NoMAD Survey
There were 127 eligible staff, and response rates declined over time from 71–43%. Table 3 outlines participant characteristics. Radar plots representing median survey scores are presented in Fig. 3: the fuller the radar plot, the more normalised practices were. Whilst median scores for question 13 (is there sufficient training?) and 14 (are there sufficient resources?) improved over time, scores for question 7 (are you willing to work with others in new ways?) and question 8 (will you continue to support the bundle?) decreased. Median scores for the remaining questions were consistently 1 (agree) across all three surveys.
Table 3
Participant Characteristics
Participants | NoMAD Survey 1 n, (%)* | NoMAD Survey 2 n, (%)* | NoMAD Survey 3 n, (%)* | Observations with RTAI Interviews, n | Semi-structured interviews, n |
Consultant | 4 (4) | 4 (6) | 3 (7) | 0 | 4 |
Specialist Registrar | N/A | N/A | N/A | 4 | N/A |
Senior House Officer | N/A | N/A | N/A | 1 | N/A |
MTI/Fellow | 4 (4) | 0 | 2 (4) | 3 | 0 |
Nurse Band 8 | 3 (3) | 3 (5) | 3 (7) | 2 | 2 |
Nurse Band 7 | 9 (10) | 7 (11) | 5 (11) | 1 | 3 |
Nurse Band 6 | 24 (27) | 24 (37) | 16 (36) | 5 | 4 |
Nurse Band 5 | 35 (39) | 19 (30) | 14 (31) | 5 | 2 |
Nurse Band 4 | 11 (12) | 6 (9) | 2 (4) | N/A | N/A |
Missing | 0 | 1 (2) | 0 | N/A | N/A |
Total | 90 (100) | 64 (100) | 43 (100) | 21 | 15 |
Response rate | 71% | 56% | 43% | N/A | N/A |
NoMAD = Normalization MeAsurement Development survey, RTAI = retrospective think aloud interviews, *Percentage of respondents |
Table 3 Participant Characteristics
Figure 3 Survey Results Radar Plot
Outcome and Process Measures
CLABSI and bundle adherence rates are shown in Fig. 4a and 4b, respectively. Whilst there appeared to be an initial decrease in CLABSI rates, this was not sustained. Adherence rates improved over time. Adherence to ANTT was consistently reported as 100%, whereas adherence to the insertion checklist and two-person insertion improved from 49–89%, and 42–87% respectively.
Figure 4 CLABSI and Bundle Adherence Rates
Observations of Practice with Dyadic Think Aloud Interviews
Twelve observations with RTAIs were performed, totalling six CVC insertions and six CVC fluid changes (see Table 3). Eleven observations took place during day shifts and one overnight. No emergency procedures were observed. Dyadic interviews lasted between 15–30 minutes.
Semi-Structured Interviews
Fifteen semi-structured interviews (see Table 3) were performed between July and August 2019, lasting approximately 50–70 minutes.
Influences on Implementation
Thematic analysis of observational, dyadic RTAI, and semi-structured interview data resulted in eight themes organised into three over-arching categories: individual, team, and organisational influences. Theme descriptors and supportive qualitative data are provided in additional file 2.
Individual influences
Bundle endorsement
Participants discussed the extent to which there was buy-in for the bundle. This included beliefs about bundle utility and beliefs about the problem (CLABSIs). Broadly, participants believed that the new practices addressed aspects of practice thought to be poorly performed previously (RTAI 2, 3, 4, 5). Nurses believed AIPPs made their work easier and reduced variations in practice, however, unclear expectations and uncertain rules caused confusion. This included the frequency of CVC access, the use extension sets, and different ‘rules’ for difference devices.
There was less collective endorsement for CVC insertion practices, specifically the two-person technique. Beliefs about its utility were mixed; some believed they had previously been doing this informally, whilst others believed it was more appropriate for less experienced colleagues. When asked how important the second person was, one participant commented:
I didn't used to think it was! But the more I've done it with a second person, actually it makes the procedure easier, if you've always got somebody to monitor the baby and comfort the baby, but also to see things you haven't seen, like when turning round and your glove touches the port hole […] I think you're more aware of your own practice when you think someone's watching you.
