Implementation of clinical practice changes in the PICU: a qualitative study using and refining the iPARIHS framework

Semi-structured qualitative interviews were carried out to prepare for a study that will examine prospective implementation of new blood transfusion recommendations for critically ill children. The interviews aimed to understand prior experiences with implementing clinical practice changes within the PICU. Ethics approval for the study was obtained from the Stanford University Institutional Review Board (IRB-47140). The completed Standards for Reporting Qualitative Research (SRQR) checklist https://​www.​equator-network.​org/​reporting-guidelines/​srqr/​ is included as additional file 1.

We conducted interviews with health care providers working in various roles from eight PICUs across the USA. We selected units to represent variation in PICU types (pediatric ICU (excluding cardiac patients, 4 units), pediatric cardiovascular ICU (CVICU) (2 units), and combined pediatric/cardiovascular PICU/CVICU (2 units)) and overall PICU size (11-32 beds) in the USA. The three types of ICUs are collectively referred to as PICUs, unless specified. A unit representative helped recruit participants. The study team and unit representative sent an initial e-mail to solicit volunteers for participation. We provided additional information about the study to respondents and contacted them if they agreed to participate following review of this information. We selected participants using a stratified, purposeful sampling strategy, as described by Palinkas [21] to achieve a sample of participants within each PICU role. One of two authors (KS and MF) conducted interviews either in person or via telephone. We obtained the participant’s verbal consent prior to each interview. Interviews were audio recorded and transcribed verbatim. We continued interviewing participants in each role until no new information was forthcoming from new participants. The qualitative team members (KS, LH, and GL) held meetings to review data and confirm when variation in responses was no longer noted.

We developed an interview topic guide from the iPARIHS framework [17]; however, added topics that focused on barriers and facilitators of clinical practice changes and blood transfusion in the intensive care unit to ensure we addressed aspects of change that are not explicit in the iPARIHS framework (Additional file 2). Participants were asked first to describe a clinical practice change that had been implemented in their ICU as a starting point to discuss the process and impact of these initiatives. Additional questioning focused on exploring PICU culture, receptivity to clinical practice change, initiating and sustaining practice changes, and experiences and perspectives on blood transfusion in the PICU. Qualitative data related to blood transfusion is not reported in this manuscript.

We used a Framework Approach [22] for qualitative analysis with the NVivo (Version 12) software package. Framework is an approach to analysis in which case-level data (rows) are summarized along thematic categories (columns) in a matrix and involves five steps of familiarization, identifying a thematic framework, indexing, charting, and mapping and interpretation [23]. The coding framework was structured around the iPARIHS constructs (innovation (the evidence-based intervention), recipients, context, and facilitation) with associated subconstructs [17]. We applied iPARIHS in this specific setting to gain insight into the combination of factors required for successful implementation in this clinical care environment [20] and move toward precision implementation to ultimately better specify and tailor implementation strategies [24]. Subconstruct definitions were not available to standardize analysis; therefore, we developed a codebook (Additional file 3) to define subconstruct items using definitions drawn from the revised PARIHS framework [25], the Consolidated Framework for Implementation Research [26], and the Theoretical Domains Framework [27]. The facilitation construct was coded broadly to include any data wherein deliberate support or problem solving was applied to promote implementation. Following coding, review, and organization of data coded within the facilitation construct, it became evident that this data broadly aligned with the implementation strategies as described in the Expert Recommendations for Implementing Change (ERIC) project by Powell et al. [28]; thus, we organized this data within the ERIC categories. We also added inductive codes while searching for emergent themes, particularly around implementation barriers and facilitators to address potential constraints of using the iPARIHS framework to structure our analysis. At the outset of the analysis, two researchers (KS, LH) selected and coded fived interviews independently using the a priori codes from iPARIHS, while creating new codes for emergent themes. The researchers met to agree on definitions and interpretations of existing codes, compare coding, and discuss emerging themes and integrate into the coding framework. One researcher (KS) coded the remainder of the interviews using the established coding strategy. We initially categorized coded data within the iPARIHS subconstructs, then, following the Framework approach, summarized data to retain context for ease of comparisons across units and roles. Following summarization, we compared and contrasted data between roles and clinical units, with careful consideration of identifying patterns as well as unique, but informative statements that provided an alternative perspective to capture variation. Strategies to ensure credibility (internal validity) of findings followed guidance by Miles et al. [29] and included linking data to categories in iPARIHS, checking for negative evidence, and checking that findings are replicable across the dataset (i.e., across more than one PICU).

