the science of spread: how innovations in care become the norm

CAL I FORNIAHEALTHCAREFOUNDATION

The Science of Spread: How Innovations in CareBecome the Norm

September 2007

The Science of Spread:How Innovations in CareBecome the Norm

Prepared for:

CALIFORNIA HEALTHCARE FOUNDATION

Prepared by:Thomas Bodenheimer, M.D.

September 2007

About the AuthorThomas Bodenheimer, M.D., is an adjunct professor in theDepartment of Family and Community Medicine at theUniversity of California, San Francisco.

About the FoundationThe California HealthCare Foundation, based in Oakland, is an independent philanthropy committed to improvingCalifornia’s health care delivery and financing systems. Formedin 1996, our goal is to ensure that all Californians have access toaffordable, quality health care. For more information aboutCHCF, visit www.chcf.org.

ISBN 1-933795-37-9

Copyright © 2007 California HealthCare Foundation

Contents4 I. Overview

6 II. A Brief Summary of “Spread” LiteratureEverett Rogers

The Tipping Point

Institute for Healthcare Improvement

Paul Plsek

Sarah Fraser

A Critique of the Spread Literature

11 III. Examples of the Successful Spread of Improvement in Health CareVeterans Health Administration

Humboldt–Del Norte Independent Practice Association

Cincinnati Children’s Hospital Medical Center

Institute for Clinical Systems Improvement

IHI’s 100,000 Lives Campaign

21 IV. Conclusion

23 Endnotes

4 | CALIFORNIA HEALTHCARE FOUNDATION

CHANGE—WHETHER IN BIOLOGY, CONSUMERgoods, or health care—generally begins in one geographic siteor organization. These focal innovations, whether they are natural processes or human inventions, may wither and die orspread widely and make history. The spread of improvementsin health care organizations is distinctly different from the diffusion of innovations in other areas of life.

For example:

n In 1918, a deadly influenza virus spread throughout theworld, killing an estimated 50–100 million people within 18 months.

n Since the mid-1990s, a fungus known as “sudden oak death”has killed up to 1 million oak trees in California.

n In 1946, regular television broadcasting began in the UnitedStates; by 1953, 25 million American homes (half of allhouseholds) had a television set.

n In 2001, Apple, Inc., introduced the iPod, which sold morethan 88 million units by January 2007.

However,

n Even though it was proven in 1601 that lemon juice preventsscurvy—a major killer of sailors in the British navy—it tookuntil 1795 for the British navy to adopt this innovation.1

n An average of 17 years is required to translate new researchevidence into practice.2

n Since 1993, with the publication of the Diabetes Controland Complications Trial, physicians have known that control of blood sugar levels reduces patients’ complica-tions. Yet in 2000, 63 percent of patients with diabeteswere in poor glycemic control.3

Why do some innovations spread like wildfire throughout thepopulation while others take endless time to catch on or neverpropagate beyond a few early adopters?

The influenza virus and sudden oak death have nothing incommon with improved diabetes care. They are biologic agentsthat have found a favorable ecologic niche, allowing them topropagate freely and rapidly, overwhelming the resistance oftheir hosts.

I. Overview

Television sets and iPods are human inventionsappealing to an innate human desire for enter-tainment. They exemplify a combination ofmarketing and unplanned person-to-personpropagation: “My neighbor got a TV set and Iwant one.” “My friends have iPods so I wantone for my birthday.” The only resistance to thespread of these inventions is the capacity of theconsumer (the intended target) to afford theproduct. The consumer need not change oldhabits or give up longstanding routines to pur-chase these items.

Improving diabetes care is an entirely differentanimal. Patients with diabetes do not know thatcare innovation even exists, and the practice teamthat implements the innovation must give up oldhabits and longstanding routines to embrace thechange. Pursuing improvement may require morework. It may lengthen someone’s workday. Itdemands better teamwork among a group ofcaregivers, and teamwork is not a slam dunk. Incontrast to Apple selling 1 million iPods, betterdiabetes care may cost a health care institutionmore money without enhancing revenues. Instark contrast to the 1918 influenza virus and theiPod, improving diabetes care lacks a powerfulbiologic or financial advantage and faces substan-tial host resistance.

In Spreading Good Ideas for Better Health Care,Paul Plsek writes that many organizations canboast of “islands of improvement,” but that onerarely finds organization-wide improvement thathas become the permanent way of doing busi-ness.4 In addition to the inherent difficulties inspreading improvement, it is not clear to manyimprovement champions how to spread the“islands of improvement” into permanent, organization-wide practices.

This report summarizes some of the thinkingtaking place within the emerging science ofspreading improvement and provides case studiesof health care organizations that have achievedsome success in the broad propagation of institu-tional change. Serious students of “spread,” theshorthand term of art for such diffusion, mightwish to consult Trisha Greenhalgh and colleagues’exhaustive review of the literature on spreadingand sustaining innovations in health services.5

The Science of Spread: How Innovations in Care Become the Norm | 5


Everett RogersIn 1962, Everett Rogers published Diffusion of Innovations,a book that has become the classic work on this topic.6

“Diffusion” is synonymous with “spread.” Rogers begins thebook’s fourth edition (1995) with a description of a publichealth campaign to introduce water-boiling in a Peruvian village, aimed at preventing water-borne infections that are afrequent cause of infant mortality. Because the health workerwho led the campaign was not able to gain the trust of the vil-lagers, the campaign failed in spite of its obvious advantages.

