Clinical intervention research in nursing

Available online at www.sciencedirect.com

International Journal of Nursing Studies 46 (2009) 557–568 www.elsevier.com/ijns

Clinical intervention research in nursing Angus Forbes * King’s College London, The Florence Nightingale School of Nursing & Midwifery, Primary and Intermediate Care Department, James Clerk Maxwell Building, Waterloo Road, London SE1 8WA, United Kingdom Received 13 February 2008; received in revised form 25 April 2008; accepted 17 August 2008

Abstract As a healthcare profession nursing has a duty to develop practices that contribute to the health and well being of patients. The aim of this paper is to discuss current issues in clinical research within nursing. The paper defines clinical interventions research as a theoretically based, integrated and sequential approach to clinical knowledge generation. The paper provides specific criteria for defining a clinical intervention together with an overview of the stages involved in clinical research from problem identification to implementing knowledge in practice. The paper also explored the extent to which nursing research was focussed on clinical issues, through a snapshot review of all the original research papers in Europe’s three leading nursing research journals. In total of 517 different papers were included and classified in the review. Of these 88% (n = 455) were classified as non-clinical intervention and 12% (n = 62) as clinical intervention studies. The paper examined the intervention studies in detail examining: the underpinning theory; linkage to previous (pre-clinical) work; evidence of granularity; protocol clarity (generalisable and parsimonious); the phase of knowledge development; and evidence of safety (adverse event reporting). The paper discusses some of the shortcomings of interventions research in nursing and suggests a number of ideas to help address these problems, including: a consensus statement on interventions research in nursing; a register of nursing intervention studies; the need for nursing to develop clinical research areas in which to develop potential interventions (nursing laboratories); and a call for nursing researchers to publish more research in nursing specific journals. # 2008 Elsevier Ltd. All rights reserved. Keywords: Clinical research; Clinical interventions; Research principles and methods

What is already known:  Nursing interventions are poorly defined.  Nursing interventions are important to patient care outcomes.  Research into nursing interventions is under developed and resourced. What this paper adds:  The paper provides criteria for defining clinical interventions. * Tel.: +44 20 7848 3367; fax: +44 20 7848 3230. E-mail address: [email protected].

 This paper provides a framework for developing integrated approaches to clinical research.  The paper provides some suggests as to how clinical research in nursing can advance.

1. Introduction As a healthcare profession nursing has a duty to develop practices that contribute to the health and well being of patients. To fulfil this responsibility nursing requires robust clinical research to show that its interventions do no harm and have a beneficial effect for sufficient numbers of patients to ensure that they are both clinically and economically worthwhile. While healthcare is an increasingly inter-professional

0020-7489/$ – see front matter # 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2008.08.012

558

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

endeavour and the practices and interventions delivered by nurses are shared across disciplines, nurses cannot absolve this responsibility to others. There are many nursing practices that are not subsumed within this common programme of research (continence, nutrition, symptom alleviation, tissue viability, etc.) and many other topics in which nurses are the main clinical care providers (e.g. self-care support, patient education, health promotion). The paper is presented in two parts: in the first part the general principles and stages of clinical interventions research are outlined; and in the second current issues in clinical interventions research within nursing are explored. The overall intention of this paper is to suggest how nursing can maximise the generation of clinical knowledge by adopting a theoretically integrated and sequential approach to knowledge generation. 1.1. Focus The focus of this paper will be on nursing research and clinical interventions. Given the breadth of activity and the multidisciplinary nature of research it is important to define the relationships between nursing research and the wider research community. Nursing research is defined in this paper as research led by nurses with the purpose of developing knowledge of relevance to the nursing profession. The contribution of nursing research to the development of clinical interventions is illustrated by the model in Fig. 1. This model identifies two core elements of clinical interventions research, the development of the interventions (the bench science) and the subsequent translation of those interventions into clinical practice (health services research). This model suggests that nursing research (or nurse researchers) contributes to both these elements. While much of the nursing research contribution is part of the common multidisciplinary development of health care knowledge, nursing research also contributes knowledge that is more

Fig. 1. The nursing contribution to clinical research.

specific to the practices of nurses. It is this latter area of contribution that is the focus of this paper (as indicated by the white arrow in Fig. 1).

2. Clinical interventions research In this first part of the paper consideration is given to some of the key principles and methods of clinical research. 2.1. Clinical interventions The first step in clinical interventions research is the development of the intervention. Defining and describing the intervention is as important as testing it (Conn et al., 2001). In nursing we might view all of our actions in giving care to patients as interventions. A patient has a leg ulcer so the nurse applies a compression dressing, this is a clinical intervention. However, from a research perspective such a definition is less than satisfactory as it lacks the precision required to ensure that any observations made about the intervention are transferable and (importantly) repeatable. It could also be argued that there should be some explicit connection between the intervention and the underpinning mechanism (theory) for the intervention. The intervention in this case could be more specifically classified as a four layer compression system with defined compression thresholds. The underpinning theory for this being that compression bandages assists venous return, diminishing peripheral backflow, reducing tissue damage and (hence) shorten healing times in venous ulcers. The intervention then becomes an action with a predicted outcome rather than a random activity. It is also an activity with a more precise label (compression regime X) and a link to prior (theory) work showing the relationship between compression and wound healing. This definition also identifies the specific population of patients for whom the therapy is suitable, people with venous leg disease. Furthermore, it can also enable comparative judgements in that the four-layer model with regime X is clearly distinct from a single layer regime or other multilayer regimes (this distinction is called granularity). Therefore, for research purposes (i.e. knowledge development) interventions need explicit and detailed description (what it is, how it works and how much of it you need, etc.). They also need to be linked to a predicted outcome (criteria that determine whether it works or not) and to the population for which it is intended. In addition the intervention may be clustered within a family of therapies or treatments for a specific problem for which there may be comparative estimates of effect. These elements form the basic facets of the classic Population Intervention Comparison Outcome (PICO) model used in identifying and synthesising evidence from studies (Sackett et al., 1997). These different facets suggest a need for systems of coding or classifying problems and treatments both to enable cross study comparisons and to ensure clarity in the way the