NURS0804, Interview 14
There was a disconnect between the intended purpose of the second person role, and how the role was enacted: the second person was actively involved in the procedure, donning sterile gowns and gloves, rather than monitoring the process using the checklist. This meant that some participants did not believe a second person was always necessary, nor that nursing involvement was legitimate. The expectations were not always clear and there were mixed beliefs about the value of specific bundled components, such as maximal barrier precautions (specifically hats and masks) and the insertion checklist. Hats and masks were infrequently worn, and the checklist was not observed being used. Semi-structured interviews revealed it was exclusively used retrospectively after the CVC had been inserted.
Whilst all participants believed that reducing infections was important this was juxtaposed by beliefs that it was “unrealistic to think you could eradicate” (DOC01, Interview 3) them, and participants questioned if CLABSIs were a “genuine” infection (DOC04, Interview 12), or “a bad infection” (NURS0803, Interview 8). There was a broad consensus that the bundle had increased awareness of infection rates, which was perceived to be beneficial.
Seeking Reinforcement
Participants were actively seeking reinforcement, checking together in observations, and reflecting during RTAIs on how procedures could have been done differently. Participants believed they received limited feedback on CVC management, and identified problems with reporting CLABSI rates:
So, I think there's various negative things that come back, in a very non-specific way, which means that no one does anything about it, because there's no ownership. […] but the only way of that being reported is a number in a governance meeting several months later, then it's not being fed back in a timely fashion, we're waiting to find out we've got a problem
DOC01, Interview 3
Whilst local infection rates were disseminated, one participant recognised that it was “difficult to extrapolate the data, for it to mean something” (NURS0801, interview 6). When shown local CLABSI rates, participants apportioned responsibility to other professions or roles. One participant noted that:
We never think it’s us, do we? We always think, oh well, that’s the doctors, that’s the surgeons. As nurses, we feel that we get the blame for when infection rates rise […] its everyone, isn’t it? and we’re all responsible for it
NURS0618, Interview 7
Only one participant reflected on their own behaviour, suggesting that “it’s probably our fault and we can improve” (NURS0509, interview 8). Whilst some participants explained that they wouldn’t change their practice unless they received specific feedback on their own practice, this was juxtaposed by those in senior roles who expressed concerns about providing individual feedback. One consultant believed it would be an “awful thing” to have the responsibility for an infection “laid at someone’s door” [SSI3, DOC01]. Participants across both professions used words such as “moral distress”, “blame”, “guilt”, and “morale” in relation to feedback on infection rates. This contrasted with data from the dyadic RTAIs, who were interviewed immediately after doing the work, who wanted reassurance that what they were doing was right.
Without feedback, participants resorted to informal self-appraisal of bundle effectiveness. Nursing participants ‘felt like’ they were administering less antibiotics, equating that with a reduction in CLABSIs, believing new practices ‘felt cleaner’. However, two participants expressed concerns of a “feedback culture” [DOC03] and receiving feedback for “just doing your job” [DOC03, NURS0610]. Finding ways to use positive reinforcement strategies was believed to be important, such as celebrating central lines that completed their journey without an infection and reporting line-free infection days.
Team Level Influences
Division of Labour
The division of work influenced bundle implementation. Nursing enactment of ANTT included clearly allocated roles which they believed made it easier to understand expectations. However, the second person role for CVC insertion was not implemented as intended: in observations, the second person was always actively involved in the procedure rather than acting as a monitor of practice. Participants reflected on how the second person could be helpful with one participant suggesting allocating roles at the start. There were some aspects of central line work where responsibility was unclear, such as the monitoring of CVC dressings. Some participants believed that the current division of labour was not appropriate; for example, one consultant believed that “some of our more experienced nurses would be better placed” to change central line dressings (DOC03, interview 13).
Participants referred to “their work” and “our work” suggesting a siloed division of labour. Whilst some participants believed the bundle had increased the collective responsibility of CVCs others believed there was no clear responsibilities for some aspects of care, such as assessing the need for the CVC:
Well, I’d like to think it’s a Consultant role to think about it all, but- I think you can’t really expect a junior bedside nurse to think about that- but I think it’s everybody’s responsibility to think, is this still required and do we still need it, and if we still need it, is it safe and are we managing them well?