Rogers finds that four main elements are needed for innova-tions to spread and take root:

1. The innovation needs to be better than the status quo, butat the same time compatible with the existing values andneeds of the potential adopters. It should be simple tounderstand, and must demonstrate success on a small scale.Further, its results need to be evident to potential adopters.

2. A communication channel must exist through which theinnovative idea is transmitted to potential adopters. Thismight be person-to-person or via mass media. Ideally, thechange agents are similar in education and social status tothe potential adopters.

3. Time is required for spread, which is measured by thenumber of people who adopt the innovation in a giventime period. An innovation may be adopted, rejected, oradopted and then discontinued. The rate of adoption isdescribed by the Rogers S-shaped curve, in which spreadbegins slowly, then takes off rapidly, and finally plateaus asthe proportion of adopters reaches 80–100 percent. Rogersclassifies people in the social system within which theinnovation is taking place as innovators, early adopters,early majority, late majority, and laggards.

4. The structure of the social system can facilitate or impedethe diffusion of innovations. The social system is the set ofinterrelated units that are engaged in activities with a com-mon goal—for example, a health care provider organization.Social systems contain individuals and units. In health care,spread may be easier in integrated delivery systems than incollections of small independent practices. Other character-istics of social systems are the norms or culture of the sys-tem, and the existence of opinion leaders who support oroppose a given innovation.

II. A Brief Summary of “Spread” Literature

The Tipping PointIn 2000, a book about spread became a nationalbestseller. Malcolm Gladwell’s The Tipping Pointcompares rapidly diffusing innovations to epi-demics and emphasizes that epidemics can rise orfall in one dramatic moment—the tippingpoint.7 The tipping point concept, according toGladwell, became popular as a description of“white flight” to the suburbs in the 1970s; whenthe number of African Americans in a neighbor-hood reached a certain level, the communitywould “tip” and most of the remaining whiteswould leave almost immediately. A technologicexample offered by Gladwell is the adoption ofthe fax machine; introduced in 1984, sales wereslow until the tipping point year of 1987, when 1 million faxes were sold. Cellular phones hadtheir tipping point, as did television sets andiPods.

Does The Tipping Point provide any lessons forhealth care improvement work, a variety of inno-vation far removed from faxes and iPods? In fact,Gladwell does provide an important insight forhealth care: the key role of individuals in thespread of innovation. Gladwell recounts the storyof Paul Revere, who rode at night through thesuburbs of Boston to warn of an imminentBritish attack. Revere was highly successful, persuading others to join in spreading the news.In contrast, William Dawes, who conducted asimilar ride through other Boston suburbs duringthe same night, mobilized few. The difference,according to Gladwell, was that Dawes lacked thepersonality to persuade townspeople to resist.Revere’s message tipped; the same message fromDawes did not. The individuals involved inspreading innovation matters greatly.

Another insight provided by Gladwell is the concept of “stickiness.” As advertisers know,some messages stick, while others do not. Onemust package information to make it irresistible,or one’s message will flop. Failure of the messageto stick can lead to an improvement failing tospread.

Health care innovators love change. But exceptfor the small number of early adopters, mostpeople in health care settings (and most settings)are wary of change. If innovators simply extol thevirtues of making a change, their message will failto stick. For a message to stick, it must speak tothe concerns of people who are neither innova-tors nor early adopters. Because innovators are sodifferent from the majority, they must be highlysensitive to whether their message is sticky for themajority. An unsticky message might be: “Thischange will save money for the health care sys-tem.” A sticky message might be: “This changewill help get you home half an hour earlier eachnight.”

The Tipping Point may have limited relevance tohealth care improvement because it generallydescribes unplanned diffusion; in health care,improving the care of patients with diabetesrequires planned diffusion. Yet changes in healthcare can tip. Take, for example, the extraordi-narily rapid and relatively unplanned diffusionof hospitalists, the new physician specialty caring for inpatients. Some health care changesthat reach a tipping point are bad; for example,Vioxx, the anti-inflammatory medication—aggressively marketed by its pharmaceutical company—also turned out to increase the risk of heart attacks. Thus far, most health careimprovements—which are more complex thanthe examples featured in Gladwell’s book—havenot achieved a tipping point.

Institute for HealthcareImprovementThe Institute for Healthcare Improvement (IHI)is the leading organization in the United States inspreading innovation in health care. IHI uses theword “spread” as a popular synonym for themore academic terms “diffusion” and “dissemina-tion.” For IHI, Everett Rogers’ Diffusion ofInnovations is the spread bible. Don Berwick,IHI’s President and CEO, published an article


titled “Disseminating Innovations in HealthCare,” applying Rogers’ concepts to health care.8

In 2006, IHI published a white paper titled “AFramework for Spread.”9 The paper makes practi-cal suggestions on preparing for spread, writingan aim statement for spread, developing an initialplan for spread, and executing the spread plan. Amore detailed discussion of spread can be foundin the “Spreading Changes” pages on the IHIWeb site.10 The site provides some descriptions of organizations that have made progress inspreading improvements.

