Research impact in the community-based health sciences: an analysis of 162 case studies from the 2014 UK Research Excellence Framework
© Greenhalgh and Fahy. 2015
Received: 6 August 2015
Accepted: 28 August 2015
Published: 21 September 2015
The 2014 UK Research Excellence Framework (REF2014) generated a unique database of impact case studies, each describing a body of research and impact beyond academia. We sought to explore the nature and mechanism of impact in a sample of these.
The study design was manual content analysis of a large sample of impact case studies (producing mainly quantitative data), plus in-depth interpretive analysis of a smaller sub-sample (for qualitative detail), thereby generating both breadth and depth. For all 162 impact case studies submitted to sub-panel A2 in REF2014, we extracted data on study design(s), stated impacts and audiences, mechanisms of impact, and efforts to achieve impact. We analysed four case studies (selected as exemplars of the range of approaches to impact) in depth, including contacting the authors for their narratives of impact efforts.
Most impact case studies described quantitative research (most commonly, trials) and depicted a direct, linear link between research and impact. Research was said to have influenced a guideline in 122 case studies, changed policy in 88, changed practice in 84, improved morbidity in 44 and reduced mortality in 25. Qualitative and participatory research designs were rare, and only one case study described a co-production model of impact. Eighty-two case studies described strong and ongoing linkages with policymakers, but only 38 described targeted knowledge translation activities. In 40 case studies, no active efforts to achieve impact were described. Models of good implementation practice were characterised by an ethical commitment by researchers, strong institutional support and a proactive, interdisciplinary approach to impact activities.
REF2014 both inspired and documented significant efforts by UK researchers to achieve impact. But in contrast with the published evidence on research impact (which depicts much as occurring indirectly through non-linear mechanisms), this sub-panel seems to have captured mainly direct and relatively short-term impacts one step removed from patient outcomes. Limited impacts on morbidity and mortality, and researchers’ relatively low emphasis on the processes and interactions through which indirect impacts may occur, are concerns. These findings have implications for multi-stakeholder research collaborations such as UK National Institute for Health Research Collaborations for Leadership in Applied Health Research and Care, which are built on non-linear models of impact.
The 2014 UK Research Excellence Framework (REF2014) was the first national exercise to measure the impact of research in the higher education sector (Box 1) . It reflected a growing policy interest in demonstrating the benefits of investment in academic research and reducing the waste that occurs when findings are not implemented [2, 3].
The literature on research impact has been summarised in several recent reviews [4–7]. Researchers often assume a direct and linear link between a study and subsequent impact, achieved through academic publications or corresponding lay summaries . Policymakers may assume that they can commission targeted research to solve policy problems. In reality, these ‘knowledge-driven’ and ‘problem-solving’ mechanisms of impact are uncommon . Clinicians rarely read published research or consciously follow guidelines ; and policymakers ask different questions and operate to very different logics, timescales and value systems from researchers [11, 12].
That is not to say that a direct and linear link between research and impact never occurs. If a research finding is simple, unambiguous and uncontested; if it aligns with people’s values and predictions (especially those of local opinion leaders) and/or with a policy window; if its implementation can be trialled on a small scale before practitioners commit; if mechanisms exist (or can easily be made to exist) to provide timely reminders and feedback on the change; if implementing the finding saves money or becomes a legal or professional requirement; and if the implementation is generously resourced and incentivised, the research finding may be taken up directly [12–14].
More commonly, impact from research occurs indirectly. Among clinicians, this happens via ‘mindlines’ (that is, collectively generated and socially shared tacit knowledge, developed in professional communities of practice) . In the policy setting, it occurs when researchers and policymakers, through repeated interaction over time, come to better understand each other’s worlds and develop goals that are compatible if not fully aligned. This process has been described variously as ‘percolation’ , ‘linkage’  and ‘pragmatic muddling through’ . In the commercial sector, it occurs through two-way secondments and ‘value co-creation’ by the entrepreneurial university [18–20]. The principle of co-creation—the collaborative generation of knowledge in its context of application by academics working with other partners, sometimes referred to as ‘Mode 2 knowledge production’ —also underpins networked models of research such as the UK’s Collaborations for Leadership in Applied Health Research and Care (CLAHRCs), in which universities and local National Health Service organisations collaborate to identify research priorities, undertake applied research and build research capacity in the service sector .
