Co-ordinated multidisciplinary intervention to reduce time to successful extubation for children on mechanical ventilation: the SANDWICH cluster stepped-wedge RCT

Characteristics of sites

The characteristics of the 18 PICUs that participated in the SANDWICH trial compared with all other PICUs in the PICANet (n = 28) are presented in Table 1. PICUs were geographically spread across England, Wales and Northern Ireland. Based on overall UK PICU characteristics, the sites participating in the SANDWICH trial were broadly representative of regions, types of hospitals, bed numbers and annual admissions.

At baseline

Baseline interviews focused on usual unit sedation and weaning practice and other relevant issues. The data collection preceded the training period to avoid potential contamination. Informed by the logic model, the interviews explored:

• current unit policy and practice regarding sedation management (e.g. staff roles and responsibilities, consistency of practice and role of ward round discussion)

• current unit policy and practice regarding ventilator weaning

• unit and hospital culture regarding change management.

These data provided important insights into local-level contextual factors that had the potential to influence trial implementation and delivery (see Appendix 2, Baseline visits interview guide).

During the intervention period

Regular informal (approximately every 6–8 weeks) telephone contact was maintained throughout the intervention phase with the unit research teams (research nurses and/or PIs) concerning the trial progress and issues arising. Conversations were guided, but participants were encouraged to raise any issues that they considered relevant. The data from these telephone discussions provided valuable stand-alone information about ongoing trial delivery, including adherence to the protocol. They were also used to identify unit-specific issues for follow-up during subsequent conversations and end-of-trial interviews with clinical staff.

End of the trial

On-site semistructured individual interviews and focus groups were held to discuss staff understanding and experience of trial implementation and delivery. Informed by the logic model, the interviews explored:

• staff receipt of the intervention in relation to both the protocol and the supporting resources, such as the training provided

• fidelity of intervention delivery

• impact of contextual factors on trial implementation and delivery (barriers to and facilitators of, such as extent/nature of multidisciplinary working on the unit, workload pressures and unit routines of working)

• wider impact of trial participation.

The participants were encouraged to consider both positive and negative aspects of trial implementation and delivery, including the nature of the intervention. To avoid potential contamination, all data collection preceded knowledge of trial outcomes (see Appendix 2, End of trial interview guide).

Participants

Participants were the participating PICU staff. They included the PI, the dedicated SANDWICH research nurse, other relevant research nurse support, the SANDWICH champions and a range of professional grades for medical, nursing and physiotherapy staff. Staff were excluded if they had not worked clinically to experience using the SANDWICH intervention.

Recruitment and consent

The selection and recruitment of participants was organised within the PICU in consultation with the SANDWICH nurse or PI. Participants were purposively selected from a convenience sample of staff who were available on the days that the researcher was visiting the unit. Recruitment entailed an initial e-mail from the researcher, comprising an introduction and the participant information sheet outlining the purpose of the interview, information on likely areas of discussion and details of participation (see Appendix 2). This was distributed to staff who were on rota on the days of the visits. Prior to the interview, participants were given an opportunity to read the participant information sheet again and ask questions. Once satisfied that informed consent had been completed, participants were asked to sign the consent form (see Appendix 2).

Data analysis

A framework-based approach was used to analyse the quantitative and qualitative data. This included a strategic focus on the four key components of the intervention and the resources, particularly the training, that were provided to support its delivery. Subsequently, patterns and trends (including exceptions) characterising the factors and processes that affected the receipt, delivery and impact of the intervention components and resources were identified. These two requirements meant that the analysis included both deductive and inductive components.

Deductively, we used the a priori identified categories (key intervention components). Adherence to these components was measured by (1) the proportion of the four intervention components performed, captured daily in the study CRF; (2) the proportion of staff trained by 8 weeks; and (3) the intervention reach, measured by the proportion of admissions screened (recruits and exclusions) divided by IMV admissions over the trial period. The average adherence proportions for each PICU were ranked and divided into tertiles and a prespecified subgroup analysis investigated the impact of adherence on the primary outcome.

Inductively, we used Braun and Clarke’s52 thematic content analysis to generate a set of themes that cross-cut the entirety of the qualitative data set. 52 (Box 2).

Individual and focus group interview data were treated as comparable, an approach regularly adopted in empirical research. 53,54 The primary responsibility for qualitative analysis lay with Joanne Jordan. To ensure confirmability and trustworthiness, a 15% sample of the data was analysed by an independent qualitative researcher (Dr Claire Kydonaki, Edinburgh Napier University) to identify key differences and similarities in pursuit of an agreed final analysis. 55

Sedation management using COMFORT scoring

Although a few participants talked about limited adherence to COMFORT scoring, descriptions suggested at least reasonable, sometimes good, adherence. Overall, compliance was considered to have improved over time as bedside nurses became increasingly familiar with the use of COMFORT scoring.

Overall, bedside nurses described being able to easily incorporate COMFORT scoring into their routine patient assessments; it tended to become part and parcel of standard practice. Any lack of adherence was associated with two main issues: first, a vulnerability to the prioritisation of other caring responsibilities, so that COMFORT scoring was simply overlooked; and, second, a more fundamental disinclination, so that, even when a bedside nurse was looking after a relatively well patient, COMFORT scoring was not carried out and/or acted on. At times, such failure was associated with a preference for patients to be kept relatively heavily sedated as a means of avoiding risk:

For me, it wasn’t an issue. However, there may have been times on very busy days that I . . . maybe didn’t do it as frequent as I ought to, but as the time went by I think I made a bigger effort to try and fulfil the requirements . . .

Bedside nurse, S01

At times, a perceived lack of practical application undermined the adherence of bedside nurses to COMFORT scoring. Some nurse participants talked about lacking motivation because their efforts failed to have any direct impact on patient care, whereas others could see that they were directly relevant to immediate/ongoing clinical decision-making, including decisions regarding ward round discussions or a SBT screen/SBT. In this context, the effectiveness of the COMFORT scoring component of the protocol could be highlighted, as this aided timely extubation:

It wasn’t pointless necessarily, but we didn’t do anything with it either. We had to write a lot to explain one of the three numbers you’d chosen, and then no one would ever look at it again.

Bedside nurse, S15

I think when the doctors started asking for it on ward round . . . that was the big thing. I think the nurses before were . . . writing the number down and nobody actually cared what this number is for. I think for doing it for the study and the doctors being involved and everyone discussing it basically gave it a sense of worth and actually the nurses then bought in . . .

Senior nurse, S11

Although most of the participants considered COMFORT scoring to have facilitated weaning and extubation, the traffic light colour-coding scheme was, at times, identified as problematic. For example, the ‘green’ category ranged between 12 and 17, but allowed for varying levels of alertness. The lower end of the range (a score of 12) was, on occasion, described as problematic in that patients could still be overly sedated, thus increasing the possibility of failing a SBT or not being suitable for extubation. In this context, several participants commented on what they considered to be a ‘disconnect’ between the COMFORT score and the SBT screening components of the protocol:

. . . I came across a few cases where the child . . . may have been in the green zone but was still sedated, and then nurses were doing SBTs and then saying that they’d failed the SBT, but actually I wouldn’t have actually done an SBT on the amount of sedation the child was on.

Consultant, S07

Ward round weaning discussion/planning

Overall, descriptions of ward round discussions and the associated development of patient sedation and weaning plans highlighted inconsistent adherence within and between units. Three main factors emerged. First, whether or not individual members of staff, particularly those leading the ward round, ensured the discussion of sedation and ventilator plans. Second, the clinical status of patients, in that a discussion was more likely to occur if a patient was considered to be approaching readiness for extubation. Third, the pressure of time, in that large amounts of information had to be shared within very tight time frames. At times, no specific reasons could be identified for the absence of discussion:

The ward round checklist, when it worked, it worked really well, and that was the part that was missing all these years, but that was the part that was probably the hardest to actually implement because there wasn’t consistency . . . but I don’t feel like that that was mirrored from a consultant point of view . . . in their ownership . . . because if they didn’t properly own it then the registrars didn’t get on board.