IHI proposes that spread is a leadership responsi-bility and makes specific arguments about howspread may be optimized:

n Top executives should designate an executivesponsor for spread, as well as a day-to-dayspread manager.

n Spread should not take place until ideas forimprovement have been tested in one or morepilot sites and the results demonstrate that theinnovation is clearly better than what currentlyexists. Spreading the wrong set of changeswould be a major setback.

n Once the leadership determines that a betteridea should be spread throughout an organi-zation (or spread to other organizations), acommunication strategy is needed with a clearmessage and people (spread agents) capable of persuading others in the target population(the sites where the new idea would be imple-mented).

n To evaluate the adoption of the new ideas,measurement and feedback are necessary toallow the leadership and target sites to deter-mine whether improvements are being madeand to provide data that encourage successfulsites and challenge slower adopters.

IHI’s spread strategy has been misunderstood bya number of health care champions. (Innovatorsin health care improvement are generally called“champions.”) IHI has organized much of its

improvement work through learning collabora-tives, which bring together improvement teamsfrom different health care organizations for about12 months to work on particular topics, such asdiabetes care or reducing patients’ waiting timesfor medical appointments.11, 12 To test whetherparticular changes actually lead to measurableimprovement, the improvement teams catalyzechanges for a small pilot population of patients.These improvement teams are at times calledupon to spread the improvement from the pilotpopulation to all the patients in their organiza-tion. This spread strategy has proved extremelydifficult, because the champions are busy withtheir clinical responsibilities and may not under-stand how to make their improvement messagesticky for the less enthusiastic majority. It isimportant to remember that the majority isquite different from the early adopters, andneed different messages and motivation toadopt an innovation.

IHI strategy, in fact, does not place the primaryresponsibility for spread on the collaborativeimprovement team. IHI’s spread strategistsunderstand that while pilot improvements are the work of the collaborative team, spread mustprimarily be the job of an organization’s seniorleaders. Organizations that have successfullyspread improvement have done so through theinvolvement of senior leaders.

IHI has relied heavily on the learning collabora-tive to initiate improvement work in health care,but does not believe that collaboratives by them-selves can spread improvement throughoutorganizations. Collaboratives focus on front-lineteams with enthusiastic innovators (champions).They are often successful in spreading ideas fromone organization to another, but have had lesssuccess in fostering broad improvement withinorganizations. Because the majority of people inorganizations are not enthusiastic champions, asingle-minded focus on champion teams is insuf-ficient to spread improvement throughout anorganization.


Paul PlsekIn Spreading Good Ideas for Better Health Care,

Plsek cites Rogers and Gladwell to argue thatonce 10–20 percent of the target population hasadopted an innovation, the tipping point hasbeen reached and it is difficult or impossible tostop further diffusion of the innovation (an ideadisputed below). Plsek also borrows from the“stages of change” concept that places people inthe categories of precontemplation (not ready toconsider a change), contemplation (willing toconsider a change), action (ready to do some-thing concrete), and maintenance (having made a change and striving to continue it). Plsekemphasizes that individuals in these differentstages need different messages to assist in thespread of new ideas.

He also draws on the concept of the complexadaptive system, which describes systems such ashealth care institutions as containing a meshworkof relationships that makes behavior hard to pre-dict, but adaptive in what the people in the sys-tem can do in response to changes in their envi-ronment. For example, 50 years ago primary carecould be successfully practiced by a lone physi-cian, whereas now the march of medical science,together with the uncertainties of reimburse-ment, have made primary care practice an almostimpossible task. This change in objective circum-stances is opening some adaptive primary carephysicians to change, while others remain in theprecontemplative phase. Different arguments areneeded to address physicians at opposite ends ofthe readiness-to-change spectrum (a spectrumparallel to Rogers’ early adopter–majority–laggardconcept). The spread of innovation thus requiresan analysis of the objective circumstances ofcomplex systems and the subjective states of theactors in these systems.

In Spreading Good Ideas for Better Health Care,Plsek offers tools that might help improvementchampions analyze the systems and individualsthat make up the spread target population.

Sarah FraserSarah Fraser has served as a consultant to healthcare organizations in the United Kingdom’sNational Health Service. Her 2004 book onspread is a practical guide with a wealth of goodadvice.13 She feels that health care improvementchampions may misinterpret Rogers. WhileRogers focuses on individuals accepting relativelysimple innovations, spread in health care requiresthe adoption of complex behaviors. Rogers discusses discontinuous change—a Peruvian villager either boils water or doesn’t. A personeither buys an iPod or doesn’t. In contrast,improvement in health care often requiresmany small changes; the changes are not discontinuous (all or nothing) but continu-ous—a series of many small changes. Moreover,while diffusion of innovation theory often dealswith things, innovation in health care is con-cerned with ideas and behaviors.

Fraser also makes the point that innovators arenot normal people. They love changing things,while most people are wary of change.Sometimes innovators cannot grasp how themajority views the world and thus are poorspread messengers—the William Daweses who fail, in contrast with the Paul Reveres whosucceed. The majority are the people who holdorganizations together—they do the day-to-daywork. They don’t run off to conferences or holdmeetings at 7 a.m. They care for patients, day inand day out. If innovators look down on thembecause they are wary of change, spread is dead.

Fraser makes the key point that spreading inno-vation must reduce costs; otherwise, it is notsustainable. In particular, it must create areturn on investment for the organizationspreading the change. Reducing costs for anemployer or insurer is fine, but unless theorganization paying for the innovation is itselfbenefiting financially, the improvement willprobably not become permanent.


Fraser also distinguishes between “let it happen”spread (iPods and MRI scanners) and “make ithappen” spread, which is more difficult but morepertinent to health care improvement work.

