understand features of ERM associated with effectiveness and successful implementation within paediatric intensive care units (PICU).
Study management
The work package was led by BRS. The study management group was responsible for defining and reviewing scope of search. JYT performed searches, data extraction, risk of bias assessment, evidence synthesis and first draft of chapter. Second screening of articles, data extraction and risk of bias performed by Dr Olivia Craw, JMc and JMen. Methodological expertise provided by DM and BRS.
Objectives
Our primary objective was to summarise the types and effectiveness of ERM interventions and outcome measures delivered to children admitted to PICUs.
Our secondary objective was to thematically identify subpopulations (if any) that benefit most from ERM or experience associated adverse or clinical events, and any patterns or gaps during implementation.
Methods
Search strategy
Searches were conducted in the bibliographic databases [Excerpta Medica Database (Embase) (via OVID), Cumulative Index to Nursing and Allied Health Literature (CINAHL) (via EBSCO), MEDLINE (via OVID), PEDro, Open grey or Cochrane CENTRAL)]. Original search was from inception to 12 October 2019 and an updated search was performed 1 November 2021, using strategies that combined, where relevant, free text and index terms for:
- children and young people (CYP);
- admitted to paediatric critical care settings;
- receiving early (within 7 days) rehabilitation and mobilisation.
The search strategy was developed in MEDLINE (via OVID) (see Appendix 2) and adapted for other databases.
Database of Abstracts of Reviews of Effects, National Health Service (NHS) Economic Evaluation Database and Health Technology Assessment database (all via Centre for Reviews and Dissemination) were searched for relevant systematic reviews to identify primary studies for the review. These were supplemented by relevant websites using hand-searching for mobilization-network.org and search terms for clinicaltrials.gov or Chinese clinical trial registry, checking reference lists of relevant studies, and forward citation-checking of included studies in Web of Science. We screened reference lists to identify relevant primary studies and contacted primary authors to find full texts of incomplete records.
Eligibility criteria
We included all completed studies published in English that met the following criteria:
- Study participants: critically ill infants, children or young people aged ≤18 years, admitted to PICUs, who received an intervention described as rehabilitation or mobilisation delivered by any health professional within ≤7 days after admission. Rehabilitation or mobilisation interventions could include but were not limited to physiotherapy (PT), occupational therapy, speech and language therapy (SLT) and bundled interventions; these included ABCDEFH bundles (spontaneous awakening and breathing trials; choice of sedation and analgesia, delirium prevention, surveillance and management; early mobilisation and exercise programmes with or without adjuncts; family engagement and empowerment; proper nutrition and humanism) so long as their application was considered within the first 7 days of admission; AND
- Outcome: at least one outcome was related to participants’ health and well-being, health service utilisation, feasibility, acceptability or intervention implementation.
- Study design: primary research studies of any designs, with >10 participants to synthesise evidence on intervention effectiveness. Case reports, case series with ≤10 patients, qualitative studies and systematic reviews were excluded if relevant primary studies were not identified in the references. Abstracts or ongoing studies identified from clinical trial registries were used to highlight the presence of future emerging research.
Screening and selection
Records identified were imported into a bibliographic referencing software programme (EndNote X9, Thomson Reuters, San Francisco, CA, USA), and duplicates were removed. One PT researcher (JT) and one clinical academic (BS or JMen) independently screened titles and abstracts for relevance against the eligibility criteria within Rayyan systematic review software.1 The full texts of relevant articles were obtained and assessed against the selection criteria by two reviewers independently (JT, Dr Olivia Craw). A wider range of publication types (abstracts and full texts) were selected to identify all possible lists of ERM interventions but were not analysed to summarise types of ERM. Reasons for exclusion were noted. Discrepancies were discussed via consensus meeting with a third author (JMc).
