Diffusion Of Innovation

By James W. Dearing and Jeffrey G. Cox doi: 10.1377/hlthaff.2017.1104 HEALTH AFFAIRS 37, NO. 2 (2018): 183–190 Diffusion Of Innovations Theory, ©2018 Project HOPE— The People-to-People Health Principles, And Practice Foundation, Inc.

James W. Dearing (dearjim@ ABSTRACT Aspects of the research and practice paradigm known as the msu.edu) is a professor in the diffusion of innovations are applicable to the complex context of health Department of Communication at Michigan State University, care, for both explanatory and interventionist purposes. This article in East Lansing. answers the question, “What is diffusion?” by identifying the parameters Jeffrey G. Cox is a research of diffusion processes: what they are, how they operate, and why worthy associate in the Department innovations in health care do not spread more rapidly. We clarify how the of Communication, Michigan State University. diffusion of innovations is related to processes of dissemination and implementation, sustainability, improvement activity, and scale-up, and we suggest the diffusion principles that can be readily used in the design of interventions.

n synthesizing many studies from differ- series of developmental, demonstration, and ent disciplines about how people re- evaluation grants from the Robert Wood John- spond to new ideas, Everett Rogers son Foundation beginning in 2003, more than was answering a call set forth by the 200 Green Houses were in operation across the sociologist Robert K. Merton: theorize, US in 2017 with 300 expected by the end of 2018.6 Ibut in empirical ways and with practical impli- Project ECHO and the Green House model cations.1 Now, fifty-six years past the first are evidence-based innovations that are spread- publication of Rogers’s book Diffusion of Innova- ing as new ways to deliver health care, but have tions, we briefly review this theory, its principles, they diffused? To assess the diffusion of an inno- and the implications for practice as a fifteen-year vation, one must attend to its denominator. In update to the book’s last edition in 2003. these examples, the number of plausible and One of the best documented if frustrating prin- potential adopting sites for either of them is ciples of diffusion is that it can take a long time. large, with 4,134 Medicare-certified rural health Consider the case of Project ECHO (Extension clinics in 2015 and 15,583 certified nursing for Community Healthcare Outcomes), previ- facilities in the US in 2016.7 In diffusion terms, ously reported in Health Affairs.2 This innovation even after fourteen years and like many other in how academic medical centers partner with health care innovations, impressive innovations rural primary care clinicians to extend specialty such as Project ECHO and the Green House mod- care began at one site in New Mexico in 2003. el still have not reached “takeoff” or a tipping By November 2017 Project ECHO reported 158 point in time on a national diffusion curve.8 sites across the US, with sixty more sites in twen- ty-four other countries.3 The program has moved from hepatitis C care to include HIV/AIDS, geri- What Is Diffusion? atrics, psychiatric medication management, and Diffusion is a social process that occurs among more.4 Or consider the Green House model of people in response to learning about an innova- nursing home care, in which “house-like” facili- tion such as a new evidence-based approach for ties are built that emphasize an open kitchen, extending or improving health care. In its classi- residents’ control in decision making, and em- cal formulation, diffusion involves an innova- powered nursing assistants.5 Underwritten by a tion that is communicated through certain chan-

February 2018 37:2 Health Affairs 183 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Diffusion Of Innovation

