Diffusion of Innovation in Health Care - California HealthCare

Diffusion of Innovation in Health Care

Prepared for:CALIFORNIA HEALTHCARE FOUNDATION

Prepared by:

Institute for the Future

Authors:

Mary Cain and Robert Mittman

May 2002

diffusion fnl guts spot 5/15/02 12:08 PM Page 1

Acknowledgments Institute for the Future (IFTF) is a nonprofit research andconsulting firm founded in 1968 and dedicated to under-standing technological, environmental, and societal changesand their long-range consequences. We help public and privateorganizations think systematically about the future by evalu-ating long-term trends and their implications, identifyingpotential markets, and analyzing policy options. Our clientsinclude Fortune 500 companies, public and private foundations,multinational corporations, and government offices and agencies. For more information about IFTF, visit our Web siteat www.iftf.org.

IFTF wishes to acknowledge Jody Ranck for contributions to this report as well as Susannah Kirsch, Jane Sarasohn-Kahn,and Charles Wilson, M.D. for peer review.

The California HealthCare Foundation (CHCF) is an inde-pendent philanthropy committed to improving California’shealth care delivery and financing systems. Our goal is toensure that all Californians have access to affordable, qualityhealth care. CHCF’s work focuses on informing health policydecisions, advancing efficient business practices, improving thequality and efficiency of care delivery, and promoting informedhealth care and coverage decisions.

The iHealth Reports series focuses on emerging technologytrends and applications and related policy and regulatory developments.

Additional copies of this report and other publications in theiHealth Report series can be obtained by calling the CaliforniaHealthCare Foundation’s publications line at 1-888-430-CHCF(2423) or visiting us online at www.chcf.org.

ISBN 1-929008-97-X

Copyright © 2002 California HealthCare Foundation


Contents

4 I. What Is Diffusion?

7 II. The Ten Critical Dynamics of Innovation Diffusion

Relative Advantage

Trialability

Observability

Communications Channels

Homophilous Groups

Pace of Innovation/Reinvention

Norms, Roles, and Social Networks

Opinion Leaders

Compatibility

Infrastructure

26 III. Epilogue

28 Endnotes


4 | CALIFORNIA HEALTHCARE FOUNDATION

HEALTH CARE IS CONSTANTLY EVOLVING. Waveafter wave of new technologies, insurance models, informationsystems, regulatory changes, and institutional arrangementsbuffet the system and the people in it. But people and institu-tions, for the most part, do not like change. It is painful, difficult, and uncertain.

Entire organizations in health care are devoted either to pro-moting innovations—selling the latest drug, imaging system,medical device, software package, or Internet site—or to preventing innovations from disrupting the status quo bycounter-detailing, keeping drug reps away from doctors,requiring certificates of need, or disallowing reimbursement.Trying to change the pace at which new ideas about healthcare spread through the system is a priority of health care professionals; such changes easily have major impacts on cost,quality, and patient satisfaction.

“Diffusion is the process by which an innovation is communi-cated through certain channels over time among the membersof a social system.” So says Everett Rogers, who masterfullyrepresents a vast literature that spans 50 years in his classicDiffusion of Innovations, now in its fourth edition. This reportdraws from that literature to describe how the dynamics ofinnovation diffusion play out in health care.

“Ideas and products and messages and behaviors spread just like viruses do.”

— Malcolm Gladwell, The Tipping Point

I. What Is Diffusion?


About the Report

This report presents the basics of innovation diffusion: the stages of adoption, including the typical “S” curve, and the types of individualswho adopt the innovation at different stages. The ten critical dynamics of innovation diffusionare explored:

1. Relative advantage. The more potential thevalue or benefit anticipated from adoption ofthe innovation relative to current practice, themore rapidly it will diffuse.

2. Trialability. The ability to try out an innova-tion without total commitment and withminimal investment improves the prospectsfor adoption and diffusion.

3. Observability. The extent to which potentialadopters can witness the adoption of an innovation by others improves its prospectsfor diffusion.

4. Communications channels. The pathsthrough which opinion leaders and otherscommunicate about an innovation affect the pace and pattern of diffusion.

5. Homophilous groups. Innovations spreadfaster among homophilous groups (those withsimilar characteristics) than among hetero-philous groups (those that differ in importantways).

6. Pace of innovation/reinvention. Some innovations are relatively stable and do notevolve much while they diffuse. Others evolvemuch more rapidly and are altered by usersalong the way.

7. Norms, roles, and social networks. Innova-tions are shaped by the rules, formal hierarchies, and informal mechanisms ofcommunication operative in the social systems in which they diffuse.

8. Opinion leaders. Individuals whose opinionsare respected (or even just listened to) by othersin a population affect the pace of diffusion.

9. Compatibility. The ability of an innovationto coexist with technologies and social pat-terns already in place improves the prospectsfor adoption/diffusion.

10. Infrastructure. The adoption of many inno-vations depends on the presence of someform of infrastructure or of other technolo-gies that cluster with the innovation.

The dynamics that govern the adoption (or lackof adoption) of new medical and informationtechnologies in the health care industry are com-plex. This report is intended to help those frus-trated with the pace of adoption to understandhow diffusion works and how to affect the paceand style of diffusion in their own organizations.

Diffusion of Innovation in Health Care | 5


Defining Innovation Diffusion

An outgrowth of sociology, diffusion research first looked at how agricultural innovations spreadfrom farmer to farmer. The first researchers in this discipline sought to explain why one farmerwould know about and adopt use of a certain pesticide while his neighbor didn’t. The master ofthis research is Everett Rogers, of the Universityof New Mexico, who has studied the dynamics of the diffusion of innovations for most of his 45-year career. His classic text, The Diffusion ofInnovations, summarizes and interprets decades of diffusion research, identifying basic patterns,categories of adopters, and factors that influencethe decision to adopt.

Two definitions create a lens through which to view technology diffusion in health care.Diffusion is the process by which an innovation iscommunicated through certain channels overtime among members of a social system.(Rogers, 5) An innovation is an idea, practice, orobject that is perceived as new by an individualor some other unit of adoption. (Rogers,11)Though often described as bureaucratic andincrementally changing, health care is also a verydynamic and innovative field. Around the globe,research scientists, private industries, academics,and governmental and nongovernmental agencies work to create new ways to providebetter care, find cures, and improve health.