Contemporary health research systems are made up of multiple stakeholders with competing interests, and are characterised by (more or less) productive conflicts and bidirectional knowledge exchange. They illustrate why the unenhanced ‘logic model’ of impact, comprising inputs (research funding) → activities (research) → outputs (e.g. papers, guidelines) → outcomes (e.g. changed clinician behaviour, new service models) → impacts (e.g. reduced mortality), is increasingly viewed as over simplistic [4, 5, 18]. Some scholars of impact (notably Buxton and Hanney, who developed the widely used payback framework ) have proposed making the logic model more ‘permeable’, for example by emphasising the interactions and feedback loops that link researchers, research commissioners, knowledge intermediaries and end-users throughout the research cycle.
Other scholars prefer to depict the research endeavour and its various stakeholders as a complex (adaptive) system [22, 24, 25]. Especially when research findings are complex and ambiguous (e.g. where they illuminate the complexity of a phenomenon rather than providing a definitive answer to a simple question); where they fit poorly with prevailing policy priorities; or where their potential stakeholders are many but agreement between those stakeholders is low (perhaps because of conflicts of interest), logic models lose their predictive power [8, 12, 16]. In such cases, it is particularly important to study the processes and interactions by which research priorities are set, studies planned and executed, data analysed, findings made public and implications debated [12, 26, 27]. If we overlook these interactions, there is a risk that any impact assessment exercise will be reductive and naïve.
REF2014 produced a unique and important database of impact case studies, each describing a body of research and impact beyond academia (Box 1). Some researchers have begun to use computerised text mining techniques to generate ‘big data’ from all 6,975 impact case studies in this dataset (save for a tiny fraction of commercially or otherwise sensitive case studies that were redacted) . In this study, we sought to complement that work by manually reading and coding a smaller (though still substantial) sample of case studies submitted to a single sub-panel (A2: Public Health, Health Services Research and Primary Care). Our research questions were (1) What kinds of research designs and impacts were described in these submissions?; (2) What models and mechanisms were invoked to account for the impacts?; and (3) To what extent did the format and scoring system support the assessment of direct, indirect and co-produced impact?
Our study consisted of two elements: a descriptive survey of all 162 impact case studies submitted to sub-panel A2 in REF2014, and a more in-depth analysis of four of these case studies selected for maximum variety in study design, range of impacts and mechanism of impact; all were considered to illustrate some aspect of good practice. Queen Mary University of London Research Ethics Committee approved the study in December 2014 (QMERC1388b). Informed consent was not required because all documents were in the public domain.
TG read and re-read a sub-sample of 30 impact case studies and used these inductively to develop a draft data extraction framework on a Microsoft Excel spreadsheet, informed by a systematic literature review undertaken in parallel . She then went through the full set of 162 cases and extracted data on study design(s), impacts, assumed mechanism of impact (coded as direct, indirect, or co-produced or ‘Mode 2’), and active efforts described by the research team to achieve impact. NF independently coded a random 20 % sample of the case studies and checked TG’s interpretations; differences were resolved by discussion.
Judgement was needed to code and categorise free-text descriptions, especially in the last category. For example, many case studies mentioned one-off coverage of the research findings by the lay press, but this alone was considered insufficient to count as proactive engagement with the media. Similarly, linkage with policymakers or front-line clinicians was counted as ‘active efforts’ if there was evidence that these links were strong and ongoing, preferably pre-dating the research and continuing through the research period and beyond it. If the research had been commissioned (e.g. by policymakers, industry or other interest groups), this ‘pull’ was also classified as a form of linkage. The 31 impact templates (which described the infrastructure for achieving impact in submitting institutions) were also coded.
For the in-depth analysis, we selected a maximum variety sample to illustrate the full breadth of research designs and approaches to impact, and within each design, selected an example ‘model of good practice’ (that is, a study whose impact had been very substantial and also evident across different categories of impact). For each of the four included case studies, we conducted an interpretive analysis considering how the research was linked to the claimed impact, taking account of the different framings and mechanisms of impact described in the introduction. In these four case studies, we also read the papers from the underpinning research and contacted the authors by email and telephone for their own account of the case. Our initial sample was five such studies but the lead author of one responded (positively) only to our initial email and ignored subsequent contact, so after checking to ensure that no further themes had been identified thus far in the excluded case, we reduced our sample to four.