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

intervention is utilised in clinical practice (Given, 2004). In mainstream medical therapies these are explicitly defined by internationally regulated classification systems and definitions of interventions based either on the formulations of medicines or explicit descriptions of procedures. Nursing interventions on the other hand are often rather loosely defined or are wrapped up in multiple interventions presented as a ‘black box’ of activities without explicit definition (Conn, 2007). While this tendency for multi-faceted approaches is, in part, a reflection of the nature of nursing (care is provided to the whole person, it is relational and responsive to multiple needs) it may also be a product of a failure to regulate and define (theoretically and empirically) the nature and actions of the underlying procedures. This is a key point, neither nursing or nursing care are interventions in themselves. An intervention is a specific activity. While that activity may be delivered in the context of nursing care it is the workings and impact of that specific activity that is of interest in interventions research. This is not intended to be atomistic or a challenge to the holistic nature of nursing it simply demands that specific activities are tested to ensure that they are beneficial to patients. Attempts have been made to define nursing interventions more explicitly. One example of this is the Nursing Intervention Classification (NIC) system (Bulecheck et al., 2008). The current edition describes (classifies) 542 nursing interventions from abuse protection support to wound irrigation. The interventions are based on ‘common nursing activities’ and are developed by expert panels. The ‘electrolyte management’ intervention is typical, being defined as the ‘promotion of electrolyte balance and prevention of complications resulting from abnormal or undesired serum electrolyte levels’. This intervention is comprised of 23 activities that may be sequentially related but are not absolute requirements of the intervention (e.g. teach patient and family about cause, contact physician if symptoms worsen, monitor cardiac arrhythmias). The classifications are independent of formal diagnostic categories and specific populations or settings, making it difficult to specify the intervention for research purposes. The NIC fails because it is trying to retrospectively define nursing interventions in terms of general care facets, rather than considering the underlying nature of the activity in a theoretically explicit way. DeJong et al. (2004) suggested a different model for defining interventions based on explicit criteria rather than through all encompassing descriptions. These criteria should be used as rules to guide the description and definition of the intervention within a research protocol. The criteria are: (1) Theoretical integrity, the protocol needs to make explicit the theoretical assumptions of the intervention, these assumptions need to make conceptual sense and be linked to established theory. (2) Domain completeness, the protocol needs to make explicit exactly what the intervention entails, no ‘black boxes’.

559

(3) Multiple dimensions, if the intervention is multidimensional then the protocol should provide a thorough description of the different dimensions within it and how these relate together. (4) Granularity, the protocol should provide a sufficient level of detail to adequately describe and locate the intervention within the wider family of interventions of which it forms part (e.g. four layer bandages and short stretch bandages are within the family of compression treatments for venous leg ulcers). (5) Parsimony and non-redundancy, the protocol defines the interventions, including complex interventions, in an efficient, non-burdensome, and non-redundant way. In other words there is a need for some consistency, avoiding the temptation to repackage the same intervention with lots of different descriptions. (6) Clinical and research utility, the protocol should be clear and explicit enough to ensure that it can be readily transferred into clinical practice and/or be replicated by other researchers. (7) Reliability, the protocol is used and interpreted similarly across different treatment settings, different users, different diagnoses, and across time. In the second part of the paper consideration will be given to the extent to which these criteria are apparent in clinical nursing research. 2.2. Clinical intervention research Clinical interventions research is concerned with a range of different types of questions: does it work (efficacy); does it work reliably with a specific population in the real clinical world (effectiveness); is it as good as another therapy (equivalence); is it safe; is it cost effective. Underpinning these questions is a sequential process of knowledge development or theory building. Dickoff and James’ (1968) seminal taxonomy of theory progression in nursing provides a very useful structure for understanding how clinical interventions research progresses. Their taxonomy identifies four levels of theory development: (I) Factor isolation—at this level the focus is on identifying a phenomena or an activity that may have some clinical relevance. The nurse researcher conducts an observational study of patients following surgery for skin grafting and identified that most pain occurred during dressing changes to the donor area. (II) Factor relating—at this level the focus is on building a theoretical context for the original observation considering how the intervention works and whether its effect is mediated by other factors. Following the initial association between dressing change and pain the nurse researcher identifies that dressing type and frequency of change impacts on pain independent to the use of prophylactic analgesia.

560

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

(III) Situation relating—at this level the focus is on organising the factors into an intervention to establish whether it can effect the target outcome and to determine the optimal formula for that effect (questions of efficacy). The nurse researcher conducts a serious of small experimental studies to assess the impact of various dressing types for the donor area and frequency of changes considering the effects on pain, wound healing and infection. Following these experiments an optimal dressing regime is identified in an explicit protocol. (IV) Situation producing—at this level the focus is on assessing the probability that the intervention will produce a predictable effect on patients with a certain problem (questions of effectiveness). The nurse researcher conducts a series of larger experiments to test out the protocol in a range of different treatment centres. The results are positive and replicated in subsequent trails. The practice is recommended in national guidelines. This process of theory development follows the different phases of clinical interventions research as detailed in Fig. 2 (Jadad, 1998). This figure also emphasises the relationship between this process and clinical practice. Clinical interventions must ultimately be related to patient problems or needs. This process for clinical research has been adapted in recent years to accommodate more ‘complex’ topics. The Medical Research Council (MRC) has produced a framework for managing complex research topics. They suggest that: ‘‘the greater the difficulty in defining precisely what exactly are the ‘active ingredients’ of an intervention and how they relate to each other, the greater the likelihood that you are dealing with a complex intervention’’ (Campbell et al., 2000a,b, p. 456). The Improved Clinical