Learning from others is an important spreadactivity. “Who else is doing this really, really well?Let’s go visit them and find out how they aredoing it.” Allowing such visits requires givinghost and visiting teams time off. While this gen-eral education is crucial, the specifics must oftenbe tweaked or reinvented from one site toanother because the available personnel, types of patients, and personalities of the caregiversdiffer from place to place.14

A Critique of the Spread LiteratureIn 2006, Fraser became impatient. Many healthcare statistics were intolerable, yet improvementwas spreading at a snail’s pace. In “Undressingthe Elephant: Why Good Practice Doesn’t Spreadin Healthcare”,15 she confronts much of theaccepted doctrine about spread.

n “Pilotitis” refers to champions or organizationsmaking an improvement for a small number ofpatients, often with extra funds or extra per-sonnel. The pilot project succeeds, but the con-ditions under which it succeeds are so differentfrom the norm that it cannot possibly propa-gate, whereupon the change agents blame the“laggards” or “resisters” for the failure of theimprovement to spread. The solution? Involvethe majority of people from the start, listen to them, and have them help design theimprovement.

n The “low-hanging fruit syndrome” refers toinitially picking a population of patients whosecare can be easily improved. The changesmade, however, may not apply to the entirepopulation. Changes should be tried on popu-lations of patients for whom the change willhave the greatest impact, rather than the popu-

lation for whom change is quick and easy.

n Innovators often push an idea as a universalsolution applicable to all organizations when infact the idea may be fine for one organizationbut completely unworkable for another.

n The notion that a tipping point has beenreached when 20–25 percent of people haveaccepted a change does not conform to reality.Improvement must be a continuous process.

n The blind acceptance of Rogers’ categorizationof people—early adopter, majority, and lag-gard—may lead to failure. “Laggards” may bethe pragmatists who keep organizationsgoing—the people who need to validatechanges. The canyon that often exists betweenvisionaries and pragmatists needs to bebridged.

n Spreading improvement requires continuousmeasurement.

n Without leaders, spread programs founder.Leaders must be courageous, curious, and passionate.

n Rather than using the phrase “spreading goodpractice,” it might be better to say “imple-menting better ideas.” This gets away from theconcept that ideas are moving from an activeperson to a passive person and views every per-son as active in the work of implementation.



THIS SECTION DESCRIBES ORGANIZATIONS THAThave been successful in spreading “islands of improvement”into large-scale successful change. The descriptions includeimplementation of better ideas both within organizations andamong organizations. These two geographies of spread overlap:spread within an organization means extending improvementsin one unit of the organization to other units, while spreadamong organizations involves taking innovations in one organ-ization and introducing them into other organizations.

Veterans Health AdministrationThe best example of spread within a large organization is theVeterans Health Administration (VHA) system.16 For decades,the VHA was widely considered the health care system of lastresort. The huge system had deteriorated so badly by the early1990s that Congress considered disbanding it.17 Yet by 2007,the nationwide system—with 154 hospitals and 8,875 clinicsserving more than 5 million patients, run and financed by thefederal government—had been dramatically improved. It nowconsistently outperforms hospitals and clinics in the privatesector, as indicated by the following data.

The VHA reports its performance using quality of care indica-tors across a broad spectrum—breast cancer screening, beta-blocker use after heart attack, cholesterol and blood pressurecontrol, smoking cessation counseling, and influenza immu-nizations for high-risk and elderly patients—that are betterthan commercial and Medicare HEDIS indicators.18

In their 2004 study, Steven M. Asch and colleagues sampled596 VHA patients and 992 non-VHA patients. After adjustingfor risk, they found no significant differences between the twosamples in the age of the patients or the number of chronicconditions. The VHA patients received significantly betteroverall care, chronic care, and preventive care. Although thetwo samples did not differ significantly in acute care, the VHA generally performed significantly better across the whole spectrum of care: screening, diagnostics, treatment, and follow-up.19

For the past six years, the VHA has outranked private-sectorhospitals on patient satisfaction in an annual consumer surveyconducted by the National Quality Research Center at the

III. Examples of the Successful Spread of Improvement in Health Care

University of Michigan, despite the fact that theVHA spends an average of $5,000 per patient,versus the national average of $6,300.20

Quality of care expert Lucian Leape of theHarvard School of Public Health stated that “theVHA is a dramatic example of what can happenif you have the will and the leadership to makechange happen.”21

How did this remarkable transformation takeplace? The simple answer is leadership. A morecomplex answer helps to clarify the concept ofspread.

In the mid-1990s, Dr. Kenneth W. Kizer, thenthe Health Under Secretary for Veterans Affairs,installed an extensive electronic medical recordssystem, decentralized decision-making, closedunderused hospitals, reallocated resources, andinstituted a culture of quality-aided measure-ment. Dr. Kizer had a great advantage comparedto other health system leaders: an integrated sys-tem with a global budget in which quality andfinancial incentives are aligned, and a group ofsalaried physicians who could be required by topleadership to make needed changes. As an exam-ple of quality and cost alignment, the VA systemhas markedly reduced hospitalizations as a resultof increasing its pneumovax vaccination ratefrom 29 percent in 1995 to 94 percent today.