As predicted by RAND Europe , the case study format generally contained rich information, allowing us to make a reasonably confident judgement about the nature and quality of the research, the depth and quality of interactions, and the nature and mechanism of claimed impact. All 162 case studies are in the public domain ; the detailed dataset extracted from them is available from the authors; our inter-rater reliability in coding was 89 %; most discrepancies were minor and stemmed from ambiguities in the coding framework.
Most case studies identified more than one target audience for the research findings: policymakers in 133 (including the National Institute for Health and Clinical Excellence, which leads most national guideline development in the UK), clinicians in 88 and industry in 15. Few (8 our of 162) explicitly identified patients and/or the lay public as an audience, which was especially surprising given that every submitting institution claimed in its impact template that ‘patients and the public’ were a key audience for its work.
In 161 of the 162 case studies, the main mode of impact depicted was direct; in nine of these there was also considered to be an element of co-production and in two, the authors proposed an element of indirect (‘enlightenment’) impact. In one case study, impact was depicted as mainly co-produced with a lesser component of direct impact. Some case studies talked in general (and sometimes unconvincing) terms about promoting public debate.
Each case study claimed impact in (on average) three different areas. In more than two-thirds (122/162), research had influenced a clinical guideline and in more than half, it had had some influence on an international, national or local policy (88/162) and/or changed clinical or public health practice (84/162). Less commonly, there was evidence of a new service model (57/162), or improvement in morbidity (e.g. disease progression, symptom severity; 44/162), mortality (i.e. clear evidence of improved survival; 25/162) or quality of life (5/162) attributable to the research. In 32 case studies, research was linked to cost saving to the health service and in three to documented profits for industry.
Much less commonly (in only 10/162 case studies) was there evidence of strong linkage with front-line clinicians, though again such links may have been assumed and gone undescribed. Surprisingly, only a minority of case studies (38/162) explicitly described efforts to ‘translate’ their research findings into formats accessible to policymakers, front-line clinicians or the lay public (for example by producing leaflets or online materials designed for the needs and learning styles of designated audiences), and only 22 of 162 described a specific training package developed for, and delivered to, a particular non-academic audience. There also appeared to be limited engagement with industry, but this may have been due to the fact that most industry-relevant studies were returned to other sub-panels.
Four in-depth impact case studies: summary of key features
Main mechanisms of impact
BELLS (Dundee) 
Randomised controlled trial
Improved cure rate
Commissioned as ‘high clinical priority’ study by Health Technology Assessment Programme
Reduced referral to hospitals
Ex ante and ongoing engagement of clinicians
Widespread involvement of clinical research networks nationally
High-impact publication in international journal
Cot death (Bristol) 
Well-established and mature programme of ‘niche’ research
Ex ante and ongoing engagement of third sector charity
Skilled knowledge translation (working with knowledge translation experts) to disseminate key messages for lay audiences
Commitment of researchers to the ‘moral work’ of linking the contribution of research participants (bereaved parents) and potential beneficiaries (new and prospective parents)
Social inequality (York) 
Systematic review of observational studies
Shifting the focus of public debate
Energetic and proactive dissemination campaign run through a newly established charitable trust
Achieving political commitment to addressing the issues
Extensive lobbying of prospective and existing politicians and policymakers
Authors’ input to commissions and working parties
Primary focus on outputs for a lay/civic audience with ‘academic’ outputs as a secondary priority
Production of cultural artefacts
Researcher commitment to ‘moral work’
Lay people in public health (Leeds Beckett) 
Co-production, systematic review, service evaluation
Lay health trainer programmes established locally
Local multi-stakeholder partnerships
Online public information resource
Emphasis on ensuring all voices were heard
Ex ante linkage with (and preferred provider status to) national policymakers
The Bell’s Palsy case study from the University of Dundee (Appendix 1) described a well-conducted clinical trial in a community setting whose findings were important, definitive and easy to understand. The study’s impact was direct and readily captured by existing data systems. It was aided by the high quality of the research, the commissioned mode of research (hence, a ‘policy pull’) and extensive ex ante links with front-line general practitioners and the emerging networks for general practice research in Scotland.
