Clinician Perspectives of Barriers to Effective Implementation of a Rapid Response System in an Academic Health Centre: A Focus Group Study

Background: Systemic and structural issues of rapid response system (RRS) models can hinder implementation. This study sought to understand the ways in which acute care clinicians (physicians and nurses) experience and negotiate care for deteriorating patients within the RRS. Methods: Physicians and nurses working within an Australian academic health centre within a jurisdictional-based model of clinical governance participated in focus group interviews. Verbatim transcripts were analysed using thematic content analysis. Results: Thirty-four participants (21 physicians and 13 registered nurses [RNs]) participated in six focus groups over five weeks in 2014. Implementing the RRS in daily practice was a process of informal communication and negotiation in spite of standardised protocols. Themes highlighted several systems or organisational-level barriers to an effective RRS, including (1) responsibility is inversely proportional to clinical experience; (2) actions around system flexibility contribute to deviation from protocol; (3) misdistribution of resources leads to perceptions of inadequate staffing levels inhibiting full optimisation of the RRS; and (4) poor communication and documentation of RRS increases clinician workloads. Conclusion: Implementing a RRS is complex and multifactorial, influenced by various inter- and intra-professional factors, staffing models and organisational culture. The RRS is not a static model; it is both reflexive and iterative, perpetually transforming to meet healthcare consumer and provider demands and local unit contexts and needs. Requiring more than just a strong initial implementation phase, new models of care such as a RRS demand good governance processes, ongoing support and regular evaluation and refinement. Cultural, organizational and professional factors, as well as systems-based processes, require consideration if RRSs are to achieve their intended outcomes in dynamic healthcare settings