Senior nurse, S10

And especially on ward round . . . SANDWICH and COMFORT was always mentioned, it always came up and you always had to report back on it. What’s your SBT and what’s the COMFORT scores and what about SANDWICH and the documentation was something that was paid close attention to, it was something that was religiously done.

Bedside nurse, S02

Spontaneous breathing trial screen

A consensus among nursing staff participants across all units was that the SBT screening criteria were easy to understand and the actual screen was simple and quick to complete. Ease of use was considered an important factor in promoting adherence over time:

It doesn’t take very long . . . it’s just a quick tick-box. So I don’t think it was a burden . . . So yeah, I think it was good.

Bedside nurse, S09

Participant accounts showed that all units had set times for SBT screens to be completed. Some units timetabled twice-daily SBT screens (typically one in the early morning and one in the afternoon/evening) and some had three- or four-daily SBT screens scheduled over a 24-hour period. Across all units, the stated rationale underpinning the selected times for SBT screens was the provision of relevant information to fit with the unit and wider hospital routines of patient care, assessment, decision-making and extubation.

Varying levels of adherence to the performance of SBT screens were described. Overall, the highest compliance was associated with the early morning SBT screen, which was typically undertaken towards the end of the night shift. Two reasons were identified. First, night staff were considered to have the requisite time available. Second, more materially, it ensured that information was available to inform the morning ward round discussion, either stand alone or in combination with that provided by a follow-up SBT. The morning ward round was regularly described as the primary forum of important decision-making regarding plans for individual patient care, including in terms of ventilator weaning and extubation:

. . . the 5 o’clock [a.m.] one . . . so by the time it’s dayshift, the nightshift have got them ready to extubate and the dayshift don’t have to do anything apart from wait for . . . usually around ward round. So at 10 o’clock [a.m.] is generally when they’d start to think about it, because ward round is done, they’ve been starved (stopped feeds) for 4 hours or so . . .

Bedside nurse, S13

The afternoon/evening SBT screens were described as less regularly undertaken. Several reasons were identified. First, because the patient may have failed a morning SBT screen or subsequent SBT, and nursing staff were instructed not to repeat either procedure until the following morning. Second, because of the need to prioritise other aspects of patient care. More fundamentally, accounts suggested a diminished impetus to undertake afternoon/evening SBT screens, on the grounds that they were much less likely to lead to SBTs/extubation. Thus, the function of the SBT screens performed later in the day/evening was seen more in terms of providing staff with insight into patient progress towards extubation and, when appropriate, allowing for the planning of such. For example, they could provide useful guidance on how likely the patient was to wean from the ventilator the following day, so that sedation could be appropriately managed overnight:

The 4 p.m., we’d always know right, OK, so overnight we’ll wean for a planned extubation tomorrow.

Bedside nurse, S06

. . . but probably less so if it’s a bit of a surprise that the patient passes the SBT screen in the afternoon. I think a lot of people are then more likely to say well, we’ll just give them another 12 hours and do it in the morning.

Research nurse, S18

More generally, not carrying out SBT screening was associated with staff awareness of specific patients being clinically inappropriate for extubation. This scenario represents one of the most frequently identified departures from the protocol, namely the non-performance of tasks for which unit staff could see no obvious benefit. An awareness of the adverse impact of such behaviour on trial delivery was often accompanied by an acknowledgement that selective performance of the SBT screens was a perfectly rationale response in the circumstances, especially when an already heavy workload was taken into account. In one unit, the subsequent measures taken by the research team to reduce such ‘unnecessary’ workload were welcomed as making the trial a more practical and relevant endeavour:

So that [SBT screen] was a bit that was less well performed. Sometimes the assessments would get forgotten, particularly if we knew that we weren’t thinking of extubating them, so they were being neuro protected or they were oscillated or whatever, they would miss out the assessment. And you’d say you’ve still got to do the assessment and say they’re not ready for an SBT. So that would tend to be less well done.

Consultant, S16

To start off with, there was some kids who were on long-term weaning plans who weren’t being planned to extubate . . . we were doing it four times a day, but then . . . the research team came with stickers and every patient who was identified as a slow wean, we only had to do it once a shift, and I think doing it once a shift wasn’t an issue.

Bedside nurse, S03

Spontaneous breathing trials

Overall, from the descriptions given by participants, the performance of SBTs and extubations following a successful SBT was inconsistent across and within units. Although some units appeared to fare better than others, of all components of the protocol it was SBTs that were talked about most often as not being undertaken, or undertaken without strict adherence to the protocol. Typically, SBTs were meant to follow as soon as possible after a successful SBT screen. As per descriptions of the vulnerability of scheduled afternoon/evening SBT screens, participants highlighted the same for SBTs. Several reasons were identified. First, participants described non-performance of scheduled SBTs owing to lack of fit with unit routines and priorities of care. Established practice led to the majority of extubations being carried out in the early part of the day. Consequently, there was a reduced impetus to carry out afternoon/evening SBTs because they were seen as less likely to have an immediate impact on patient care. Second, hospital-wide routine practice means that afternoons are customarily given over to other aspects/priorities of care, such as off-unit tests and investigations. Consequently, patients and/or unit staff could be engaged in these other care processes. Third, staff accounts emphasised a drive to make the most efficient use of staff time and other resources, so that when a SBT was thought to be superfluous it could be missed:

. . . because we go around at night and I actually did then find myself saying ‘and can I say that there’s no point in this child having an SBT tonight’, but it had taken me a few experiences to realise that I had to say it.

Consultant, S07

. . . but it almost felt a bit pointless if they had failed just a few hours ago and not much else has changed in the interim.

Bedside nurse, S15

Consequently, and again in line with SBT screens, participant accounts across all units confirmed that SBTs scheduled for the (early) morning were both more consistently performed and, when performed, more likely to lead to patient extubation. The reasons identified for this all emphasised the importance of the fit of the morning SBT with unit resources (primarily staffing) and established routines of patient care. This is not to say that SBTs were not carried out later in the day. For example, participants talked about SBTs/extubation being performed in the afternoon following morning SBTs at which patients had been assessed as not quite ready for extubation, on the understanding that by the afternoon they would pass the SBT and be extubated. More generally, participants also talked about the importance of the afternoon SBT in promoting extubation:

I don’t think it influences the 9 o’clock decision very much, but what it does do is it allows us to think, ‘hang on a minute, we’re not ready for 9 o’clock, let’s try the 1 o’clock one,’ instead of saying ‘well let’s see what happens’ and then you leave it and then it comes 5 o’clock and you think ‘oh, 9 o’clock/10 o’clock tomorrow’. So I think definitely the ones that you may not have pushed to get extubated between 4 o’clock or so, the 1 o’clock assessment thing is a difference.

PI, S17

A second factor adversely affecting adherence to SBT performance concerned a straightforward impulse to follow preferred clinical practice. For the most part, this issue was discussed in relation to medical staff, but, on occasions, bedside nurses were implicated. Three main forms of non-adherence were described: non-performance of a SBT following a successful SBT screen, performance of a SBT but in ways that deviated from the protocol and non-performance of an extubation following a successful SBT:

. . . well some of the consultants come round first thing in the morning, but they’d just change the settings, they wouldn’t necessarily have looked at that chart [for SBT screen result] . . . So they went round and put their rate down or put them on to CPAP [continuous positive airway pressure] anyway, regardless of what the score was . . .

Bedside nurse, S02

. . . I think our overall position, as a unit and the nursing staff, was they felt more comfortable waiting until the patient was on less than that before going for a spontaneous extubation.