The case study from the University of Bristol on sudden infant death syndrome (Appendix 2) illustrates the rare but effective approach of developing proactive links with a ‘niche’ patient charity and paying careful attention to high-quality knowledge translation for a lay audience. It also shows how researchers’ personal, emotionally engaged, ethical commitment to reducing harm through research can provide impetus and justification for the time spent on impact activities.
Passionate commitment to ethical implications was also a feature of Wilkinson and Pickett’s study of income inequality from the University of York (Appendix 3). Strikingly—and very unusually in our sample of 162—the societal impact of this research (promoting debate, influencing political and policy decisions in the UK and internationally) seemed to have been the authors’ primary objective, while publication in ‘high impact’ journals appeared secondary.
Our last case study, Developing the Evidence Base on Lay Health (Appendix 4) was selected because it was the only one in this sample of 162 that was based primarily on co-produced (Mode 2) research. Raw scores for individual impact case studies were not published, but the overall impact submission (template plus three case studies) from Leeds Beckett did not score well. The prescribed format of the impact case study (describe the research and then describe the impact) may have made it difficult for the authors to emphasise the strengths of this work. As the account in Appendix 4 illustrates, the authors did not depict a linear and causal link between ‘upstream’ research and ‘downstream’ impact (this was an impossibility inherent to the Mode 2 design). The authors’ emphasis on principles (inclusivity, inter-sectoral partnerships) and activities (relationship-building, ongoing dialogue with policymakers) may have been unfavourably compared with ‘hard’ metrics of impact in other case studies that had been narrativised in logic model frameworks.
This detailed review of the impact case studies submitted to a single sub-panel in REF2014 showed that the underpinning research for cases submitted to this sub-panel was overwhelmingly university-led and consisted mainly of clinical trials, meta-analyses and modelling studies. Impacts were mostly in the form of changes to guidelines and policies, and the main mechanism of impact was depicted as direct knowledge-into-practice, often achieved via ex ante linkage with policymakers and representation on guideline development groups. Few case studies described patient-relevant changes in morbidity, mortality or quality of life or documented whether the evidence-based guideline was actually being followed. Only a handful placed significant emphasis on indirect impact or collaborative knowledge production. There was a striking mismatch between institutions’ claims to have engaged with ‘patients and the public’ (universally made in impact templates) and the limited range and depth of activities oriented to achieving this described in the individual case studies.
This study illustrates how detailed manual analysis of a small sample of case studies can complement the more automated ongoing analysis of the wider REF dataset. Text mining of the REF dataset has already produced useful quantitative data (such as frequency statistics on different countries to which impact had spread) . There is, however, a trade-off between breadth and depth, given that automated analysis of a massive dataset is unable to tease out meaningful narratives or detailed mechanisms. Further research could fruitfully explore where and how in-depth qualitative analysis could inform and complement text mining approaches and vice versa.
The structure of the impact case study in REF2014 (a four-page document divided into ‘summary’, ‘underpinning research’ and ‘impact’ along with space for references and corroborating sources) arguably implied a direct and linear link between a programme of research and its subsequent impact. Perhaps because of this, and also because of the context of REF2014 (UK higher education’s opportunity to demonstrate that it was internationally competitive and value for money), almost all impact case studies in our sample were presented using a linear, ‘logic model’ framing.
To some extent, this framing worked in that it allowed research teams to draw out the link between particular research studies and resulting changes in clinical guidelines, health policies (international, national or local) and, to a lesser extent, health outcomes and/or some form of cost saving. The narrative form also allowed research teams to express their passion for the topic, explain why it mattered and convey how their research did ‘moral work’ (reducing harm and/or achieving social justice). But as others have predicted previously (see Background), the implicit logic model framing seemed to both invite and reward ‘hard’, quantitative experimental studies and computational models that had clear, measurable and attributable short-term impacts (most commonly, incorporation into guidelines).