Within the iPARIHS recipients and context subconstructs, providers identified unique sub-themes of the PICU environment that increased complexity around clinical practice change and had important implications for implementation. These include added complexity with team-based care, high-stakes care that limited receptivity to change, variable readiness for change, limited bandwidth, and the emotional toll of the PICU environment; each sub-theme is further described in Table 1. Of note, receptivity to change related to providers being amenable to or interested in change conceptually, whereas readiness to change occurred when providers had the appropriate resources (time, skills, support) to operationalize the change. When providers were receptive to or ready for a change to occur, change was facilitated. Conversely, when providers were either not receptive to a change or not ready to integrate the change into practice, this served as a barrier to change.

Providers reported that the stimulus for change in the PICU, located within iPARIHS innovation and context constructs, typically originated from three sources: (1) from a motivated individual or group with a specific interest working within the unit, (2) from a source outside the unit, or (3) as a result of a negative or sub-optimal clinical outcome (iPARIHS Construct: Context; Sub-constructs: Local level Culture, Local level Past experience with innovation and change, Organizational level Priorities, Organizational level Culture). Based on prior experience or personal interest, individuals or groups proposed change initiatives to improve care. Outside sources included the hospital, hospital system, or, infrequently, a regulatory agency; associated changes were usually intended for the entire hospital or organization. Often, these changes did not account for PICU workflow or unit dynamics and were more difficult for providers to both accept and integrate into practice (iPARIHS Construct: Innovation; Sub-constructs: Degree of fit with existing practice and values). Providers disliked these changes “imposed” on the PICU and preferred when they were modified to fit PICU practice. Providers of all professions noted that changes were made in response to adverse outcomes with significant or unexpected morbidity or mortality (iPARIHS Construct: Context; Sub-constructs: Local level Culture, Local level Past experience with innovation and change). In these cases, clinical practices that were felt to be causal were often modified or abandoned despite prior demonstration of safety or success as providers were highly motivated to avoid future negative events: “There’s a lot of anecdote as well. Like there was that one bad case one time, so we don’t ever want to try that again” (Site 1 CVICU, Attending 2). Changes made in response to adverse outcomes were often made quickly and sometimes without obtaining provider input.

Prior to initiating change, unit leaders made the decision to devote resources and pursue a change. Identified through the iPARIHS recipients and context constructs, there was variation in who made the decision to pursue a change in practice. PICU physician and nursing leaders were often noted to work together to decide on making changes; however, in some cases, nursing or physicians made decisions independently without input from other professions (iPARIHS Constructs: Recipients, Context; Sub-constructs: Power and authority, Local level Culture). One NP noted: “A lot of policies and changes to nursing care are driven by nursing leadership. The medical team leadership is not necessarily involved in those changes…[it’s problematic because] we don’t learn about new standards of care until the nurse actually prints out a new copy of it” (Site 1 Combined PICU/CVICU, NP 12). Half of units had unit-based committees (multi-professional or nursing only) that informed leadership’s decision to make a change. Other professional groups (respiratory therapists, etc.) and subspecialists were routinely omitted in many units from the decision to make a change, though there was a broad understanding that obtaining feedback, making modifications, and obtaining buy-in from all professions as change moved forward was essential (iPARIHS Construct: Context; Sub-constructs: Local level Culture, Local level Past experience with innovation and change).