Stakeholders In Health Care

Medical and information technology adoptiondecisions differ when made by individuals ororganizations. Beyond that distinction, the number of stakeholders potentially affected byany technology adoption decision varies greatly.Once a clinician decides to use a new device or piece of technology, the clinician must often consider not only the impact on the patient and on the practice but also what it means for reimbursement, health care policy, and the organization in which the clinician works. A shortlist of stakeholders involved in a technologyadoption decision are:

� The policy makers and regulators who evaluate the safety and efficacy of the technology;

� The payer, such as Medicare or other insurer, who decides whether payment willbe made for use of the technology;

� The provider organization (in the form ofphysicians and hospitals) that must decidewhether to provide the technology and thenalso get the proper training and education to use it appropriately;

� The patient who must know enough about the technology to give consent for its use; and

� The vendor company that researches, develops, and sells the technology.

Change Agents

The vendor representative often plays an impor-tant role in innovation diffusion as the changeagent. A change agent is an individual who influ-ences clients’ innovation-decisions in a directiondeemed desirable by a change agency (Rogers,27). For example, there are almost 60,000 phar-maceutical sales representatives who are perfectexamples of change agents. Their job is to influence physician decisions in favor of usingtheir drug over another by providing the physicianeducational materials, free samples, junkets, etc.,in an effort to win them over.




1. Relative Advantage

The decision to adopt a technology is influenced by (1) theability of a potential adopter to judge whether the benefits ofusing the innovation will outweigh the risks of using it, and (2) whether the innovation improves upon the existing tech-nology. The more benefit people anticipate from adopting theinnovation relative to what they now do, the more rapidly it will diffuse. According to Rogers, “the degree of relativeadvantage can be expressed as economic profitability, socialprestige, or other benefits.”

Examples of relative advantage as a driver of diffusion� Some technological advances result in huge improvements

in how health care is provided. In 1895, Roentgen discov-ered x-rays during his study of light phenomena, and within six months the first diagnostic radiograph was performed. X-rays gave physicians the ability to look inside the bodywithout cutting it open. Though by the end of 1896 thenegative effects of exposure to radiation were starting to bedocumented, the value of being able to see the structure andfunction of internal organs outweighed the disadvantages ofradiation. X-ray technology diffused rapidly and is still theimaging modality that generates the greatest revenue today.

A more modern example is minimally invasive surgery,whereby surgeons use specialized instruments, including avideo camera, to enter the body via small incisions andmove through the tissue to the affected area without having to expose the entire body cavity. Cholecystectomy,the surgical removal of the gallbladder, has been radicallychanged by minimally invasive surgery techniques. Theamount of trauma to the patient, surgical complications,and time in the hospital with laparascopic cholecystectomyis much lower than with a conventional cholecystectomy.

II. The Ten Critical Dynamics ofInnovation Diffusion


� In many metropolitan markets, having the latest technology is necessary for health systemsto be competitive. Having a high-tech repu-tation increases prestige and is important for attracting the most innovative physicians,as well as for attracting patients. For somepatients getting the best and most up-to-datecare means judging health systems by the cur-rency of the facility and staff. Such patientsmight choose the hospital with magnetic resonance imaging facilities (MRI) over onewith no MRI facilities.

� Relative advantage promotes a technologywhen the innovation is easier to use than theprevious method, often reducing duplicativeand inefficient practice. In the 1980s, computer-based practice management systems wereintroduced into physician practices to stream-line and standardize the business end. Thesesystems automated and consolidated severalaccounting functions previously calculated byhand. Relative benefits increased as regulationand health insurance policies increased infor-mation demands on office staff.

Relative advantage is not always clearly expressedwhen a technology diffuses.

� Relative advantage does not apply when thebenefits of a technology are controversial orare not explicit. For example, the NationalInstitutes of Health and the American Cancer Society have released conflictingrecommendations about the appropriate agefor mammography screening. Lack of con-sensus has confused the public and healthinsurers and hindered the diffusion of mam-mography for middle-aged women.

� The risk of using a technology can be so greatthat its advantage does not seem worth it. For some women with a family history ofbreast cancer, genetic tests can detect specificgenes that significantly increase the risk ofcontracting the disease. With no cure forbreast cancer, for some patients prophylacticmastectomy is the preventive measure. Manywomen choose not to take the genetic testbecause the perceived benefit of knowing islower than the stress caused by knowing, notto mention the perceived risk of discrimina-tion by health insurers.

Lessons for Technology Diffusion � Understand the end user of the technology,

the person who is measuring the relativeadvantage of the new technology, vis á visother decision-making and decision-influenc-ing entities. Though MRIs are very valuablediagnostic tools for many specialty physicians,the technologist using the MRI, the chieffinancial officer who decides how to fund thepurchase of the MRI, and the chair of ortho-pedic surgery, for example, will all weigh in onthe purchasing decision.

� Recognize the impact of significant behaviorchange and be able to illustrate how a newtechnology will offer significantly greater bene-fit if it is expected to be adopted.

� Consider the business case for the adoptionof a new technology. A calculation of thereturn on investment (ROI) helps potentialadopters recognize the tangible benefit ofusing the technology. For intangible benefits,success stories and testimonies from otherswho have adopted the technology are a stan-dard part of most marketing packages.



2. Trialability

Trialability is the ability to try out an innovationwithout total commitment and with minimalinvestment. In classic diffusion research, the easier it is to try out an innovation without hav-ing to commit fully to it, to discard an existingway of operating, or to invest heavily in tech-nology or training, the better the prospects for adoption and diffusion. Trying out an innovationor new technology allows potential adopters to reduce their uncertainty about the risks and benefits. Even when the weight of evidenceargues for or against the benefits of adopting atechnology, personal experience (one way or theother) can overcome the evidence.

ExamplesThe free samples that pharmaceutical companiesdistribute to physician offices are the classicexample of using trialability to promote adoption.Prescription drug manufacturers benefit from thetrialability of their products in two ways—first,as a new drug is introduced, free samples makephysicians aware of the product. Second, oncethe drug is well accepted, free samples helpphysicians introduce patients to a new drug.Beyond drug samples, there are many health caretechnologies that lend themselves to sampling of some sort:

� At professional association conferences andtrade shows for medical and information tech-nology professionals, technology vendors givedemonstrations that walk individuals throughthe mechanics of the innovation and a hands-on trial using it on a dummy or model.