nels over time among the members of a social sion to reject an innovation socially confirmed.12 system.9 The typical dependent variable in In the case of voluntary adoption decisions, diffusion research is time of adoption, though acceleration in the rate of diffusion is usually when complex organizations are the adopters, the result of influential members of the social subsequent implementation is a more meaning- system making the decision to adopt and their ful measure of change. Diffusion can be assessed decision being communicated to others, who among individuals such as members of Con- then follow their lead. To use the example of gress, organizations such as health care insur- efforts to reduce tobacco use, while a small sub- ers, or larger collectivities such as cities and set of tobacco taxation policy experts, child wel- states. Exhibit 1 illustrates the relationships be- fare specialists, or mayors may make careful as- tween rates of adoption and how we characterize sessments of the evidence and other attributes of diffusion under different scenarios, including an innovation, most of their eventually adopting when innovations are introduced and do not dif- peers do not. When opinion-leading individuals fuse. When time-of-adoption data are graphed and organizations adopt an innovation, social cumulatively, an S-shaped curve is common, systems convert from one normative state (such with an initial slow rate of adoption giving way as smoking in public being acceptable) to anoth- to a rapidly accelerating rate, which then slows er (smoking being unacceptable).When opinion as fewer nonadopters remain within the social leaders do not adopt an innovation, systems do system in question. Not all instances of diffusion not change. Diffusion is an atypical outcome, play out this way, especially in policy diffusion— since the vast majority of innovations fail to dif- where time to adoption can be shorter because of fuse, never accelerating up an S-shaped curve.13,14 the occasional convergence of national attention This can be a wholly warranted result, since an to a problem, financial incentives, readiness for innovation is defined simply as that which is change among elected officials, motivated and perceived to be new—not necessarily better— organized groups, and an innovative solution by potential adopters. Unworthy innovations that is perceived positively.10 sometimes diffuse, and effective innovations As exhibit 1 suggests, several contextual as- are often stymied. pects of diffusion typically go unstudied. Com- Over time through waves of innovations, peting or complementary innovations are impor- diffusion changes societies. Sometimes these tant, since potential adopters usually have a changes manifest as differences in knowledge, choice in what to adopt. Failures are important, disproportionate access to government and com- since most innovations do not diffuse. Decelera- mercial services, and worsening inequality be- tion is important in two ways, since the decision cause resource-rich communities tend to adopt to adopt an innovation often means abandoning innovations early relative to poor communities.15 a prior one,11 and nonadopters have their deci- In this special issue of Health Affairs, for exam-

Exhibit 1

The context of diffusion

SOURCE Authors’ analysis. NOTE Each curve represents a separate hypothetical innovation.

184 Health Affairs February 2018 37:2 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. ple, researchers report that rates of adoption of they may engage in a secondary search for the annual wellness visits were lower among prac- evaluative judgments of trusted, expert, and ac- tices caring for poor communities.16 Resource- cessible others—that is, opinion leaders—who rich communities with greater concentrations of are more discriminating and less susceptible to professionals exhibit greater capacity to acquire influence.23 The seeking of advice or the model- and make use of innovations.17 Even when low- ing of one’s behavior on what others do is a income communities also benefit from innova- heuristic that often reflects an emotional desire tion adoption, gaps between the haves and for status and that allows the decision maker to the have-nots can widen. A study of forty-four save time while reducing uncertainty. Taken criminal justice policies and their diffusion from together, an innovation’s attributes and social 1960 to 2008 shows that states are more likely to influence can be thought of as psychological adopt policies that benefit privileged segments and sociological barriers that serve to protect of the population and weaken marginalized seg- the potential adopter from unworthy innova- ments.18 Fortunately, diffusion principles can be tions. At the level of the social system, this man- used in ways that stimulate the spread of inno- ifests as no or partial diffusion, or a very slow rate vations specifically in low-resource settings,19 a of adoption. strategy known as purposive dissemination or Needs and motivations differ among people designing for diffusion.20 according to their degree of innovativeness (ex- hibit 2). Based on Everett Rogers’s meta-review of empirical studies,9 the first to adopt (innova- Factors That Affect Diffusion tors) tend to do so because of excitement over Diffusion or the lack thereof is often well ex- novelty and feeling unconstrained by social plained by three general sets of variables: each norms; the next to adopt (early adopters, some innovation’s set of pros and cons, or attributes; of whom are opinion leaders) do so because of a the characteristics of adopters, especially poten- measured appraisal that an innovation’s advan- tial adopters’ perceptions of opinion leaders’ re- tages outweigh its disadvantages; and the subse- actions, or social influence; and the larger social quent early and late majorities adopt because and political context, including the salience of they feel social pressure to do so. Laggards are, issues related to the innovation, how proponents like innovators, less susceptible to social pres- and opponents frame the meaning of the inno- sure and feel free to take their time. Campaigns vation, and the timing of its introduction. Per- to spread evidence-based innovations often tar- haps unsurprisingly, given diffusion’s many con- get particular messages to the degree of innova- cepts, diffusion studies have helped form the tiveness (or readiness to change) of potential basis for a number of other areas of study,21 such adopters on the basis of data from formative as dissemination and implementation science in evaluations. Innovativeness reflects individual health.22 thresholds for change: To adopt an innovation When a person learns about an innovation that themselves, those who adopt early require few in they think may have important consequences their reference group to have already adopted; for them or those they serve, uncertainty about how to respond typically leads to a search for Exhibit 2 further information, so the potential adopter can better assess whether the innovation’s attri- Distribution of adopter innovativeness based on time of adoption butes warrant further exploration. The following pros and cons are well codified: cost, or the per- ceived monetary, time, or other resource ex- pense of adopting and implementing an innova- tion; effectiveness, or the extent to which the innovation is perceived to work better than what it would displace; simplicity, or how easy the innovation is to understand and use; compatibil- ity, or how well the innovation fits with estab- lished ways of accomplishing the same goal; ob- servability, or the extent to which outcomes can be seen; and trialability, or the extent to which the adoption decision is reversible or can be managed in stages. Whether or not people engage in such a cost- benefit assessment, if the innovation continues SOURCE Modified from Rogers EM, Diffusion of innovations (see note 9 in text). NOTES This exhibit is to seem promising and consequential to them, based on Everett Rogers’s meta-review of empirical diffusion studies. SD is standard deviation.

