Developing and evaluating interventions that are applicable and relevant to inpatients and those who care for them; a multiphase, pragmatic action research approach

Later we include a detailed account of how AR was used in this study. For now however a potted AR history or background seems appropriate, especially as this history articulates with the broader social concerns in addition to the clinical realities of modern health care. Detailed accounts of the structural procedures complete with diagrammatic representations of the cyclical nature of the AR process are ample [5‐7]. This paper is more about how AR can be regarded as a legitimate clinical intervention and how it may be usefully applied in the broadest translational sense.

We begin by providing some antecedents from the early beginnings and original intentions of AR to its more recent manifestations in health-based research. This brief chronicle places the root of AR as a democratic form of social practice and alludes to its shift to a technical process of improving professional practice with little or no social intent. We acknowledge that this type of evaluative work serves a purpose, though it is generally regarded as a deviation from the social mission of AR. At a time when public service (including public health) is framed by a neoliberal discourse, with emphasis on the individual, competition, and market driven policy, we are concerned that AR could be narrowly used as a simple analytical or even benignly reflective tool. However we would argue that in clinical and medical settings not only is this inadequate, it would squander the opportunity AR offers to truly observe, improve, and evaluate routine clinical practice as it happens with the capacity to bring about change through the cyclical protocol; and further, to engage both patients and those involved in their care in this process.

Though some trace the origins of AR back to John Dewey’s work in the 1930s [8], this tends to focus more what Schön (1983) might call reflection-on-action [9]. This epistemological lineage is not insignificant; however Kurt Lewin is generally credited with coining the term action research [10]. He came to prominence in the 1940s for his work, which attempted to bring together “the experimental approach of social science with programs of social action in response to major social issues of the day” (p.29) [11]. In particular, Lewin saw AR as having great potential to improve the position and wellbeing of minority groups in post-war America, and his work within ‘race relations’, (a term that must seem somewhat anachronistic), across the country is a testament to his commitment to social betterment [10]. His skill in setting up projects that involved deep levels of consultation through the now well-recognised steps and spiraled progressions of AR was a major achievement of the time and universally acknowledged today. He was well recognised for work with groups where, through a ‘planning – fact-finding – execution’ sequence, behaviour changes could be brought about that would advantage the group whether it was food consumption practices or group solidarity and factory worker rights in an increasingly industrialized and urbanized America. The connection of this history to health care may not be immediately obvious. However, malnourished elderly hip fracture inpatients may be considered as stranded skeletons hiding in hospital closets quietly waiting to be rescued [12]. In other words this group tends to be somewhat marginalized within the context of health care and to that end, the applicability of Lewin’s work to this field appears justified.

There is no doubt that this democratized approach to social inquiry as advocated by Lewin and others around the time challenged the scientific orthodoxy of the day, and in many respects in spite of its broader acceptance in contemporary research circles, it probably still does.

It is not difficult to see how AR might apply to patient care in clinical settings. As Vallenga et al. suggest AR is “suitable because it is a form of research enabling practitioners and consumers to participate in the development of knowledge which they themselves will subsequently use or will be used in their care” (p.81) [8]. In other words the conditions of and for clinical care are best observed, developed, implemented, refined and evaluated through the collaborative and cyclical reflective structures shared between practitioner teams and the patient. What is especially important about Vallenga et al’s position here is that it frames the idea of ‘practice’ in a particular way. Practice as it is being used here is very much about the co-construction of knowledge that can be recruited to affect changes to care. In other words it is ‘care’ that is the practice, not the technical components of the procedures to bring it out. This approach clearly brings into focus the patient and further highlights the need for those who care for them to dynamically work together across systems, processes, and interventions to procure effective multidisciplinary and multimodal care. So as McTaggart describes, practice is not some automated procedure, repetitive task, or the implementation of a technique (or even policy for that matter) [13]. Hence when we talk about AR as a way to improve practice we would caution being lulled into a liberal discourse of autonomy, which tends to undermine any sense of responsibility. Habermas has a term that perhaps is more fitting [14]. He talks of ‘mature autonomy’. As McTaggart points out, this co-joining of autonomy and responsibility fits more closely with the idea of practice as a social endeavor and in the case of this study, that responsibility was both to and with patients as a way of improving and reporting quality care within the constraints of routine clinical practice [13]. As McIntosh suggests this approach to the improvement of practice requires a shift from evidence-based medicine that invariably establishes a series of routines and interventions based on clinical trials to evidence based practice that is better informed by pragmatic trials [15]. At the same time McIntyre would argue that this distinction between evidence based medicine and evidence base practice is an example of how the idea of practice should be separated from the idea of institutions [16]. In the case of this study we could argue that medicine (or indeed health) might be perceived as an institution (given its power this is not too much of a stretch of the imagination). Care, on the other hand, is the practice that goes on before, within, and beyond the context of that institution (be it in a hospital clinic or health centre). We explore this relationship later in the paper.