Data extraction
Data were extracted using a standardised, piloted data-extraction form in Excel. Information within the following domains was extracted:
- Study – author, year of publication, country, study design using an algorithm for classifying studies.2
- Patient demographics – age, sex, admission diagnosis, the severity of illness and, comorbidity using established criteria3 or paediatric scoring tools such as Pediatric risk of mortality (PRISM III), the Pediatric logistic organ dysfunction (PELOD), the Pediatric Cerebral Performance Category (PCPC) and the Pediatric Overall Performance Category (POPC). We also considered the following prognostic factors when assessing non-randomised studies: age, sex, weight or body mass index (BMI) in percentile, baseline severity, comorbidities and admission diagnosis on the intervention.
- Intervention details – definition of ERM, type of interventions, the volume of ERM (time-to-initiation, duration, number of sessions), implementation strategies such as safety and progression criteria, involvement of health professionals or availability of organisational support.
- Study comparators and outcome – components of usual care, primary and secondary outcomes (where specified) and assessment time points. When outcome measures were not specified or reported, the outcomes most proximal to the health domain were considered the primary outcome.
Data were extracted (JT) and independently verified by co-authors (Dr Olivia Craw, JMc and JMen). We used all eligible studies, abstracts or full texts that reported any intervention to summarise types of interventions. We only included full-text reports where ERM was initiated within the first 7 days of admission to PICU to synthesise ERM outcomes.
Quality assessment of individual studies
Risk of bias was assessed in randomised controlled trials (RCTs) the using Cochrane Risk of Bias tool version 24 and in non-randomised studies using the Risk Of Bias In Non-randomised Studies – of Interventions (ROBINS-I) tool.5 One reviewer (JT) assessed the methodological quality of studies and this was independently verified by a second (Dr Olivia Craw, JMc).
We evaluated the reporting quality of studies using the Consensus on Exercise Reporting Template (CERT).6 Due to the nature of the study interventions included in this review, participants, providers and assessors were aware of the intervention, which can affect compliance, outcome assessment or intervention fidelity. To understand implementation, we grouped studies that provided information on different aspects of delivering ERM, such as core content of ERM, who commonly delivers it, mode, timing, frequency of delivery, and the adaptation process for tailoring ERM.
Data synthesis
A narrative synthesis of all included studies was undertaken. Due to the heterogeneous nature of paediatric populations and interventions, meta-analysis was not appropriate. All outcomes were grouped as short-term (≤6 months post-discharge) or intermediate-term (≥6 months post-discharge) outcomes.
Results
Characteristics of included studies
As shown in Figure 1, 2580 unique records were screened for relevance, and 62 relevant full-text articles were assessed for inclusion. Twenty-six of these were excluded, mainly due to the ineligibility of the study design or outcomes. Eighteen of the 36 studies that met the eligibility criteria were abstracts, and 18 had full-text reports. Most were conducted in North America,7–25 Australia,26 Belgium,27,28 Brazil,29,30 Italy,31 Japan,32–34 the Netherlands,35 Turkey36,37 and the UK.38–41 One study was conducted across 15 countries in Europe42 (see Appendix 2, Table 32).
FIGURE 1 PRISMA diagram. (Figure not reproduced here; see the original article.)
Five studies15,18,21,22,24 were conducted across multiple PICUs, and two studies15,43 used controlled designs. Three studies were prospective cohorts,24,41,42 2 were interrupted-time series,7,27 9 non-comparative studies,8,10,13,22,30,31,34,36,40 11 before–after studies9,11,12,17,23,25,32,33,35,38,44 and 9 non-concurrent cohort studies.14,16,18,19,21,26,28,37,45
Types of interventions identified
Of the 36 studies that evaluated PICU rehabilitation, we identified two broad categories of early rehabilitation or mobilisation (ERM): non-mobility and mobility interventions. Non-mobility interventions mentioned in included studies were pain and agitation assessment,23 sleep hygiene/delirium screening,17,22,23,35 ERM screening checklist,26 cuddles,18,40 SLT8,12,15,21 and chest PT.21 The majority were mobility interventions and included mobility goals/orders,22,23,35,40,41 out-of-bed exercises,15,17 in-bed cycling,11,19,20,46 edge-of-bed mobility,40 bed-mobility exercises,15,17,21,32 interactive boxing,7 physical therapy8,10,12–16,18,21,30,32 and OT.8,10,12–16,18,21,30 When usual care15,20,24 was used as a comparison, it consisted of positioning,
Interventions were commonly administered from 24 to 72 hours after PICU admission. The volume of sessions varied widely, but most sessions were delivered twice daily. In some situations,32 information on initiating ERM delivery was unavailable. Five studies7,11,19,26,43 used single-component interventions. One study13 reported the number of encounters or admissions but provided no information about patient characteristics. Multicomponent interventions (12/16 studies) consisting of PT, OT and SLT were more commonly explored.