February 2018 37:2 Health Affairs 185 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Diffusion Of Innovation

those who wait need more of their contacts to have adopted. Whileeasytoconfuse, Motivations and time of adoption can be pre- dicted by each adopter’s structural position in dissemination activity the network of advice-seeking and advice-giving relationships that tie a social system—an organi- and diffusion zation, community, or virtual network—togeth- processes are wholly er. The pattern of diffusion often begins on the periphery of a relational network, as the first to distinct. try the innovation experiment with it. Central members of the network—the opinion leaders— observe the periphery and then adopt the inno- vation if they judge it to have important advan- tages over current practice. And the many others between the center and the periphery then follow care and Medicaid Innovation established agree- by paying attention to what highly connected ments with accountable care organizations to opinion leaders do and advise.24 This form of partner with hospitals and practices to spread social contagion, an outside-inside-outward pro- the principle of rewarding value over volume; gression of adoption, when graphed cumulative- or partnering with distribution networks as a ly, reflects the S-shaped diffusion curve. pathway to scale, in much the same way as the Potential adopters also perceive the relevance Agency for Healthcare Research and Quality uses of innovations when others like themselves health extension networks to help small primary adopt, even if they are not relationally con- care practices institute preventive cardiac care nected. This sort of imitative effect can result in the EvidenceNOW innovation.28 All of these from having the same job title, same type of pathways to scale still rely on the activation of employer, common training, same hometown, demand from providers or patients as essential or shared beliefs or practices—all of these can for sustained scale-up success.29 lead potential adopters to reject or adopt inno- So diffusion is a form of social activation that vations since homophilous others have done so. may or may not occur after the dissemination of Modelers, forecasters, and experimentalists information or scaling up of services or products have spent considerable time testing the effects has occurred. Diffusion can also occur without of both heterogeneous differences among units organized, intentional dissemination. of adoption and homophilous characteristics of social system members on the decision to adopt25,26 and have shown, for example, that Implementation Science And lagged introductions of innovations across coun- Diffusion Processes tries can actually accelerate diffusion by allowing Implementation science is the study of what hap- potential adopters in later-adopting countries to pens before, during, and after an innovation’s better assess early adopters’ experiences with an adoption occurs, especially in organizational innovation.27 settings.30 Many studies of implementation fo- cus on the period before dissemination, on field- based tests of external validity to understand the Triggering Of Interest And Demand extent to which an evidence-based innovation is While easy to confuse, dissemination activity effective under realistic practice conditions and and diffusion processes are wholly distinct. thus a good candidate for dissemination. A Dissemination refers to activities by proponents smaller proportion of implementation research or intermediaries to inform others of an innova- concerns postdissemination behavior, partly tion, often in terms of segmenting targeted because of the oft-occurring lag for diffusion audiences. Information about an innovation is to occur. transmitted or advertised in what is usually a An implementer is someone who will change one-to-many process using social, mass, or spe- their behavior to use an innovation in practice. cialty media channels—though simply making In complex organizations, the users are often not information available is probably more com- the choosers of an innovation—which can make mon.With innovations that require complex im- the study of implementation fascinating, since plementation, dissemination of information is motivation to use an innovation in practice can joined with the establishment of branch offices, be absent or can even contribute to sabotage. in much the same way that health care providers Historically, little attention to implementation open new clinics; licensing affiliate organiza- has been a major limitation of diffusion re- tions as franchises, much as the Center for Medi- search, most of which focused on physicians,

