Searching the Embase, MEDLINE, CINAHL and Cochrane Library databases yielded 142 references and a further 18 references were identified through grey literature sources (Fig. 1—Search methodology and outcome). Of these eighteen full text articles were assessed for eligibility and 5 studies were included (Table 1—Study characteristics). The studies had diverse study designs, including qualitative methods [20, 27], mixed methods [28], a randomised control trial (RCT) [29] and a pilot cohort study [30], making direct comparisons challenging. The heterogeneity of study designs and small number of studies meant a quantitative meta-analysis was not possible. Three of the studies were generated from the same research laboratory [20, 27, 28] which may introduce bias into the overall findings. Clear rationales were provided by the researchers for the choice of methods for the qualitative and mixed methods studies [20, 27, 28]. The results from the different methods used in the mixed methods study [28] were integrated and differences and similarities between the data sources were discussed. Both qualitative studies used theoretical frameworks to help interpret and map their data. One limitation of the mixed methods study [28] was the small number of patient–nurse–SLP triads which included only one acute stroke patient triad which restricts the generalisability of findings. A potential limitation of the qualitative studies was lack of reflexivity from the research team regarding their role and the consequent potential for bias. Several methodological limitations were identified in the pilot study [30]. Limitations included a lack of information about how outcomes were being measured and minimisation of bias, participant characteristics were not discussed meaning confounding factors were unable to be identified, follow up of participants was not complete, and the author was unable to answer the research question due to recruitment challenges. The primary limitations of the RCT study were the low statistical power and small effect sizes, the challenges of blinding research staff, and the failure to split all the findings into stroke and non-stroke populations within the experimental and non-experimental group due to the small sample size.

Search methodology and outcome

The findings are presented in alignment with the identified CFIR domains, with quotations from the included studies. The word limit precludes including all the supporting information which is available in the Supplementary Material (Table 4). A summary of the identified barriers and facilitators mapped onto the CFIR are presented in Table 2.

All included studies cited the Frazier Free Water Protocol developed by the Frazier Rehabilitation Institute [13] as the origin of the innovation. Lack of evidence and clinical guidelines for use of the FWP in the acute stroke setting was perceived as a barrier to implementation and reinforced the avoidance of its use [20, 27, 28]. Incorporating the FWP into hospital policies and procedures was felt unlikely to happen until there was more empirical evidence on the safety and use in the acute stroke population [27]. Acknowledging limitations for statistical power, in one study no significant group differences were reported for eligible stroke and trauma patients in positive outcomes, and negative clinical indicators did not differ significantly between the control and experimental groups [29]. None of the patients in the pilot study [30] had any medical compromise and recipients reported their quality of life improved. Short length of stay made it challenging to follow up patients for a sufficient length of time to enable measurement of negative and positive outcomes [29, 30].

A complex interaction of factors affecting implementation success were identified [28]. These factors included patient selection and protocol design. Facilitative factors included guidelines to minimise risk of adverse consequences and maximise patient safety [20, 27]. As part of the implementation design one study incorporated the Aspiration Precaution Oral Care program and used in service and written material to educate nursing staff on the study purpose and implementation [29], whilst others included a tracking sheet for oral care and water intake as part of their innovation bundle [30]. Conversely speech and language pathologists (SLPs) were found to design the FWP to be as safe as possible by giving teaspoons of water and if there was no evidence of aspiration following instrumental assessment. Although the FWP can be adapted according to the environment and specific patient condition and needs [13], these adaptations were felt to counter the original intention of the FWP and made it “impossible to assess safety and efficacy outcomes” (P.119) [28]. One study tailored the selection criteria and excluded patients with brain stem strokes [30], referred to as diagnosis-based exclusion criteria [20], whilst others adapted the FWP design by including ice chips [29].

Exclusion criteria were given as reasons for difficulty recruiting to trials [20]. Poor functional status (mobility, cognition or respiration) and SLP red flags (significant oral, swallowing and secretion issues) warranted exclusion by many participants. Risk factors such as levels of alertness, impulsivity, delirium and fatigue were considered to potentially increase risk. Nurses felt that patients on fluid restrictions would be unsuitable. In contrast dietitians felt the amount of water consumed was unlikely to affect a fluid restriction, and that changes could be made around the patient’s non oral feeding to avoid compromising the fluid restriction. Moving forward Murray et al. [20] propose consideration of broader patient function selection criteria alongside modifiable context specific factors.

Barriers included local conditions such as changes to the health system and the hospital admitting more patients than the ward could manage resulting in greater utilisation of agency and bank nurses who lacked familiarity with the acute stroke environment [27]. National guidelines and recommendation by the National Health Institute for Health Care Excellence (NICE) for research of the benefits of the FWP versus NBM or thickened fluids was identified as a way forward for research trials to be conducted in the acute stroke setting [20]. The updated NICE guidance ‘Stroke rehabilitation in adults’ [31] makes a recommendation for research to investigate the use of the FWP, particularly in studies with a larger number of participants. This is an opportunity to reassess the exclusion criteria and address the evidence gap.

The cohesiveness of the stroke ward and the frequent contact between nursing staff and patients and regular monitoring were positive characteristics of the acute stroke unit setting [27, 28]: "Generally, I feel we do reasonably well here cause obviously we are the stroke ward.” N3, P.116 [28]. The heavy workload and time intensity of acute care were identified as challenges for SLPs to educate clinicians and for implementation. This would result in a potential lack of adherence to the FWP guidelines or not implementing the FWP at all [27]: " My concern would be more about when do you actually fit it in, that you can go on with the workload that you’ve already got to do. That’s why I feel like some nurses just wouldn’t do it.” N4A, P. 291. The transient work force impacted on requirements for ongoing education for rotated staff and agency nurses [27]. One study found it difficult to educate and train all shifts of nursing and assistants that may be involved as this changed daily [30], whist time and organisation for ongoing staff education were perceived as a significant barrier to implementation by others [27].