Patient and study characteristics
Out of the 36 studies that met our eligibility criteria, 18 full-text records evaluated at least one ERM outcome (as defined by the study authors) within 7 days after admission; the results reported here are for these 18 full-text publications. The study population consisted of day-old children to ≤18 years, with sample sizes of 12–722 participants. In almost all studies (n = 17/18), patients were admitted with a mixture of medical and surgical diagnoses – respiratory, neurological and cardiac conditions.
Outcomes
Among 18 studies that evaluated ERM, the feasibility and safety of ERM alongside process outcomes were the most frequent outcomes considered. See Appendix 2, Table 32.
Adverse events
Fifteen studies7,10–14,17–19,24,25,32,36,42,43 provided information on adverse events (AEs). The most common event was tachycardia/desaturation.13,14,24,42 Three studies reported haemodynamic changes7,15,42 or tube removals.7,10,42 Other events mentioned include pain,7 fall,7 behavioural changes,24 excessive secretions24 and discontinuation of therapy.14,15,43
Evaluation of early rehabilitation and mobilisation interventions
Feasibility in randomised controlled designs
Only two studies used randomised controlled designs, both judged as having a moderate risk of bias.15,43 These studies15,43 evaluated the feasibility of ERM as a primary outcome. The consent rates were 60%15 and 94%.43 One RCT, in 58 children aged 3–17 years with brain injury, showed that physical therapy was delivered 80% of the time in the usual-care arm and 100% among patients receiving early protocolised ERM.15 In addition, patients receiving early protocolised rehabilitation received less post-PICU rehabilitation, but there were no differences in functional or quality of life (QoL) outcomes at 6 months.15
In a pilot randomised trial with 30 children aged 3–17 years,43 the primary end point was feasibility defined as (1) the ability to enrol at least 75% of eligible patients, (2) an accrual rate of 1–2 patients a month and (3) a 30-day follow-up of >75%. The consent rate was 94%, and the 30-day follow-up rate was 87%. The median time from randomisation to delivery of mobility PT was not different between patients in the control standard mobilisation group [2.3 hours, interquartile range (IQR) 1–20] and standard mobilisation plus in-bed cycling (2.5 hours, IQR 0.9–11 hours). In this study,43 the authors found no difference between arms (0.17, IQR −0.01–0.36) among 24/30 patients (80%) who developed new functional difficulties in PICU [Paediatric Evaluation of Disability Inventory-Computer Adaptive test (PEDI-CAT)]. In the usual-care arm 7/10 (70%) did with median of 0.4 (IQR: 0.3–0.6), compared with 17/20 (85%) in the cycling arm (median: 0.6, IQR: 0.4–0.7). Overall, only 10% of these patients fully recovered functional ability at 1 month, and mobility was the slowest to return.43 No differences were identified for other outcomes evaluated in these studies.15,43
Feasibility/prevalence of early rehabilitation and mobilisation in observational prospective and retrospective studies
Single-centre studies
Most studies (n = 16) reported improvements in early mobilisation rates, demonstrating the feasibility of ERM. One study completed enrolment 1 month earlier than anticipated with an 85% consent rate.19 Another study12 reported more PT and OT ERM consultations. Betters et al.10 also reported higher ERM consultations (median 30, IQR 29–45 minutes). Alqaqaa et al.8 reported higher mobilisation rates among non-ventilated patients (mean difference of a day) compared to no change for mechanically ventilated patients. In another study,17 mobilisation and OT consultation rates increased, but this change was not significantly different for PT consultations.