One example of how the VHA system spreadimprovement is its adoption of advanced access.Advanced access is a scheduling system thatallows patients who call for an appointment toget the appointment on the same day. It is animportant innovation for patients with chronicconditions, because appointment delays are oneimportant reason why patients with chronic con-ditions fail to receive proper management. As aresult of this innovation, the national averagewait time for an appointment at the VHA systemfell from 60 days in 2000 to 25 days in 2004.24

The VHA started the advanced access process in1999, when the organization’s national leadershipdecided that appointment delays were an unac-ceptable quality problem. The central leadershipinitiated an internal collaborative of 134 teamsfrom all 22 Veterans Integrated Service Networks(VISNs, or regions) throughout the UnitedStates. The collaborative demonstrated that teamsin particular sites could improve timely access tocare. Many organizations would have stopped atthat point. But the VHA leadership understoodthat the great bulk of sites in the VHA systemhad been untouched by the improvement. Onlythrough centrally mandated change could everypatient seeking care in the VHA system benefitfrom advanced access. Accordingly, the top lead-ership mandated advanced access for each of the22 VISNs.

The key strategy was to require local leaders totake on the task. Each VISN designated aspread team responsible for the improvement,including facility directors with organizationalclout. As long as those local leaders could showimprovement in average waiting time for anappointment, it did not matter how the leadersaccomplished the change. Different VISNsachieved results in different ways. Some heldregional collaboratives. Many had meetings inwhich pilot sites demonstrated how theyachieved their successes. Information—how toinitiate and sustain advanced access—was key,as were champions who could persuade theskeptics.


EHR Helps Spread Improvements

An important tool for spreading improvementin the VHA is the electronic health record, orEHR.22 The EHR can remind physicians andcare teams about needed preventive andchronic disease studies throughout the entiresystem, not just in a few early adopter clinics.The electronic data allows for routine perform-ance measurement that is both a quality-enhancing and a spread tool. In diabetes care,for example, processes of care, and HbA1c and LDL-cholesterol control, were better forpatients in the VHA system than for patients in commercial managed care.23

The VHA leadership did not tolerate the statusquo. Their message was clear. “We’re going toreduce delays in patients getting appointments.”Everyone in the huge VHA system knew thatthis was a priority. Some providers viewedadvanced access as more work; the leaders neededto match the messages they created to the con-cerns of the front-line providers and to enlist asmessengers those providers whose day-to-daywork was positively rather than negativelyimpacted by the change.

In some organizations, change bubbles up fromthe front lines. It is either seen as valuable andadopted by the leadership and successfullyspread, or it is not embraced centrally and dies.In the VHA, advanced access did not bubble upfrom the base; the idea came from outside theorganization, was adopted by central leaders, andmandated for the entire organization. While thechange was chiefly top-down, a bottom-up aspectexisted in which individual sites could re-inventhow the change was accomplished to match theirneeds or desires. Moreover, even though thechange was mandated from the top, little couldbe done about sites that resisted, which some did.

As a result of the VA work, a “Checklist forSpread” was developed.25 It includes such questions (paraphrased here) as:

n Is improvement in this area a strategic priorityfor the organization?

n Is there an executive who is responsible for thespread of the improvement?

n Is there a person or team in the leadershipgroup who will be involved in the day-to-dayspread activities?

n Will the leadership supply resources needed forsuccess (personnel, information technology,tools, etc.)?

n Has the advantage of adopting the change beendocumented and communicated in an easilyunderstood package to all potential adopters?

n Is there a successful site that has implementedthe change in a way that is scalable throughoutthe organization?

n Are there credible messengers who can per-suade potential adopters to implement theinnovation?

n Is there a clear plan to communicate the inno-vation throughout the organization and toassist different sites in making the necessarychanges?

Humboldt–Del Norte IndependentPractice AssociationIt is relatively uncommon for significant qualityimprovement work to take place in small primarycare practices, because these practices have fewresources and little time available for improve-ment work. It is even less common to observesignificant quality improvement taking place inthe primary practices of an entire community.Through the leadership of family physician Dr. Alan Glaseroff, however, improved diabetescare spread across the medical community ofNorthern California’s Humboldt County over abrief three-year period. While the VHA exampledemonstrated spread within an organization, the Humboldt example represents spread amongdifferent organizations.

The Humboldt–Del Norte Independent PracticeAssociation (IPA) was formed to give physiciansclout in negotiating contracts with HMOs. TheIPA includes 240 physicians, plus 140 mid-levelpractitioners and mental health providers work-ing in 26 practices and five community healthcenters—virtually all the medical practices in thecounty. Some 84 of the physicians are in primarycare.

Dr. Glaseroff employed the structure of the IPA to spread a diabetes registry that includes themajority of patients with diabetes in the county.The registry is most pertinent to the 84 primarycare physicians, who care for most patients with


diabetes. Registries are critical tools for improv-ing chronic disease care because they allow prac-tices to look at all their patients with a specificchronic disease, to determine which patientsneed laboratory or other studies and bring theminto care for those studies, and to focus addition-al resources on patients with poor disease control.Registries create a paradigm shift for physicians,helping them to consider the care of their entirepanel of chronically ill patients rather than limit-ing their focus to individual patients who happento have an appointment. Moreover, registry datacan be fed back to physicians to inform them ofhow well they are performing, using such meas-ures as the percentage of patients with diabetes ingood versus poor control.

How did Dr. Glaseroff succeed in spreading thediabetes registry so widely in a short time period,given that most physicians had no familiaritywith a diabetes registry?