Whilst one interpretation of our data is that impact is largely linear and best achieved through quantitative empirical studies (hence, such study designs are ‘stronger’), another is that the more diffuse impacts from fields such as social science and policy research could not be captured, so institutions made a strategic decision not to submit them as impact case studies. The design of this study allows no conclusions about research whose impacts were not submitted to REF2014. But (given that sub-panel A2 potentially covered a very broad church of research designs relevant to primary care and public health) the dearth of designs grounded in the social sciences (such as qualitative studies, evaluations of natural experiments and policy analyses) in impact submissions is consistent with previous claims that such work rarely produces direct and readily measurable impacts . The format of the REF impact case study (strong emphasis on measurable impacts that could be tracked back to the study reported in the ‘research’ section) allowed direct but not indirect flows of influence to be demonstrated. This may have created a bias towards biomedical and epidemiological research and away from health services and policy research (in which impact is inherently less linear, more complex and multi-stakeholder, and thus harder to demonstrate).
None of the case studies in this sample framed impact in terms of the productive (and organic and reciprocal) interactions among organisations and sectors described by impact scholars who have taken a ‘complex systems’ perspective [8, 21, 22, 24–27]. Indeed, perhaps our most significant empirical finding is the mismatch between the sophistication of theoretical approaches to assessing impact published in the specialist ‘research on research’ literature (see ‘Background’) and the direct and linear way in which the research-impact link was actually depicted and scored in REF2014 submissions.
Almost entirely absent from this sample of impact case studies was any acknowledgement of power differentials or conflicts of interest in the research-impact link. Few case studies describing co-produced research were submitted, but as the example in Appendix 4 illustrates, even those that were submitted did not highlight clearly how power was shared during the research or the extent to which power was redistributed as a result of that research (so-called emancipatory benefits ). Again, this may have been because the REF template did not ask for such information and its linear format implicitly discouraged discussion of such matters. Yet the emerging evidence from CLAHRCs suggests that issues of power and conflict loom large in co-production models of research, and that any impacts are likely to depend heavily on how these issues are handled [22, 31].
Our findings thus suggest that that the high impact scores for Medicine in REF2014, whilst commendable up to a point, are no cause for complacency. The final report from REF Main Panel A strongly upheld the usefulness of the case study format and cautioned against reducing any future exercises to automated metrics . But that conclusion does not mean that the outline structure used in REF2014 is optimal. In particular, our data suggest (though they fall short of proving) that those whose research had indirect, long-term and (therefore) more contestable impacts and those involved in participatory (co-production) models may have been discouraged from submitting impact case studies in this exercise. We know of no comparable exercises in other countries that could help answer the important question: might a different structure for reporting impact support the submission of a broader range of research?
On the basis of our findings, and with a view to ensuring that future research assessment exercises do not inadvertently create a perverse incentive to focus on ‘safe’ impacts at the expense of wider engagement activities, we suggest some changes to the design of the impact case study. First, systematic reporting of the processes and activities oriented to achieving impact should be a requirement. Indeed, in studies such as health policy research where impact is largely indirect, these processes and activities should be allocated a substantial proportion of the overall score. Research teams should not be penalised for impact activities (such as building relationships with policymakers, professional bodies or citizens) that are worthwhile in the long term but unlikely to map to specific, measurable impact metrics in the timescale of the assessment.
Second, we suggest that impact case studies from Mode 2 research collaborations such as action research partnerships (Appendix 4) or CLAHRCs should be assessed differently from more conventional study designs such as randomised trials. In particular, assessment should recognise that in such designs, ‘research’ and ‘impact’ are not always separate and sequential stages but may be two dimensions of co-produced activity; and that activities oriented to effective governance may be crucial to the success of this activity.
Third, given the importance shown in this study of individual researcher enthusiasm and commitment in achieving research impact, it may be worth reconsidering the current model of assigning the entire impact score to the researcher’s current institution. If researchers move institutions, they are likely to continue their efforts to follow through on past research undertaken elsewhere. And such work would be in the public interest—but it will not bring benefit to the new institution.
Our findings highlight both the strengths and limitations of the prevailing assessment system and raise questions about impact measurement more generally. REF2014 both inspired and documented significant efforts by UK researchers to achieve impact. But whilst these efforts, and the short-term direct impacts described in most impact case studies, should be celebrated, further thought needs to be given to how indirect and longer-term impacts can be fully and fairly captured.
We are grateful to the authors of the four impact case studies reviewed in depth for participating in correspondence about them and to Richard Hobbs for helpful feedback on an earlier draft of this paper. The study received no formal financial support but NF’s time was supported by an NIHR Senior Investigator Award to TG.