Choong and colleagues11 showed higher lower-limb activity during in-bed cycling [mean ± standard deviation (SD) 266.47 ± 166.12] versus during non-intervention times (mean ± SD 20.94 ± 15.26, counts/20 minutes, p < 0.001) different to baseline. In another study,21 the median time to mobilisation was 2 days (IQR 1–6) compared to 1 day for non-mobility interventions (IQR 1–3). Likewise, the frequency of physical therapy improved in another study,32 with more patients achieving their rehabilitation goals.
Multicentre studies
In a multicountry study,42 the prevalence of PT- and/or OT-provided mobility was 39% [95% confidence interval (CI) 34.7 to 43.9%] and did not differ according to baseline neurodevelopmental function level (PCPC ≤ 2.22% vs. PCPC ≥ 3.26%, p = 0.331), while PT or OT consultations were higher in a different study, ordered in 68/128 (49.6%) within 2 days (1–5 days) after PICU admission.24 The prevalence of out-of-bed mobility was nearly two-thirds (87/110; 63.5%), passive range-of-motion 13.9% and no activity 19.7%. Out-of-bed mobility was common among non-mechanically ventilated children (48/56; 85.7%) and those under 3 years (71/100; 71%), while older children (18/37; 48.6%) were commonly not actively mobilised.24 Similarly, consultations for all admissions were increased from 25% pre-implementation to 56% (p < 0.001) post implementation,25 while consultations within 2 days increased from 34% to 67% (p < 0.001) and within 3 days, from 21% to 30% (p = 0.02) for at least one PT and/or OT mobility or 29% to 35% (p = 0.29) for mobilisations.
In another multicentre study, ~70% of children18 received at least one mobilisation intervention. For children on mechanical ventilation (MV), one-third of those <3 years and ~50% of those ≥3 years were mobilised using passive range of movement (ROM) (72%).18 Out-of-bed mobility was achieved 70% of the time, but less frequently among mechanically ventilated children, 47% (95% CI 44% to 49%).
Secondary or clinical outcomes in observational prospective and retrospective studies
Single-centre studies
Choong et al.19 found improved functional ability measured using PEDI among 28% and 42% of participants at 3 and 6 months. Only 22% of those with a pre-existing chronic condition and 14% with functional limitations returned to baseline levels at 6 months. In comparison, 60% of previously healthy children and 58% of children with normal baseline function regained full functional abilities. The overall mortality rate was 3/33 (9%), and 19/33 (63%) were readmitted within 6 months after PICU discharge.19
Alqaqaa et al.16 reported small improvements in length of stay [LOS; average LOS pre-WeeMove = 6.25 days vs. post-WeeMove LOS = 5.23 days] and time spent intubated (pre-WeeMove was 27.86 hours, post-WeeMove = 25.09 hours).16 In contrast, Abdulsatar and co-authors7 noted that PCPC scores were worse (mean ± SD change of 1.08 ± 1.0, p = 0.02) among two-thirds of patients who received ERM.7 There was no improvement in grip strength or physiological status, despite increased activity levels (mean ± SD, upper-limb (UL) activity pre: 9.36 ± 4.12 vs. post: 57.12 ± 46.60 counts) and higher carer satisfaction.7
Colwell and co-authors reported higher adherence among younger patients (p = 0.04), with higher baseline severity of illness (p < 0.001) when mobilisation sessions were goal-directed (p < 0.001). However, there were no significant differences in mobilisation rates (pre: mean = 0.86 vs. post: = 0.84) or AEs 14/560 (2.5%, p = 0.18).13 In the study by Wieczorek et al., nearly half of the children (48/100) received at least one ERM intervention by day 3 of admission, and the proportion of children receiving at least one in-bed activity increased by 18% from post intervention (p < 0.001).17 However, there was no change in passive ROM, only an increase in active interventions (57% vs. 26%; p < 0.001), especially among children ≥3 years. There was also a slight increase from 0% (0/39) ambulation while orally intubated to 10% (4/40) post intervention.17
Multicentre observational studies
Studies described a lower amount of mobilisation among younger patients with higher baseline disability13,17,18 or severity of illness on admission.13,18 One study21 showed that older, less sick children admitted during the winter who were not mechanically ventilated, sedated or receiving neuromuscular blockade were more likely to receive mobility interventions. Similarly, consultations within 3 days were higher among males, older children and those with lower baseline function, without an indwelling endotracheal tube (ETT) or urinary catheter, who had adequate family support.18