First, Dr. Glaseroff made adoption of the dia-betes registry easy for primary care physicians.The IPA built the registry and IPA personnelinput clinical data into the registry so that theindividual practices did not have to perform thiswork. The IPA taught medical assistants in physi-cian practices how to use the registry. The IPAgave computers and free Web access to practices.A nurse practitioner from the IPA traveled topractices to assist them in making the workflowchanges needed for improved diabetes care.

As a result, almost all the primary care physiciansare now using the registry, which has resulted ina striking improvement in diabetes care inHumboldt County. In California’s pay-for-per-formance system, the IPA had the third bestrecord in the state—the only small IPA toachieve such results.

Second, the IPA tailored the messages used to persuade physicians to adopt the registry,employing different messages for physicians with differing concerns. Perhaps 20–25 of the

84 primary care physicians were early adopters,willing to take some chances in order to improvequality. For them, the message of improved quality was a sticky message. A larger number ofphysicians was persuaded by the message that theregistry could make diabetes care easier.

The third message, for more reluctant physicians,took advantage of the pay-for-performance initia-tive that came to California in the past threeyears; virtually all physicians saw that the registrycould help them obtain more pay-for-perform-ance dollars. The three arguments—improvequality, make work life easier, and get paidmore—were each persuasive for different seg-ments of physicians.

Now that the registry is in wide use, furtherimprovements are being spread; in particular,assisting patients to become more active partnersin the management of their diabetes, encouragingpatients to adopt healthier behaviors throughgoal setting and action planning, and addressingdepression, which frequently accompanies dia-betes and other chronic illnesses. The registry is also being expanded to include preventive services, as well as hypertension, hyperlipidemia,and depression.

The IPA did not begin its diabetes improvementwork with a small pilot that would hopefullyspread throughout the county. Rather, the IPAused a “big bang” approach—similar to theInstitute for Clinical Systems Improvement’smass movement model (see below). To initiatethe improvement work, the IPA called a commu-nity meeting attended by many physicians, andexplained how the registry might work in theirpractice. The registry was rolled out to as manypractices as possible, and the kinks were workedout in different practices at the same time. Anurse practitioner visited practices and listened to the successes and challenges of each practice.In addition, a leadership council met weekly tolearn from the successes and to work to solvechallenges.


The Humboldt County experience was a cam-paign rather than a collaborative. Like the100,000 Lives Campaign discussed below, theIPA took an idea tested in other parts of thecountry and, using appropriate messages, per-suaded practices to sign on. The IPA providedassistance and problem-solving advice. In con-trast to a collaborative, which tries out untestedconcepts in a pilot site, the IPA’s campaign tar-geted the entire primary care infrastructure of thecounty in one fell swoop.

Cincinnati Children’s HospitalMedical CenterCincinnati Children’s Hospital Medical Center isa not-for-profit pediatric academic medical centerwith 475 beds and 15 patient care sites through-out the region. Cincinnati Children’s is widelyconsidered one of the leading health care organi-zations in the United States. The medical centerhas introduced a number of family-centered careimprovements such as Family First Rounds,which allows family members to join physiciansfor clinical discussions at the child’s bedside,rather than the traditional mode of doctors dis-cussing hospitalized children among themselvesand later informing the family members (whooften feel left out). Families are involved inimprovement work at the top leadership leveland as part of specific improvement teams.

Cincinnati Children’s regularly measures a widevariety of performance indicators, includingadverse drug events per 1,000 doses; bloodstreaminfections per 1,000 catheter days; ventilator-acquired pneumonia per 1,000 ventilator days;surgical site infections per 100 procedure days;percent of eligible patients receiving evidence-based care for seven medical diagnoses; risk-adjusted cost per discharge; appointment waitingtimes for specialty outpatient clinics; and specificmeasures on quality of chronic disease care forattention-deficit hyperactivity disorder, autism,cystic fibrosis, diabetes, inflammatory bowel dis-ease, juvenile rheumatoid arthritis, and asthma.

Cincinnati Children’s received a PursuingPerfection grant26 and used this improvementprocess to reduce hospital admissions for severalcommon childhood illnesses by 15 percent and to improve the care of children with cystic fibro-sis. In surveys, about 90 percent of families statethat they can get an outpatient appointment assoon as they want it, that they are involved indecisions about their child’s care, and that theyreceive understandable answers to theirquestions.27

The medical center’s acclaim was not earnedovernight. For years, the organization’s top lead-ers have stimulated an organization-wide cultureof improvement. Improvement is a topic at boardmeetings and senior management conferences,and everyone in the organization is expected tosupport excellence. The CEO and board oftrustees set strategic priorities for improvementbased on feedback from families and front-linecaregivers. The leadership provides front-linechampions with the time and funds to visit andlearn from other organizations. Each seniorleader is responsible for one or two of theapproximately 20 improvement teams, and theircompensation is tied to achievement of theirteams’ goals. Financial analysts determine the fiscal impact of improvement projects.28

One area in which Cincinnati Children’s hasachieved excellence is the treatment of pediatricasthma.29 The organization launched a physician-hospital organization in 1996. One aim of thePHO was to spread successful improvementmodels among primary care practices, includingcommunity physicians. The PHO included 165primary care practices, ranging in size from oneto 12 physicians. The Cincinnati Children’s PHOasthma improvement initiative began in 2003with the goal of improving asthma care to 13,000children in 43 primary care practices (30 percentof the region’s pediatric asthma population). Aquality improvement team was organized in each practice; network meetings and conference callspromoted the spread of successful improvement


strategies. A measurement tool was developed tohelp practices assess their progress. Parents fill outa survey via computer or on paper in the waitingor exam room, and the physician completes thedata. A data entry person in each practice entersthe data. As of March 2006, the proportion ofthe asthma population who received a flu shothad increased dramatically to 63 percent. Some87 percent of patients had an asthma action plan,and 97 percent of persistent asthmatics were oncontroller medications. Asthma-related hospitaladmissions over 12 months dropped to 12 per10,000 patients (compared with 24 for non-PHOpractices, and 27 nationally).