DOC03, Interview 13
Both professions were unsure if involvement in CVC insertion was a legitimate nursing role. Competing workload priorities between teams also caused tension. The organisation of CVC fluids within nursing teams also caused discomfort, and there was a shared experience of guilt among nurses:
There’s this feeling that you’re a failure if you hand over your patient to the night shift and go ‘I haven’t done any fluids. Haven’t done any infusions’. The response you get sometimes from some are like ‘What?! What have you been doing?!’ […]
NURS0703, Interview 9
Knowledge regarding CVC care was role specific; not everyone had access to the same knowledge. For example, one nurse believed that they were unable to advise on CVC insertion because “its not our job” (NURS0617), whilst a consultant recognised that nobody is “thinking holistically” (DOC03). A lack of shared knowledge is likely to make monitoring practice difficult and participants believed that multi-disciplinary training could help improve this.
Surveillance and Monitoring
There was variation in bundle surveillance, and informal monitoring- such as checking together-was frequently observed:
someone got their gloves on and they put their hair behind their ears and someone’s, “Oh what have you just done?” and started again! (Laughs) So, it’s not like it has to be judgmental, it’s just, “Oh, do you realise you’re -?” “Oh, I did not realise” …
NURS0515, RTAI 8
However, not all participants engaged in informal monitoring, despite participants’ beliefs that behaviour may change if they are being observed, or that “corners may be cut” (NURS0804, interview 14) if no one is watching. One nurse explained that the process is “automatic, you know what you’re doing” whilst another recognised that that she doesn’t “watch that intently what other people are doing” (NURS0610, interview 11). One participant suggested incorporating more formal surveillance into their practice:
We don't really have very much feedback, and whether we should be observing each other, having an outside observer…
NURS0704, Interview 4
Monitoring of CVC insertion appeared to be ad-hoc, with one consultant feeling that “sometimes the co-ordinators around just to keep an eye” (DOC13, interview 15) whilst another reflected that he “can’t say they [CVC practices] are being done properly” (DOC04, interview 2). In CVC insertions, the second person was implemented as an assistant rather than as a monitor of asepsis using the checklist prospectively.
Participants reflected that it was difficult to question poor practice, despite believing this was important. When concerns were raised, some believed they were not always listened to. One solution was to give permission- to make it acceptable- to raise concerns, thereby providing a clear expectation of behaviour:
[…] and you can say to the nurse or the doctor with you, 'Keep an eye on it for me because I might not notice. Please tell me if you see that something's become desterilised,' and then that just takes that onus away from someone because actually then you can say it's all right, I'm expecting you to say that to me.
DOC01, Interview 3
Learning and Teaching
Participants across both professions described learning from each other, which was believed to be important for sharing tacit knowledge. Participants descriptions of learning were reminiscent of the ‘see one, do one, teach one’ approach which was reflected in observations. Participants believed that less experienced colleagues may not have learnt the high-risk moments for asepsis lapses, such as incubator portholes, something that was learnt through doing. Whilst nursing participants believed they were all taught a standardised approach to ANTT, some found it hard to teach others who may have been taught different techniques in different workplaces.
Whilst there were few references to formal teaching resources in the dyadic RTAI interviews, participants in semi-structured interviews referred more explicitly to educational resources. Cascade training was believed to bring challenges such as introducing variation. This could create uncertainty, with one nurse recognising that “we make it confusing” by teaching new staff different things (SSI8, NURS0509). Participants felt that it was important to learn the right way to perform a task the first time it is taught and to not pick up ‘bad habits’ (SSI12, DOC04). Visual aids were felt to improve implementation and refresher training was identified as being useful. One consultant believed that CVC education needed “a team” (SSI12, DOC04). Learning from others continued to be referred to as how practices were taught, triangulating with RTAI data:
We just run through whether the person is able to do it and let them do it on their own, which we shouldn't be doing.
DOC13, Interview 15
This highlights the absence of formal surveillance, suggesting the two-person technique-which was observed being used in all central line insertion observations- was not always used.
Organisational influences
Resources
Resource availability, including easy access to equipment and the appropriate hand scrub, were identified as barriers. Insufficient staffing and competing demands were frequently cited as a barrier. The feeling of ‘being busy’ was a potential explanation for poor CVC hub decontamination:
…and so particularly with the clean for 30 seconds, dry for 30 seconds and they’re standing there anyway everything’s a hurry, we’re busy, busy, busy, clean, dry, yeah, that’s 30 seconds.
NURS0610, RTAI3
There was a perceived need to “get on” with tasks with nurses feeling they need to “fit it all in”, despite also recognising that it was a “24-hour service” (NURS0515, RTAI8). Those inserting lines used words such as “whip one in” and “crack on” in keeping with a perception of busyness. Interruptions were also a potential cause of lapses in asepsis.