Predictors of early rehabilitation and mobilisation
Three studies evaluated predictors of ERM using multivariate analysis.18,24,42 The adjusted odds ratio (aOR) for out-of-bed mobility was negatively associated with the presence of an ETT (aOR, 0.13; 95% CI 0.08 to 0.2), a urinary catheter (aOR, 0.28; 95% CI 0.14 to 0.57), opioid infusion (0.42; 95% CI 0.24 to 0.73) and severe baseline disability (PCPC 4 vs. 1) (aOR, 0.59; 95% CI 0.4 to 0.87). Longer PICU LOS and lower nurse-to-patient ratio (1.82; 95% CI 1.2 to 2.8) increase the odds of out-of-bed mobilisation. For children <3 years old, family presence was associated with out-of-bed mobility (aOR, 4.55; 95% CI 3.1 to 6.6) while being older predicted PT or OT (aOR, 3.1; 95% CI 2.01 to 4.79).18 In another study,24 the presence of ETT or infusion among children ≤3 years reduced the likelihood of receiving therapist-provided out-of-bed mobility [odds ratio (OR) 3.62; 95% CI 1.49 to 8.82]. However, this improved when the family were present.24
Ista and colleagues42 showed that older age (2.28, 95% CI 1.23 to 4.22), moderate baseline disability (defined as PCPC: 3 vs. 1) (2.12, 95% CI 1.02 to 4.56), severe baseline disability (PCPC: 4 vs. 1) (2.24, 95% CI 1.14 to 4.40), having a central venous line (CVC) in place (aOR 1.63, 95% CI 1.02 to 2.62) and family presence (aOR 5.13, 95% CI 2.55 to 10.32) increased the odds of receiving a mobility session. However, the presence of a urinary catheter reduced the chance of mobilisation (aOR 0.46, 95% CI 0.22 to 0.92). MV through an ETT (aOR 0.29, 95% CI 0.12 to 0.68), being admitted for a surgical reason (aOR 0.58, 95% CI 0.35 to 0.95) and the presence of a urinary catheter (aOR 0.39, 95% CI 0.19 to 0.81) reduced odds of out-of-bed mobility, but this improved with family presence (aOR 7.83, 95% CI 3.09 to 19.79).42
Quality assessment
Controlled trials
Two studies15,43 randomly assigned participants into groups and were assessed using RoB v2. One study43 ensured allocation concealment using a computer-generated sequence and reported adequate sample size considerations. In the other study,43 sample size calculations were not applicable; consequently, outcomes were possibly underpowered. We did not identify attrition bias for both studies, outcomes assessment was similar at baseline, and co-interventions were similar across groups.
Prospective and retrospective studies
We used the ROBINS-I tool to assess the quality of observational studies (see Appendix 2, Table 33). Overall, most studies were judged to have a serious or critical risk of bias. Three studies (3/18)18,24,42 were judged to have a low risk of bias. Nine10,12,13,15,19–21,25,32 were judged to be at moderate risk of bias, while in four there was serious risk.11,14,16,31 In some studies bias was judged as critical,8 or information reported was insufficient17 and could not be assessed (Figure 2).
FIGURE 2 Pictorial presentation of quality assessment using ROBINS-I. (Figure not reproduced here; see the original article.)
The primary reason for downgrading studies was bias due to blinding or poor consideration of baseline confounding. There was no indication of selection bias during enrolment; studies were judged as adequate except in three retrospective studies.14,17,31 The lack of a clear definition of ERM hierarchy limited the evaluation of demonstrable effects on objective clinical outcome measures. This made it challenging to determine intervention superiority. Broadly, intervention categories of non-mobility and mobility were consistent across studies, which were judged to be at risk of misclassification bias half of the time (9/18). We assessed the impact of bias due to deviations from the intended response as adequate in most studies. In some studies,7,8,10,11,13,14,16,17,21,32 the technique for handling missing data was unclear or not reported. We did not identify any evidence of selective outcome reporting or errors due to outcome measurements. Most studies demonstrated congruence between previously defined analyses and outcomes reported. However, none of the studies published a protocol. Overall, most studies did not provide definitive evidence of the effects of the intervention. However, consistent evidence across all studies supports the feasibility of ERM as an intervention in PICU, while physical therapy was the most common intervention considered.