� New clinical (and even administrative) infor-mation systems can be tried out with littlecommitment through Web sites and applica-tion services providers (ASPs) that do notrequire fully installing or converting to thenew system.

� Simple, stand-alone medical devices that donot require extensive training or need to interact with a complex system, such as newsyringes, airway devices, or surgical staples or adhesive strips to hold together skin edges, are, like pharmaceuticals, easy to sample.

� Even when a technology fits into a moreextensive or complex system, some compo-nents of it may be trialable, such as newreagents in an automated lab system or newcontrast media for imaging systems.

� New medical procedures or disease manage-ment strategies may be trialable if they do notrequire a great deal of coordination of care. A single practitioner or office can try them outmore readily than a complex health system.

These last examples point to types of innovationsthat are not easy to try; other approaches to pro-moting diffusion will be more successful.

� Innovations that require a large amount oftraining before they can be successfullydeployed are not very trialable. New combina-tions of medical devices and surgical tech-niques, such as keyhole surgery on a beatingheart or arthroscopic repair of cartilage in kneeand shoulder joints, do not lend themselves to easy trials.

� Innovations that require hands-on experienceto notice the difference may not be very trialable. A new grip on a hand-held device or improved flexibility in a laparoscopic tool may be something a potential user must experience first-hand.

� Technologies that require a large capital expenditure, such as complex radiologicalinstruments, integrated information systems,or telemedicine for remote surgery, are notreadily trialable, though it is possible to bringpractitioners to demonstration sites.



� Innovations that fit into complex legacy systems—human, organizational, technologi-cal, or informational—are difficult to samplebecause they require adaptation and haverepercussions throughout the system. The slowdiffusion of electronic medical records (EMRs),for instance, has come in part because,although it is easy to demonstrate the functionsand benefits of an EMR, it is very difficult toget a practitioner to try it out in a clinical setting because it must fit in with so manyother systems. Similarly, consumer-orientedquality measures interact throughout thehealth system, making them difficult to try.

Lessons for Technology Diffusion� Look for opportunities to carve out some

part of a system that is more trialable, even if an innovation does not obviously lend itselfto trials. Divide the overall process into com-ponent pieces, some of which can be attacked.

� When designing a complex new technologyor system, think about which components of it could be tried out without committing to the full innovation. Make sure that thebenefits are tangible and even more powerfulwhen deployed in the context of the completesystem.

3. Observability

Observability is seeing how an innovation worksby watching someone else use it and then acknowl-edging that the technology is safe and/or bene-ficial. The extent to which potential adopters canwitness the adoption of an innovation by otherscan determine its prospects for diffusion. Themore obvious the evidence of improved experi-ence, increased functionality, and better outcomes,the more likely it will be adopted by new users.

ExamplesA non-medical example of observability is info-mercials. An infomercial is a program that is anadvertisement for a product. During an infomer-cial, an actor shows how much better, faster, andmore efficient a new technology is by demon-strating its use in comparison to the traditionalmethod. The audience is only asked to believewhat it sees and is offered to try it, often with amoney-back guarantee. Health care technologiesare often demonstrated in a similar way:

� During internship and residency programs,physicians are taught by example. When a surgeon is learning a new procedure or howto use a new device, he or she usually watchesa more experienced person using the device,often by watching the operation and thenrepeating the procedure.

� Manufacturers of very complex and expensivemedical devices advertise their innovationsmuch more discriminately than with infomer-cials. Many new medical technology demon-strations are conducted by the vendor of thedevice or therapy in the office of the targetedclinician or at seminars, often with extra finan-cial or other forms of incentive thrown in toencourage adoption of the technology.



Some innovations are not easy to observe andtherefore may diffuse more slowly. It is hard to observe an individual’s experience with infor-mation technology and even harder when theobserver does not know what to look for.

� Innovations that cannot be demonstrated easily with immediately visible results are notvery observable. Some innovations improvethe long-term health outcomes but show noinstant change in results. Examples are devicesor drugs to lower the risk of primary or recur-rent events, such as cholesterol-reducing drugsto prevent atherosclerosis.

� If a new technology creates observable resultsthat require additional expertise to interpretthem and make them understandable, thattechnology may be slower to diffuse. The morecomplex the results, the less obvious theybecome. Examples are novel devices that arelikely to benefit a small, but important, sub-population within a disease category, such asan implanted cardiac defibrillator.

Lessons for Technology Diffusion� Make the invisible visible with viral mar-

keting. One marketing challenge is makingnon-observable adoption more visible, which iswhat pharmaceutical companies have done byplastering their logos on free items for physi-cians. It is also the key idea behind viral mar-keting. By adopting a free Hotmail email serv-ice and allowing Hotmail to put its footer onan individual’s outgoing email, the individualsays “I’m a hotmail user and proud of it.” Theuse of the Hotmail product makes an act that typically is not observable—the choice ofemail service—more observable.

The Basics of the S-Curve

The S-curve is by now a quite intuitive modelof how technologies and other innovationsspread through a population. It is based in atradition of agricultural research that looked at,for example, the uptake of a hybrid corn varietyamong farmers in an Iowa county or thespread of irrigation technology in California’sCentral Valley.

In essence, the S-curve model shows that anyinnovation is first adopted by a few people (or organizations). As more use it, others seeit in use and, if the innovation is better thanwhat went before, others begin to use it.Once the diffusion reaches a level of criticalmass, it proceeds rapidly. At some point, theinnovation reaches a part of the populationthat is less likely to adopt it, and diffusionslows to a point of saturation. Figure 1 tracesthis S-curve.

Figure 1. The Diffusion S-Curve

Source: Institute for the Future


Late majority

Laggards

Early majority

Early adopters

Innovators


The Basics of the S-Curve, Continued

The S-curve implies a hierarchy of adopters, andRogers divides this population into five cate-gories, based on when different groups are likelyto adopt the technology along its S-curve path.Table 1 summarizes these groups. The shares ofthe population in each group are an arbitrary

Table 1. Hierarchy of Technology Adopters

division that indicates the rough scale of thatgroup in the total adopter population. (Note thatthe population base in this analysis is people who adopt. There also are non-adopters, who are not included.)