Much of the improvement was attributed toimproved patient/parent self-management skillsand collaborative goal-setting between parentsand providers, using the asthma nurse educatormodel with planned care done during a pre-visitsession and post-visit follow-up care. Another keytool is the asthma registry for the entire PHO,which identified specific patients not engaged inperfect asthma care, allowing practices to contactthose patients and make the necessary improve-ments in care. The PHO found that the keys tothe spread of excellent asthma care included:

n Identifying the population and creating a registry;

n Collecting data and feeding it back to prac-tices;

n Self-management skills training for families;

n Collaboration among practices;

n Practice-level and PHO-level leadership; and

n Pay-for-performance—financial rewards forpractices with improving asthma measures.

The importance of registries in chronic diseasemanagement is widely recognized, but the role ofregistries in the spread of improvement is less evi-dent. Cincinnati Children’s viewed its Web-basedPHO asthma registry as a spread tool. Practicescould see their own data compared to that of

other practices; those that lagged behind werestimulated to improve. Pay-for-performancestrengthened that incentive.

Cincinnati Children’s President and CEO JimAnderson makes clear what should be expectedof top leaders:

“One should not tolerate processes that donot work. If you fix them, the savings morethan offset the costs, and the outcomes aresuch that the marketplace finds them veryattractive. Thus, not only can you deliver ahigher-quality product at a lower cost, butalso you can charge more for it. The busi-ness case is pretty compelling. There is nobusiness reason not to tackle the issues.Coupled with the fact that this would pro-duce better medical outcomes… there issimply no reason other than cultural barri-ers or inept management for us to continuein an environment where we deliver carethat is not as good as it could be.”30

“When I meet with other CEOs fromother institutions, I am concerned that (thequality movement) does not seem to havemuch traction. There are exceptions, but inmy experience, not many are committingto the revolution. I think people feel beatendown by day-to-day challenges, and are notlooking at the opportunity for transforma-tion as one that can be productively pur-sued. I realize that it is a big job, but wejust need to do it.”

A message from Cincinnati Children’s warns thatimprovement does not take place in a year ortwo. It takes time and a sustained commitment.Improvement work exemplifies the unity ofopposites: impatience to get things done andpatience to keep getting things done.


Institute for Clinical SystemsImprovementIn 1993, Minnesota’s Institute for ClinicalSystems Improvement (ICSI) was created inresponse to a Minnesota employers’ grouprequest for quality improvement within provider organizations.31 ICSI’s founders wereHealthPartners Medical Group, Mayo Clinic,and Park Nicollet Health Services. ICSI isfunded by Minnesota’s health plans.

ICSI is all about spread. Dr. Gordon Mosser,until recently ICSI’s executive director, describesa common story: a champion brings forth anew idea, senior management approves, theinnovation is implemented in one hospital unitor clinic site, and no other units or sites pick itup. ICSI’s purpose is to overcome that scenario.ICSI promulgates spread both within andamong organizations.

ICSI’s first several years focused on the creationof clinical practice guidelines and the develop-ment of concrete quality metrics to measure eachprovider organization’s performance. Duringthose first years, ICSI had three members—HealthPartners, Mayo, and Park Nicollet. ICSI’sstrategy was to start with these prestigiousprovider organizations, hoping that others wouldchoose to join this elite club. Thus the initialspread strategy was to challenge providers toemulate the highest performers in the state, andto engage provider organizations through thedevelopment of practice guidelines and qualitymeasures.

ICSI uses the word “propagation” rather thanspread, in part because propagation is an agricul-tural term that would have traction in an agricul-tural state. Propagation can take place both with-in an organization and among organizations.ICSI is concerned with both.

Spread among organizations. The most impor-tant plank in ICSI’s spread platform is therequirement to adhere to certain performancestandards to join ICSI and remain in the organi-

zation. Membership in ICSI requires a commit-ment to engage in four improvement projectseach year. Each project needs to be of substantialimportance and must be tracked with perform-ance measures. These commitments must bemade by an applicant organization’s highest senior leadership.

ICSI’s Board of Directors, which includes repre-sentatives of member provider organizations,health plans, and consumers, chooses two ICSI-wide improvement topics each year; two of eachmember’s improvement projects are the ICSI-wide topics, with the other two chosen by themember organization. ICSI-wide topics haveincluded the prompt access to appointments; andthe care of patients with diabetes, depression, andcongestive heart failure. Over the years, a fewmedical groups who joined ICSI were unable tomeet their commitments and were asked to leavethe institute.

The ICSI model for propagation of improve-ment, then, is a disciplined model in whichmembership in ICSI is voluntary, but forproviders joining ICSI, improvement work ismandatory.