186 Health Affairs February 2018 37:2 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. farmers, consumers, and other autonomous de- Policy diffusion among the states accelerates cision makers for whom adoption served as a with more federal attention to a problem area reasonable proxy for use. In clinics and other and its policy alternatives.42 types of organizations, the extent and quality Policy diffusion studies have also shown the of implementation and the responses of clients importance of types of intermediary actors, such and constituents are outcomes at least as impor- as professional associations, in diffusion proc- tant as initial adoption. The same can be said esses.43 Policy entrepreneurs are a particularly about the sustained use of innovations after notable type of actor with the ability to pollinate implementation and continued outcomes for political jurisdictions with innovations.44 A poli- patients or other end beneficiaries. Sustainabili- cy entrepreneur combines the functions of a ty is the subject of increasing study by implemen- bridge who ties together disparate groups with tation science and organizational change that of a champion who represents an innovation scholars.31 from one city or state to high-level decision mak- ers in other jurisdictions. Effective policy entre- preneurs are able to talk about innovations as Government Policies As Innovations solutions to public policy problems in ways that Policies have been long studied as innovations are politically palatable.45 Policy entrepreneurs in the diffusion tradition, starting with a seminal have been state representatives, leaders of US study about the spread of traffic-safety nonprofit community organizations, and well- legislation among the states32 to hundreds of known experts within a profession. They work diffusion studies about policies concerning edu- to exploit political windows of opportunity; cation, health, civil rights, and lotteries.33 While frame solutions to problems in politically palat- studies about policy diffusion among the states able ways; and join together disparate individu- suggest rapid imitation once diffusion begins, als, groups, and networks to diffuse policies. the diffusion of policies sometimes demon- strates the same S-shaped curve as do other types of innovations in their cumulative distribution Fidelity, Reinvention, And over time,34 with long latency periods before me- Adaptation dia and public attention are able to propel policy Fidelity is the extent to which an innovation adoption—as was the case with the issue of HIV/ is implemented by others in the way intended AIDS in the 1980s.35 Researchers often concep- by its developers. Fidelity is often measured as tualize more or less time-ordered stages of the correspondence between how a program is policy consideration, adoption, and growth or delivered in tests before scale-up and how the scale-up,36 though such stages have become com- program is later offered by implementing part- pressed over the past century as communication ners in the field.46 Innovation developers differ technology has enabled faster and faster aware- in the degree to which they modify innovations ness of innovations.10 before dissemination, and how much they seek Policy diffusion researchers have found that to maintain control over potential modifications beliefs about an innovation’s effectiveness can by practice-based implementers. Although a be more important than knowledge of actual out- strict adherence to the original procedures comes, again suggesting that who has previously may be desirable to maximize effectiveness in adopted an innovation can be more important the new setting, implementers often make for decision makers than what was previously changes—knowingly or not—to better fit an in- adopted and what effects it had.37 This type of novation to their organization and clients. result echoes the importance of imitation and Fidelity can be affected in the process of diffu- mimicry in studies of other types of innovations sion in two ways: reinvention and adaptation. in other eras and in other countries.38 Policy Reinvention refers to changes made by an inno- diffusion studies show that national policy and vation’s developer to an innovation before its media attention can drive policy consideration at dissemination or scale-up to increase its likeli- the state level,39 as a contextual effect,18 though hood of being adopted and effectively imple- there is evidence that policy attention and enact- mented. These changes often take the form of ment in neighboring states and gubernatorial lessening a “perfect” but costly innovation so agenda-setting can be stronger predictors of that it produces enough benefit to justify its dis- state policy adoption.40 There is also consider- semination to more beneficiaries. For example, able evidence that local successes in cities and the YMCA of the USA reinvented its Diabetes states can become noticed and highlighted at the Prevention Program from a one-on-one counsel- federal level and then diffuse back out broadly to ing intervention led by a medical professional the states as new programs and policies, often to a group intervention facilitated by YMCA with the incentive of funding mechanisms.33,41 personnel—which lowered the program’s cost