Category Characteristics Percent of Adopters

Innovators � Venturesome 2.5

� Cosmopolite

� Geographically dispersed contacts

� High tolerance of uncertainty and failure

Early adopters � Well-respected opinion leadership 13.5

� Well integrated in social system

� Judicious and successful use of innovation

Early majority � Deliberate 34.0

� Highly interconnected within a peer system

� Just ahead of the average

Late majority � Skeptical 34.0

� Responsive to economic necessity

� Responsive to social norms

� Limited economic resources

� Low tolerance for uncertainty

Laggards � Traditional 16.0

� Localite

� Relatively isolated

� Precarious economic situation

� Suspicious

Source: Everett Rogers, Diffusion of Innovations. New York: The Free Press, 1995.

In general, early adopters have more years ofeducation, higher social status, greater empathy,greater ability to deal with abstractions, greaterability to cope with uncertainty, more social and

professional contact, more cosmopolite contact,more change agent contact, more exposure to mass media, and a greater degree of opinionleadership than later adopters (Rogers, 276).


4. Communications Channels

Diffusion of innovations is a social process thatdepends on new ideas being communicated froman individual who knows about the innovationto an individual who does not. Health careprofessionals and consumers are bombarded byinformation from many sources, each trying topersuade them to try a product, a procedure,or a point of view. Many specialized communica-tions channels have evolved.

ExamplesWhen the locus of power in health care was clearand concentrated in the hands of physicians, thecommunications channels were simple to discernand exploit: fund research that was reported inmedical journals; buy advertising in those samejournals; send detail people to the offices to promote products; show up at meetings of pro-fessional societies; and wine and dine the opinionleaders who have influence in each local com-munity of physicians. As power has dispersed tomanaged care organizations, group practices, and consumers, the communications channelshave become more complex and interact witheach other in unexpected ways.

� The Internet is the single most important new communication channel to develop in thepast century. It has sped up and democratizedthe dissemination of medical information. The medical literature, traditionally theprovince of only trained professionals or themost diligent and educated consumer, hasbecome open to all.

� Increasingly, consumers are taking responsibilityfor their own health care (some voluntarily,some pushed by cutbacks in insurance cover-age). They are thirsty for medical knowledge.Many channels have evolved to influence theirdecisions, including increased coverage ofhealth technology in the popular press, a hostof Web sites on health care, and a rash ofdirect-to-consumer (DTC) advertising of newpharmaceuticals.

� As consumers become better informed abouttheir medical conditions, they themselves havebecome a communications channel to physi-cians about medical innovations. This shift incommunication pattern can be upsetting tophysicians, some of whom report that con-sumers come armed with the latest research orinfluenced by the latest DTC ad, demandingmedically inappropriate treatments.



� A weakness of traditional, peer-reviewed jour-nals is the time lag between a new finding and when it reaches health care professionals.New channels are opening up to speed thatprocess. Online e-journals now provide earlyrelease and discussion of findings. This newcommunications channel may raise questionsabout the integrity of information that maynot have undergone the rigors of traditionalpeer-review.

� Personal contact remains a powerful commu-nication channel. Even as consumers andmanaged care organizations have gainedpower, pharmaceutical companies have tripledtheir sales forces. Detailing still works.

� Groups that want to carry a message havelearned to piggyback their messages on exist-ing communications channels. Managed careorganizations (MCOs) have, for example,negotiated with drug companies to have theirrepresentatives carry MCO-determined mes-sages (in addition to their samples and market-ing materials) to the MCO’s physicians.

Not all communications channels are equallyeffective and some have either not been adoptedwidely or are quite ineffective.

� Email, a staple of communications in manybusiness areas, has been very slow to penetratethe doctor-patient relationship. Estimates ofhow many doctors use email to communicatewith their patients range from 10 percent to30 percent. Despite its popularity with con-sumers, a range of legal, ethical, and financialissues prevent physicians from adopting emailwidely. Like the telephone, which was slow to diffuse into medical practice initially, but eventually became a universal communicationschannel, email likely will reach large penetra-tion in the future.

� Accrual rates of adult cancer patients into clinical trials remain very low, on the order of2 to 5 percent, compared with childhood can-cer patients, some 70 percent of whom reachclinical trials. Diffusion of the best cancertreatments, often represented by the practicesin clinical trials, depends on patients becominginformed about their direct and indirect benefits. But many cancer doctors, for whom clinical trials are sometimes unrewarding and burdensome, often neglect to tell patientsabout the option to participate in trials orindicate that treatment will be determined by“drawing a card.” A key communication link is broken, slowing diffusion.

� Scientific innovation depends on the free flowof information. The academic tradition ofrapid publication of interesting results hasbegun to give way to more protective behavior,in which results are not published or are published incompletely until final intellectualproperty rights are resolved. Breaking or slow-ing this critical communication link risksimpeding scientific progress.

Lessons for Technology Diffusion� To inform people about an innovation, select

mass media and “cosmopolite” sources. To persuade people to adopt the innovation,closer links and interpersonal channels aremore effective.

� To communicate more complex messages,select interpersonal communications channels. Describing surgical techniques orcomplex medical devices works best one-on-one with a trusted source.



� In order to select the right communicationchannels, select the right target audience. Forexample, promoting administrative IT systemsto physicians has limited effectiveness. Practiceadministrators or CIOs are the primary target,and they have their own, often separate, com-munications channels.

� Identify individuals who are “Connectors.”Malcolm Gladwell, in his book The TippingPoint, identifies a type of individual who hasmany acquaintances (not necessarily withstrong ties to them) and propagates informa-tion freely through an established network.People, ideas, and innovations that get theattention of Connectors have a much betterchance of diffusing. They are a key communi-cation channel.

5. Homophilous Groups

The degree of similarity among group membersacross which an innovation diffuses will affect theease and speed with which the diffusion takes place. Innovations spread faster amonghomophilous groups (those with similar charac-teristics) than among heterophilous groups (those that differ in important ways).

In general, communication is more effectivewhen the source and receiver share commonmeanings, beliefs, and mutual understandings.Communication among like individuals will be more effective because individuals feel more comfortable interacting with others likethemselves. However, differences in training,social status, beliefs, and language can lead tomistaken meanings, thereby causing messages tobe distorted or to go unheeded.

Physicians are a relatively homophilous group.The medical profession is a guild with rituals,hierarchy, and rules of professional etiquette thatgo well beyond their common schooling. Physi-cians all take the Hippocratic oath, a promise to do no harm. They are trained with a relativelystandard curriculum that is focused on clinicalskills in diagnosis and treatment. As a group,they enjoy the same high status in society.

