Application of normalisation process theory in understanding implementation processes in primary care settings in the UK: a systematic review

NPT is a middle-range sociological theory that conceptualises implementation, embedding, and integration of innovation in healthcare settings [16]. It provides ‘a set of sociological tools to understand and explain the social processes through which new or modified practices of thinking, enacting, and organising work are operationalised in healthcare and other institutional settings’ [16]. By emphasising the interactions between contexts (encompassing organisational and technical structures), actors (including individuals and groups), and objects (such as clinical practices and procedures), it facilitates examination and understanding of the translational gap between evidence, policy, and practice [17]. It focuses in particular on the work required of stakeholders to embed and normalise innovations in routine practice. More recently it has been used to account for the influence of the characteristics of innovations and contexts themselves in facilitating or impeding this work [18].

Initially developed as an applied theoretical model, the Normalisation Process Model (NPM) sought to assist the understanding and evaluation of factors that act as barriers or enablers for routine incorporation of complex healthcare innovations into practice. While the NPM explained factors that promote or inhibit the distribution of work among stakeholders and supportive resources, it did not address how stakeholders understand, engage with, and evaluate the innovation [19]. To overcome this, Finch et al. developed three further constructs to account for how stakeholders understand and make sense of practice, engage and participate with the innovation, and reflect on or appraise its effects [19]. Through this development, the NPM became NPT which was then further elaborated through an increasingly sophisticated account of the features of context (in particular its ‘elasticity’) and the intervention (its ‘plasticity’) that could themselves influence the viability of actors’ work to normalise newly introduced innovations [18]. Definitions of NPT’s key constructs and sub-constructs are summarised in Table 1.

The use of NPT in health research is growing. Originally used to evaluate e-health and tele-health interventions, its use has spread to a diverse range of health-related settings and interventions [20, 21]. For example, a review by McEvoy et al. found that many authors endorse the use of NPT both in the analysis of wider implementation processes and in guiding recommendations for future implementation work [20]. In our research on the embedding of new practices around education for self-management for a long-term condition (Type-2 Diabetes), we were concerned with the work that individuals and groups have to undertake in order to embed innovation and for it to be sustained into routine practice; NPT pays attention to these dynamics with a contextual focus and robust theoretical basis. We therefore focussed on understanding how NPT has been operationalised in primary care in the United Kingdom (UK) to support the implementation and embedding of complex interventions.

Researchers [20, 22] have advocated for an increased understanding of how NPT might be used to shape implementation processes in ways that promote integration and embedding of complex interventions. A recent qualitative systematic review by May et al. [21], has provided a valuable characterisation and exploration of the contribution of NPT to the implementation of healthcare interventions in variety of health systems. However, the complexity of developing and implementing healthcare interventions in systems with unique characteristics, such as the UK National Health Service (NHS), has not yet been an explicit focus.

Further attention has been drawn to the use of NPT in NHS primary care by our own development and evaluation of a package of practical advice and solutions to increase uptake of structured management education for Type-2 Diabetes Mellitus, by addressing barriers and enablers to uptake at patient, healthcare professional and organisational levels.

Primary care in the UK and the system within which it operates is distinctive from health systems in other developed countries, in terms of its clinical function, organisational arrangements, workforce complexion, and funding.

Organisationally the NHS comprises many organisations and structures. Primary care providers deliver frontline contact for non-urgent physical and mental health complaints and conditions. These include general practitioners (GPs), pharmacists, dentists, and opticians. Where a condition requires specialist treatment or investigation, patients may be referred to another healthcare provider based in a hospital or the community.

While the NHS is traditionally thought of as a public entity, primary care organisations “straddle the public-private sector interface” [23]. For example, general practices are usually small businesses owned by GPs, which combine public sector funding (from general taxation), private ownership by GPs and personal profit. GP partners own practices and are subject to both the benefits and risks of investment and changing revenue streams; salaried GPs are employed by those practices (rather than by the NHS). In addition, a mixture of large commercial organisations or independent businesses provide pharmacy services. This organisational scene is volatile: larger businesses are increasingly encroaching on general practice, there are various arrangements for the provision of out-of-hours GP services, and primary care networks bring together individual practices into larger groupings with a view to increasing the flexibility and responsiveness of general practice, community pharmacy and other primary care services to patients [24].

Primary care in the UK faces major challenges, including increasing workloads, an ageing population, the need to manage increasing complex and multiple medical conditions in the community, and a focus on informing and providing choice to patients in healthcare decision-making [25]. These challenges are further compounded by reductions in capital investment, the need to increase access to care and communication between and within organisations, medical recruitment, and problems in the recruitment and retention of nursing staff [25].