Consultant, S04

Regarding the performance of a SBT in ways that deviated from the protocol, the main manifestation concerned the use of settings other than ‘5 and 5’ (PEEP 5 cm H₂O, Psupp 5 cm ₂O above PEEP). Several contributory factors were identified. First, patient clinical status and/or other characteristics (typically age and size) were identified as contributory factors. One of the most frequently cited cases was neonates, for whom these settings were considered inappropriate. Second, another factor identified was medical staff avoiding the use of the ‘5 and 5’ settings when this involved a significant drop from the ventilator support that patients had been on during their successful SBT screen. Participants also talked about, and articulated support for, the use of higher settings for patients who, for clinical reasons, were planned to be weaned onto non-invasive support. Finally, another factor identified was the persistence of medical staff in the use of settings other than ‘5 and 5’ for which no obvious reason was apparent:

A lot of people wouldn’t [even] do the 10 on 5 either. They wanted higher CPAP [continuous positive airway pressure]. I didn’t really understand why we wouldn’t at least try the 10 on 5 and then if we fail then we could maybe try 12 on 6 or whatever. There was no even attempt at that with some of them.

Senior nurse, S18

We did implement a lot of modified SBTs . . . 7 on 7 or, you know, and that was considered an SBT, do you know what I mean, where it wasn’t your exact criteria.

Registrar, S02

For me, it was just conceptual, so I didn’t really mind if it was 5 or 6. If someone left it on 6 I think I even allowed it a few times and just said ‘well, effectively we did it’.

Consultant, S03

A second form of deviation concerned the performance of SBTs that were ended prematurely. Essentially, the reported issue was the reluctance of staff to allow a patient time to overcome any initial difficulties that could have potentially settled. Frequently, this reluctance was traced back to inexperience on the part of staff involved:

. . . sometimes just the adjustment change from going from BIPAP [bilevel positive airway pressure] to CPAP [continuous positive airway pressure], the baby is going to have to get used to that. So you might get an increase work in breathing initially and then it will calm down, but some of the juniors would be like oh, my God, their rate has gone up to 80, and they would stop it . . . they’re [the bedside nurse] scared that they’re going to fail and they don’t know how to cope with it because they’re new . . .

Senior nurse, S16

I think the biggest problem we have here with mode of ventilation is that people leave the apnoea alarms very low on spontaneous mode, particularly the babies, they would get 15 seconds before they’re having an apnoea, and then they’re failing, when actually, if you put it to a realistic amount, you can allow them to pass.

Physiotherapist, S18

Non-performance or delayed performance of an extubation following a successful SBT was highlighted by participants across all units. Doctors were sometimes described as failing to articulate their reasoning in any detail, which caused the nurses additional workload and some frustration. In such cases, none of the reasons (codes) on the trial documentation was relevant, leaving the bedside nurse with no option but to use the ‘other’ code. The accounts of some participants also highlighted non-performance of, or delays to, extubation following a successful SBT in cases in which trial exclusion criteria did not apply. In addition, on a limited number of occasions, consultant participants associated delayed or non-performance of SBTs with a failure on the part of bedside nurses to inform them that a patient had passed the SBT screen. One consultant stressed the responsibility of nursing staff to remind medical colleagues of the need to undertake trial-related activity:

And sometimes doctors are like ‘oh, they’re not ready’, and you’re like ‘give us a better reason’ and then you’ve got to put down ‘doctors have said not ready’. That’s not a proper reason.

Bedside nurse, S15

Although in my experience the nurse doesn’t tell you the child has passed the SBT . . . No, you have to ask them . . . They’re supposed to do it night and day, night and day, so they know . . . Still we have to remind them.

Consultants, S05

Finally, although rarely acknowledged explicitly, some participant accounts suggested that SBTs could be performed in circumstances in which a decision to extubate a patient had already been made. In many cases, this decision would be based on a process of weaning the patient down to ‘extubated-able’ settings and then carrying out a SBT as a final check or justification. This meant that, rather than being used to potentially expedite extubation, the SBT was being ‘shoehorned’ into existing unit practice:

And, to be honest, most of the times we would have decided overnight whether we’ll be extubating or not, so we would just go ahead and do an SBT and that would be sort of extra information, or it would give us more confidence that child has passed an SBT and child is ready for extubation. As I said, the decision would be done the night before . . . So most of the decision will be done on the night before and it would be just sort of an extra precautionary exercise.

Registrar, S06

It’s used as an extubation predictor. So when we think the child is ready to extubate, we do the SBT, not when we’re going, ‘I’m sure this kid could be weaned a bit more, can we do an SBT?’.

Consultant, S08

The option of running a SBT for up to 2 hours was discussed as problematic on three main fronts. First, a lack of explicit direction regarding the action to be taken at the end of a 2-hour SBT meant that staff could be left confused, including in respect of how/when to record the ‘ending’ of the SBT. In the absence of extubation, a number of scenarios were described: typically a patient remained on the SBT settings (sometimes for extended periods), was returned to the pre-SBT settings or returned to reduced support. Second, it allowed the focus of attention to move away from the patient’s weaning progress as other priorities of patient care took precedence, or staff simply forgot about the SBT:

. . . but I think we had lots of kids that were on for far more than the 30 minutes and they weren’t extubated at 2 hours . . . I think it was an acceptance of OK, we can extubate this child, we’ll do it whenever we’re ready to do it, as opposed to actually they’ve passed the 30 minutes and we should try and take the tube out by 2 hours or within the next few hours. That wasn’t the case. It was ‘we know they’ve met it; we know we can take that tube out; we’ll take the tube out when it suits us’.

Physiotherapist, S03

Third, on a few occasions, although participants acknowledged the need for the protocol to be sufficiently flexible to take account of unit routines and (changing) priorities, they also highlighted its failure to stipulate a requirement for extubation to be performed immediately following a successful SBT:

. . . there was no closed loop feedback back to the medic or the ANP to say ‘you know we’ve been SBT 5 on 5 for the last 2 hours, why are we not extubating this child?’ So it’s . . . like ‘oh, I forgot’ . . .

PI, S03

Enhanced (bedside) nurse understanding of weaning

Across all units, there was widespread endorsement of the trial as enhancing nurse, particularly bedside nurse, understanding of the nature and process of ventilator weaning. As part of this process, bedside nurses saw themselves as having an improved awareness of relevant clinical indicators and what they mean, and what the appropriate response/management should be. Enhanced nurse understanding was echoed by other members of unit staff. More experienced staff tended to associate the learning achieved by less experienced nurses with several positive outcomes. These included the fact that this learning was an important dimension of continuing professional development and that it meant a greater likelihood of bedside nurses involving themselves, and being involved by others, in the weaning process:

It’s definitely for the junior ones, that now they know which criteria you use to decide . . . when you extubate . . . when we started SANDWICH, they didn’t have [that] knowledge . . . they [can] go and escalate that to the doctor and say ‘oh my patient is actually on setting . . .’

Research nurse, S15

Enhanced nursing confidence/autonomy

In line with an awareness of enhanced nurse understanding of the weaning process, the trial was widely associated with improved confidence and autonomy. Bedside nurses were described by colleagues and by themselves as more confident to pursue autonomous practice and to engage in meaningful discussion about this practice with others, especially senior colleagues. Throughout, the crucial factor identified was the shared understanding provided by the protocol. Enhanced confidence was seen as particularly beneficial for more junior/inexperienced nurses, who were considered more likely to be reluctant to raise issues with medical staff. In some accounts, a ‘positive loop’ was identified, whereby increased confidence provided by the trial documentation encouraged junior nurses to initiate discussion with medical staff. When these discussions were positive, subsequent discussion was encouraged:

But I can honestly say that I don’t find any downside to SANDWICH, because I think of the positive things about how we work, the nurses’ mindset is now a part of the working narrative. I like the way they escalate their concerns.

Ward manager, S16

And I think it does. It has empowered the junior staff, it helps them in the ward round, they can talk about the COMFORT score, they can explain where they’re at on the screening tool and I think it has really helped with the junior staff, feeling a bit of self-worth.

Senior nurse, S11

. . . it gave you a bit more backup than just saying I don’t think they’re adequately . . . It’s made it easier to have that objective overview between various members of the team.