Implications for policy makers •
Appreciate the importance of local, contextual factors, and model elements in implementing rapid response systems (RRSs). • Organisational policy should ensure communication and negotiation via ongoing monitoring, evaluation, and coaching of health professionals. • Ongoing training and evaluation of physicians' roles in RRSs is critical to ensuring patient safety. • Creation of smaller dedicated RRS teams that inhabit these roles for a longer period will enable ongoing training and support for the physician role and consolidation of skills. • Prioritise inter-professional education and teams to increase understanding of the unique role and contribution of professional groups to the clinical encounters.
Implications for the public The rapid response system (RRS) concept focuses on the 'rescue' of patients showing abnormal signs and symptoms, preventing adverse clinical events. The way in which clinicians operate within such a system depends partly on their perception of its value as a tool for patient safety, as well as ways in which they engage and effectively communicate within and between professional disciplines. Failing to activate a RRS can risk patient safety and lead to adverse health outcomes. This study has identified an absence of ongoing training and evaluation of physicians' roles in the RRS and the importance of teamwork and communication in ensuring patient safety.
• What factors in your ward make it easy/difficult to care for 'sick' patients whose condition deteriorates? The RRS had been in place for 5 years at the time the study took place and received between 250-400 activations per month. Purposive sampling was used to recruit nurse and physician participants who were employed at this site and had current knowledge of and actively participated in RRSs. 19 A qualitative design was used to elicit perspectives of participants. We intended to facilitate discussion and narratives of experiences to understand clinicians' meanings and motivations that informed their actions. Given the centrality of inter-professional perspectives of teams in our study, six discipline-specific and multi-disciplinary focus groups were undertaken during April and May 2014 to identify registered nurses' (RNs) and physicians' perceptions and experiences of the RRS. 20 Focus groups were used to generate dynamic discussion and responses to participants' comments, prompt memories, and refine opinions already expressed. As nurses and physicians have their own distinct cultures, histories, and approaches to teamwork, conducting several discipline-specific focus groups allowed investigation of roles and practice and for open dialogue and disclosure of potentially diverse perspectives. 15 Owing to time constraints, some clinicians were unable to attend discipline-specific groups and chose to attend a multi-disciplinary group comprising both physicians and nurses. This choice allowed for individual narratives as well as responses and elaborative comments from others within each type of group. A literature review and preliminary discussions with key stakeholders informed development of the semi-structured topic guide (Box 1). 2,21,22 Topics included barriers and facilitators to caring for deteriorating patients, RRS experiences, operating within and outside of the RRS protocol, and perceived need for protocol changes.
The Rapid Response Model Track and trigger systems are recognised both nationally and internationally as best practice models. They take many forms with triggers typically incorporating numerical (aggregate weighted) scoring, vital sign parameters or combinations of both. 23,24 The rapid response model utilised in the study is a state-based multi-tiered vital sign parameter track and trigger system. 23 Individual tiers are activated when a pre-determined set of clinical observation and vital sign variables are breached (track), which then 'triggers' the response of the appropriate level of rapid response team (RRT). 25 The two tiers, 'Clinical Review' (Tier 1) consist of more sensitive trigger indicators (early warning signs), while 'Rapid Response' (Tier 2) contains less sensitive indicators indicative of late warning signs. Indicators are derived from research outcomes of the 'SOCCER' study, 26 each attracting differing levels of clinician response (Table 1). This allows a degree of individual facility autonomy based on RRS structure, resourcing, and geographic location. Tier parameter criteria can be modified to create individual patient customisation, affectively making indicators more or less sensitive to system activation over the standardised criteria. The response processes are primarily based around initial medical response (in the Rapid Response tier) coming from admitting medical teams, or dedicated facility physicians out of normal business operating hours.
Although not alone in adopting this type of response model, the majority of peer facilities more popularly initiate this level of medical response in the first (Clinical Review) tier, dispatching a critical care lead medical emergency team (MET) 13 when Rapid Response criteria are breached. 27 The Clinical Review tier is generally responded to and managed by unit RNs in the study facility who perform a thorough A-G (airway, breathing, circulation, disability, exposure, fluids, glucose) patient assessment within 30 minutes, initiate required interventions within their scope of practice, and escalate to the second tier if their assessment reveals possible or actual clinical deterioration. The admitting team model was chosen by this facility as it allows admitting physicians to initially manage the patient's deterioration, thus, decreasing workload demands on individuals as the RRS response load is spread across many speciality teams, rather than just a single MET. This model was also intended to allow admitting teams opportunity to develop skills in identifying and managing clinical deterioration themselves through experience rather than relying on the MET for 'rescue' in every RRS situation. The admitting, or after-hours team registrar (a physician who has obtained full registration with the Medical Board of Australia with at least 3 years' experience working in public hospital service), 28 is required to respond to all second tier calls within 30 minutes of activation. A junior resident medical officer (physician who has obtained full registration with the Medical Board of Australia) 28 is allocated to each clinical floor and is also required to attend. A third tier (Code Blue) is embedded within the Rapid Response tier and activates the MET from ICU if clinicians feel that immediate critical care assessment is required, there has been no physician response from a rapid response activation, or the patient is not showing sign of stabilisation or improvement 1 hour after rapid response intervention.

Recruitment
We sent invitations to attend focus groups to all nurses and physicians employed at the site via administrative email distribution lists. In addition, advertisements posted on hospital notice boards sought clinician volunteers. Although this method enabled significant reach, it precluded our ability to establish a response rate. Individuals were included if they were a nurse or physician employed at the study site and currently worked in clinical environments where the RRS operated. Interested potential participants contacted the principal researcher who provided additional oral and written study information. Recruitment ceased upon data saturation. As the principal researcher was a senior nurse within the facility and had a working relationship with many of the potential participants and a significant role within the RRS, an external experienced clinician and researcher (JLP) conducted the focus groups to minimise researcher and response bias. This individual, also a senior nurse, was neither known to participants, nor was a usual collaborator of the principal researcher, but had an understanding of and previous affiliation with the facility. Another experienced researcher moderated one group due to schedule conflict of the principal moderator; this person also performed the role of scribe in the other groups to record observational notes. Participants were informed that the principal researcher would not be attending the focus groups, but would have access to the recordings and conduct analysis. They were assured that names and identifying information would be removed from transcripts and demographic information would only be reported in aggregate form. They were also assured that the principal researcher would take steps to ensure confidentiality of participants including secure storage of data and act in accordance with established ethical frameworks. Prior to focus group commencement, all participants provided written informed consent including permission to audio record proceedings.