Quality of consensus on exercise reporting
We described techniques considered during intervention design and excluded items not relevant to this review. These items included item no. 3 (descriptions of individual or group exercises), item no. 9 (use of home equipment; discharge interventions were not considered) and item no. 12 (setting of exercise delivery; all studies were conducted in PICUs). See Appendix 2, Table 34 for details.
Quality of consensus on exercise reporting reporting
Most studies (12/18)7,10,16–18,20,21,24,25,31,32,42 provided details on the type of ERM intervention to aid replication. Some studies (4/18)18,21,25,32 provided training or engaged multidisciplinary teams (MDTs) to facilitate ERM delivery. Organisational strategies were sometimes applied across MDTs to facilitate PICU culture change. Some studies (12/18) explicitly mentioned using qualified providers such as physiotherapists, nurses or other therapists to supervise sessions and ensure intervention fidelity.10,11,13–15,17,18,21,24,25,42,43 However, the detail provided was insufficient to explain how differences in experience levels, treatment approaches and therapists’ behaviour in these circumstances influence outcomes.
Intervention components were generally tailored and not standardised. Sometimes, it was unclear what aspects of the interventions were usual practice or complementary during ERM delivery. In most studies, information on how deviations from study protocols were handled (i.e. regression or progression) and how these events may have affected outcomes was unavailable. Personalising the volume of ERM was a common concept across studies used to improve compliance. However, since interventions were tailored to tolerance levels, we did not assess intervention fidelity. Some studies provided a detailed description of how compliance or adherence was measured. Strategies mentioned include rehabilitation sheets11,19,20,32 and electronic records during PICU ward rounds.17,18 Two studies7,11 used an objective outcome to measure compliance – ACTi-graph accelerometers. This outcome measure can be used as a benchmark for future studies, incorporating routinely collected outcomes to increase transferability.47 Labour-intensive or ad hoc approaches for determining adherence or compliance, such as caregiver verbal confirmation following direct observation, charts or checklists, may limit the implementation of rehabilitation and future attempts at service evaluation.
Early rehabilitation and mobilisation change techniques
No study provided explicit details about applying complex intervention theories when designing interventions, while details on the implementation processes were inconsistently reported. Therefore, it is unclear to what extent interactions between intervention volume of ERM and health systems produce positive effects. It is not impossible to envisage a situation where contextual factors such as how the intervention works, for example the presence of a local champion or an ERM enthusiast, staff turnover, population demographics, or existing PICU culture, affect outcomes, either positively or negatively. No study provided details on motivation strategies or precise details on how interventions were personalised. Safety guidelines (underpinned by clinical stability), verbal feedback and tolerance levels to determine progression were commonly used across studies. Eight studies10,14,17,18,21,24,25,42 provided information to enable replication. Hence, these studies can be used as a springboard to undertake detailed intervention mapping when designing ERM manuals. Overall, key aspects of intervention delivery were poorly reported, such as co-interventions, strategies for tailoring interventions and motivating patients.
Discussion
This review aimed to summarise evidence on the effectiveness of ERM research within PICUs. We narratively synthesised evidence to improve interpretation of effectiveness given variation in ERM implementation. We identified a broad range of activities, categorised as non-mobility and mobility interventions. Other interventions identified but not considered in this review include undefined ERM,48–50 music therapy51 and neuro-psychological training.52 This review suggests that interdisciplinary multicomponent interventions, sometimes delivered as a bundle, are feasible, safe, acceptable and possibly beneficial to patients. The programmes mainly were designed using safety criteria, were goal-directed and tailored. Although the rate of intervention-related AE reporting across all studies was low, we cannot rule out selective reporting as none of the included studies had published their statistical analysis plan a priori.