Bedside nurse, S13

In most units, the SBT screen gave bedside nurses a designated role in ventilator weaning for the first time. The descriptions of enhanced confidence focused on the SBT screening criteria providing bedside nurses with ‘objective’ evidence regarding a patient’s clinical status and readiness to be weaned, which bedside nurses could use to support multidisciplinary discussions. In addition, the fact that the protocol required the medical team to make their reasoning explicit with regard to the non-execution of a SBT and/or extubation gave bedside nurses the authority to seek explanations:

I don’t know if it’s just now that I’m more experienced, whereas I was quite new before it started, but I definitely would think more about whether they could be extubated, whereas before it was just the doctor’s job and I didn’t really think about it at all, whereas now it makes me think I might bring up a conversation whereas I never would have started a conversation.

Bedside nurse, S05

A similar, but more understated, enhanced confidence was apparent among more senior nursing staff, according to participant accounts. In one unit, senior nurses highlighted the potential of SANDWICH to formalise/cement their greater involvement in patient weaning as part of standard unit practice. In this context, they described themselves as the frequent ‘drivers’ of weaning, but not the technical implementers. Based on the role that they were assigned as part of SANDWICH, namely that they could formally commence a patient on a SBT, there was at least the potential for this to continue. Such a role was seen as representing an important advance in terms of nurse involvement in weaning:

We aren’t the ones that fiddle with the buttons, but I do think we’re quite often the drivers of weaning, but we may not be the ones that are able to actually alter the ventilator . . . For this [SANDWICH] we were able to do an SBT.

Senior nurse, S06

In general, medical staff participants acknowledged that the trial had, to a greater or lesser degree, enhanced nursing confidence and autonomy of practice. This was regarded as a positive development. That said, reluctance on the part of nursing staff, particularly less experienced nurses, to assume greater responsibility was noted as an important obstacle. More widely, some participants described nurses, particularly junior nurses, as inherently reserved in their interactions with senior, typically medical, colleagues. Several bedside nurses also commented on their own reticence. Failure on the part of nurses to question medical colleagues was seen as an impediment to trial implementation, in that it could result in the perpetuation of deviations from the protocol:

. . . involve them in the decision process and to make them proactive in the raising that children were ready, and this has not happened . . . Not really, because I just keep on asking them ‘have you done the . . .?’. So you had to prompt them, so if I was to go and say ‘have you done the bedside checklist?’, and they’re like ‘oh no, I’ll do it now. Passed! Great, what shall we do now?’. So it had to be very led by us.

PI, S04

Enhanced multidisciplinary communication/collaboration

There was widespread awareness of trial participation having improved multidisciplinary communication and collaboration among unit staff. The crux of this improvement centred on the requirement to discuss weaning plans and the shared language provided by the trial to aid these discussions. Nursing staff talked about how they could use this language to articulate their understanding of individual patient sedation weaning progress and/or requirements. They described enhanced communication from doctors regarding their intentions and actions concerning patients. For some participants, a key forum in which enhanced multidisciplinary communication played out was the daily handover/ward round:

I do like [that] they listen when we say they passed it or they failed it, they do listen to that decision. They don’t usually go ‘oh, well, let’s take the tube out anyway’.

Bedside nurse, S15

. . . especially in handover when the doctors are talking about weaning the pressures, whereas sometimes they wouldn’t even tell you, they would just come and wean without you knowing sometimes, I think. They’ve changed yes. So I think sometimes having that conversation if they have failed the SBT . . .

Bedside nurse, S18

Other members of staff tended to endorse the experiences and understandings of nursing staff outlined above. For example, physiotherapists talked about the trial as having facilitated a greater role for their own team because they learnt more about how to maintain appropriate sedation and/or were able to directly influence patient weaning plans based on the use of trial criteria and language. Medical staff, typically consultant participants, were also aware that SANDWICH had provided a shared language of weaning and, through this, enhanced staff communication and collaboration. At times, their reflections were couched in a strategic context of their vision for unit multidisciplinary working. At other times, accounts focused on an awareness that multidisciplinary communication had been improved precisely because staff were required to reflect on their clinical decision-making and to account for it to others:

. . . because I thought the SANDWICH trial was absolutely brilliant, it really helped inform us of what was going on, we felt we were more part of the team as well . . . I think it was just the fact that we could immediately see, because we learnt more about the sedation scoring and what have you . . . but suddenly we had that information, we’d learnt about it a little bit . . .

Physiotherapist, S06

I used to have 10 conversations in a day about the sedation, it used to do my head in . . . and those conversations just went away, people just did the score . . . But what I found when we were doing the trial is the nurses were doing it because they had to, and then instead of coming to me and saying to me ‘can I do this, can I go up, can I go down’, whatever, they would just do it, and that made a massive difference . . .

Consultant, S11

So having common words that we can use, SBT readiness, is really helpful and it makes communication more effective, efficient. So, I found it positive.

Registrar, S13

Enhanced sedation management

One of the most frequently identified positive outcomes of the trial was the increased interest in, and attention paid to, patient sedation. From their diverse roles and responsibilities, unit staff talked about this process in different ways. Underlying all of their accounts was an acknowledgement that the use of COMFORT scores and related target setting in ward round discussion had, to a greater or lesser degree, improved unit practice. This does not mean to say that problems were not identified with both the COMFORT scoring tool itself and the implementation of the component of which it was a part.

Nurses’ (typically bedside and research) accounts tended to emphasise that the ‘point’ of COMFORT scoring, including its application in sedation management, had been made much clearer through participation in the trial. Implicit in some accounts was the idea that, prior to SANDWICH, COMFORT scoring was being carried out in the absence of a full comprehension of its value and purpose. One aspect of the COMFORT scoring system that garnered widespread endorsement was the ability to assess pain. This was seen as enabling a more comprehensive consideration of all potential factors having an impact on how settled or comfortable a patient could appear:

I think the COMFORT scoring itself was fantastic, and we’d do pain scoring at the same time, so we could then compare it to see whether the patient was adequately sedated or whether they needed any analgesics.

Research nurse, S14

In terms of unit practice around sedation management, COMFORT scoring was regularly described as having made several material improvements. First, it had encouraged a greater focus on and proactivity concerning patient sedation. Second, it had encouraged clarity and consistency by providing a shared multidisciplinary language (e.g. COMFORT scores and associated target setting) with which to communicate and/or act in relation to sedation management. For both of these reasons, although problems continued to be identified in all units to a greater or lesser extent, a sense of moving away from keeping patients ‘overly’ sedated was evident:

On the sedation part, I thought that was quite helpful in focusing the whole team’s approach to sedation. I thought that the groupings, if you like, and the colour coding and being able to discuss if they were in the green, for example . . . it did improve the focus everybody, but particularly the bedside nurse on sedation, and give them some additional empowerment to change things.

Consultant, S07

Enhanced weaning practice and related patient care

To a greater or lesser degree, participants in all units talked about the implementation of the trial as having contributed to enhanced weaning knowledge, practice and related patient care. Several relevant processes were identified. First, staff were repeatedly prompted to consider and/or pursue weaning, which encouraged an increased pace that would otherwise be missing. Typically, the core issue was one of having encouraged a move away from a gradual ‘step-down’ approach to one in which patient could be moved forward more dynamically. Even in cases in which, for whatever reason, patients failed a SBT, weaning was understood to have been progressed in other ways:

But what we found was if they failed the SBT, instead of going back to the pressures they were on, we went back halfway up, so even that helped. OK, maybe they’re not ready to go down from 20 to 5, but OK, they want that back, take it up to 16.

PI, S18

Overall, I think it’s been really, really good. It’s changed the mindsets, it’s changed the language, it’s allowed people to positively challenge the way we work in a ‘let’s try something new approach’.

Yes, because it made people question and think, as opposed to just going along.