The NHS performs well in managing chronic illnesses, such as diabetes, despite the fact that healthcare spending in the UK is lower than average for comparable countries [26]. This reduced spending has implications for the implementation of interventions, particularly embedding and adopting novel and often increasingly complex healthcare interventions, such as those set out by the Long Term Plan [24]. Adopters of an intervention play a significant role in the replication and translation of an intervention [27]. Achieving replication of an intervention may require staff to adapt and integrate the intervention into pre-existing or new ways of working, potentially requiring the development of new skills, or cultural change [27, 28]. Furthermore, contractual incentives for GP participation in budget holding, partnerships with other organisations to provide co-ordinated care and the sharing of financial risks have further changed the primary care system, meaning that the practicalities of intervention implementation require tailoring to and learning from the context [27, 29].

Given this complexity, it is important to understand how NPT is being applied in order to contribute to an improved understanding of the complexities of intervention implementation, particularly in contexts and interactions that are resource constrained. Therefore, the purpose of this study was to investigate how NPT has been applied to primary care settings in the UK and to consider how it may be used to inform and assess implementation processes. To meet this aim, the objectives were to review the NPT literature in order to: a) understand what types of UK primary care interventions use NPT; b) explore how NPT was operationalised in practice in these examples; and c) examine how authors reflect on the use of NPT in UK primary care settings.

A systematic search of eight relevant databases (British Nursing Index, Ebsco Host Academic Premier, International Bibliography of Social Science, PubMed, Science Direct, Scopus, Web of Science, and Zetoc) was conducted for the period 1 January 2012 to 1 April 2018 for all relevant English-language publications. The search strategy is shown in Table 2.

Hand searching of the reference lists of included studies and relevant systematic reviews was conducted to identify further relevant papers. Citations were managed using RefWorks and divided equally between two authors (JT and LH). Titles and abstracts were screened against the defined inclusion criteria (Table 3). Where uncertainty arose as to whether a paper met the inclusion criteria, it was selected for full-text screening. All papers meeting the inclusion criteria were downloaded for full-text screening by the first and second authors to ensure consistency. Any differences were resolved by discussion among the authors.

A structured data extraction instrument was developed. Authors (JT and LH) independently extracted data from all included papers using the tool and resolved any discrepancies in discussion, referring to the original papers. This enabled concerns or disagreements to be resolved through joint discussion. Extracted data comprised information on author(s), date of publication, study design and methods, sample and setting, topic, and implementation stage (shown in the populated form: see Table 4). Full text articles were then imported into NVivo 10 for analysis.

The CASP checklist for qualitative research was used to assess 30 of the 31 articles. Among the three eligible articles describing randomised control trials (RCTs), two reported qualitative elements and were therefore assessed using the CASP checklist for qualitative research, and one was assessed using the CASP checklist for RCTs. Studies were categorised according to the checklist guidance.

Analysis aimed to characterise and describe the use of NPT in UK primary care research, to explore how authors apply NPT throughout the implementation process, and understand how authors reflect on using NPT in UK primary care implementation research.

Initial analysis focussed on understanding and describing the characteristics (e.g. topic, methodology, methods) of the articles eligible for inclusion in this review. Next, we identified how NPT was applied and the stage of the implementation process it contributed to in each article. Frequencies were calculated for these and other study characteristics.

Articles were then reviewed to explore how NPT was operationalised in UK primary care research. A framework comprising four a priori items, relating to the four NPT constructs (Coherence, Cognitive Participation, Collective Action and Reflexive Monitoring) was chosen to sensitise researchers to the data. Analysis commenced with familiarisation with the data. This involved two authors (JT and LH) reading and re-reading the included articles. Data were individually extracted by JT and LH from any part of the paper detailing the use of any of the items.

Next, data relating to authors’ accounts of applying each NPT construct were extracted to examine whether these resonated with the description of the constructs put forward by the developers of NPT in their original exposition of the theory [16, 30]. Finally, all articles were reviewed to ascertain the presence of authors’ reflections on NPT use in primary care settings. Where these were identified, the researchers coded and organised these into broad themes.

The database and hand searches identified 325 articles, 31 of which were eligible for inclusion in this review. Fig. 1 details the process undertaken and provides information on the number and reasons for the exclusion of articles at each search stage.