February 2018 37:2 Health Affairs 187 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Diffusion Of Innovation

and broadened its reach.47 Adaptation refers to changes made to an inno- Getting off on the vation by implementers who serve intended ben- eficiaries. Adaptations are made by staff in re- right foot in the sponse both to the immediate context of a health stimulation of a care or public health organizational setting and to changes in the external environment that can diffusion process is make or break the sustained applicability of an innovation for improving health and health care. important. Developers who share or cede control of the im- plementation of an innovation, sometimes in- sisting on fidelity to its core components while encouraging customization of peripheral com- ponents, can achieve diffusion through ongoing course corrections and allowing the implemen- novation’s ability to achieve positive outcomes tation strategy to evolve, as exhibited in the across a diversity of sites—needs to be assessed twenty-year history of Health Leads reported (ideally on the basis of theory as well as data) in this issue of Health Affairs.48 Health Leads from the vantage points of stakeholders who will has successfully integrated social needs into clin- implement the innovation.54 Other measures of ical care partly as a result of developers’ willing- readiness also should be assessed, including how ness to cede control. This result—that degrees of potential adopters perceive the attributes of the decentralized control can increase the rate and innovation and the availability of implementa- reach of innovation diffusion—is found in stud- tion support in anticipation of demand from ies of educational and public health innova- providers and patients.55 tions, too. Formative assessment of advice-seeking net- Feedback from field-based implementers so works among potential adopters of an innova- that ongoing results can contribute to an evolv- tion is an important key to the stimulation of ing implementation strategy need not end with diffusion. Such data can statistically and visually developers. The sharing of real-time insights identify which few potential adopters are partic- from implementers to other implementers is a ularly influential when the vast majority of key takeaway lesson from the Center for Medi- others are deciding whether or not to adopt, as care and Medicaid Innovation’s experience, as illustrated in the work of the Translating reported in this special issue.49 Performance im- Research in Elder Care group, based at the Uni- provement methodology does not suit all inno- versity of Alberta. A recent formative study by vations, but health care services in particular this group assessed advice-seeking ties across seem well suited to the incorporation of stake- 958 nursing homes in nine of Canada’s eleven holders’ perspectives into service redesign.50 En- provinces and territories. The results identified abling and supporting adaptation by stakehold- opinion leaders within each jurisdiction, as well ers can produce sustained use of innovations as advice-seeking ties across provinces, so that because of a stronger sense of ownership by im- future resources can be focused on intervention plementers,51 as long as adaptations are fidelity with small proportions of influential individuals consistent.52 and organizations for eventual system change.56 Getting off on the right foot in the stimulation of a diffusion process is important. Diffusion Using Diffusion Concepts To Affect processes often exhibit path dependence, where- Rate And Reach by initial conditions determine how rapidly and Purposive dissemination, or designing for diffu- to what extent an innovation will spread.57 Relat- sion, means taking additional steps early in the edly, the timing of dissemination can be critical process of creating an innovation to increase its to diffusion.58 If potential adopters are attending chances of being noticed, positively perceived, to a different type of problem than the innova- adopted, adapted, and implemented—and, thus, tion addresses, waiting to disseminate can be the successfully crossing the research-to-practice right decision. chasm.53 First of all, one wants to be certain that Learning about and addressing barriers to dif- an innovation should be diffused and that, in so fusion for both end beneficiaries and the health doing, its reach is extended to those communi- care practitioners who serve them is important. ties and population segments where need is Many health care innovations require multiple greatest and capacity is sufficient to adopt and levels of adoption—for example, by a chief medi- implement the innovation to good effect. In pur- cal officer and organizational sponsors, clinical posive dissemination, external validity—the in- chiefs, head nurses, and patients and families.