PICU senior nurses, S03

A second process raised was the enhanced consistency and clarity of weaning practice. The protocol was described as promoting a more structured, systematic and standardised approach, which was based on shared goals and understanding how these goals should be reached. This standardisation was discussed as allowing members of staff to advance weaning according to their roles and responsibilities. This was particularly the case for bedside nursing staff, who described themselves, and were described by colleagues, as proactively advancing weaning. Systemisation was also considered to have helped remove at least some of the inconsistency introduced into the weaning process by the different approaches and preferences of individual members of staff (particularly consultants). In this context, some consultant participants talked about the protocol as removing some of the uncertainty of weaning decision-making. The more structured approach was further associated with improved patient weaning, in that it encouraged staff to plan ahead. Participants talked about being prompted to consider, and prepare for, what might come next. They also talked about a similar planning ahead by their colleagues:

But I think SANDWICH is actually good in that way for extubation, because before that it was all consultant based, if this is the consultant that was on, you would know whether the child is going to or not going to, but SANDWICH has made that more consistent.

Bedside nurse, S02

. . . because it gave everybody a structure on what to do when you come to extubation, isn’t it? Instead of waiting for the consultant to sort of say the magic word, it was all sort of lined up to be done.

PI, S17

And also being a bit more prepared as well. If you know that your patient has passed the SBT and we’re most likely going to extubate, then I might think OK, let me go and get a non-invasive ventilator or some oxygen or whatever so you’re prepared . . .

Bedside nurse, S15

At times, participants linked the protocol with improved patient care beyond the immediate context of ventilator weaning. For example, time spent unnecessarily on ‘nil-by-mouth’ was seen as having been reduced, in that this could be stopped as soon as a patient failed a SBT, rather than having to wait until extubation was attempted. From a physiotherapist perspective, participants talked about improved rehabilitation care. For example, the fact that the trial had encouraged a more ‘light touch’ approach to sedation meant that physiotherapy interventions, such as chest rehabilitation, could not only be initiated earlier, but also were more likely to be effective:

I like that our patients are much lighter on sedation . . . it means, as physios, we have to show that education on how we can still treat patients even if they are awake, we just treat them slightly differently.

Physiotherapist, S16

Contribution to (the development of a) unit research culture

When discussed, most participants considered that participation in SANDWICH had made a positive contribution to the unit’s research culture and capacity. This could include, for example, the development of an appropriate infrastructure, growing a (national) research profile, increasing research interest among unit staff and/or promoting the skills and/or team identity of the research team both within the unit and more widely. All were discussed as having promoted the unit’s practical ability to engage in research. At times, an underlying interest in/support of research among senior members of staff was considered to have helped trial implementation, including in terms of its encouragement of other members of staff.

In one such unit, building on SANDWICH, the unit team had been able to secure additional funding from the hospital trust and local clinical research network to continue to fund research nurses and, thus, to participate in other major studies. In another unit, involvement in SANDWICH had helped the research team in their discussions with hospital management concerning the need for the team to be based on substantive posts, rather than the current fixed-term contracts, which were considered unconducive to strategic research planning and implementation:

And people valuing research as well, like understanding that’s a way of life now and we have to sort of get on board with it. We’ve been doing a lot of projects, which has been good as well.

Senior nurse, S11

So this is another thing that we’ve had a discussion with, with senior hospital management, just about the lack of substantive contracts . . . you’ve got research staff . . . they want to stay but the funding’s not there for them to stay or they don’t actually have a substantive post . . . and then you go back . . . Based on how heavy research can be for us . . . it’s part of the vision of the hospital. Paediatric intensive care has a lot of research going through it, so if that’s the case why don’t we have more permanent [staff].

PICU research nurse, S03

Support from the study research team

Overall, the SANDWICH research team was described as having provided effective support throughout the trial. Of perceived central importance were the benefits derived from ongoing communication between the trial and the unit research teams. Here, the trial implementation manager was consistently acknowledged to have delivered not only well-organised, comprehensive training, but also prompt, well-informed answers to queries arising throughout the entirety of the trial. As a single individual with responsibility for managing a very large study, a substantial workload burden was recognised. In addition, the monthly PI teleconferences, attended by unit PIs and research staff, were valued for allowing unit-specific problems and/or queries to be addressed, and unit-to-unit learning to take place. In this context, a research nurse from one unit balanced the perceived disadvantage of being randomised to the intervention phase relatively late in the trial with the learning gained from the experiences of others during the PI teleconferences, which helped pre-empt and/or manage issues arising. Finally, the material resources provided, including, for example, the COMFORT scoring chart, were frequently described as being of considerable practical utility, and the sundry items, such as SANDWICH pens and lanyards, were understood to have encouraged a positive response to the trial:

I think the availability and the visibility, [name of implementation manager] being there, was useful certainly, it raised the profile, and also having the little notepads and the pens and the posters, sort of like a publicity drive, it put that on people’s agenda . . . So I think all those things were helpful.

PI, S18

I can certainly compare it to a few other national studies, the support is so much better in SANDWICH, I’ll definitely say.

Consultant, S15

Intervention training

Overall, the training was described as relevant and comprehensive and, for this reason, to have been important in enabling staff to understand the purpose and content of the protocol and to discharge their roles and responsibilities. The vast majority of participants considered that, in the absence of the training, trial implementation would have been severely compromised. A limited number of participants considered the training not to have been entirely necessary to trial implementation. Significantly, they tended to be more experienced/senior nurses. In a number of units, participants talked about either using or hoping to be able to use components of the training as part of future in-house unit training and/or current guidance. Notwithstanding an acknowledgement of the need for and benefits of the training, it was frequently discussed as intensive and time-consuming. The online component in particular attracted some criticism over the length of time it took to complete:

I didn’t find it too bad, personally. I know a lot of people were saying it was long winded but I didn’t find it too bad.

Bedside nurse, S17

I think we were all very well informed about SANDWICH and how it would work, I thought the training package was very good, so I felt very well informed.

Physiotherapist, S11

Of the two delivery formats, most participants favoured face-to-face training as more enjoyable and effective in terms of delivering knowledge. Clinical and research nurses talked about the training provided at the bedside and/or in small groups as particularly helpful owing to its practical nature and focus, as well as allowing more meaningful engagement with what was being taught, especially because participants were able to ask questions and resolve emerging issues. By contrast, the online component could be described as lacking an immediate ‘real-world’ relevance. The same preference for/appreciation of the practical benefits of training is reflected in participants’ comments regarding the value of the training becoming apparent after the trial was fully implemented. A few participants considered the online training to be more beneficial than that provided face to face. Again, such individuals tended to be more senior/experienced nurses:

. . . definitely face to face was far more successful in terms of how people understood the process.

Bedside nurse, S13

Not particularly useful or necessary [online training] . . . I think probably perhaps just from a very personal point of view, I’m more of a doer. So I have to keep doing it to understand it.

Bedside nurse, S04

In some units, ensuring that the requisite numbers of staff were trained in the time provided was described as difficult, sometimes extremely so. This tended to be the case in larger units. At times, special measures were necessary to ensure completion. These included research nurses relieving bedside nurses so that they could complete the computer-based component during their shift. Where the research nurses had the support of key colleagues, for example unit managers who facilitated the completion of online training during working hours, meeting the required staff training quotas in the allotted time was considered to have been much easier:

It just got ‘we need to get you all done! I’ll have your patient, just go and do it’.

Research nurse, S07

. . . it was made easier, because the matron was lovely . . . we were able to persuade her and she agreed to add the SANDWICH training into the induction for this ward, which was mandatory and which she assigned staff hours to, which meant that we had groups of 10 to 15 people that were sitting in a room and that for half an hour, 45 minutes we could teach SANDWICH to.