No papers were excluded during the quality assessment process: all 31 articles were deemed to be of sufficient quality according to our application of each checklist (see Additional file 1).

The diverse ways in which NPT has been used to develop and evaluate primary care interventions are summarised in Table 4. Briefly, twenty-nine studies were described in 31 articles, and published across 14 peer-reviewed journals.

Most papers were published towards the end of the search period: from 2016 onwards (Fig. 2). Twenty-eight articles reported studies conducted within the UK only (England n = 22, Scotland n = 5, and Wales n = 1) [31‐58]. The three remaining papers, which concerned cross-cultural communication, were conducted in European countries in addition to the UK (England and Scotland) [59‐61].

Qualitative research designs were most frequently reported (n = 25), followed by mixed-method studies (n = 4). Quantitative and document analysis designs were utilised once each. Among the 25 articles using qualitative methods only, ten used semi-structured interviews [32, 39, 44, 45, 47, 49, 50, 55, 57, 58]; one utilised focus groups [60]; seven involved a combination of semi-structured interviews and focus groups [34‐38, 52, 56]; six were ethnographic studies [33, 46, 48, 51, 53, 61]; and one used in-depth interviews [41]. The mixed-methods studies used varying combinations of semi-structured interviews, focus groups, survey data, patient data, and narrative and systematic reviews of the literature [31, 40, 42, 43]. The document analysis article adopted a mapping approach and the quantitative paper utilised a cross-sectional survey [54, 59].

While all 31 studies were located within primary care settings, four studies also included secondary care, community health services, or a combination of these [36, 37, 46, 55]. Samples in the studies comprised patients, carers, healthcare professionals, administrators, commissioners, policymakers, and government and non-government agencies. Sample sizes ranged from 12 to 419 participants.

NPT allowed for the identification of multiple influences across individuals and organisations, which both challenge and enable intervention implementation, embedding, and sustainability. A strength of NPT is that it takes account of the inter-relationship between social and structural constraints and individual agency – something which may be missed when individual behavioural change theories are used [50]. However, one paper suggests that focussing on the individual and collective agency of actors who are delivering the intervention risk neglecting the agency of those who receive the intervention – particularly at a patient or service user level [44]. The use of additional theories may also extend aspects of NPT to enable a better understanding how interventions are received, operationalised, and regarded by all stakeholders within a service. For example, Band et al. used NPT to characterise the everyday use of a blood pressure self-management intervention [31]. Combining behavioural analysis (NPT and the Behaviour Change Wheel), intervention planning (including the evidence base), and logic modelling, the authors claimed they achieved an in-depth and multifactorial understanding of the barriers and facilitators most relevant to the intervention and the user population [31].

The authors of twenty-five articles reflected on their use NPT [31‐45, 47‐53, 55‐61]. The majority of these articles (n = 16) reported only advantages of using NPT, while three reported only disadvantages, and six reflected on both advantaged and disadvantages. Following coding, authors’ reflections were organised into 12 themes, shown in Table 5 with example quotations.

The key findings of this review are that: (i) NPT provides an effective and flexible method for understanding a diverse variety of interventions implemented in UK primary care settings; (ii) NPT offers researchers the tools to understand the theoretical and practical challenges of implementation design and evaluation across, and within, complex health systems, such as UK primary care; (iii) NPT provides a constructive framework for explaining critical implementation processes for interventions focussed on the management of chronic health and multi-morbid conditions; and (iv) NPT appears helpful in understanding the implementation and evaluation of interventions in resource-constrained contexts.

This review identified 31 articles reporting on 29 studies of complex healthcare interventions and implementation processes in UK primary care settings. Eighteen [4, 32‐35, 37, 39, 41, 43, 44, 46‐48, 50, 51, 56‐58, 61] of the articles identified in this review were included in a systematic review of 130 papers detailing NPT use in feasibility studies and process evaluations of complex health care interventions, authored by May et al. [21]. The articles in this review used NPT to evaluate, and to a lesser extent develop interventions. All but one of the articles used qualitative methods to collect and analyse data in ways that facilitated intervention and implementation planning and the identification of the dynamics of embedding interventions at the individual and organisational level.

As found in previous reviews [20, 21], the use of NPT in primary healthcare settings has largely focussed on feasibility and process evaluation studies. Among the articles in this review exploring NPT use in UK primary care settings, none used NPT to explore the entire implementation course of an intervention. However, this is not uncommon; May et al. [21] in their recent review identified only one longitudinal study [62].