Effectiveness through Behavioural Research Group (ICEBeRG) (2006) identified a number elements which can vary within an intervention: content (what is done); intensity (frequency and depth to which it is done); method of delivery (face-to-face, telephone, computer); duration of follow-up (six months, one year, or two years); and context (primary care or secondary care). They point out that varying only these five elements produces 288 combinations. The MRC complex evaluation framework is in many ways similar to the standard phases of clinical research outlined in Fig. 2, with an added emphasis on developing strong underpinning theory and managing multi-faceted interventions. The framework has four phases:  Phase I, the modelling phase: in which the intervention is modelled or simulated to increase knowledge of the components of an intervention and their interrelationships (this work can be qualitative or quantitative).  Phase II, involves exploratory trials focusing on the acceptability, feasibility and efficacy of the intervention. This work will include considering the patients experiences (preferences) and different versions of the intervention may need to be tested to achieve optimal effectiveness.  Phase III requires a definitive RCT designed appropriately to manage the complexity of the study this may involve factorial designs, cross-over studies or cluster trials. The key is to insure that all the main confounders are covered and that in the case of interventions with multiple components there is a potential to compare the different elements.  Phase IV, long-term follow-up and replication. Campbell et al. (2007) have advocated some refinements to the MRC framework emphasising the need to:

Fig. 2. Clinical intervention research (sequential knowledge development).

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

 define and understand the clinical problem;  conceptualise the problem using established theory to identify the argument for the relationship between the intervention and the problem;  collect diverse evidence to enable more expansive assessments of confounding variables;  optimise the intervention prior to any trial work;  optimise the method and system of measurement to ensure a valid assessment. Campbell et al. (2007) advocate an iterative process prior to a definitive trail to ensure that the intervention works at its most efficient level and that the methods are sensitive to the intervention while remaining true to the objective defined by the study problem (see Fig. 3). Clearly, there are many examples of nursing interventions that are multi-faceted in nature. It is cautioned, however, that before nurses adopt the complex evaluation approach that they question the true complexity of the intervention. There is a distinction between interventions that are inherently complex (such as patient education models, is it the content, is it the teacher, is it the length of the course and so on) and those that simply package lots of activity together for the sake of expediency or the lack of prior thinking and development. Complexity emphasises the need for careful and detailed protocol development (Given, 2004; Conn et al., 2001). One of the key considerations in complex research is noise management. This management can take two forms: firstly reduction, minimising the amount of information examined at anyone time; secondly selection, using designs (e.g. factorial, or crossover) and analysis techniques (e.g. covariates, or sensitivity analysis) that enable the competing effects of the intervention to be tuned in and out of the

Fig. 3. Iterative process: optimizing therapy and method in relation to context and problem definition.

561

evaluation model. These approaches both come at some cost, however, as either it takes many single studies to build up knowledge or the studies need to be large and multi-faceted (with issues of patient recruitment). Finally, while it is not the intention of this paper to give details of research designs or methods there are two current requirements that are important in clinical research, both of which relate to rules for executing research to help minimise bias and improve quality. The first rule is the adoption of the CONSORT (Consolidated Standards of Reporting Trials) standard for reporting clinical trials (Moher et al., 2001). These standards stipulate the minimal level of reporting required for quality clinical trials with clear statements on participants and outcomes (including adverse outcomes). The second rule is the requirement that all trails are registered at the beginning of the study rather than after the findings are no known to prevent publication bias (the tendency not to publish research findings).

3. Clinical intervention research in nursing This second part of the paper will critically examine some examples of current clinical intervention research in nursing, relating these examples to the principles for clinical intervention research outlined in part one of the paper. To embed the discussion in current nursing research and to identify some specific examples of clinical intervention research in nursing a ‘snapshot’ review of all the original research papers published in Europe’s three leading nursing research journals (International Journal of Nursing Studies (IJNS), Journal of Advanced Nursing (JAN), and the Journal of Clinical Nursing (JCN)) for the year 2007 was undertaken. This snapshot review involved examining the abstracts of all the published studies within these journals. The studies were classified as being either clinical interventions research or non-clinical interventions research (if it was not clear from the abstract then the full paper was examined). Clinical interventions were defined as explicitly described actions, treatments or technologies (physical, psychological and/or social). Sub-classifications of the focus of the research were used as detailed in Table 1, these classifications were generated iteratively as each paper was evaluated. These classifications were used to produce summary statistics of the type of research published. The full papers of all the intervention studies were read by the author (A.F.). In addition, studies categorised as being phase III clinical intervention research were assessed against the following criteria (based on the principles for clinical interventions research outlined in part one of the paper):  Clear statement of underpinning theory (or previous work on mechanism);  Evidence of a clear protocol and granularity (generalisable and parsimonious);

562

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

Table 1 Classification criteria for nursing research Category

Description

Clinical nursing interventions research Descriptive intervention Evaluate intervention phase I Evaluate intervention phase II Evaluate intervention phase III Evaluate intervention phase IV

Empirical work describing a specific nursing activity or intervention Earl development work (e.g. pilot work or proof of concept) Test efficacy of nursing intervention Test effectiveness of nursing intervention Long-term follow-up of nursing intervention

Non-clinical nursing research Descriptive patient studies Descriptive professional Descriptive of nursing Evaluate nursing Tool development (research) Tool development (clinical) Care system System evaluation Educational evaluations

Studies describing patient issues/problems/experience Studies describing professional nursing issues Studies describing nursing care in general (no specific activity) Studies evaluating nursing roles rather than activities/interventions Development or refinement of research tools (e.g. QoL measures) Development or refinement of tools for clinical assessment Studies describing care systems (structures, processes, services) Studies evaluating care systems (structures, processes, services) Studies evaluating professional education programmes

 Evidence of sequential knowledge development (linkage to previous work and explicit phase of knowledge development);  Concordance with CONSORT standards and evidence of safety (adverse event reporting).