Examples� Most clinicians join professional associations

during their careers. Associations count on thehomophily of their members in communicat-ing with them and offering services such asconferences, continuing medical education,and special certification. Associations canencompass a broad range of a profession, suchas the American Medical Association for doc-tors or the American Nursing Association fornurses, or it can encompass a much smaller,more specialized community, such as theAmerican Society of Clinical Oncologists.

� Most associations also publish a scientific journal that tracks the latest advances in clinical and laboratory research. Beyond themost widely-read general medical journalssuch as the Journal of the American MedicalAssociation or the New England Journal ofMedicine, researchers try to publish in specialtyjournals with specialist physician readers. Theresearchers and readers are both consideringthe homophily of the specialty as a way to filter out less pertinent information.



Though physicians are very similar, sometimeshomophily is simply not there.

� Executive leadership at health systems is a mixof physicians and administrators. Though theprimary incentive to adopt new technology isto improve patient care, the heterophilousnature of the roles in health system leadershipand their expertise also influences their deci-sions. Physicians who leave clinical practice tobecome full-time administrators are likely tothink as an administrator and be so consideredby the practicing physician community.

� An example of physician heterophily that hasthwarted health care quality advocates is practice variation. Though physicians all overthe U.S. are trained using a similar curricu-lum, practice behavior varies quite drasticallydepending on the practice standards in thegeographic area. Quality advocates attempt toovercome discrepancies in local clinical cultureby disseminating clinical practice guidelinesand protocols, using the appeal of adherenceto generally accepted practices.

Lessons for Technology Diffusion� To use homophily as a technology promoter,

understand the degree of homophily in thetarget group. Though some groups may sharetraining and biases and appear homophilous,when introduced to something that they arenot familiar with as a group, such as informa-tion technology, their homophily as clinicians may be overcome by their heterophily as ITconsumers.

� Look for other homophilous groups beyondphysicians. Examples of other groups thatmay share similar needs with each other arepractice administrators, health system strategicplanners, nurses, specialty nurses, interestgroups, and patient advocacy groups such asAIDS or breast cancer. Communication within these groups may speed up diffusion of information.

� Put the right person in front of your targetaudience. People are more likely to be per-suaded by like individuals. Doctors respectother doctors because they share training, have the same commitment to heal, and feelthe shared pain of increases in managed care administrative burdens, etc. It is valuableto speak to potential adopters from their ownperspective and show that you understandtheir daily challenges.



6. Pace of Innovation/Reinvention

Some innovations are relatively stable and do not evolve as they diffuse. Others evolve morerapidly and are altered by users along the way.The degree of reinvention of any innovationaffects both its pace and style of diffusion. Someinnovations lend themselves readily to beingappropriated and altered by their users, and candiffuse rapidly even as they change, becomingalmost unrecognizable as they reach new popula-tions. Others are much less mutable—either theyare used as intended or are not used at all.

ExamplesInnovations that are general, not specialized, verycomplex, and that are process-oriented often getreinvented to a larger extent than innovationsthat are not.

� Physicians are trained to be autonomous, scientific thinkers. As such, many medicaltechnologies get reinvented and reinterpretedin the hands of practicing doctors. Off-labeluses of prescription drugs are increasinglycommon, and compendia are published thatlist off-label uses, as well as those that areapproved. Consumers also will find new applications for existing drugs, as evidenced by the recent publicity about women usingViagra. Surgeons are ever eager to add a per-sonal signature on a new surgical procedure,e.g., off-pump cardiac surgery.

� Other medical technologies, such as lasers,often are reinvented. Sometimes this results inextending the range of conditions for which thedevice can be used. An example is radiosurgery.Initially applied to only two neurosurgicalconditions, it is being used today in a range ofapplications within and outside of the nervoussystem, such as epilepsy and abdominal cancer.

� Reinvention is often a necessary step for diffusing a flexible technology into a complexsocial environment. Most hospitals will insistthat their administrative procedures areunique, and that a standardized informationsystem would not fit the local culture. In fact, 70 to 90 percent of the transactions andprocedures typically are very similar to those of other hospitals, and only a small degree of customization is required. But the act of customizing—of reinventing—increasesacceptance of the technology. A similar phenomenon takes place with practice guide-lines. Local providers must put their stamp on standard frameworks to feel that they arelocally adapted.



Stages in the Innovation—Decision Process

The decision to adopt an innovation takes time.There is inherent uncertainty to a new untestedalternative. There are actions and choices thatone must make in order to evaluate the new ideaand decide whether or not to put it into practice.Rogers maps this Innovation—Decision Processinto five steps.

The Innovation—Decision Process is the processthrough which an individual (or other decision-making unit) passes from first knowledge of aninnovation, to forming an attitude toward theinnovation, to a decision to adopt or reject, to implementation of the new idea, to confirma-tion of this decision.


Knowledge

Knowledge occurs when the decision-maker is exposed to an innovation’s existence andgains some understanding of how it functions. Acquiring knowledge occurs throughout theinnovation—decision process.

Persuasion

Persuasion occurs when the decision-maker forms a favorable or unfavorable attitude towardthe innovation. Persuasion is when the decision-maker becomes psychologically involved andstarts to feel something about the innovation. Early information received or past experiencewith a similar technology affects the attitude towards the innovation.

Decision

Decision occurs when the decision-maker engages in activities, such as partial trial of theinnovation, that lead to a choice to adopt or reject the innovation.

Implementation

Implementation occurs when the decision-maker puts an innovation into use and overt behavior change happens. The new user seeks information about how to obtain the innova-tion, thinks about what problems might be encountered, and seeks support in putting theinnovation in place.

Confirmation

Confirmation occurs when the decision-maker seeks reinforcement of an innovation—decisionalready made, or reverses a previous decision to adopt or reject the innovation if exposed to conflicting messages about the innovation. At this point, the decision-maker seeks to avoida state of dissonance or to reduce it if it occurs.

Source: Institute for the Future; adopted from Everett Rogers, Diffusion of Innovations. New York: The Free Press, 1995.


Reinvention may be a signal that there is some-thing wrong with the technology as initiallyreleased. The fact that it must be adapted andreinvented may indicate flaws in the originaldesign.

� Pharmaceutical manufacturers must track their products post-approval to be certain thatthere are no severe adverse side effects notobserved in pre-approval drug trials that onlyare manifested once the product is in wide use.In some cases, such effects result in redesign or reapplication of the drug, for instance limit-ing its use to specified populations or requiringa diagnostic test to monitor side effects. Anexample is the routine ordering of a liver func-tion panel for patients shortly after a statin isfirst prescribed.