While unit leaders made the decision to institute a change in the unit, in many cases, individual providers made an independent decision to support the change as part of their practice. While some changes were noted to be mandatory, providers who did not support a change could avoid using it in many cases, as there was not direct accountability. The individual provider’s impression of a proposed change impacted their willingness to comply with or resist the change. This impression was determined by four sub-themes within different iPARIHS constructs: (1) the evidence behind the change (iPARIHS Construct: innovation; Sub-constructs: Underlying knowledge sources), (2) rationale for change (iPARIHS Constructs: innovation, recipients; Sub-constructs; Relative advantage, Values and beliefs; Motivation), (3) provider-level factors (iPARIHS Construct: recipients; Sub-constructs: Values and beliefs; Motivation; Skills and knowledge), and (4) competing interests and time (iPARIHS Constructs: recipients, context; Sub-constructs: Time, resources, support; Local level Culture) (Table 1). Provider level-factors were intrinsic to the individual provider that influenced perception of and receptivity to change. We identified three factors within this sub-theme: provider experiences, provider beliefs, and balancing benefit with effort. Provider experiences refer to how receptivity to change and implementation were dependent on a provider’s level of experience or prior experiences working in the PICU. Provider beliefs were ideas providers held about their own practice or change in the PICU that impacted their receptivity to change. Finally, all providers perceived that change required an additional investment of time and effort. Each provider explicitly or implicitly recognized that they balanced the “costs” of a change (time, effort) with the perceived benefits of the intervention. Table 1 describes each of these sub-themes and factors in detail and provides an example of how providers discussed them.

We gained perspective on the iPARIHS facilitation construct by asking providers about barriers and facilitators of change as well as prior experiences with change. Broad coding for this construct yielded a wide range of data elements describing actions and strategies used to promote implementation. These data were often linked with one or more of the innovation, recipients, and context constructs, which made it difficult to characterize elements within a single iPARIHS construct given the unique PICU context. The actions and strategies that constituted facilitation mostly fit within the ERIC categories of planning, restructuring, education, quality management [30], as detailed in Table 2. These facilitation strategies have key features that are unique to the PICU environment which we will describe.

Sub-optimal levels of engagement around new evidence and need for practice change (iPARIHS recipients, context) among some providers were an important barrier to changing existing clinical practice. This often stemmed from professional expectations and requirements. Nurses were not expected to stay abreast of new evidence which impacted receptivity to change, as described by one nurse: “Not all the nurses have made it part of their professional practice to stay updated on evidence… Practice changes that come from new evidence can be a little more challenging for that group” (Site 3 PICU, Nurse 25). Support given to nurses and other non-physician providers to spend time working on practice changes was also limited: “One of the challenges is that they [hospital administration] call our work [on change] ‘non-productive time’…they don’t allot a lot of resources to it. Some people are willing to go above and beyond without getting paid, but it’s not very well resourced” (Site 3 PICU, Nurse 25). As such, lack of expectations or incentives to work on change in these roles hindered progress.

Providers felt identifying a champion within each professional role was necessary to ensure a resource existed to support each group (iPARIHS recipients, context). Champions were stated to be knowledgeable about and promote use of the change, remind and reinforce why change was being made, and in some cases, assist with implementation. As in many implementation efforts, achieving buy-in from both leadership and front-line staff was essential; in addition, for nurses, creating true ownership of change was noted to be particularly powerful in ensuring consistent application of the change.

Providers emphasized the need to have easy access to information and tools required for change at all times (iPARIHS innovation, context). Different units had various established options (electronic, hard copy, signage), but providers differed in their opinions about which of these worked best. Because multi-professional education about change was often completed at a time remote from actual implementation, additional “just-in-time” training at the time of change “roll-out” was viewed as very helpful, particularly by nurses.