Table 2 Intervention types  study phase Phase

Intervention types

n

I

Acupressure Virtual reality distraction (pain management) Clinical protocol Educational intervention Psychological interventions Combined education and psychological intervention Patient centred recording system (diary) Clinical management models

1 1 1 3 1 1

Dressing system (pressure relief) Exercise programme Catheter care intervention Acupressure, acupuncture and reflexology Psychological intervention Educational intervention Group education Combined education and psychological intervention Telephone counselling Dietary/nutritional intervention Patient positioning

1 2 1 3 2 1 3 4

Psychological interventions Educational intervention Combined education and psychological intervention Group education Group education and psychological intervention Tele-follow-up Cannula care intervention Acupuncture Pressure relief Physical exercise intervention

1 1 3

Only phase III studies were included in this extend examination so that the extent to which they were related to prior theory and earlier work could be consider, given that sequential theory development is a major theme in this discussion. 3.1. Clinical interventions research in nursing In total 517 different research papers were identified in the three journals, of these 88% (n = 455) were classified as non-clinical interventions and 12% (n = 62) as clinical intervention studies. The intervention studies were classified as: descriptive (n = 13, 21%); phase I (n = 10, 16%); phase II (n = 22, 35%); and phase III (n = 17, 28%). No phase IV studies (long-term follow-up or repetition studies) were identified. The types of interventions identified are summarised in Table 2 (the descriptive examples are not included as they were not examined in the same level of detail as the empirical studies). The most commonly identified interventions at all phases were psycho-educational interventions of different types. In terms of classic nursing activities there were examples of: pressure relieving interventions; mobility and activity therapies; diet and nutrition; pain management; and clinical procedures (catheter and cannula care). There were also examples of alternative therapies such as the use of acupuncture, acupressure and reflexology. Of the 49 intervention studies less than a quarter (n = 12) were interventions unambiguously specific to nursing care.

II

III

1 1

1 2 2

3 1 2 1 1 3 1

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

The extent to which the phase III studies observed the criteria reflecting the theoretical principles of clinical interventions research (as detailed above) is summarised in Table 3 and expanded upon below: (1) Clear statement of underpinning theory Most of the studies identified some underpinning theory, with the majority detailing the theoretical principle the intervention was based on (e.g. self-efficacy theory). A few studies described the mechanics of the theory relating the intervention to the mechanism of effect. Shin et al. (2007), for example, provided a very detailed model of how a particular acupressure point could influence nausea and vomiting. Duimel-Peeters et al. (2007) provided a detailed account of the different mechanisms of action for different approaches to pressure relief. An example of a study where there was little evidence of any theoretical underpinning was provided by a study examining an intervention for diabetes, comprising: treatment intensification, web-based interaction, text messaging and professional advice (see Kim, 2007a,b; Kim and Jeong, 2007). There was no linkage between these different elements and it is completely unclear what each added to the observed effect. Indeed it is likely that most of the effect was related to the intensification of treatment independent to any other elements of the intervention and as the control group were not intensified by other means, such as routine clinic based contact, no claims for the independent effect of tele-management can be made. As such the findings and conclusions of this study are extremely misleading. There are two factors to consider here: firstly, the need where appropriate to make explicit the established theoretical principles that the intervention is based on; and

563

secondly the need to make explicit the theoretical mechanism of the intervention (following DeJong et al.’s (2004) criteria and the recommendations of Campbell et al., 2007 for complex research). For the former it would be important to state what the theory was and how it was used within the intervention. For the latter (theoretical mechanisms) the nurse researcher needs to make clear how the intervention actually works (and if there are multiple components how these link together). (2) Evidence of a clear protocol and granularity As indicated in Table 3 the majority of the studies detailed clear protocols for the interventions. However, these protocols were very internal to the study and did not use more generalisable terminology. This was particularly true of the psycho-educational interventions in which there were multiple formulations of materials, delivery methods, frequency and duration. There was also very little evidence of granularity, so that it is difficult to use the finding from these studies to generate enduring transferable knowledge. For example, in Glindvad and Jorgensen’s (2007) study, while we are given a clear description of the intervention contents no attempt has been made to categorise the intervention. The description outlines an educational topic guide with a follow-up telephone call. The intervention includes a distinction between oral and individual education but it is not clear what this distinction is. The distinction should be explicit, didactic information versus a specified interactive (or adult learning) model of education. The telephone element intervention could have been defined as a follow-up or booster education for postoperative pain management. Olsson et al.’s (2007) study of a care pathway for rehabilitation for post-hip fracture provides another

Table 3 Number of studies meeting theoretical criteria de Study

Underpinning theory

Linkage to previous work

Evidence of granularity

Clear protocol

CONSORT standard

Phase of theory

Evidence of safety

Lee and Yen (2007) Kim (2007a) Kim (2007b) Reid and Courtney (2007) Maneesakorn et al. (2007) Jiang et al. (2007) Lii et al. (2007) Webster et al. (2007) Shin et al. (2007) Vanderwee et al. (2007a) Tseng et al. (2007) Glindvad and Jorgensen (2007) Lindskov et al. (2007) Duimel-Peeters et al. (2007) Yip et al. (2007) Paul et al. (2007) Vanderwee et al. (2007b)