188 Health Affairs February 2018 37:2 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Formative evaluation along the entire supply Conclusion chain that needs to coordinate for the dissemi- The research and practice paradigm known as nation, supply, delivery, and support of an inno- the diffusion of innovations offers a ready set of vation can reduce barriers before launch.59 This concepts and approaches that can be used to includes attention to perceived incentives, both explain receptivity to health care policies and monetary and intrinsic, which can be tailored to practices by individuals and organizations. address types of stakeholders where formative Diffusion principles can also be operationalized evaluation suggests that barriers to adoption to accelerate the rate of adoption and broaden are high—thus contributing to a climate for the reach of health innovations. ▪ change.60

NOTES

1 Dearing JW, Singhal A. Communi- 8 Morris ZS, Wooding S, Grant J. The p. 91–117. cation of innovations: a journey with answer is 17 years, what is the 20 Dearing JW, Smith DK, Larson RS, Ev Rogers. In: Singhal A, Dearing question: understanding time lags in Estabrooks CA. Designing for diffu- JW, editors. Communication of translational research. J R Soc Med. sion of a biomedical intervention. innovations. Thousand Oaks (CA): 2011;104(12):510–20. Am J Prev Med. 2013;44(1 Suppl 2): Sage; 2006. p. 15–28. 9 Rogers EM. Diffusion of innova- S70–6. 2 Arora S, Kalishman S, Dion D, Som tions. 5th ed. New York (NY): Free 21 Estabrooks CA, Derksen L, Winther D, Thornton K, Bankhurst A, et al. Press; 2003. C, Lavis JN, Scott SD, Wallin L, et al. Partnering urban academic medical 10 Boushey GT. Policy diffusion dy- The intellectual structure and sub- centers and rural primary care namics in America. New York (NY): stance of the knowledge utilization clinicians to provide complex Cambridge University Press; 2010. field: a longitudinal author co-cita- chronic disease care. Health Aff 11 Greve HR. Fast and expensive: the tion analysis, 1945 to 2004. Imple- (Millwood). 2011;30(6):1176–84. diffusion of a disappointing inno- ment Sci. 2008;3:49. 3 Project ECHO: the global ECHO vation. Strateg Manage J. 2011; 22 Norton WE, Lungeanu A, Chambers network [Internet]. Albuquerque 32(9):949–68. DA, Contractor N. Mapping the (NM): University of New Mexico 12 Miner AS, Kim JY, Holzinger IW, growing discipline of dissemination School of Medicine, Project ECHO; Haunschild PR. Fruits of failure: and implementation science in 2017 Nov 9 [cited 2018 Jan 16]. organizational failure and popula- health. Scientometrics. 2017;112(3): Available from: https://echo.unm tion-level learning. In: Baum JAC, 1367–90. .edu/wp-content/uploads/2017/11/ Miner AS, Anderson P, editors. 23 Aral S, Walker D. Identifying influ- ECHOSuperHubs_and_Hubs_2017 Advances in strategic management, ential and susceptible members of 1109.pdf volume 16. Greenwich (CT): JAI social networks. Science. 2012; 4 Madore A, Rosenberg J, Weintrab R. Press; 1999. p. 187–220. 