Both professions felt that nurse staffing was a significant barrier to involvement in CVC insertions, with medical staff not wanting to increase nursing work. Insufficient staffing was believed to have become “normalised” (DOC01, interview 13) yet another participant felt it was “an easy card to play” (DOC03, interview 12). Some nurses felt that if nurse: patient ratios were improved, they could assist with CVC insertions.
There were differences in the extent to which staffing and workload was considered an acceptable reason for practice variation. For CVC insertions, being busy was considered an acceptable reason to not have a second person, whereas for CVC access, it was not an acceptable reason for nurses to not adhere to hub decontamination:
Yes, maybe you could be excused for not having a second pair of hands, but I don't think you can say, 'I was busy, so I didn't clean the hub for so many seconds', because I think, yes, you're busy, but then you can't compromise on safety just because you're busy, so I don't think that's acceptable.
DOC13, Interview 15
The two-person insertion and checklist were not seen perceived as an essential safety process.
Environment
Observations revealed specific environmental challenges for asepsis, with limited space and increased traffic around sterile fields. Incubator portholes provided a physical obstacle for those inserting central lines. Participants suggested having a dedicated space and equipment trolleys, to limit interruptions and separate the task from competing cognitive demands.
Culture
Social norms- such as habits and rituals-alongside professional hierarchies, influenced how the work was done. This ultimately effected bundle endorsement. The process of performing ANTT was ritualistic:
but when you put those gloves on it's like a costume to like, 'This needs to be done sterilely. I mustn't touch anything,'
NURS0707, Interview 4
This shared mental model made it easier for nurses to informally monitor practice. Participants recalled examples of normalised deviance- socially accepted ways of working that were dissonant with beliefs about how the work should be performed. Participants shared examples such as not waiting for one minute for hub decontamination (observations 3, 5, interview 1, 7, 10, 11 14, 15), not checking CVCs “as closely as we should be” (NURS0610, interview 11), “wandering around the unit” in sterile gloves and that a second person “may join you halfway through” (NURS0804, Interview 14). One consultant felt “it was vital” to lead by example (DOC01, Interview 3) whilst another reflected that he could not expect others to wear a hat and mask if he didn’t (DOC04, interview 12).
Hierarchies influenced surveillance. Nurses believed there were clear lines of escalation with a “strong” senior nursing team that would raise concerns. Whilst one nurse felt that her opinions were listened to since wearing a senior uniform (NURS0610, interview 19), others felt that they “nag, nag, nag, and then just give up” (NURS0618, interview 7). Hierarchies may influence asking for help:
I think that the nurses probably find it okay to ask. The medics, I'm not sure. I feel there's a little bit more… Going back five or 10 years, I think we were a little bit a better team and we weren't necessarily a hierarchy as much. There's a bit more of a hierarchy at the moment. NURS0703, Interview 9
Making it acceptable to ask for help was believed to be important, though it was recognised that culture was difficult to change. Participants suggested giving colleagues permission to raise concerns may make it more socially acceptable. Changing the culture was strongly believed by all participants, across both professions, to be important to improve bundle adoption. Participants referred to creating a more open culture as an important part of improving CVC care.
A Conceptual Model of Implementation
A conceptual model was developed from the data using the constructs of NPT (Fig. 5). Whilst all participants believed reducing infections was important- there was collective buy in- there were inter and intra- professional differences in coherence- how participants made sense of both the purpose, and the components of, the bundled practices. It was not always clear how the new practices were different to previous ones. There was also limited inter-professional cognitive participation; whilst the nurses had a shared mental model for central line fluid changes, this was not universal, sometimes resulting in tensions between professional groups. Nurses were unclear of the expectations for CVC insertion and felt unable to challenge practice or use the checklist. The role of a second person became endorsed as a teaching role rather than a safety role, meaning that nursing involvement was not legitimised, nor that a second person was necessary for experienced colleagues. The division of labour was therefore not as intended and there was limited collective knowledge around CVC practices. Ultimately, it was not always clear what best practice looked like, or who should do it.
This uncertainty meant that it was sometimes difficult to monitor practice and there was limited collective action around the new practices. This resulted in informal mechanisms for appraising the new practices and feedback to participants on infection rates was not always meaningful, or accessible, to those doing the work. Limited individualised feedback on practice limited the extent to which reflexive monitoring could occur, and participants across both professions were actively seeking reinforcement.
Figure 5A conceptual model of implementation