As Whitehead, Taket & Smith suggest, AR is increasingly regarded as not only a legitimate approach to research in health and medical settings, but also as a particularly effective process for supporting organisational change [17]. Within clinical settings changes to organisational practice are the key to improved health care. Indeed as we have already suggested, clinical or randomised trials can be somewhat insensitive to the nuances of service delivery in health where clinical significance is of far greater importance than statistical significance where the outcomes are patient related and change is demonstrable; no matter how small or large and predominantly based on the clinicians knowledge of the patient [18]. However as Whitehead, Taket & Smith continue, AR has been slow to catch on in the context of heath research, probably for the reasons of perceived weakness to which we alluded to earlier [17]. This poses a serious problem since as Meyer suggests, “barriers to the uptake of the findings of traditional quantitative biomedical research in clinical practice are increasingly being recognised” (p. 178) [19]. As Meyer says, the attraction of AR is that it represents a form of inquiry whereby researchers work with and for people, rather than conduct research on them, and in this sense it represents a form of democratized research entirely consistent with Lewin’s original premise. In other words it is the style of the research rather than its methods that are different. This also dispels the myth that AR is confined to or synonymous with qualitative research. In this project a number of data gathering techniques was used through the AR process including various numerical measures all of which contributed to a broad canvas of patient care, the changes in practice and the consequences of those changes.

Discussions of pragmatism invariably start with the question “will a proposed intervention work in (so-called) real life?” Hence where explanatory trials measure symptoms or markers, pragmatic trials have as their focus a range of outcomes that focus on the patient’s wellbeing. In other words pragmatic trials using a range of data sources aim to work with patients rather than ‘on them’. As we have already argued, this is entirely consistent with AR. As Patsopoulos [20] points out, rather than distinguishing between explanatory and pragmatic trials (PT), it might be better to see them as points on a continuum; both indispensible but different in what they reveal. Hence the naturalistic (a word sometimes used synonymously with pragmatic – though we are cautious here) setting of a clinic provides opportunities to see what works in practice [20].

It is interesting to note that in the 2013 Australian Government Strategic Review of Health and Medical Research, pragmatic trials (much less AR!) barely warrant a mention [21]. There is an acknowledgement in the document that a broader range of research activities should be encompassed and that practitioner research should be encouraged. However, these are hardly ringing endorsements of alternative paradigms, serving to underline the somewhat tepid enthusiasm for research paradigms that sit outside of RCT convention. Baker [22] acknowledges that part of the problem is that there is “no immutable formula for successful implementation of innovations” (i30). Yet he makes a case to suggest that, (for example), case study methods are under-utilised as a way of bringing about change to care practices, and that the knowledge created through evidence based practice solutions have to be built upon the way such solutions can be implemented. In essence he is arguing for research and practice to be far more closely aligned; indeed we might even argue that they are one and the same thing. This is entirely consistent with an AR approach to change.