This study formed a section of a dissertation entitled ‘Research Impact in the Community Based Health Sciences: What Would Good Look Like?’ for TG’s MBA in Higher Education Management at UCL Institute for Education, supervised by Professor Sir Peter Scott and part-funded by TG’s NIHR Senior Investigator Award.
The impact case studies that form the dataset for this study are already in the public domain.
The manuscript is an honest, accurate, and transparent account of the study being reported; no important aspects of the study have been omitted; any discrepancies from the study as planned have been explained.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Higher Education Funding Council for England. 2014 REF: Assessment framework and guidance on submissions. Panel A criteria. London (REF 01/2012): HEFCE; 2012.Google Scholar
- Walshe K, Davies HT. Health research, development and innovation in England from 1988 to 2013: from research production to knowledge mobilization. J Health Serv Res Policy. 2013;18:1–12.View ArticlePubMedGoogle Scholar
- Glasziou P, Altman DG, Bossuyt P, Boutron I, Clarke M, Julious S, et al. Reducing waste from incomplete or unusable reports of biomedical research. The Lancet. 2014;383:267–76.View ArticleGoogle Scholar
- Raftery J, Hanney S, Greenhalgh T, Glover M, Young A. A systematic review of impact measures in health technology assessment. Under Review for Health Technology Assessment. 2015.Google Scholar
- Penfield T, Baker MJ, Scoble R, Wykes MC. Assessment, evaluations, and definitions of research impact: a review. Research Evaluation. 2013. rvt021. doi: 10.1093/reseval/rvt021.
- Milat AJ, Bauman AE, Redman S. A narrative review of research impact assessment models and methods. Health Res Policy Syst. 2015;13:18.View ArticlePubMedPubMed CentralGoogle Scholar
- Grant J, Brutscher P-B, Kirk SE, Butler L, Wooding S. Capturing research impacts: a review of international practice. Documented briefing. Rand Corporation. 2010. http://www.rand.org/content/dam/rand/pubs/documented_briefings/2010/RAND_DB578.pdf. Accessed 16 September 2015.
- Meagher L, Lyall C, Nutley S. Flows of knowledge, expertise and influence: a method for assessing policy and practice impacts from social science research. Res Eval. 2008;17:163–73.View ArticleGoogle Scholar
- Weiss CH. The many meanings of research utilization. Public Adm Rev. 1979;426–431.Google Scholar
- Gabbay J, May A. Evidence based guidelines or collectively constructed “mindlines?” Ethnographic study of knowledge management in primary care. BMJ. 2004;329:1013.View ArticlePubMedPubMed CentralGoogle Scholar
- Davies H, Nutley S, Walter I. Why ‘knowledge transfer’ is misconceived for applied social research. J Health Serv Res Policy. 2008;13:188–90.View ArticlePubMedGoogle Scholar
- Contandriopoulos D, Lemire M, Denis J-L, Tremblay É. Knowledge exchange processes in organizations and policy arenas: a narrative systematic review of the literature. Milbank Quarterly. 2010;88:444–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Grol R, Grimshaw J. From best evidence to best practice: effective implementation of change in patients’ care. The Lancet. 2003;362:1225–30.View ArticleGoogle Scholar
- Greenhalgh T, Robert G, Macfarlane F, Bate P, Kyriakidou O. Diffusion of innovations in service organizations: systematic review and recommendations. Milbank Quarterly. 2004;82:581–629.View ArticlePubMedPubMed CentralGoogle Scholar
- Hanney SR, Gonzalez-Block MA, Buxton MJ, Kogan M. The utilisation of health research in policy-making: concepts, examples and methods of assessment. Health Res Policy Syst. 2003;1:2.View ArticlePubMedPubMed CentralGoogle Scholar
- Lomas J. The in-between world of knowledge brokering. BMJ. 2007;334:129–32.View ArticlePubMedPubMed CentralGoogle Scholar
- Lindblom CE. The science of“ muddling through”. Public Adm Rev. 1959;79–88.Google Scholar
- Hughes A, Martin B. Enhancing impact: the value of public sector R&D. CIHE & UKirc, 2012. http://www.ncub.co.uk/impact. Accessed 16 September 2015.