Our findings indicate that NPT provides a consistent representation and explanation of the processes of intervention implementation in UK primary care, irrespective of the focus of the intervention. This suggests that authors are able to use NPT flexibly to depict theoretical and practical problems in intervention design and evaluation. Furthermore, the use of NPT, in either combination or isolation, is seen by authors to offer valuable contributions to the implementation of interventions targeting long-term conditions or multiple morbidity within a complex and resource-constrained system.

Despite a number of authors experiencing challenges with operationalising constructs, this review indicates that NPT constructs can be harmonised to inform prospective evaluation studies and this is demonstrated by the integration of NPT with other theoretical models [31, 40, 43]. Furthermore, this review has identified that NPT is able to move beyond organisational interventions, to those that seek to mobilise a range of stakeholders in various primary care settings, through the successful explanation of the outcomes of such intervention studies.

The flexibility of NPT is further highlighted by the use of constructs as sensitising concepts about implementation, with findings related back to the underpinning theory [41] and by the shaping of analytical frameworks to drive implementation refinement [43, 44]. In McEvoy et al.’s review, the authors cautioned of the danger of applying NPT as a “conceptual straitjacket” [20]. In this review, we have identified articles offering similar cautions. Nevertheless, the fact that findings may fall outside NPT’s constructs should not be considered problematic in itself, since the originators of NPT suggest that the theory can be deployed in multiple ways [17], and should be used heuristically rather than mechanistically. The use of NPT in implementation studies provides researchers with a holistic understanding of routines of new practice, as well as the implied and inferred activities that occur in making sense and apportioning meaning of interventions.

In contrast to the review by McEvoy and colleagues [20], this paper highlights the use of NPT prospectively, both in terms of intervention development and in the design of data collection and analytical instruments. In common with findings in the recent systematic review by May et al. [21], several authors considered NPT to lack consideration of the patient perspective or the perspective of those receiving, rather than delivering, the intervention. The originators of NPT do not restrict its focus to professional activity; indeed, Carl May and colleagues have explicitly applied it to the burden placed by the treatment of disease on patients and carers in developing the concept of ‘minimally disruptive medicine’ [63, 64]. The exclusion of patients and carers from NPT-informed analysis may arise because they are not centrally involved as stakeholders in implementation processes. It also, however, appears to reflect a view among some researchers that the theory pertains primarily to professional work. Finally, this review points to an increased awareness of the use NPT in whole-system analysis, with a number of articles reporting the inclusion of multiple stakeholder perspectives.

While a comprehensive search was undertaken, it is possible that some studies were missed, particularly if they did not explicitly state that they took place in primary care. However, it is debatable whether this would have influenced the overall findings. It is also important to note that our search strategy included only the English spelling of ‘Normalisation’ (as in the original usage) and as such, those articles choosing to use the American English spelling (Normalization) will not have been included in those articles retrieved. However, it is likely that at least some of these missed published articles, using the American spelling of ‘normalisation’, will have been retrieved using the abbreviation ‘NPT’ in our search. The a priori analysis plan in this study included three distinct phases. Firstly, we identified and characterised studies that had utilised NPT in the development and evaluation of interventions in primary care. Next, the fidelity of NPT constructs and sub-constructs use by authors was explored. Finally, we focussed on authors reflections on NPT use in UK primary care settings. While other researchers may be able to replicate this approach to reviewing the NPT literature, it is to some extent an inherently subjective process, and differing experiences, backgrounds and interests may mean that alternative conclusions may be drawn. While our findings may add to the evidence base concerning NPT and its use in healthcare, this review covers only UK-based primary care studies, and so extrapolation to other healthcare systems and settings may be limited.

The majority of included articles lacked description of the researcher-participant relationship, including bias and reflexivity, and while authors generally stated that ethical approval had been obtained, discussion of ethical or study-related considerations were limited. Limited reporting of these issues may be related to restrictions in publication word counts. Researchers might also wish consider ways to include reflexivity in their reporting of implementation studies. Authors’ rationales for the use of NPT was also lacking, meaning that it was not possible to ascertain if other theories may have been considered. In the future authors may wish to explain why NPT was selected, whether in preference to or in conjunction with other theories, identifying limitations and strengths of NPT in relation to other theories or frameworks that assist in implementation development and evaluation. This would assist researchers in assessing the relative merits of NPT and alternative theoretical frameworks. Similarly, researchers may wish to expand on the challenges of adopting NPT. This would assist others in implementation planning, and in making decisions about whether normalisation of an intervention would be feasible in the real world, and potentially across contexts.