U – – U U U U U U U U – – U U U U

– U U U U U U – U U U – – U U U U

– – – U U U – – U U – – – – – U U

U – – – U U U U U U – U U U U U U

U – – – U – – U U U – U U – U U U

– – – – – – – – – – – – – – – – –

NA NA NA NA – – NA U – U – NA NA U – – U

564

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

good example of why these criteria are important. This pathway incorporates: a psychological component (motivation and anxiety management); an intensive physical rehab component; and a place of care component (the patients remain on the orthopaedic ward). None of these elements are described in the context of other specific therapies such as motivational interviewing or a specific method of rehabilitation. In addition, there is no explicit theoretical reason why the components of the care pathway have been selected or placed in the pathway. Therefore, as with Glindvad and Jorgensen’s (2007) study, it is not possible to classify the care pathway in a way that infers any transferable meaning. Therefore, it is not an intervention, but rather an adopted set of practices in a particular setting. Such looseness not only thwarts transferable knowledge it also impedes the final criteria suggested by DeJong et al. (2004) that of reliability, making it very difficult to replicate these interventions. Clearly it may not always be possible to define an intervention as part of wider set of interventions. The intervention may be completely novel. Nevertheless DeJong et al.’s parsimony criteria should be observed with the researcher ensuring that the intervention is detailed efficiently in a non-burdensome and non-redundant way. A tip here may be to look through previous studies and existing classification models including established medical and nursing index terms to find the correct terms or to challenge the previous terms in way that is clear to future researchers. In both the examples cited above, the common problem was the complex multi-faceted nature of the intervention. The studies that tested more specific interventions tended to use more precise definitions, Shin et al. (2007) gave a detailed definition of their acupressure point and Vanderwee et al. (2007a,b) gave specific classifications of the type of pressure relieving positions used. These studies describe the interventions in an unambiguous way clearly locating their intervention within the therapeutic area and in a way that makes explicit the theoretical content. This distinction does not mean that nurse researchers should avoid complex topics but they should approach them in an appropriate way following the complex evaluation principles described in part one of the paper. In the Glindvad and Jorgensen’s (2007) study, for example, why test the interactive education and the telephone follow-up simultaneously. Compare the two education interventions first and then the telephone booster, sequential knowledge development. (3) Evidence of sequential knowledge development Most of the phase III studies as indicated in Table 3 cited previous work, although this was often through parallel inference (i.e. its been tried in that setting so why not this) rather than previous developmental phases of the specific intervention being presented. None of the

studies explicitly identified what phase of theory development their work was aiming to establish, although this was generally implicit in their designs and study questions. If nursing intervention research is to progress it needs to be much more integrated, with each study informing the next in a more cohesive way. Such sequential theory building is essential in ensuring the generalisability (or transferability) of an intervention, in reducing research errors and in managing the complexity that often comes with nursing interventions. There were some examples from each study phase illustrating both the benefits and limitations of following, or not following, this sequential approach. The Phase I study: Miller et al. (2007) examined the acceptability and feasibility of a diary to promote oral self-care behaviour in patients undergoing chemotherapy. This small study established whether patients would use the intervention, and through a smaller qualitative component what they felt about the intervention. The study also gave some insights into the mechanism of the diary: one mechanism was knowledge, patients knew what to do; another was that the patients felt more involved in their care, which was motivating (factor relating theory). Another very important observation in this study was the identification of people who did not want to participate in the study, with the common reason being that they did not perceive their oral health to be a problem. This type of information on patient preference is vital in building up an intervention (early phase I development) and as identified by Campbell et al. (2007) it is particularly important in the area of complex interventions. This research team can now make some decisions about what to do next, to either continue with the intervention but identify that it will only work with patients who are concerned about their oral health (using some test of this as an inclusion criteria) or they may also need another intervention that will motivate those who do not perceive oral health as important into accepting the main intervention. If this intervention had been set up in a larger study without this phase 1 theory development work, the practice may have been adopted without being optimise resulting in the exclusion of patients from its benefits (or the trail might have failed due to high attrition levels and the intervention would be lost). These early inquiries are also important in helping to reduce a common problem in nursing research, the underpowered study with the resulting risk of Type II errors (Polit and Sherman, 1990), by allowing an assessment of effect magnitude (size) and assessment of clinical impact (is the intervention worthwhile). Underpowered studies are not just a minor nuisance their findings are very misleading and ethically there is the issue of involving patients in studies which have little possibility of identifying the true effect.