� Many IT systems are reinvented after installa-tion, particularly when the design team didnot initially involve the right individuals in theprocess. This reinvention may reflect poordesign and installation, but often is the resultof insufficient after-sale training and support.Training and support cost money and time,and so are often overlooked. Change-manage-ment experts attest that training is a criticalsuccess factor to successful diffusion of new IT systems.

Lessons for Technology DiffusionDissatisfaction is the mother of reinvention and reinvention is a source of extremely usefulfeedback from users. Several devices can be used to exploit that knowledge.

� Put in place active listening posts. Many ITcompanies have mastered the art of creating,nurturing, and listening constructively to usergroups as a source of early warning aboutproblems in a system as well as users’ prioritiesfor new features.

� Monitor medical technologies very carefullyfor instances of potentially dangerous misuses. Early detection can trigger productchanges or education campaigns that avert significant liability down the road.

� Look for the “work-arounds” that usersemploy to make a technology work. Work-arounds are extremely common with infor-mation systems users who often do the firstthing that works for them, never learningoften easier ways to operate.

� Don’t be offended by reinvention. It is not a sign that the innovation was inadequate or that users did not really understand thedesigner’s intent. It is typically a sign that theinnovation is being put to even greater usethan was intended.

7. Norms, Roles, and Social Networks

The norms, roles, and social networks of medi-cine are very important to the diffusion of newtechnology. According to classic diffusion theory,the configuration of the social networks throughwhich innovations diffuse help govern the paceand extent of diffusion. Studies in the mid-1960sof physicians’ prescribing patterns of tetracycline,for example, showed that doctors with more extensive social networks—those on hospitalstaffs or who attended hospital staff meetings,those who practiced in groups, those who con-sulted with other physicians about their cases,and those who were cited as being close friendsof other doctors—adopted the drug much morerapidly than doctors who were relatively moresocially isolated.1

Norms of behavior and expectations about rolesare strongly ingrained in health care professionalsand affect how new ideas and technologies dif-fuse into practice. These norms and expectationscan be used to target the appropriate social net-works to help diffuse a given innovation.



Examples� Norms that are instilled into physicians in

medical school are very difficult to change.Some of the geographical variation in surgeryrates such as hysterectomy and mastectomy,for example, derive from differences in howthe procedures were taught at different teach-ing hospitals around the country. A single pro-fessor’s ideas about when surgery is appropriatecan bring about great variations in cost andquality of care. However, doctors in trainingmay learn more than one way to manage a particular problem and what they practice isoften determined by local practices because“that’s the way we do it.”

� Medical and other professional societies are a key node in health care’s diffusion networks.An endorsement by a professional society orinclusion of a drug or procedure into a society’s practice guidelines can speed diffusionand lock a practice in (or out).

� Practice guidelines (PGs) themselves are agood example of the need to choose the rightsocial network and a group with appropriatenorms to maximize the pace of diffusion. Early practice guidelines were derided bymany mainstream physicians as “cookbookmedicine” that threatened to erode physicianautonomy in the service of the financial incentives of managed care. As more providersrecognized the problems of practice variationand as more professional societies producedtheir own PGs, physicians were more willing toget on board, as long as their own autonomywas not threatened. They found a compromisein many specialties by turning to a group—nurses—whose social norm is compliance withroutines and orders set by someone else, forwhich they had a delivery role. Once nurseshad responsibility for the implementation ofPGs, diffusion was more rapid.

The power of social networks and norms ofbehavior in medicine also has acted to retardinnovation and to reward poor practice patterns.

� Although many steps have been taken over the past decade to improve medical qualityand to remove dangerous physicians frompractice, a sort of medical omertà—a code ofsilence protecting members of the guild—stillexists. That social norm, professional courtesy,and loyalty to the social network can slow the diffusion of better practices.

� Despite progress in the professionalization ofmany parts of health care delivery systems, the conventional medical hierarchy in whichphysicians reign supreme still persists in manysettings. This affects the diffusion of manytechnologies. Investments in the latest imagingsystem (which may be a revenue center) prevailover administrative information technology(which may be viewed as a cost center).

Lessons for Technology DiffusionAn acute awareness of the social settings intowhich a new technology or other innovation dif-fuses is essential for affecting its pace. In the caseof PGs, the first managed care organization tostop targeting doctors and to start targeting nurses made an insightful breakthrough. Someideas on social networks:

� Pay explicit attention to the physical and virtual networks of the groups you wish toreach. In the past, social networks and geo-graphic proximity were closely linked. Most of a clinician’s contacts lived or worked close by.With the advent of the Internet and with thesustained mobility in American society, net-works are more far-flung. Look for long links,members of a network who are geographicallydistant; and weak links, people who are onlyacquaintances, not close friends. Both are likely to have a social network that divergesfrom the tighter network and are often asource of innovation.



� Be aware of opportunities to leverage exist-ing or create new social networks. Peoplewith AIDS, cancer, and many other diseaseshave gravitated to support groups and onlinecommunities. Often these communities havebeen a rapid channel to diffuse informationabout new medical technologies and for mobilization of activists. Many pharmaceuticalcompanies have put in place online communi-ty spaces for people with a disease their drugstarget. One drug company that produces cancer drugs helped the American Society forClinical Oncology to create a new network ofstate oncology societies. These societies servedas a dissemination channel for ASCO’s infor-mation into community practice.

8. Opinion Leaders

Opinion leaders are key actors in the diffusion of innovations. Change agents who want to promote a new idea or technology must payattention to those individuals whose words andbehaviors influence that of their peers. In general,opinion leaders have greater exposure to newideas through the media, by being more cosmo-polite than others, by having more exposure tochange agents, having higher income and educa-tion, and by having wider social networks thanmost people.

Opinion leaders have been a key vector of diffu-sion for many medical and information technol-ogies, and a large amount of effort is dedicated toidentifying, informing, and convincing them tobecome early adopters.

Examples� Drug and device companies have perfected

the art of identifying the opinion leaders in agiven community or specialty and of courtingthem through their sales forces. They also have access to detailed databases showing prescribing patterns in different communities.Although the norms of how to influencephysician leaders have shifted (free trips toHawaii are out; unrestricted educational grantsare in), their importance has not diminished.