Through exploration of the iPARIHS framework’s constructs and identification of the themes described above, we developed a model for change implementation in the PICU (Fig. 1). This model reflects the iPARIHS constructs and subconstructs and extends beyond iPARIHS to illustrate the relationships between PICU-specific factors that influence change, including the hospital and PICU environment that influence change appreciably. These relationships are not reflected in iPARIHS, but likely crucial to understand when planning implementation in the PICU. As a process model [31], our model describes the manner in which research is translated into practice within the PICU; this process includes multiple phases.

Ideas are initially proposed by individuals or groups (three sources of change, as described in Section 3.2). The decision to pursue change in the unit is made by unit leadership (unit leadership: Section 3.3). Our interviews did not explore the factors that influence how leadership selects changes to pursue; however, available unit resources, unit needs and priorities, and mandates at the hospital level likely contribute to these decisions. Individual providers also choose to support or resist a change based on multiple factors (individual determinants, Section 3.4), including the evidence and rationale for the change, their own beliefs and experiences, the potential benefit versus the effort required to change, and time required in light of other competing interests. While initial subthemes such as evidence and rationale for change are necessary, they are superseded by subsequent subthemes. As such, regardless of the evidence or rationale for a change, implementation will not occur unless providers perceive that the benefits of a change exceed the effort required and they have adequate time to operationalize the change. Change subsequently requires facilitation (Section 3.5), implementation, and later, efforts to ensure sustainability (Section 3.6).

Individual determinants and unit/team processes interact with one another and influence the selection of facilitation activities which are operationalized primarily as implementation strategies; likewise, facilitation activities impact these factors and are linked in a bidirectional fashion. Facilitation efforts enable implementation; whether or not a change is implemented is dependent on individual determinants and unit leadership and processes. This relationship is unidirectional. Finally, after initial implementation, if a change has been integrated into workflow, it may be sustained. The PICU environment exerts an influence on this entire process of change, impacting how individuals perceive proposals to change and their ability to integrate changes into standard work. To a lesser degree, the hospital may also influence change, often when hospital-wide modifications to practices are also mandated in the PICU. We acknowledge that this model is idealized in the linear arrangement; in actual practice, implementation steps are unlikely to be sequential, and implementation efforts may stall and require teams to go back to a previous step before a change is integrated into workflow. This model emphasizes the factors and their relationships that impact implementation and may be used for implementation planning in this setting.

The iPARIHS framework was generally helpful as a determinant framework in exploring implementation; we found the innovation, recipients, and context constructs overall were a useful scaffold to structure thinking about implementation in the PICU. Harvey and Kitson place facilitation central to the process of implementation in iPARIHS, as it “activates implementation through assessing and responding to characteristics of the innovation and recipients within their contextual setting” [17]. While the other main iPARIHS constructs are expanded for users with detailed subconstructs, facilitation is somewhat under-defined. The characteristics of a facilitator are described; however, the actions that constitute facilitation are not detailed, and there is no sub-classification of the types of activities or strategies that facilitators might utilize to enable and support change. Lack of clarity around the action of facilitation constituted somewhat of a “black box” in the framework, despite the fact that it remains crucial for implementation. We aimed to further describe facilitation by identifying activities that “activated implementation” through our interviews and subsequent qualitative analysis. While this work does not go as far as to propose a comprehensive set of facilitation subconstructs, we provide insight into facilitation activities that may be useful in the PICU, which are presented in Table 2. A strength of our work is the delineation of the activities important for facilitation, which in this PICU environment consisted primarily of activities that could be linked with known implementation strategies. By developing a more comprehensive and clear understanding of the facilitation construct, we aim to advance the framework and potentially inform future identification of important facilitation subconstructs within iPARIHS. We offer this critique based on our qualitative data and suggest that to further improve the framework, a more detailed description of components or features of facilitation would be useful for framework operationalization and in determining where facilitation barriers exist. Moreover, identifying these activities may be key to informing the selection of implementation strategies, which continues to be an active area of inquiry within implementation science [34] and holds practical value to those translating evidence into practice.