Research nurse, S09

Overall, the medical participants (typically consultants) who discussed the training materials considered them to have been well structured and informative, and to have enabled a necessary awareness and understanding of the intervention, especially given its complexity and spread. At times, some queries were raised in relation to two issues, namely the length of time taken to complete the (especially online) training and some of the relevance of the content in the context of participants’ medical training and knowledge:

I think, like with everything, it’s very hard to make it specific for certain people, and there were bits of it that I didn’t need but I understood why it was as it was, because without going into everybody’s background how do you know what they do and don’t understand.

Consultant, S02

Trial ‘champions’

Overall, the appointment of designated champions was thought to have facilitated trial implementation. That said, some difference of opinion emerged concerning the perceived extent of their contribution. These differences were associated with three main issues: the commitment of the particular individuals appointed; the availability of the individuals on the unit; and, irrespective of their individual commitment and/or availability, other unit pressures. For the research nurses, the support of champions in getting the requisite numbers of unit staff trained in the time available was of considerable importance, particularly in the larger units. For clinical staff, especially bedside nurses, the value of champions was associated primarily with their routine presence on the unit, meaning that relevant support and advice were available as and when required. In this context, they were seen as particularly effective at times when research nurses were unavailable. To a lesser degree, the value of the champions was located in their physical presence acting as an automatic prompt to consider the SANDWICH intervention.

In terms of individual commitment, some champions were described as expending considerable time and effort to ensure that staff understood and adhered to the protocol. Others were described as playing a ‘token’ role only. At times, the importance of the role played by champions in the early stages of the trial was highlighted, as they were able to answer initial queries and correct misinterpretations, often immediately. Precisely because of the need for availability, a few participants suggested that in the absence of a fixed presence of champions or research nurses on the unit floor, deficits in unit staff knowledge of, and adherence to, the protocol were more likely to persist:

My champions were great, they did so many of them [training] . . . they really picked up the slack, and so did [name of PI] and all our medics. The training, I don’t think really was much of a problem.

Research nurse, S10

I think after a lot of support by its champions, who wanted to see it done properly, we got into good practice, we got into good routines.

Bedside nurse, S06

. . . that didn’t work . . . it’s around consistency; there needs to be someone consistently, every day, every shift, to support them, otherwise it just doesn’t work.

PI, S01

Nature of the intervention

Overall, the intervention was considered well designed, the written materials user-friendly and the supporting resources (e.g. lanyards and pens) visually appealing and extremely useful. The pens, in which a COMFORT score chart was embedded, were the focus of particular praise. For all of these reasons, the design and content of the intervention (in its entirety) were widely endorsed as having promoted implementation. In the context of acknowledged difficulties involved in any practice change, participants identified two core properties as having been particularly valuable. First, the intervention was relatively straightforward to use. Second, the intervention was flexible, so that it allowed for independent clinical decision-making.

In terms of ease of use, several related features were discussed. First, the tasks to be performed by bedside nursing staff, namely the COMFORT score assessments and SBT screens, were routinely described as relatively undemanding. Participants talked about both as not only making sense to them, but also having the relevant clinical indicators as unambiguous and simple to assess. Second, trial paperwork was routinely described as user-friendly, so that completing the COMFORT scoring and SBT screening sheets was not experienced as overly burdensome. The fact that each component of the protocol was clearly set out in the bedside documentation, outlining what was to be undertaken, when and how, meant that bedside nurses could more easily accommodate it within their working practice:

To be honest, none of it took very long, it wasn’t a huge workload . . .

Senior nurse, S15

So in terms of the bundle, it was a lot of information but I think it was handled well . . . and the way that it was presented was really helpful, the colours and the tables and just the way that the bundle, it just made it easy to use, very user-friendly.

Senior nurse, S03

Some deviation from the above understandings was apparent. A few bedside nurse participants talked about the trial having added to an already heavy workload. In the context of having other, more critical, patient charts to complete, the need to complete additional paperwork was unwelcome. That said, accounts never suggested anything more than a degree of annoyance and that the paperwork was intentionally not completed.

In terms of protocol flexibility, where discussed, this was identified as particularly important for buy in from unit consultants, as it guaranteed clinical independence. In this context, consultant participants expressed no sense of being constrained in their clinical decision-making or having been aware of any such feelings on the part of colleagues. On much rarer occasions, the importance of the same independence was highlighted among nursing staff. In addition, participants pointed to the fact that units had not been constrained in terms of other weaning-related initiatives that were either ongoing at the time or planned for introduction during the lifetime of the trial. The fact that the protocol had been developed by specialists in the field was considered to have further promoted its acceptance by the medical team, as had the extended and inclusive process of development, which enhanced the feeling of ownership of the intervention:

I think if you hadn’t been flexible you wouldn’t have gotten so many people on board, so many units on board. We were particularly reassured by the fact that when you said, ‘OK, if you already planned to introduce your nurse weaning, whatever, it’s OK, continue, we won’t stop you doing that’ . . . that flexibility helped.

PI, S18

. . . some of our colleagues didn’t like putting children under pressure support . . . but when this came in and they looked at it as that it is being reviewed by many experienced intensivists, they did buy into it and there wasn’t any resistance.

Consultant, S15

Although the protocol was acknowledged to have been appropriately flexible, on a more fundamental level it was discussed as inherently unable to deal with the wide variation in patient clinical status, complexities and requirements. The consensus that emerged from participant accounts was that the more complicated a patient’s clinical status, for example in cases of comorbidity and/or underlying fragility, the less appropriate (the use of) a protocol. There were simply too many clinical variables, typically nuanced and often conflicting, that needed to be taken into consideration and which protocols were unable to accommodate:

The problem with the small baby, you know, when you put them on 5 on 5 it doesn’t work for them. So we always fail because they’re going to cough with this sort of tiny tube and this 5. So we couldn’t modify that as an SBT to put a pressure point of 8 or 10. So that wasn’t age appropriate.

Senior nurse, S16

. . . they were having to . . . say why we’re not doing this on patients who it’s not eligible for, and that increased resentment . . . there would still be a discussion about their weanability because of the paperwork, even though it was ridiculous . . . why would we even start the discussion of ‘are they eligible for an SBT?’, their chest is open . . .

Senior nurse, S01

Unit research team

The vast majority of units possessed a dedicated research team comprising nurses trained in paediatric intensive care. The presence of this team was discussed by its members and by unit staff generally as crucially important to trial implementation in several respects. First, a dedicated team was able to maintain a physical presence in the unit. Research nurses talked about themselves, and were described by their colleagues, as regularly visiting the unit floor to undertake various trial-related activities, including reminding staff of the need to complete relevant tasks. Furthermore, their mere presence was considered to act as a visible prompt:

But what I think worked well was . . . even just seeing us on the floor, like ‘oh we have to do COMFORT scores today, they’re in’.

Research nurse, S07

Second, the team could tailor their work pattern in ways that allowed them to monitor adherence, particularly regarding recording data, and undertake activity to make good any identified deficits. This could include, for example, regular ‘spot checks’ on adherence and ‘chasing up’ missing information. Probably more than any other aspect of their work, the ‘chasing’ of information, whether physical or electronic, was considered particularly time-consuming and frustrating. Deficits in record-keeping meant that research teams had persistent difficulties in capturing essential trial data:

The research nurses had to do so much more, the delivery was really difficult, because of the poor compliance with the paperwork we were always chasing stuff, we were always having to go to the wards to track down the SBT . . . we would spend days looking for charts.

Research nurse, S01

Third, across all units, the majority of research team members were nurses who worked part time clinically. These nurses regularly talked about actively progressing the trial when working on the unit floor by, for example, reminding colleagues of the need to perform trial tasks, providing practical advice and/or actually undertaking tasks when colleagues were busy with other aspects of patient care. The fact that they were themselves responsible for the bedside care of patients prompted several to suggest that their efforts to promote adherence were inherently mindful of the context in which bedside nurses worked:

It was good that we got to do it as a unit, as opposed to research nurses from obviously the research department coming down, because they wouldn’t understand all the ventilation.