Procedure
One-hour focus groups took place on weekdays at the designated health facility in a private meeting room to enable attendance of target groups. Throughout the focus groups, the moderator noted newly emerging topics and points in need of clarification that were re-visited prior to concluding the sessions along with a summary of main points. This step enabled participants to verify the moderator's understanding and interpretation of reports, thus, acting as one method to verify findings.

Analysis
All focus groups were audio recorded and transcribed verbatim to facilitate thematic content analysis. 29 Analysis began with the principal researcher closely reading each transcript and listening to the audio recordings to get a sense of the proceedings and context. Transcripts were analysed using the general inductive approach. 30 Inductive coding began with line-by-line reading and coding of raw data without a pre-specified framework to remain open to emergent topics and multiple meanings within the text. Coded text was grouped into categories of material reflecting similar topics. Categories were then synthesised into themes and independently reviewed by two additional researchers (JRT and MD). To facilitate analytical rigour, three analysts (1) principal researcher (experienced clinician perspective and context/topic expert), (2) principal moderator (experienced, yet detached clinician perspective and witness to focus group processes), and (3) external qualitative researcher (methodological expertise) posed contradictory viewpoints and new insights and contributed to consolidation of themes. This analytical triangulation facilitated capture of key aspects of the themes assessed to be most important and useful in answering the research questions.

Results
Thirty-four health professionals (21 physicians, 13 RNs) took part in six focus groups over a five-week period ( Table  2). Each group was comprised of two to five participants with the exception of the registrar group, which included 15 participants. Four groups were discipline-specific and two groups were multi-disciplinary. Participants included both junior and senior RNs and physicians. Participants held differing skill levels and clinical experience ranging from less than one year to greater than 10 years (Table 3). Physicians had worked in both admitting specialty teams and facilitywide 'after hours' roles. The majority of participants were under 30 years old and had worked at the study facility for less than five years. Analyses of focus group data yielded a range of organisational and systems-level factors shaping the ways in which health professionals experienced and negotiated care for deteriorating patients within the RRS environment. The themes that reflect systems or organisational-level barriers to an effective RRS include (1) responsibility is inversely proportional to clinical experience; (2) actions around system flexibility contribute to deviation from protocol; (3) misdistribution of resources leads to perceptions of inadequate staffing levels inhibiting full optimisation of the RRS; and (4) poor communication and documentation of RRS increases clinician workload.
Responsibility Is Inversely Proportional to Clinical Experience Interns and resident medical officers (hereafter, junior physicians) reported feeling unprepared and out of their . Patient safety was a concern for nurse participants, especially in relationship to junior physicians altering the RRS calling criteria. Despite protocol mandating changes can only be made by senior physicians (registrar level and above), junior physicians altered criteria at times; a strategy perceived to avoid registrar attention.
"The only one who should change the RRS criteria is the Registrar, and that should be done in consultation with the team anyway. They [junior physicians] shouldn't just be doing that…" [RN]. Also concerning to nurses was the enactment of alterations by physicians who were not medically familiar with patients with complex medical care needs. In many speciality areas, nurses perceived that the RRS physicians were operating outside their area of expertise and were, therefore, not cognisant of the specific care needs of some complex specialty patients. Not having time to review the patients' medical records before initiating changes to their treatment amplified these concerns. It was also perceived that medical records frequently lacked adequate detail, context and clarity to enable full, detailed assessments and management paths.
"I think it's unfair for clinicians who aren't familiar with the patient to have to make that decision in such a short period of time, and I think it's a lot of pressure" [RN].