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

The phase II study: A phase II study undertaken by Nakagami et al. (2007) examined the efficacy of a pressure ulcer prevention dressing in a group of older people acting as their own controls (an intervention pressure point with the dressing, compared to a matched control pressure point without the dressing). The study identified the underpinning mechanism for pressure ulcer development in older people (factor isolating theory); they also described the relationship between these factors and the intervention, the dressing reduces friction (and hence sheer forces) and contains an agent called ceramide (factor relating theory), although it is not clear which of these elements is the most active. The study was set up to test whether the dressing would prevent erythema of the pressure areas (situation relating theory) and reported a significant reduction in observed erythema in the intervention group. However, the problem with this study is that researchers seem to think that this was job done, as they go on to claim that the dressing is safe and effective in preventing pressure ulcers. This claim cannot be justified. While safety could be inferred, as there were no adverse events, the sample was small and the potential for a rarer reaction to the dressing cannot yet be ruled out. The claim of effectiveness is more tenuous as none of the patients developed a pressure ulcer in either the control or intervention areas it cannot be said that the intervention prevented anything but the erythema. If they had an empirically established model that predicted the relationship between erythema and likely ulceration then the claim would have some validity, but without this evidence the claim cannot be supported. Given that the safety and efficacy of this intervention is not yet established, should we be advocating it as treatment for patients? It could be said that this is such a low risk intervention that given its potential benefits we should use it without any further investigation. However, such a casual approach to nursing interventions can lead to science by the back door, with nurses experimenting in discriminately on patients in practice, when with a little more investment we can be far more confident about the knowledge of the intervention. Hence, the story needs a conclusion (larger scale RCT) and even an epilogue (repetition studies, statistical meta-analysis or follow-up studies). Sequential theory building demands that knowledge generation is ongoing. The story of the intervention should not stop part way through or even begin at the end. The phase III study: By the time a phase III study is considered the researcher should be clear about the: theoretical principles behind the intervention; the mechanism of action for the intervention; in multifaceted interventions, the contributing effects and dynamics of the different components within the intervention; and a clear estimate of the effect of the intervention. A good example of what such a study might

565

look like was provided by Jaing et al.’s (2007) study. This study assessed a multi-faceted cardiac rehabilitation intervention. While this was a complex intervention it was developed within an explicit theoretical context. They chose the elements for the intervention based on previous work showing a dose dependent relationship between the target behaviour and specific cardiac outcomes (factor relating theory). There was also evidence of granularity as they adopted a classification system based on a previous model of cardiac rehabilitation. The intervention, although multi-faceted, manipulated these established variables and measured outcomes with proven sensitivity to those variables (situation relating theory). Thus, while the intervention is multi-faceted it is underpinned by integrated theory, satisfying DeJong et al.’s (2004) criteria that the different elements of multi-faceted interventions need to be linked together. Without these links it could not be regarded as a single or even a composite clinical intervention it would be a series of parallel interventions modelled simultaneously. A wide range of different methods are needed in supporting these different stages of inquiry to enable the proper development and testing of interventions. Both qualitative and quantitative studies are important and there is much to be said for a mixed-method approaches particularly in relation to more complex interventions. While the RCT should remain the definitive test for many interventions in phase III level assessments, other methods may be equally valid where the population, intervention or outcomes can be better understood with an alternative approach (see Grocott et al., 2002). As suggested in part one of the paper a good design is very important in managing the ‘noise’ (the confounding, compounding and extraneous effects) associated with complex interventions. (4) Concordance with CONSORT standards and adverse event reporting The majority of the studies were CONSORT standard (now a requirement of most journals). The studies that were not up to the standard were generally older or weaker. Such as Kim’s study (Kim, 2007a,b; Kim and Jeong, 2007) which disregarded patients deemed noncompliant from the intervention (i.e. the analysis was not undertaken on an intention to treat basis), emphasising the importance of these standards in reducing bias. A few of the studies provided data on adverse events but generally only when this was an explicit outcome (e.g. pressure ulcers), although most of the interventions were non-invasive or physical in nature. 3.2. Clinical research in nursing—the way forward? It would be inappropriate to make too many claims about the state of intervention research in nursing simply by looking at the output of three European nursing journals in one year. There are many potential biases to consider

566

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

particularly the fact that nurses publish their research in other disorder specific and multi-professional journals. Indeed there is now in many countries an imperative for researchers to publish in high impact journals and these tend to be non-nursing journals. Nevertheless as a snap shot of nursing research activity it does provide some insight into the general range of nursing research. Given the relatively short history of nursing as an academic discipline, the most encouraging observation to make is that there is clearly a lot of research activity and diversity of knowledge generation. However, on the evidence of the content reviewed only a small proportion of nursing research is concerned with developing clinical interventions. Nearly 90% of the studies were non-clinical in nature. These non-clinical intervention studies were largely comprised of studies describing patient problems, issues and experiences and those describing professional issues (see Chart 1). The former clearly indicates a desire to understand the issues and problems affecting patients. Such knowledge should provide a powerful stimulus to the search for interventions to address those problems. As suggested in Fig. 2 clinical intervention research should start here with clearly defined clinical problems. This stimulus clearly needs harvesting given the lack of intervention studies. One of Florence Nightingale’s many sound bites is apt at this point ‘I think one’s feelings waste themselves in words; they ought all to be distilled into actions which bring results’.

The latter (research into nursing) is perhaps a greater concern. A common put down of nursing research is that it is nurses conducting research on themselves, and there is little to counter this jibe in the materials observed within these journals. While nurses, as the largest professional group in health care, rightly attract questions about workforce issues, there is a disproportional amount of nursing research time being spent on this type of research. There are of course many reasons why this type of research is prolific in nursing, with the often cited difficulties of getting funding and access to patients being the most likely factors Meerabeau (2006). However, the danger here is that knowledge generation in relation to what nurses should do to improve the well-being of patients will become stagnant or controlled by researchers outside the profession. As stated at the outset of this paper, this conflicts with the duty of the profession to patients and would be detrimental to the long-term future of nursing as a distinctive therapeutic force in health care. It is unhealthy for any practice-based profession not to be fully engaged in developing the interventions it applies to patients. In such circumstances nursing is nothing more than a delivery system (an intervention itself) passive to the evidence of others. In addition, a number of general limitations have been highlighted in relation to the clinical nursing intervention studies examined, including: a lack of prior theoretical modelling; inadequate and inconsistent description/categorisation of interventions; a lack of theoretical integration and

Chart 1. Non-clinical research.