Some participants felt that the FWP would fit into their daily practices [27]: “Just basic nursing care” N8C, P.291. In contrast the fast pace, high turnover of patients and overall caseload impacted on nursing availability to follow through all recommendations for safe implementation, in particular oral care [20, 30]: “That's going to be potentially another thing for nurses to have to do…they might be like, oh, well, it's just too hard, we're not going to give you your water or something. So if we're adding extra things for the nursing staff to do, I mean that could hinder the successfulness of the water protocol as well.” S DN3 P.639 [20]. Additional barriers were delegation of tasks to other staff for dependent patients, limited resources to record water intake [28, 29] and SLPs perceiving nurses prioritising other duties over implementing the FWP [28]: “We expect nursing staff to do it, but they don’t always have time. Sometimes that gets put down lower on the priority list.” SLP2, P.115.

The use of established systems to educate nurses about changes to current practices and processes to disseminate new protocols and staff expectations that they will learn new processes and procedures were factors that promoted implementation: "We're always learning something new and we're always implementing new activities on the ward." N3A, P.290 [27]. In one study SLPs provided education about the FWP guidelines to families and caregivers, and in house training to nursing staff and physicians before initiation of the FWP, for which attendance was documented [29]. Education involved skilled instructional education, demonstration and handout and teach back, and individualised education sessions to patients, caregivers, and nurses. Others reviewed the protocol with the patient and family, and nursing staff and nursing assistants [28]. Written material was provided, and nurses' competencies were checked after being educated. Peer support and modelling from supervisors and peers were identified as facilitators to implementation [20, 27]. “So everything I did I talked about with my supervisor and she often prompted me…Having someone that was more experienced who could suggest when it would be appropriate, absolutely was helpful in confidence.” SLP6B, P.291 [27]. Not having a clear written protocol was the most significant barrier to implementation [20, 28].

Clinicians' attitude to risk aversion, their beliefs and previous experiences influenced clinical decision making and implementation of the FWP design [20]: “As a general health service, we’re risk averse. And the logic of water and the FWP and the conditions you recommend water go against some of those built-in risk averse concerns that we have.” SSLP3, P.639. The culture of routinely providing thickened fluids prevented clinicians considering the FWP. [20, 27].

Perceived facilitators were recipients’ quality of life through comfort, normalisation, preferences for care [20, 27] and hydration, although to a lesser degree [27]: "I think it's important that we allow people to have the most normal life that they can have and if a little bit of water makes life more normal, then I think from that perspective it's quite an important thing.” DN4B, P.290 [27]. Not being ‘dry’ was a motivating factor for recipients; "I gotta have the water…I'll dry up if I don't get the water" (P3, P.117) [28]. Aspiration and development of chest complications were perceived negative outcomes, but not all participants were concerned with aspiration "if the aspiration isn't developing into anything." DN4B, P.290 [27]. Despite patients reporting their preference for water and SLPs stating patient choice was important, patient choice was often outweighed by factors associated with the patient’s medical condition and safety [20] with the focus only shifting to the person’s quality of life in palliative care or comfort situations.

Knowledge of the FWP intervention and a high degree of dysphagia expertise were perceived as implementation facilitators. Prior experience of the FWP made it quicker and easier to implement [27]. Greater experience was associated with increased confidence in the ability to determine patient suitability and implement the FWP in the acute setting [20]. Fewer years of experience were associated with more caution and risk avoidance. Knowledge and working with the stroke population was perceived to contribute positively to implementation [27]. In one study a stroke medical officer felt confident in trialing FWPs with all stroke patients if the patient was comfortable and not experiencing episodes of choking or excessive coughing. This was not a common opinion [20]. Clinicians’ understanding of stroke were perceived as factors that may perpetually reinforce the predominant practice patterns of not implementing the FWP in acute settings [20].

Lack of clarity of SLP instructions limited nursing staff ability to implement the FWP as intended. Incomplete and unclear documentation by nursing staff made it challenging for SLPs to monitor the impact of the FWP on the patient [28]. The sporadic and informal approach with which SLPs recommended the FWP was perceived to contribute to lack of awareness by other professionals [27]. Lack of stroke specific skills by new, student and agency nurses were concerns for misinterpretation of the FWP recommendations. The potential for misinterpretation also extended to family members. Most nurses were felt to lack the oral care skills required. Nurses forgetting to offer water and following measures to mitigate aspiration risk were examples where the FWP was not implemented as intended [28].

In one study almost two thirds of participants felt confident they could implement the FWP according to their role requirements and it was an extension of their current role [27]. Potential sources of implementation support were families to assist with positioning, oral care and monitoring outcomes. A common barrier was staff availability to deliver the intervention as intended. The level of staffing resource required to position and supervise patients, and the provision of oral care required, might preclude implementation [20, 27, 30]:"But the bit that I don't see time for…is the mouth care and the rest of it. If they have to sit there and supervise sips of water, I can just see them [nursing] hand them a cup of water and walk out the door. …So steps are going to get missed." DN2A, P.290 [27].

Deliverer and recipient receptivity and desire were motivating factors. Nurses discussed how they would be available for the FWP: "I think we're obliged to make time. It's our job, and all of the allied health team have got their job to do, so personally I think you've got to make time" SN5, P.640 [20]. Patients preferred and had a desire to have access to water [