Misdistribution of Resources Leads to Perceptions of Inadequate Staffing Levels Inhibiting Full Optimisation of the Rapid Response System
Introducing the RRS increased participant awareness of patient deterioration, but also generated a perception of further workload burden. Both nurses and physicians expressed concern that the RRS generated an increase in workloads, often without any additional resources to assist. This variability highlights diversity in practice despite working within the same systems. Participants discussed how inadequate technological tools, such as information management systems, were contributing factors to communication barriers and variation in handover practices. They believed that establishing better ways of identifying patients who received RRS calls or had calling criteria modified, would lead to better clinical handover and prioritisation of sicker patients on rounds. The following excerpt depicts one participant's description of sharing information as being reliant on clinician memory and note taking in the absence of appropriate electronic tools.
"As far as I am aware, there is no formal list or computerbased system, [rather] it's a matter of people noting it down and taking a sticker [containing patient details] and presenting it at the handover. I think that works relatively well, it's not very formal. " [Junior physician]. Although described as adequate by participants, this manual system of RRS had the potential to miss identifying priority patients and those needing monitoring more closely. Completion of documentation of altered criteria was on single, loose paper forms placed in the front of patient's bedside medical records alongside vital sign observation charts. Clinicians discussed how these forms have at times become misplaced or difficult to locate if not in the correct location every time, potentially resulting in unnecessary RRS due to poorly documented changes.
" [Junior physician]. The above excerpt illustrates the impact of poor documentation of patient management plans on the ability of subsequent clinicians to meet their workload demands.