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

evidence of sequential knowledge development. So how might clinical research in nursing move forward and begin to address some of these problems? To hopefully stimulate further consideration of the issues raised in this paper, some potential remedial strategies are set out below. Firstly, to promote a greater level of integration between studies and to ensure a common set of criteria for categorising nursing interventions, there should be a European consensus (statement and guidance) on interventions research in nursing (North America has already developed its own agenda under the auspices of the Council for the Advancement of Nursing Science (CANS)). This consensus could consider issues to do with the classification (granularity) of interventions and guidance on the stages needed for developing interventions. The consensus could form the basis of an ongoing collaboration of nursing researchers sharing expertise and helping to modulate (harmonise and encourage) research activity. This collaboration could also provide guidance on methods and on the selection of common outcome measures for specific topic areas. Such a consensus would hopefully improve the quality of the studies being developed and ensure a common framework to enable a greater integration of knowledge across the profession. The final function of this collaboration could be to identify some key areas for interventions research relevant to nursing, such as: pain management; dressing products; pressure relief; falls prevention; restraining techniques, etc. Secondly, to promote a greater of level of integration in the approaches and methods used in clinical nursing research there should be a register of nursing intervention studies. This register could provide a dual function: firstly, it would allow ongoing monitoring and standardisation of research protocols; and secondly, it would enable researchers to see the range of work being conducted in different areas (encouraging symmetry of definition). In addition this register could be structured to reinforce the consensus statements in relation to classifications and stages of knowledge development. By encouraging compliance with a consensus statement the quality of studies could be better modulated, with better quality studies being more likely to gain research funding and generate transferable knowledge. Thirdly, the need for nursing to develop clinical research areas in which to develop potential interventions. As illustrated in the model detailed in Fig. 1, bench sciences are important in developing interventions. In nursing there is a need for clinical areas where ideas and interventions can be developed to ensure that they are constructed to achieve optimal impact. This need could be addressed with the creation of virtual laboratories for nursing science. Such laboratories would allow nurses to develop and optimise nursing treatments. These laboratories could be based in designated clinical areas with nursing scientists actively engaged with day to day practice working with a scientific gaze in an ethically managed

567

setting to identify and model the effects of interventions in early pre-clinical studies. Finally (and perhaps most contentiously), nursing researchers particularly when addressing topics core to nursing practice should aim to publish in nursing journals. Presently many nurse researchers spurn the nursing journals in favour of multi-professional or medical journals. While this largely reflects the need to present their work to peers within a disorder area rather than the professional domain, it suggests that as a community we do not value our own professional journals (and if we do not value them how can we expect others to do so). An investment in these journals by researchers, particularly when the focus of the topic is on activities specific to nursing care, would provide a stronger forum for knowledge development.

4. Conclusion In conclusion, this paper has advocated that clinical interventions research is all about generating clinical knowledge that responds to specific patient (or health care) problems or needs. That such knowledge needs to be developed incrementally to establish not just whether something works, but also why, for whom and in what circumstances it works. Clinical interventions research needs integrated studies with each adding more knowledge sequentially building up the level of theory surrounding the intervention. Something which demands a degree of standardisation in the way interventions are described, either following a previous protocol or in justifying and distinguishing a new protocol. Whether the intervention is simple or complex this requires a careful and explicit assessment of the way it works to minimise any unnecessary elements, maximise the beneficial elements and minimise or eliminate any harmful elements. This paper has also considered the current level of intervention research in nursing. The snapshot review of European nursing journals suggested that very little nursing research activity is focussed on developing nursing interventions. Furthermore, there was little evidence that research was being undertaken in a theoretically integrated way. While more specific funding programmes to support clinical intervention research of topics relevant to nursing would help improve this lack of activity, the paper has argued that there is much that the community of nurse researchers can do to address some of these problems themselves. Finding effective solution to patient problems should be the primary impetus for the nursing profession and for the research community within that profession.

Conflict of interest No conflict of interest.