337(6092):337–41. Project ECHO: expanding the ca- 13 Lomas J. Words without action? The 24 Kerckhoff AC, Back KW, Miller N. pacity of primary care providers to production, dissemination, and im- Sociometric patterns in hysterical address complex conditions [Inter- pact of consensus recommenda- contagion. Sociometry. 1965; net]. Cambridge (MA): President tions. Annu Rev Public Health. 28(1):2–15. and Fellows of Harvard College; 2017 1991;12:41–65. 25 Meade N, Islam T. Modeling and Mar [cited 2017 Dec 12]. (Cases in 14 Berwick DM. Disseminating inno- forecasting the diffusion of Global Health Delivery). Available vations in health care. JAMA. 2003; innovation—a 25-year review. Int J from: http://www.globalhealth 289(15):1969–75. Forecast. 2006;22(3):519–45. delivery.org/files/ghd/files/ghd- 15 Buchanan A, Cole T, Keohane RO. 26 Centola D. An experimental study of 036_project_echo_case.pdf Justice in the diffusion of innova- homophily in the adoption of health 5 Zimmerman S, Bowers BJ, Cohen tion. J Polit Philos. 2011;19(3): behavior. Science. 2011;334(6060): LW, Grabowski DC, Horn SD, 306–32. 1269–72. Kemper P. New evidence on the 16 Ganguli I, Souza J, McWilliams MJ, 27 Kumar V, Krishnan TV. Multina- Green House model of nursing home Mehrotra A. Practices caring for tional diffusion models: an alterna- care: synthesis of findings and im- underserved less likely to adopt tive framework. Mark Sci. 2002;21: plications for policy, practice, and Medicare’s annual wellness visit. 318–30. research. Health Serv Res. 2016; Health Aff (Millwood). 2018;37 28 Larson RS, Dearing JW, Backer TE. 51(Suppl 1):475–96. (2):283–91. Strategies to scale up social pro- 6 Baker B. Rebooting the nursing 17 Bhatti Y, Olsen AL, Pederson LH. grams: pathways, partnerships and home. Politico [serial on the Inter- Administrative professionals and the fidelity [Internet]. East Lansing net]. 2017 Jan 11 [cited 2017 Dec 21]. diffusion of innovations: the case of (MI): Diffusion Associates; 2017 Sep Available from: https://www citizen service centres. Public Adm. [cited 2017 Dec 12]. Available from: .politico.com/agenda/story/2017/ 2011;89(2):577–94. http://www.wallacefoundation.org/ 01/nursing-homes-of-the-future- 18 Boushey G. Targeted for diffusion? knowledge-center/Documents/ 000269 How the use and acceptance of Strategies-to-Scale-Up-Social- 7 Henry J. Kaiser Family Foundation. stereotypes shape the diffusion of Programs.pdf State health facts [Internet]. Menlo criminal justice policy innovations 29 Simmons R, Fajans P, Ghiron L. Park (CA): KFF; 2017 [cited 2017 Dec in the American states. Am Polit Sci Scaling up health service delivery: 5]. Available for download from: Rev. 2016;110(1):198–214. from pilot innovations to policies https://www.kff.org/other/state- 19 Green LW, Gottlieb NH, Parcel GS. and programmes [Internet]. Geneva: indicator/total-rural-health-clinics/ Diffusion theory extended and ap- World Health Organization; 2007 ?currentTimeframe=0&sortModel= plied. In: Ward WB, Lewis FM, edi- [cited 2017 Dec 12]. Available from: %7B%22colId%22:%22Location tors. Advances in health education http://www.who.int/immunization/ %22,%22sort%22:%22asc and promotion. Vol. 3. London: hpv/deliver/scalingup_health_ %22%7D Jessica Kingsley Publishers; 1991. service_delivery_who_2007.pdf