Discussion
This study highlighted multiple factors influencing clinician's abilities to operate effectively within the RRS environment.
Protocol deviation was evident to varying degrees by both disciplines, though as reported in the literature, it is not a unique observation that nurses are more likely to adhere to protocols than physicians, 31 perhaps a manifestation of their professional training and views of role and scope of practice. This reflected consistently with nurses seemingly having greater understanding of the RRS process than their medical colleagues. The study, however, revealed potential reasons for the occurrence of some protocol deviation. The initial information given at commencement of employment pertaining to the RRSs structure and process was less likely to be retained by physicians than nurses. Though both disciplines received identical education, senior nurses and clinical nurse educators in the clinical setting were essential in ensuring embedment of RRS knowledge and operation within the nurse culture. In contrast, an absence of ongoing support, training and evaluation of physician's roles in the RRS was a key finding and influenced functioning within the RRS. While a primary aim to involve and up-skill the patients' admitting teams, barriers pertaining to the study sites' model type were evident. Relying solely on admitting medical teams (and over-extended after-hours physicians) for primary tier medical response, at times, translated to an inconsistent and desultory RRS. Physicians, still in various stages of training, participate in these response roles for short periods, limiting both development into the role and establishing peer relationships with nurses from other clinical units. This inconsistent exposure was further complicated by a need to orientate large numbers of physicians into the responder role without support of a targeted, formal curriculum. Existing literature discusses failures of RRS, 13,32,33 yet studies seem scarce on examining the direct effectiveness between a variety of efferent (response) limbs models, tending to generically conclude suitability should be based on individual healthcare facilities goals and resources. 13 While many options exist around composition and resourcing of RRTs, pros and cons are evident regardless of choice. ICU without walls 34,35 is one concept that utilises the expertise of a trained critical care physician or team. Its small, targeted group make-up would enable easier training into rapid response roles. It would also lend to more consistent exposure to other acute areas of the hospital, theoretically supporting more effective peer relationship development outside the ICU. Similarly, MET's and ICU Nurse Liaison models 36 would have correlative benefits. While perhaps not encouraging the 'enabled' up skilling of non-critical care clinicians to the same degree as admitting team models, they do afford greater opportunity for consolidation of RRS skill and role development. The admitting team model was not unsuccessful in identifying and managing deterioration, the study participants engaged the system, though model design did cause discord around understanding and the perceived availability, functionality and efficiency of appropriately positioned resources. It was apparent the deployment of resources used in any RRS is a major factor when determining implementation and ongoing system success. Investigation into RRS team composition and resourcing 6 found that teams operated 24 hours a day, yet only 25% were funded, meaning resources were stripped from one area to service another. This no doubt causes extra burden on clinicians left to cover redundant positions during that time and can result in multiple forms of deviation of protocol as evidenced in this study. There was discussion amongst participants around METs being a better option for the facility, who cited physician training, knowledge and workload as the main reasons for efficient processes. RRT makeup is still contentious within the literature with some studies showing the importance of physician inclusion, 37 while others show beneficial results of nurse led ICU liaison/critical care outreach. 36,38 The nursing unit team environment played an important role in support and ongoing re-enforcement of RRS utilisation. Additionally, two nurses noted the system's ability to provide statistical evidence of workload and patient acuity. This evidence can help to highlight discrepancies between workforce supply and demand. Physicians' experiences reflected managing multiple competing demands, learning at various institutions with differing systems, and accelerated advancement to team member roles within the RRS. These topics were of greater concern in the junior physician groups where most agreed. Unlike nurses, these physicians do not have large support teams with senior colleague (consultant level) and educator guidance. This appears to be repeated nationally 6 and is accentuated in situations of patient deterioration where consultant physician level guidance and support would be of most benefit. As many of these individuals are training for specialties there are anxieties about competencies and further opportunities. 39,40 This may have lent to situations of escalation avoidance witnessed by nurses, who believed physicians needed to be seen as being able to manage and were not comfortable with patients deteriorating 'under their watch.' Efferent limb response demands more than just high-level skills in clinical assessment and management. Effective RRS implementation requires stronger development in responder role clarity and effective teamwork, yet there is often limited attention to this critical dynamic, both within the team and between peer relationships. 41 The rapid response physician is required to enter unknown situations, while often unfamiliar with the patient or specialty, communicate with colleagues from different disciplines, make clinical decisions, frequently change the management of what is seen as 'another team's patient' and take responsibility for the change. This responsibility imposes significant burden on physicians, many of who are relatively inexperienced. The study provides strong support for responder development of the non-technical clinical skills required to effectively perform within RRS roles; in particular, advanced communication, leadership, and teamwork being primary assets. Future research should focus on investigating the impact and efficacy of differing RRS model types. Of particular interest, a focus on the impact of differing responders, their professional composition, level of seniority and area of origin on influencing optimal rescue of deteriorating patients. The impact each has on existing staffing and resources would also be invaluable in helping already overloaded clinicians cope with further demands of these and other imposed systems.
Ongoing development and evaluation of RRS team training is also required to ensure responding clinicians are confident and capable, not only with clinical skills, but also with ability to work in teams and effectively lead in what are, quite commonly, difficult circumstances for patients, families and fellow clinicians. Literature is still scant on the development of training specifically aimed at RRTs. Initial evidence from investigators such as Theilen et al 42 show promising advantages in weekly multi-disciplinary simulation training, citing responder supportiveness and clinical, teamwork and communication skills as essential elements within the curricula. Large multi-centre studies to help support this evidence are required to ensure both simulation and training content are the most effective ways to train our RRTs. Within the study site, improvements in technology are developing to aid clinicians with patient management. Electronic activation and documentation of RRS calls will prompt clinicians to better document patient clinical events and management plans while also allowing for integration of this information to other systems. Production of clinical handover alerts of these patients to proceeding shifts of clinicians for example, enables identification of patients most at risk, allowing for prioritization of rounding and closer observation. The advancement and increasing use of technologies such as these, continuous smart vital sign monitors with automated RRS activation, and technologies allowing patient bedside point of care recording, will all add to future tools for clinicians, assisting in patient deterioration prevention through swifter, more accessible and adaptable information. Add to this, increasing advancements in integrated health records allowing continuation of patient information between primary and acute health facilities.

Limitations
Generalisability of this study is limited due to the single site. Some participant demographics are absent as a result of participants not supplying all information. The selfreport and recall nature of this study is a limitation, but the qualitative approach has allowed elucidation of critical, nuanced factors influencing system implementation and ongoing optimisation.

Conclusion
Study participants viewed the use of the RRS overall as an enabling tool for keeping patients safe, but also highlighted discrepancies and weaknesses exist in the system, particularly around choice and distribution of resourcing. The ways in which clinicians operated within this system was complex, multifactorial and non-standardised, sometimes with unintended consequences. This study adds to an emerging body of data emphasising the importance of considering local, contextual factors, as well as model elements. 43 Workplace processes, cultural and professional factors and systems are important considerations in implementation of RRSs. Failing to consider teamwork, communication and inter-professional dynamics impede activation of critical elements of the RRS.