568

A. Forbes / International Journal of Nursing Studies 46 (2009) 557–568

References Bulecheck, G., Butcher, H., Dochterman, J. (Eds.), 2008. Nursing Interventions Classification. fifth ed. Elsevier, St. Louis. Campbell, M., Fitzpatrick, R., Haines, A., Kinmonth, A.L., Sandercock, P., Spiegelhalter, D., Tyrer, P., 2000a. Framework for design and evaluation of complex interventions to improve health. British Medical Journal 321 (7262), 694–696. Campbell, N., Murray, E., Darbyshire, J., Emery, J., Farmer, A., Griffiths, F., Guthrie, B., Lester, H., Wilson, P., Kinmonth, A.L., 2000b. Designing and evaluation complex interventions to improve health care. British Medical Journal 334, 455–459. Conn, V., 2007. Intervention? What intervention?. Western Journal of Nursing Research 29 (5), 521–522. Conn, V., Rauz, M., Wipke-Tevis, D., Maas, M., 2001. Designing effective interventions. Research in Nursing and Health 24, 433–442. DeJong, G., Horn, S.D., Gassaway, J.A., Slavin, M.D., Dijkers, M.P., 2004. Toward a taxonomy of rehabilitation interventions: using an inductive approach to examine the ‘‘black box’’ of rehabilitation. Archives of Physical Medicine Rehabilitation 85, 678–686. Dickoff, I., James, P., 1968. A theory of theories: a position paper. Nursing Research 17, 197–273. Duimel-Peeters, G., Halfens, R., Ambergen, A., Houwing, R., Berger, M., Snoeckx, L., 2007. The effectiveness of massage with and without dimethyl sulfoxide in preventing pressure ulcers: a randomized, double-blind cross-over trial in patients prone to pressure ulcers. International Journal of Nursing 44, 1285–1295. Given, B., 2004. Research for nursing practice: what do we tell practitioners about nursing interventions? Research in Nursing and Health 27, 293–295. Glindvad, J., Jorgensen, M., 2007. Postoperative education and pain in patients with inguinal hernia. Journal of Advanced Nursing 57 (6), 649–657. Grocott, P., Cowley, S., Richardson, A., 2002. Solving methodological challenges using a theory-driven evaluation in the study of complex patient care. Evaluation 8 (3), 306–321. Improved Clinical Effectiveness through Behavioural Research Group (ICEBeRG), 2006. Designing theoretically-informed implementation interventions. Implementation Science 1, 4–12. Jadad, A., 1998. Randomised Controlled Trials. BMJ Books, London. Jiang, X., Sit, J.W., Wong, T.K.S., 2007. A nurse-led cardiac rehabilitation programme improves health behaviours and cardiac physiological risk parameters: evidence from Chengdu, China. Journal of Clinical Nursing 16, 1886–1897. Kim, H.-S., 2007a. Impact of web-based nurse’s education on glycosylated haemoglobin in type diabetic patients. Journal of Clinical Nursing 16, 1361–1366. Kim, H.-S., 2007b. A randomized controlled trial of a nurse shortmessage service by cellular phone for people with diabetes. International Journal of Nursing Studies 44, 687–692. Kim, H.-S., Jeong, H.-S., 2007. A nurse short message service by cellular phone in type-2 diabetic patients for six months. Journal of Clinical Nursing 16, 1082–1087. Lee, J.T., Yen, H.-W., 2007. Randomized controlled evaluation of a theory based postpartum sexual health education programme. Journal of Advanced Nursing 60 (4), 389–401. Lii, Y.-C., Tsay, S.-L., Wang, T.-J., 2007. Group intervention to improve quality of life in haemodialysis patients. Journal of

Nursing and Healthcare of Chronic Illness (in association with Journal of Clinical Nursing) 16 (11c), 268–275. Lindskov, S., Westergren, A., Hagell, P., 2007. Journal of Nursing and Healthcare of Chronic Illness (in association with Journal of Clinical Nursing) 16 (11c) 368–376. Maneesakorn, S., Robson, D., Gournay, K., Gray, R., 2007. An RCT of adherence therapy for people with schizophrenia in Chiang Mai. Thailand Journal of Clinical Nursing 16, 1302–1312. Meerabeau, E., 2006. Poor relations?: nursing and medicine in the english academy. Higher Education Quarterly 60 (1), 52–73. Miller, M., Taylor, A., Kearneya, N., Paterson, G., Wells, M., Roe, L., Hagen, S., Maguire, R., 2007. Evaluation of the feasibility and acceptability of an Oral care diary by patients during chemotherapy. International Journal of Nursing Studies 44, 693–701. Moher, D., Schulz, K., Altman, D., 2001. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA 285 (15), 1987–1991. Nakagami, G., Sanada, H., Konya, C., Kitagawa, A., Tadaka, E., Matsuyama, Y., 2007. Evaluation of a new pressure ulcer preventive dressing containing ceramide 2 with low frictional outer layer. Journal of Advanced Nursing 59 (5), 520–529. Olsson, L.-E., Karlsson, J., Ekman, I., 2007. Effects of nursing interventions within an integrated care pathway for patients with hip fracture. Journal of Advanced Nursing 58 (2), 116–125. Paul, F., Jones, M., Hendry, C., Adair, P., 2007. The quality of written information for parents regarding the management of a febrile convulsion: a randomized controlled trial. Journal of Clinical Nursing 16, 2308–2322. Polit, D., Sherman, R., 1990. Statistical power in nursing research. Nursing Research 39, 365–369. Reid, C., Courtney, M., 2007. Journal of Nursing and Healthcare of Chronic Illness (in association with Journal of Clinical Nursing) 16 (7b) 197–206. Sackett, D.L., Richardson, W.S., Rosenberg, W.M.C., Haynes, R.B., 1997. Evidence Based Medicine. How to Practice and Teach EBM. Churchill Livingstone, New York. Shin, H.S., Song, Y.A., Seo, S., 2007. Effect of Nei–Guan point (P6) acupressure on ketonuria levels, nausea and vomiting in women with hyperemesis gravidarum. Journal of Advanced Nursing 59 (5), 510–519. Tseng, C.N., Chen, C.C.H., Wu, S.C., Lin, L.C., 2007. Effects of a range-of motion exercise programme. Journal of Advanced Nursing 57 (2), 181–191. Vanderwee, K., Grypdonck, M., Defloor, T., 2007a. Non-blanchable erythema as an indicator for the need for pressure ulcer prevention: a randomized-controlled trial. Journal of Clinical Nursing 16, 325–335. Vanderwee, K., Grypdonck, M.H.F., De Bacquer, D., Defloor, T., 2007b. Effectiveness of turning with unequal time intervals on the incidence of pressure ulcer lesions. Journal of Advanced Nursing 57 (1), 59–68. Webster, J., Lloyd, S., Hopkins, T., Osborne, S., Yaxley, M., 2007. Developing a Research base for Intravenous Peripheral cannula re-sites (DRIP trial). A randomised controlled trial of hospital inpatients. International Journal of Nursing Studies 44, 664–671. Yip, Y.B., Sit, J.W.H., Fung, K.K.Y., Wong, D.Y.S., Chong, S.Y.C., Chung, L.H., Ng, T.P., 2007. Effects of a self-management arthritis programme with an added exercise component for osteoarthritic knee: randomized controlled trial. Journal of Advanced Nursing 59 (1), 20–28.