Research nurse, S17

. . . and also as research nurses, us also having clinical involvement and being the ones who go and speak to the bedside nurses as something to implement, but hearing it from people who do their job, so we understand, I guess, what they’re doing naturally and how you can adapt that in or fit it in in a way that makes sense to them where they don’t feel overwhelmed with additional work to do. So I think that helped.

Research nurse, S03

Overall, the research teams considered SANDWICH to have been a labour-intensive initiative; for some, the time required was more, sometimes much more, than they had originally anticipated. Their articulation of this issue tended to be less in terms of an overt complaint and more in terms of an awareness that, however welcome the trial’s funding of research nurse time, more time was required. This was particularly the case in larger units and/or those that were research active, as in both the teams (often understaffed) were under considerable pressure to support numerous studies. In this context, several nurses talked about being able to undertake core SANDWICH activity only (primarily data collection), with limited time available for ‘extra’ supportive work, such as spending time on the unit encouraging adherence to the protocol:

. . . like the resources taken to implement this have been so much more than what I think I ever thought this study was going to be, and as soon as you step back to give other studies a fair chance, you pretty immediately see that it goes down.

Research nurse, S08

Across all units, participants endorsed the efforts of the research teams to ensure successful trial implementation. Not only were they acknowledged to have extended, sometimes very considerable, efforts, but also their physical presence on the unit was discussed as meaning that they could provide invaluable hands-on, face-to-face advice and guidance. This presence was regularly noted as sufficient to remind staff to complete trial activity that might otherwise have been missed. In addition, the targeted measures taken to improve and support compliance were regularly validated:

I think it’s worked better than I thought it would. I think the main reason it’s worked better than I thought it would, would be the commitment of the research nursing team.

PI, S18

I think if your research nurses are not good then this project is not going to be successful . . . I think that we were very lucky to have a very committed set of nurses.

Consultant, S15

Fit with established hospital and unit organisational and patient care routines

In different ways, participants from all units talked about established hospital and PICU organisational and patient care routines as having an impact on trial implementation. Here, descriptions tended to be couched in an awareness that it would be extremely difficult for any protocol to compensate for, or over-ride, these routines.

Unit culture and staff customary working

The preceding section addressed the impact of existing unit routines on trial implementation. This section continues that general theme, focusing on participant understanding of underlying culture and customary ways of working because these were thought to both facilitate and impede trial implementation. ‘Culture’ is broadly conceived, encompassing the values, attitudes, beliefs and other ‘ways’ of thinking and doing, all of which were associated with unit staff’s response to the trial and the subsequent delivery of the trial.

With regard to deviation from the protocol in relation to sedation management, a number of factors were discussed. First, in terms of deviation, some participants talked about a preference among bedside nurses, especially the more inexperienced, to keep their patients ‘flat’. Some bedside nurses themselves acknowledged such a preference based on the greater patient safety it afforded. At times, a need to keep patients well sedated because of clinical need was highlighted, such as patients with neurological injuries. In all such cases, although the COMFORT scoring component of the protocol was considered to have improved sedation management, including in terms of encouraging a more ‘light touch’ approach to sedation and/or addressing patient pain rather than automatically increasing sedation, an impetus to oversedate was still identified. On occasions, oversedation was explicitly linked to patient failure of a SBT screen or SBT, with or without the appropriate use of trial exclusion criteria:

I still think they’re oversedated, because, there’s a terrible fright that they accidentally extubate, which, without any doubt, is dangerous, and there is literally a phobia in using other ways of making sure that patients don’t pull their tube out, like muffling . . .

Consultant, S16

I do personally like a well-sedated child. People say because it’s easier, not just because of that. I think as a bedside nurse, it can be really hard when you’ve got an awake child and sometimes that line is very fine of getting sedated enough that they’re awake to a point of being able to extubate but being settled and sedate . . .

Bedside nurse, S17

. . . because everyone likes a still patient . . . they would just put they’re low conscious level . . . so they’d still kind of use one of the SANDWICH kind of get out clauses without really thinking like ‘oh, it’s because they’re too sedated’, . . . but it would be on ward round, mainly when [name of PI] was doing it, if I’m honest. It would be like ‘right, they failed because they’re too sleepy, let’s start waking these people up’.

Research nurse, S10

With regard to potential deviation from the protocol in relation to ventilator weaning, again a range of factors were discussed. In terms of deviation, one such was a preference for weaning patients onto progressively lower ventilator settings before (being considered for) extubation and/or routinely ‘bagging’ a patient in preparation for extubation. Both approaches were talked about as established practice within the unit (to a greater or lesser degree) and, therefore, difficult to change, at least in the short to medium term. Another preference was a unit-wide preference to move a patient onto NIV as quickly as possible, which was considered likely to encourage deviation from the trial ‘5 and 5’ settings, because staff would automatically use higher settings even in situations in which a patient was being fully extubated. Finally, in the context of actions taken following a successful SBT in which extubation could not be performed, participants could highlight a consultant preference for patients to be returned to full ventilator support, given their aversion to leaving patients on Psupp for any length of time:

Yes, because people have worked here for 25 years and that’s always what they’ve done. So they’d be like ‘what’s the point in doing this, because that’s what we’re going to do anyway’.

Bedside nurse, S14

. . . I think, probably for most of us we were too careful . . . I don’t know if it’s because it’s cardiac or it’s the culture or whatever . . . a lot of people would think no, I have to slowly wean this child, gradually come down and then make sure that the child is OK by gradually coming down . . . No one will go from 22 over 8 to 5 over 5, and I think this intervention has proven us that we are probably too cautious with the extubation and the assessment if it’s standardised.

Consultant, S05

Another aspect of unit practice that was understood to facilitate or hinder the implementation of ventilator weaning in the SANDWICH intervention concerned feeding policy. In some units, participants described the routine practice of stopping feeds in the early morning (typically for a period of 4 hours) as having contributed to timely extubation when the early-morning SBT was successful. By contrast, in other units participants talked about how a lack of planning had, on occasion, meant that successful SBTs could not be followed up by extubation because the 4-hour feed withholding period had not been commenced in advance of/in line with the timing of the SBT:

For a few years now what we do is we stop all feeds at 6 o’clock for every child. So we only feed them over 20 hours anyway. So by the time you’ve done the ward round at 9.00, 9.30, you’ve made the decision, you’ve got half an hour and they’ve had 4 hours of starvation.

PI, S17

Well, we usually have to wait 4 hours from stopping feeding, so we usually then, if they managed on CPAP [continuous positive airway pressure] then we’d prepare them for extubation, but then you’d have to wait 4 hours, you have to stop feeding them.

Bedside nurse, S13

Participants could also talk about more ubiquitous features of unit culture given that these affected trial implementation and outcomes. Thus, in the context of discussing how likely the trial was to achieve a reduction in the overall length of time that patients spent on a ventilator, participants discussed a pre-existing culture of proactive practice, encompassing nursing and/or medical staff. The underlying message was that, irrespective of how well the unit adhered to the protocol, an existing culture of weaning patients quickly to the point of extubation and/or being prepared to extubate on relatively high ventilator settings would probably reduce impact in terms of trial outcomes:

The other thing is, I think, by personality, I think our consultant group are not a very cautious group in the sense that . . . they will just get on and wean and extubate. They’re not the kind who are nervous about doing procedures or nervous about extubating. Of course, there’s a spectrum, but as a group we are more towards the early extubators than the later extubators.

PI, S18

. . . we probably already had a culture of not giving it a go recklessly but actually being fairly aggressive in terms of our extubation policy.