- Thorp H, Goldstein B. Engines of innovation: the entrepreneurial university in the twenty-first century. Chapel Hill. Chapel Hill, North Carolina: University of North Carolina Press Books; 2013.Google Scholar
- Ramaswamy V, Gouillart F. Building the co-creative enterprise. Harv Bus Rev. 2010;88:100–9.PubMedGoogle Scholar
- Gibbons M, Limoges C, Nowotny H, Schwartzman S, Scott P, Trow M. The new production of knowledge: the dynamics of science and research in contemporary societies. London: Sage; 1994.Google Scholar
- Kislov R, Waterman H, Harvey G, Boaden R. Rethinking capacity building for knowledge mobilisation: developing multilevel capabilities in healthcare organisations. Implement Sci. 2014;9:166.View ArticlePubMedPubMed CentralGoogle Scholar
- Buxton M, Hanney S. How can payback from health services research be assessed? J Health Serv Res Policy. 1996;1:35–43.PubMedGoogle Scholar
- Kok MO, Schuit AJ. Contribution mapping: a method for mapping the contribution of research to enhance its impact. Health Res Policy Syst. 2012;10:21.View ArticlePubMedPubMed CentralGoogle Scholar
- Long JC, Cunningham FC, Braithwaite J. Bridges, brokers and boundary spanners in collaborative networks: a systematic review. BMC Health Serv Res. 2013;13:158.View ArticlePubMedPubMed CentralGoogle Scholar
- Upton S, Vallance P, Goddard J. From outcomes to process: evidence for a new approach to research impact assessment. Res Eval. 2014;23:352–65.View ArticleGoogle Scholar
- Spaapen J, van Drooge L. Introducing ‘productive interactions’ in social impact assessment. Res Eval. 2011;20:211–8.View ArticleGoogle Scholar
- Hinrichs S, Grant J. A new resource for identifying and assessing the impacts of research. BMC Med. 2015;13:148.View ArticlePubMedPubMed CentralGoogle Scholar
- REF 2014 Results and Submissions. http://results.ref.ac.uk. Accessed 16 September 2015.
- Jagosh J, Macaulay AC, Pluye P, Salsberg J, Bush PL, Henderson J, et al. Uncovering the benefits of participatory research: implications of a realist review for health research and practice. Milbank Quarterly. 2012;90:311–46.View ArticlePubMedPubMed CentralGoogle Scholar
- Fitzgerald L, Harvey G. Translational networks in healthcare? Evidence on the design and initiation of organizational networks for knowledge mobilization. Soc Sci Med. 2015;138:192–200.View ArticlePubMedGoogle Scholar
- Higher Education Funding Council. Research Excellence Framework 2014: overview report by Main Panel A and sub-panels 1 to 6. London: HEFCE; 2015. Panel A overview report.pdf. http://www.ref.ac.uk/panels/paneloverviewreports/. Accessed 16 September 2015.
- The Equality Trust. https://www.equalitytrust.org.uk. Accessed 16 September 2015.
- The Equality Trust: Education. https://www.equalitytrustr.org.uk/education. Accessed 16 September 2015.
- South J, Cattan M. Developing evidence for public health policy and practice: the implementation of a knowledge translation approach in a staged, multi-methods study in England, 2007–09. Evidence & Policy: A Journal of Research, Debate and Practice. 2014;10:379–96.View ArticleGoogle Scholar
- Changing clinical practice in Bell’s Palsy: the impact of a clinical trial highlighting the impact of evidence for primary care. http://results.ref.ac.uk/DownloadFile/ImpactCaseStudy/pdf?caseStudyId=39367. Accessed 16 September 2015.
- Bristol research leads to a worldwide fall in the number of cot deaths. http://results.ref.ac.uk/DownloadFile/ImpactCaseStudy/pdf?caseStudyId=40166. Accessed 16 September 2015.
- The Impact of Social Inequality: changing the public and policy debate. http://results.ref.ac.uk/DownloadFile/ImpactCaseStudy/pdf?caseStudyId=43415. Accessed 16 September 2015.
- Developing evidence based practice on lay health roles. http://results.ref.ac.uk/DownloadFile/ImpactCaseStudy/pdf?caseStudyId=13837. Accessed 16 September 2015.