February 2018 37:2 Health Affairs 189 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Diffusion Of Innovation

30 Brownson RC, Colditz GA, Proctor state attention to tobacco and vac- public health programs. Am J Public EK editors. Dissemination and im- cines. J Health Polit Policy Law. Health. 2011;101(11):2059–67. plementation research in health: 2014;39(3):565–89. 52 Wiltsey Stirman S, Gutner CA, translating science to practice. 2nd 41 Shipan CR, Volden C. Bottom-up Crits-Christoph P, Edmunds J, Evans ed. New York (NY): Oxford Univer- federalism: the diffusion of anti- AC, Beidas RS. Relationships be- sity Press; 2018. smoking policies from U.S. cities to tween clinician-level attributes and 31 Wiltsey Stirman S, Kimberly J, Cook states. Am J Pol Sci. 2006;50(4): fidelity-consistent and fidelity- N, Calloway A, Castro F, Charns M. 825–43. inconsistent modifications to an The sustainability of new programs 42 Boushey G. Punctuated equilibrium evidence-based psychotherapy. and innovations: a review of the theory and the diffusion of innova- Implement Sci. 2015;10:115. empirical literature and recommen- tions. Policy Stud J. 2012;40(1): 53 Dearing JW, Kreuter MW. Designing dations for future research. Imple- 127–46. for diffusion: how can we increase ment Sci. 2012;7:17. 43 Douglas JW, Raudla R, Hartley RE. uptake of cancer communication 32 Walker JL. The diffusion of innova- Shifting constellations of actors and innovations? Patient Educ Couns. tions among the American states. their influence on policy diffusion: a 2010;81(Suppl):S100–10. Am Polit Sci Rev. 1969;63(3): study of the diffusion of drug courts. 54 Leviton LC. Generalizing about 880–99. Policy Stud J. 2015;43(4):484–511. public health interventions: a mixed- 33 Karch A. Democratic laboratories: 44 Mintrom M, Norman P. Policy en- methods approach to external va- policy diffusion among the American trepreneurship and policy change. lidity. Annu Rev Public Health. 2017; states. Ann Arbor (MI): University of Policy Stud J. 2009;37(4):649–67. 38:371–91. Michigan Press; 2007. 45 Keck ME, Sikkink K. Activists be- 55 Flay BR, Biglan A, Boruch RF, Castro 34 Baumgartner FR, Jones BD. Agendas yond borders: advocacy networks in FG, Gottfredson D, Kellam S, et al. and instability in American politics. international politics. Ithaca (NY): Standards of evidence: criteria for 2nd ed. Chicago (IL): University of Cornell University Press; 1998. efficacy, effectiveness and dissemi- Chicago Press; 2009. 46 Pérez D, Van der Stuyft P, Zabala nation. Prev Sci. 2005;6(3):151–75. 35 Rogers EM, Dearing JW, Chang S. MC, Castro M, Lefèvre P. A modified 56 Dearing JW, Beacom AM, AIDS in the 1980s: the agenda- theoretical framework to assess im- Chamberlain SA, Meng J, Berta WB, setting process for a public issue. plementation fidelity of adaptive Keefe JM, et al. Pathways for best Journalism Monographs [serial on public health interventions. Imple- practice diffusion: the structure of the Internet]. Available from: ment Sci. 2016;11(1):91. informal relationships in Canada’s https://www.researchgate.net/ 47 Alva ML, Hoerger TJ, Jeyaraman R, long-term care sector. Implement profile/James_Dearing2/ Amico P, Rojas-Smith L. Impact of Sci. 2017;12(1):11. publication/263807905_AIDS_ the YMCA of the USA diabetes pre- 57 Greve HR, Seidel M-DL. The thin red in_the_1980s_The_Agenda- vention program on Medicare line between success and failure: setting_Process_for_a_Public_ spending and utilization. Health Aff path dependence in the diffusion of Issue/links/5825d2b908ae7ea5 (Millwood). 2017;36(3):417–24. innovative production technologies. be7b623a/AIDS-in-the-1980s-The- 48 Onie RD, Lavizzo-Mourey R, Lee TH, Strateg Manage J. 2015;36(4): Agenda-setting-Process-for-a-Public- Marks J, Perla RJ. Integrating social 475–96. Issue.pdf. needs into health care: a twenty-year 58 Calantone R, Di Benedetto CA, 36 Mangham LJ, Hanson K. Scaling up case study of adaptation and diffu- Rubera G. Launch timing and launch in international health: what are the sion. Health Aff (Millwood). 2018; activities proficiency as antecedents key issues? Health Policy Plan. 2010; 37(2):240–47. to new product performance. J 25(2):85–96. 49 Perla RJ, Pham H, Gilfillan R, Global Scholars Market Sci: Bridging 37 May PJ. Policy learning and failure. J Berwick DM, Baron RJ, Lee P, et al. Asia World. 2012;22(4):290–309. Public Policy. 1992;12(4):331–54. Government as innovation catalyst: 59 Talke K, Hultink EJ. Managing dif- 38 Conell C, Cohn S. Learning from lessons from the early Center for fusion barriers when launching new other people’s actions: environmen- Medicare and Medicaid Innovation products. J Prod Innov Manage. tal variation and diffusion in French models. Health Aff (Millwood). 2010;27(4):537–53. coal mining strikes, 1890–1935. Am J 2018;37(2):213–21. 60 Parston G, McQueen J, Patel H, Sociol. 1995;101(2):366–403. 50 Batalden M, Batalden P, Margolis P, Keown OP, Fontana G, Al Kuwari H, 39 Karch A. Vertical diffusion and the Seid M, Armstrong G, Opipari- et al. The science and art of delivery: policy-making process: the politics Arrigan L, et al. Coproduction of accelerating the diffusion of health of embryonic stem cell research. healthcare service. BMJ Qual Saf. care innovation. Health Aff (Mill- Polit Res Q. 2010;65(1):48–61. 2016;25(7):509–17. wood). 2015;34(12):2160–6. 40 Pacheco J, Boushey G. Public health 51 Scheirer MA, Dearing JW. An agenda and agenda setting: determinants of for research on the sustainability of

190 Health Affairs February 2018 37:2 Downloaded from HealthAffairs.org on February 07, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org.