Consultant, S07

In a different context, participants could associate perceived limitations in unit multidisciplinary collaboration with deficits in trial implementation. This could involve, for example, consultant acceptance or rejection of information and/or advice given by bedside nurses based on the perceived knowledge and expertise of the particular nurse involved, and/or senior nurses’ ‘interference’ in the process of communication between bedside nurses and medical staff. More broadly, a culture of bedside nurses not being involved in any aspect of ventilator weaning (except titrating oxygen levels) could also mean that they could be limited, or even excluded, by others and limit or exclude themselves from discharging in their trial role:

. . . I think there’s a slightly challenging culture here for hierarchy . . . We’re quite fortunate, the consultants are very approachable, but there is sometimes a bit of a chain of command . . . some of the other team leaders are not necessarily approachable . . . if they didn’t buy into that [SANDWICH], then they would say [to a bedside nurse] ‘oh, I wouldn’t bother talking to your doctor about that. I wouldn’t take that any further’, and that’s then a break in the chain of communication and so then they would be less likely to . . . I think that’s more of a culture here that’s the problem . . .

Senior nurse, S01

But also, if a junior nurse will say something at one time and get shot down, they will never speak about it again, and I think consultants need to recognise that if a tool that the unit is supposed to have adopted is being spoken about, then they should honour it.

Research nurse, S08

By contrast, participants talked about an existing culture of robust multidisciplinary collaboration as facilitating trial implementation. Various manifestations of such working were described: first, sharing clinical roles and responsibilities, including capitalising on the involvement of nursing staff in patient care, and, second, a shared predisposition among staff to work in support of one another. For example, in several units participants talked about senior managers being extremely supportive of allowing training to be undertaken in working hours and/or ‘rewarding’ staff for their participation by giving hours in lieu. In another unit, senior staff members talked about a sustained effort on their and their colleagues’ part in supporting the trial team, including undertaking regular spot checks on COMFORT scoring and SBT screens. Supportive working could even extend to adherence to the protocol among staff who were personally opposed to its use, but who did not want to undermine the unit as a whole. Elsewhere, participants described a positive ‘circle’ of behaviour whereby the observation of colleagues’ commitment encouraged the same commitment on the part of other staff. Thus, bedside nurse participants described their motivation at seeing senior members of staff proactively championing the trial. Typically, the senior staff in question were unit consultants, but other senior nursing staff were also mentioned:

We have a very good working relationship with our consultants, it’s very much if they did say something on ward round, I could completely disagree and it will either be a discussion . . . we’ll come to a compromise. I’ve had it multiple times where a consultant said ‘OK, let’s try it your way first and then we’ll see’ . . . and all of our consultants are like that, we are very lucky.

Research nurse, S10

You have to want to do it, don’t you? You have to have a willingness, and I think what was also very important is that the whole MDT [multidisciplinary team] were engaged in it and supporting it. So it wasn’t just like us nurses and we had resistance from other teams or anything, you know, the doctors wanted to do it, the nurses wanted to do it, the allied health professionals, the physios were aware of it, you know, everyone was aware of the terminology that we had to use, didn’t they?

Senior nurse, S11

Final authority rests with consultants

As foreshadowed elsewhere in this chapter, to a greater or lesser degree, participant accounts across all units stressed the final authority of consultants regarding patient care, and consequent vulnerability of the trial to their willingness to adhere to the protocol and to support the same adherence by other members of staff. In some units, the medical team was described as greatly facilitating unit-wide adherence by demonstrating manifest interest in, and support of, the trial and/or proactively encouraging adherence among colleagues. In other units, unit doctors were variously described as uninterested or even obstructive. In both contexts, as members of staff in positions of final authority, the approach of individual consultants and/or of a unit’s consultant body in general was understood to have played an important role in either promoting or impeding a unit’s adherence to the protocol:

The consultants were really on board with the study and really drove us to try and do it, and also the doctors would come round and they’d say, ‘well, we have to at least see if they’d like to try’. So yes, I think the consultants were really on board, and the research team would come round and double check.

PICU bedside nurse, S13

I think people sort of didn’t change, and I really wished the doctors had pushed it a bit more. I felt like it had to come from them a bit more, because I kept saying like it’s a really safe way, like you know if you’re really not sure then put them on this, and you know, if it doesn’t work then you’ve still got the tube in.

PICU research nurse, S14

Well, one of the consultants was pretty anti it, to speak frankly, and didn’t believe in it, and I think that then sowed seeds for other medics who think ‘well, if they think it’s ridiculous well why would I bother?’. And then that permeates down to other people as well.

PICU senior nurse, S01

Potential consultant deviation from the protocol was identified in respect of three trial-related activities: failure to adequately discuss/set plans for patient weaning during ward rounds, failure to perform a SBT and failure to extubate following a successful SBT. A number of contributory factors were discussed. Participants could describe what they understood to be straightforward consultant lack of interest in research generally. Participants also reported that some consultants' clinical care preferences led them to deviate from the protocol and/or ensure timely extubation. A wide range of examples emerged from participant accounts. These included insisting that they personally assess all patients before progressing, as per the protocol; instructing that a patient not be put on a SBT when, according to the protocol, a SBT should be performed; ‘skipping’ a SBT and moving straight to extubation; and/or refusing to authorise/perform an extubation when, according to the protocol, this was recommended. At times, some frustration was expressed in that the nurses could be left feeling that they had wasted valuable time:

. . . as far as I’m concerned, when I take care of my patients on the unit, it [SANDWICH] hasn’t made any difference, because my criteria for weaning and extubation are my own and I go around patients and turn the knobs and I do it all myself; and SANDWICH or no SANDWICH, I just turn the knobs and I extubate on the basis of criteria that are not just about how the ventilation is weaned but also how the patient behaves and is awake and all these things . . .

PICU consultant, S16

I think the ward round part of it, I had very differing experiences, depending the consultant that was on leading the ward round . . . I think the consultants had varying levels of interest. I mean, if there’s people interested in research then they were fully throwing themselves into it, other ones, you’d really have to encourage them to even use it.

PICU senior nurse, S03

Some participants focused on a different aspect of consultant final authority. They suggested that, although the protocol had promoted confidence and autonomy of practice among bedside nursing staff, especially the less experienced nurses, nurses’ role in weaning was ultimately dependent on that sanctioned by the unit consultants. If the unit consultants were trusting of an individual nurse, their opinion was more likely to be accepted and acted on:

I think it in a way it did give more autonomy and more independence to the nurse to say . . . I’ve done my COMFORT-B, you know, they’re adequately sedated or . . . my SBT is failing because they’re oversedated or something like that, but I still don’t think that maybe the medics would trust . . .

PICU bedside nurse (experienced), S07

Staffing policy and recruitment

In several units, participants associated ongoing staff recruitment issues with limitations in trial implementation. A lack of a full complement of consultants and/or frequent changes to the medical team were discussed as adversely affecting the unit’s adherence to the protocol for two main reasons: first, because existing consultants were under considerable pressures, and, second, because locum consultants were inevitably less versed in unit activity, such as SANDWICH, and more likely to bring greater diversity in terms of individual practice. The same issues could be identified in relation to the nursing workforce. Participants frequently talked about the unit skill mix as compromised because of a dependency on junior nurses, many of whom had been (relatively) recently recruited. Elsewhere, a dependency on bank and/or agency nurses was discussed as problematic:

So we’ve had a lot of locum consultants recently. I don’t think that probably helped, because they always have different ideas, regardless. I think our main consultants mostly sing from a similar hymn sheet, mostly.

Bedside nurse, S17

Yes, we’ve got quite a junior skill mix in our unit at the moment and we’ve had quite a lot of new starters over the last 12 months, and I think sometimes with the more junior nurses, it was the same as anything that’s kind of new to practice, it was, for them, having it on their time management radar, if that makes sense.

Senior nurse, S04

Planned

• All patients.

• Adjusting for the covariates age, severity of illness (PIM3 score), respiratory compared with other diagnostic grouping, type of admission (planned/unplanned) and reason for admission (surgical/medical).

Post hoc

• Change in the calculation of intervention costs: the intervention cost per patient was estimated by dividing total intervention costs by the number of unique patients in the intervention arm only.

• Change in the calculation of total respiratory complications: all post-extubation NIVs were excluded and unplanned extubations were excluded if they were successful/well tolerated (defined as not requiring reintubation within 48 hours).