However, a shift in research paradigms may be emerging. Traditional models have defined RCTs as the gold standard design for evidence generation, and limited pragmatic, qualitative, prognostic or observational studies to pawns in the hierarchy of evidence [23, 24]. However, clinicians, funding bodies, academics and publishers may be beginning to detect chinks in the armour of RCTs, and in response to the need to economically justify research agendas and support evidence based practice, are starting to recognise the need to also consider studies focused towards relevancy and applicability rather than overwhelmingly prioritising analyses of cause and effect under highly controlled, artificial conditions [25]. This climate change is promoting, in line with Lewin’s [10] theory, the unfreezing of accepted norms and acknowledgement that a variety of study designs focused towards the patient, the question(s) under investigation, and applicability and relevancy of the research to routine clinical practice; rather than continuing to promote highly reductionist or post-positivist focused research as the standard by which all others are judged [26]. One such illustration is an extension to the Consort Guidelines for reporting pragmatic clinical trials which provide a clear mandate for clinicians and researchers to justify pragmatically focused trials as both meaningful and relevant to patients, clinicians, and the broader healthcare community [27].

As an example of how pragmatically focused AR can be used in clinical settings, not only to change practice but to develop an evidence base for change, we present here a study conducted from November 2010 to September 2012. The setting is a clinical environment under change to develop best practice whilst incorporating generally understood conventions for care based on scientific trials. We allude to the results and outcomes in general terms as these have been published and presented elsewhere [28‐31]. Of equal importance in our view are the processes by which care practices were developed and adjusted on the basis of the AR cycles.

Phase I built on baseline data and observed whether commonly applied nutrition screening and assessment processes may present as barriers to nutrition care in hip fracture and included two diagnostic accuracy studies and a prospective, consecutive case series [29, 30, 32, 33]. Phase II reported a further prospective, consecutive case series to quantitatively investigate protein-energy intakes in post hip-fracture inpatients. This phase aimed to quantitatively report observed and reported barriers to inpatient protein-energy intake. Phase III built on earlier results including an explanatory mixed methods study to further expand and explore additional patient and clinician barriers and facilitators to nutritional care [28]. These three phases provided baseline data for comparison with post-intervention data.

Following observation of barriers and potential facilitators to nutrition in hip fracture inpatients, the multidisciplinary treating team identified potential improvements to routine clinical practice. These were trialed, adjusted as required, and then embedded into routine clinical practice as a new multidisciplinary, multimodal nutritional model of care. This was considered the ‘intervention’.

A quasi-experimental before-and-after pragmatically focused study was then used to compare patient and healthcare outcomes following implementation of the new model of care with baseline data associated with the individualised nutrition care approach [31].

The study design presented in this paper is not without challenges (Table 1). To enable waiving of informed consent and to maximise relevancy to routine clinical practice, outcomes measures are restricted to those routinely available or measured within the context of multidisciplinary clinical care. However, the authors suggest having less, but highly relevant outcomes, is beneficial to those working in the field. Similarly, interventions must be made as part of improvements to routine clinical practice. Inevitably, this limits the opportunities for comparison to before-and-after or across different population groups rather than randomisation. Again, an ideally constructed randomised intervention that has only indirect relevance to clinicians may be less valuable to clinicians than a simple pre-post design that is conducted within the context of routine clinical practice. A multiphase, pragmatic AR approach further allows the flexibility to evaluate the impact of multidisciplinary and multimodal interventions; however, we acknowledge that the opportunities to describe cause and effect or isolate the impact of specific aspects of interventions are limited. The impact of applying an AR process itself must also be taken into account. However this should be considered an asset of AR, rather than a deficit. Multiphase studies require skillful connection of strands over multiple phases and are also recognised as difficult to define as part of the one program. The complex nature of the study design make it more difficult to disseminate results given the conventions of post-positivist studies that tend to favour reductionism by detailing highly selected specific variables and determinism through claims for cause and effect. In addition the proclivity for variable measurement, statistical analysis and reporting (quantitative techniques) may lead to relegation of ‘other’ research as less important or even irrelevant [37, 39]. However, there is clearly an increased demand for and interest in studies relevant to clinical practice and this is reflected in changing publication patterns in journals. We would argue that the tenets of pragmatic trials (in this case framed by AR) are an equal partner in the cause of translational research.