Given the limited description of interventions, intervention manuals and process data would be essential to understand the complexity of ERM. There are also organisational factors that need to be considered when implementing ERM. What remains unclear is the number of organisational levels that should be targeted and how. Other issues to consider include mechanisms of effect (moderators – participants’ responses to and interactions with the intervention and mediators that affect intervention outcomes in unexpected ways). Administering staff training was a common feature we identified across studies to ensure consistency. However, nursing capacity required to deliver the intervention and family involvement need to be carefully considered. Family involvement is crucial when considering non-mobility ERM for children under 3 years old.
Evidence
We found that children under 3 years of age admitted to PICU tend to be given passive and active in-bed activities. Children 3 years and older were primarily involved in out-of-bed activities when this was considered safe. In addition, one study18 reported higher ERM incidence among male children, but this finding was not consistent across studies. Hence, it is unclear what interventions should be administered to these age groups and whether certain combinations of interventions are superior to those given to their older counterparts.
We found equivocal evidence suggesting that other factors such as time to admission, tube presence, MV, or sedation also influenced developmental or mobility goals. Only four studies13,18,24,42 evaluated baseline severity of illness (measured using validated tools) and comorbidities that affected clinical and functional outcomes in multivariate models. Consequently, the effect of residual confounding or chance on the estimated intervention effect remains unknown. Overall, the evidence about subgroup effectiveness was indicative of clinical pragmatism but otherwise inconclusive.
Comparison with the broader literature
Our findings reflect the evidence in previous systematic reviews of ERM interventions within adult and paediatric PICU.53–56 The evidence emerged from North American PICU contexts, and its transferability to the UK settings remains uncertain.38,39,46,57 Besides variation in practice, interactions within complex health systems have additional issues. As an additional complexity, due to the nature of interventions considered in this review, it is difficult to determine if intervention effects are additive, multiplicative (biologically plausible) when combined or neutralise each other and plateau.
Most outcomes of feasibility or acceptability were exploratory findings, and the strength of the evidence for objective clinical or functional outcomes remains unclear. Nevertheless, PT and OT, sometimes with SLT, were frequently implicated in the exposure-mechanism-outcome pathway. Furthermore, barriers reported reflect previous literature and support the need for activity orders.
Summary of findings to inform the paediatric early rehabilitation and mobilisation during intensive care study
- Optimal aspects of intervention delivery – timing, content, active ingredients, dose–response relationships, progression and implementation strategies – have yet to be established.
- Improvements in functional independence (though small and sometimes inconsistent) have not been matched with improved PICU-acquired weakness or survival measures. ERM appears safe but requires long-term studies.
- Standardised definitions for ERM, safety and core outcome measures will improve comparability across studies. Authors should consider the effect of an intervention on core outcome sets recommended for paediatric critical care. These include four domains: global cognition, emotional, physical and overall health. Four child-specific outcomes of health-related QoL, pain, survival and communication have also been recommended.47 Benefits so far indicate some improvement in QoL.
- ERM has been demonstrated to be feasible and acceptable within PICU. There are still uncertainties about the effectiveness of ERM interventions; only two studies used randomised designs. The evidence uncertainty is worse for objective outcomes such as PICU LOS. Overall, non-mobility rather than mobility interventions seem to be preferred in children 3 years and younger compared to their older counterparts.
- The lack of a well-defined ERM protocol is a significant barrier to ERM implementation. There is no clear evidence on the impact of bundles of care or behavioural interventions incorporated with ERM. We identified several studies that evaluated the feasibility of ERM in PICU, some of which report improved QoL as a longer-term outcome. However, the evidence for effectiveness is inconsistent, uncertain and needs further testing. Most studies were quality-improvement studies, which may be the best methodology for evaluating ERM within PICU until a better consensus on intervention components is achieved. As an alternative, nested controlled trials embedded within longitudinal studies or routine data collection can be used to evaluate the effectiveness of these interventions. Data from such studies might also enable mediation analysis to understand key intervention components and mechanisms of action.