� Celebrity endorsements of drugs, nutritionalsupplements, or health care facilities are another form of opinion leadership gearedtoward consumers.

� Patterns of opinion leadership vary dependingon the content. Few physicians are opinionleaders on the business of running a practice,medical matters, and on information systems,for example. In the mid-1990s a number ofpharmaceutical manufacturers and healthinsurers thought that, because they understoodthe patterns of physician opinion leadership in their respective fields, they could act as a channel to sell information technologies ordisease management methodologies. Theyquickly found that their sales networks did nottranslate very effectively.

� For a number of reasons, the diffusion ofinformation systems in health care has beenmuch more fragmented than for medical tech-nologies. Thousands of different practice management software packages exist, few ofthem with any appreciable market share. The failure to get widespread agreement onwhich IT system to use, especially in smallpractices, is due in part to the relative scarcityof physician opinion leaders for guidanceabout information systems. This is less of aproblem for hospital information systems andin larger practices, where IT professionals(with their own patterns of opinion leader-ship) make those decisions.



Lessons for Technology DiffusionApart from the obvious lesson to identify andcourt opinion leaders, there are several other les-sons from the literature on diffusion.

� Do not mistake early adopters for opinionleaders. Although opinion leaders typically doadopt many innovations before the main-stream does, they cannot afford to get too farahead of the pack. Opinion leaders serve toreduce the uncertainty of others in their socialsystem about adopting new technology orbehaviors. This also means that, althoughopinion leaders should be courted, they shouldnot be co-opted. If they adopt innovationsthat fail too often, they risk losing leadership.

� Work hard to identify the relevant opinionleaders. With the advent of the Internet, opin-ion leadership has, to some extent, becomedemocratized. It is more difficult than beforeto tell the qualifications of an opinion-holder,because anyone can put together a slick Webpage. More than ever, it is essential to be partof, and to map explicitly, the social networksof authentic opinion leaders in health care.

� Be on the lookout for Mavens. MalcolmGladwell, in The Tipping Point, (Little, Brownand Company, 2000), identifies The Maven asa type of opinion leader who is very importantto the diffusion of new ideas. Mavens are indi-viduals who study markets and technologies togather information on the latest, and the mosteffective technologies, sharing that informationwith their peers. Mavens have a reputation forbeing knowledgeable and are often deferred toby those in their social network. The Maven isa special type of opinion leader—one who hasexposure to a wider range of fields and infor-mation than most other opinion leaders.

9. Compatibility

The familiarity of an innovation, its compatibilitywith the existing environment and behavior, isstrongly linked to its diffusion. The more aninnovation can integrate and coexist with tech-nologies and social patterns already in place, thegreater its prospects for adoption and diffusion.If the innovation is consistent with a potentialadopter’s past experiences, existing values andneeds, the decision to adopt is facilitated.

Precedent behavior or an installed base of tech-nology increases compatibility and improves the likelihood of diffusion. The more compatible an innovation is, the less change in behavior itrequires.2 A new imaging modality will diffusebecause an installed base of imaging technologyis there, and a new or improved drug therapy willdiffuse along the same path of its predecessors.

Exogenous influencing factors, such as financialreimbursement for using the technology, caninfluence the pace of diffusion. Medical technol-ogy diffusion often “follows the money” in thatthe lack of a clear reimbursement method for anew technology is a significant barrier even if thetechnology offers a perceived competitive advan-tage to the physician or hospital.



Examples � Plug and play technologies, those that mesh

with current standards and do not requirelearning new behavior, are the easiest to diffuse. IV solution bags, tubes, and valveswith standard sizes and interchangeability areplug and play technologies.

Another example of plug and play technologyin medicine is the introduction of Palm Pilotsand other handheld devices with prescribingcapabilities for physicians. Assuming that the clinician using the device already knowshow to use it, the introduction of a palm-based prescription program requires no moreeffort to learn than the new software andreduces error and duplication.

� Most new medical technology improves uponan earlier version of the technology. Ground-breaking, revolutionary technologies that arethe first to do something fundamentally differently, such as x-ray technology, are likelyto be incompatible with current practice. Such technologies are often the first in whatthen becomes a new class of technologies and all technologies that follow it build uponthe experience with the previous generation of technology.

Since the discovery of the x-ray, new imagingtechnologies have diffused by building on apotential adopter’s experience of the previousgeneration. New users of CT scans had expectations of results that were based on their experience with x-ray, and new users ofMRI had expectations built on their experi-ence with CT scans.

� The federal government wanted to promotethe use of telemedicine in rural areas in theU.S. to address geographic shortages of clini-cians and specialty services. In 1997, Congressincreased the amount Medicare reimbursed fortelemedicine services, bringing about a growthin telemedicine programs.3 By creating a fund-ing stream for telemedicine, the governmentencouraged the use of the technology and spedits diffusion among providers and vendors.

The concept of incompatibility helps illustratewhy some classes of technology diffuse morereadily than others.

� The use of information technology has beencorrelated with education and income, that is,the more education and income one has, themore likely they are to use personal computersand the Internet.4 As individuals who havehigh levels of education and income, physi-cians are assumed to have an affinity for infor-mation technology. In fact surveys show thatphysicians do use information technology, with70 percent using the Internet and 75 percentusing a personal computer.5 But only a quarterof physicians use email to communicate withpatients and only 22 percent use electronicmedical records.6 As a rule, physicians do notuse computers in their daily workflow.

� An innovation can be incompatible when anorganization or government controls andrations the technology budget. In restrictedbudgeting environments, such as with Canadaand the United Kingdom, a new technologywill diffuse only after cost-benefit analysisshows significant improvement over existingtechnology and then only minimally, rationingthe amount of technology available. Such governments do not spend as much as the U.S.on new medical technology; physicians are less likely to refer patients for high-end tech-nological procedures and patients who do getreferred must wait longer for the procedurethan in the U.S.



Lessons for Technology Diffusion� Understand current behaviors and values.

Look for currently existing behaviors that aresimilar to the behavior being introduced. Look for and understand both formal andinformal hierarchies and power relationships(e.g., nurses are trained to take orders fromphysicians but have the dominant role in dailypatient care).

� Innovations that reduce hassles are morelikely to be successful. Features that gobeyond the primary function of the technolo-gy and make it easier and more pleasant to use will increase its compatibility.