Implementation strategies identified through assessment of facilitation in our work in the PICU have been described by others in similar ICU environments. de Vos et al. and Stevens et al. noted the importance of engaged staff and adequate resources to facilitate change [5, 15]; in our study, facilitation of change was impacted by variable provider engagement and the time and personnel allocated to work on change. Other authors have reported the utility of some of the same facilitators in the ICU environment including the need for multiple educational modalities and a team of motivated champions [7, 8].

The process model that we have put forth organizes the iPARIHS constructs to reflect how change occurs in the PICU setting and potentially informs how groups might identify barriers and facilitators and plan implementation strategies. We did, however, find there were iPARIHS subconstructs that were infrequently discussed by providers (i.e., goals, organizational absorptive capacity) and may not be as important in this setting. We also identified novel concepts not included in the framework and interplay between constructs that had important implications for implementation.

Some of the iPARIHS subconstructs were difficult to separate from one another in the PICU setting. Within the recipients construct, motivation and values and beliefs were often inter-related. Within innovation, usability and clarity were related, and within context, we found that discussions of past experiences with innovation and change were linked closely to mechanisms for embedding change, as many efforts attempted to integrate changes into PICU workflow. While this did not complicate our analysis, the fact that these and other sub-constructs often “traveled together” or were interlinked should be noted.

There were a few frequently discussed concepts that did not easily fit within any of the iPARIHS subconstructs. These findings may reflect the fact that we created a process model from the determinant iPARIHS framework, but could potentially be considered useful additions to the framework. Providers’ emotions were conceptually distinct from motivation or values and beliefs and had an important influence on the work environment and the stressors that might prevent acceptance of change. Another example was the importance of having regular opportunities to use a change in actual or simulated practice for nurses; however, there was no subconstruct within innovation that captured this concept. We observed a few instances where factors within one construct significantly influenced another. For example, the concept of loss of autonomy with standardization of care fit within the construct of the innovation; however, recipients were also impacted. In examples like these, factors influencing implementation in the PICU did not fit neatly within single iPARIHS constructs; understanding the links between constructs has implications for implementation planning.

The iPARIHS framework lacks formal subconstruct definitions. Guided by other existing frameworks, we developed more explicit subconstruct definitions that may be used in future studies to guide a standard approach to qualitative analysis using this relatively new framework (Additional file 3). To further build on iPARIHS, we posit that standard definitions for subconstructs be developed to ensure consistent use of terms. This is particularly important for use of the framework within the research setting, as the need for uniform definitions has been noted frequently in implementation science [35]. We also suggest that further exploration of the concept of facilitation, with potential for development of subconstructs to further define components or actions. This will provide users with context to evaluate and operationalize plans around this critical element of the framework.

An aim of the larger study was to investigate implementation of blood transfusion recommendations, as such, we enrolled participants who ordered or administered transfusions but did not include other providers who may play a role in implementation of other PICU practice changes (respiratory therapists, pharmacists, patients, etc.). While it was not within the scope of this project to include the perspectives of all allied health professionals who work in the PICU, we perceive that our model of change, which is based on a number of different pediatric ICU models, is likely fairly comprehensive, but which may benefit from additional nuance that the inclusion of these groups might offer in future studies. While we aimed to ensure that PICUs were representative of the scope of US practice, it is possible that not all variations in PICU culture are represented and that the data presented is not fully generalizable to other PICUs; however, using the iPARIHS framework helped to ensure that we captured characteristics of the change process that are broadly transferable across settings. Barriers, facilitators, and implementation methods noted by interviewees reflect their opinions; in reality, different elements of the PICU setting may be more important to optimize implementation; however, this analysis serves as a starting point for guiding and testing implementation in this setting.