� Mimic things from other parts of life. Soft-ware developers have created computer interfaces with icons that represent functionsanalogous to those in our daily lives (e.g., theword processing software icon for cut is scissors and for paste is a jar of paste, etc.).

10. Infrastructure

The adoption of many innovations depends onthe presence of existing infrastructure that cansupport it. Fax machines are an example of infrastructure’s role in technology diffusion; theirrapid adoption depended on telephone lines,an infrastructure already in place.

Some technologies diffuse in clusters and areinterdependent on one another to succeed. Theuse of email depends, for example, on a clusterthat includes personal computers, email software,modems, voice and data communications networks, servers, and common communicationstandards.

Examples� After CT scans were introduced into hospital

radiology departments, they required computersfor image display and digital image storage.Though computers had been used in radiologydepartments for administrative functions, suchas scheduling, display and storage functionsrequired different features and created demandfor a different type of machine. When MRItechnology came along, it depended on thatsame display infrastructure.

� Advances in genomic technology promise personalized medicine and gene therapy some-time after the year 2010. Between now andthen, the platform technologies for applyinggenetic and genomic knowledge to drug designand development, such as target validation,must mature. Only then will the majority ofnewly approved biopharmaceuticals originatefrom drug discovery based on targets generatedfrom genomic-based research.



A lack of infrastructure or the wrong kind ofinfrastructure can retard technology diffusion.

� To the dismay of health IT promoters, thelegacy infrastructure of paper patient files isstill in place. IT continues to be stored in sep-arate, non-integrated databases, used primarilyby non-clinical staff and, other than digitalimaging and telemedicine, it is used mostly for administrative and billing functions. The pervasiveness of paper-based infrastructure hasmade it difficult for IT to diffuse beyondadministrative and clinical functions.

� Regulatory timelines and procedures can besignificant barriers to diffusion. Pharmaceuticaland medical device manufacturers build thetime needed for regulatory approval into theirdevelopment, and are affected adversely by thetime taken up by requests for additional infor-mation. The longer it takes for a blockbusterpatent-protected drug to make it to market,the more money a drug company loses in mar-ket exclusivity. At the opposite end of thisspectrum, the FDA instituted a controversialfast-track approval process to overcome thisbarrier in special circumstances.

Lessons for Technology Diffusion� Look for opportunities to plug and play. To

do this, it is important to understand howtechnologies are connected to each other, whatare the relevant standards and potential contexts,and how the technologies fit into the dailyworkflow of clinicians. Entrepreneurs creatinginnovative new IV devices and tubes under-stand every use of them, who uses them, howthey are stored and what remains challengingabout the current state of the technology.

� Understand current and future regulatoryconstraints and competing patent protections.

� Look for leapfrogging technology. In somecases, it is possible to overcome infrastructurebarriers by jumping to a different level of technology. An example of this is the diffusionof cell phones in parts of Asia where there are no telephone lines.




THERE WILL CONTINUE TO BE MANY BENEFICIALhealth care technologies that should diffuse, but will not. And there will be others—me-too products that offer no realinnovation, health information systems that complicate ratherthan simplify life, expensive new therapies that do not extendlife or improve quality—that diffuse despite their shortcomings.This tour through the basics of innovation diffusion shouldhelp make clear why getting new ideas and technologies put togood use continues to be both frustrating and rewarding forinnovators and users alike.

“Getting a new idea adopted, even when it has obvious advantages, is often very difficult.”

— Everett Rogers, Diffusion of Innovations

III. Epilogue


Quick Tips for Technology Diffusion

Relative advantage

� Understand the end user of the technology

� Recognize the impact of significant behavior change

� Consider the business case for the adoption of a new technology

Trialability

� Look for opportunities to carve out somepart of a system that is more trialable

� When designing a complex new tech-nology or system, consider which components could be tried out withoutcommitting to the full innovation

Observability

� Make the invisible visible with viral marketing

Communications channels

� To inform people about an innovation,select mass media and “cosmopolite”sources. To persuade people to adopt theinnovation, closer links and interpersonalchannels are more effective

� To communicate more complex messages, select interpersonal com-munications channels

� In order to select the right com-munication channels, select the right target audience

� Identify people who are “Connectors”

Homophilous groups

� To use homophily as a technology promoter, understand the degree ofhomophily in the target group

� Look for other homophilous groups beyond physicians

� Put the right individual in front of your target audience

Pace of innovation/reinvention

� Put in place active listening posts

� Monitor medical technologies very carefully for instances of potentially dangerous misuses

� Look for the “work-arounds” that usersemploy to make a technology work

� Do not be offended by reinvention

Norms, roles, and social networks

� Pay explicit attention to the physical and virtual networks of the groups youwish to reach

� Be aware of opportunities to leverage existing or to create new social networks

Opinion leaders

� Do not mistake early adopters for opinion leaders

� Work hard to identify the relevant opinion leaders

� Be on the lookout for “Mavens”

Compatibility

� Understand current behaviors and values

� Innovations that reduce hassles are more likely to be successful

� Mimic things from other parts of life

Infrastructure

� Look for opportunities to plug and play

� Understand current and future regulatory constraints and competingpatent protections

� Look for leapfrogging technology



Endnotes

1. Rogers, Everett M. Diffusion of Innovations.Fourth Edition, The Free Press, New York, 1995

2. Ibid.

3. American Telemedicine Association.“Telemedicine Reimbursement Approved byCongress; Agreement Hailed as Victory forAdvocates of Telemedicine.” 1997. www.atmeda.org, and PRNewswirewww.prnewswire.com/cgi-bin/stories.pl?ACCT=104&STORY=/www/story/7-31-97/289253&EDATE=

4. National Telecommunications and Informa-tion Administartion, U.S. Department of Commerce. A Nation Online: HowAmericans are Expanding Their Use of theInternet, 2002. www.ntia.doc.gov

5. Mills, Robert, J. “AMA Survey Finds Upsurgein Physician Usage and Regard for Internet”American Medical Association, May 2001.www.ama-assn.org/ama/pub/article/1616-4692.html

6. Landro, Laura “Unhealthy Communication:When it Comes to E-Mail, the Doctor is Not In” Wall Street Journal, 2/11/02http://online.wsj.com/article/0,,SB1013033700555906000.djm,00.html



diffusion final cover 5/10/02 2:19 PM Page iv

Documents

Diffusion of Innovation in Health Care - California HealthCare