ICSI’s Tipping Point

ICSI reached a turning point—one could call it a tipping point-—around 2000. In its first sev-eral years, provider organizations joined to beat the table with Minnesota’s most prestigiousproviders. After 2000, not being in ICSIbranded a provider as second-rate. Moreover,as public reporting of quality and pay-for-performance were widely embraced byMinnesota’s health plans, provider organiza-tions experienced external pressure toimprove. ICSI remains cautious in acceptingnew members, not wishing to dilute its highstandards.

Discussions have taken place within ICSI’s leadership regarding the definition of success.Currently, ICSI’s culture rewards trying toimprove; if a provider’s performance measures donot change, there are no consequences as long asa sincere effort was made. A more stringentrequirement would involve ICSI setting goals; forexample, the percent of patients with diabetes ina provider organization achieving HbA1c levelsbelow 7. ICSI would evaluate an organizationbased on movement toward the goal. Thus far,ICSI has not made measurable improvement arequirement.

In addition to ICSI’s requirements for conduct-ing improvement work, ICSI provides mandato-ry training for all new members’ improvementleaders. Training involves a one-day meetingthree times a year, monthly conference calls,coaching sessions, and homework. ICSI trainersalso travel to medical groups and work with lineparticipants. Sessions cover the science andmethods of improvement, including measure-ment, and specific topics such as improvedaccess, the use of clinical practice guidelines, theChronic Care Model, and building care teams.

Improvement work often uses the collaborativeprocess, with a number of organizations joiningan “action group” on, for example, improvedaccess, diabetes, or depression. Member organiza-tions can conduct improvement work on theirown or join an action group; those that join arerequired to send in measures every month, andcan be ejected from an action group if they fail toreport their measures. In action groups, memberscan share forms, protocols, and successes withone another-–a process helpful in spreadinginnovation.

To summarize, ICSI has been successful inspreading improvement from one organization toanother by becoming the only game in town andby requiring a serious commitment to improve-ment as the ticket to get into the game.

Spread within organizations. Since mostimprovement is initiated at a pilot site within alarger organization, the problem of spreading theimprovement from the pilot to the entire organi-zation is a universal one. According to Dr.Mosser, a former director of ICSI, five models(sales, parallel play, central driver, billboard, andmass movement) describe how different providerorganizations in ICSI have worked on intra-organizational spread.

The sales model involves the pilot site having achampion who is effective at persuading others in the organization. After the pilot site achieves asuccessful improvement, the champion goes toother sites in the organization to describe theimprovement and to persuade others to adopt it.This model is particularly applicable to organiza-tions with strong autonomous physicians whoresist change mandated from central leadership.

The parallel play model is relevant to academicprovider organizations, which generally containphysician leaders who do not accept data unlessit comes from a randomized controlled trial. Inthis model, several pilot sites work on the sameimprovement project; if they all produce similarresults, the skeptics have more difficulty under-mining the improvement.

In the central driver model, top leadership decideswhat the organization needs and chooses pilotsites with a strong probability of success. Whenthe pilots achieve results, the top leadership man-dates the change for the rest of the organization.

The billboard model relies on broadcasting newsof success in the pilot site; this model acceptsthat different units are autonomous and may ormay not adopt the improvement.

The mass movement model does not rely onsequential propagation from early adopters to themajority; instead, change is promoted by leadersapproaching all units at once through inspira-tional messages and mass communication.


All these models require improvement work tocome from an organization’s senior leadership,who can decide which spread model, or combi-nation of models, might work best. Ideally, seniorleadership decides on a spread model at thebeginning of improvement work, not after a pilotsite has completed its work.

In ICSI’s experience, certain improvement cam-paigns worked better than others. Advancedaccess was successfully adopted and sustained bya number of medical groups because the methodsdeveloped by advanced access founder MarkMurray were concrete, understandable, andmeasurable. In contrast, some chronic diseaseimprovement work was more difficult becausethe Chronic Care Model does not prescribe astep-by-step formula for improvement. The ICSIprovider organizations most successful in improv-ing chronic care were those that implemented aregistry, assigned personnel to run the registry,and used the registry to ensure that every patientreceived all indicated care. Improvement is easierto propagate if simple, concrete steps can accom-plish the change.

Improving depression care has been a recent cam-paign for ICSI, using the PHQ9 questionnaire toscreen for depression, and having a trained per-son designated to follow up. The former changeis easy; the latter is difficult because health plansdo not reimburse for such follow-up. Propagatingsimple changes is necessary for spread, but if per-sonnel are not available or there is no businesscase, it may not be sufficient for success. A recentICSI initiative has been to work with healthplans to develop new reimbursement codes topay for evidence-based depression care.

ICSI leadership realizes that a “culture of quality”must be adopted in order to “till the soil” toallow propagation. Traditional organizational cul-ture is based on physician autonomy. If clinicalpractice guidelines dictate that a physician usecertain medications and order certain diagnosticstudies, what happens if the physician does not

comply? Under current organizational culture inmost health institutions, nothing would be done.Public reporting and pay-for-performance mightcreate incentives for such physicians to changetheir ways, but few organizations would requirethat the guidelines be followed. ICSI has dis-cussed the possibility of requiring physicians to“do the right thing,” but has not mandated sucha drastic culture change. ICSI is developing acurriculum to train its member organizations onissues related to organizational culture. Spreadinga culture of quality is far more complex thanspreading a diabetes registry.

ICSI is one of the nation’s most successfulimprovement organizations, fostering spreadboth within and among health care providerorganizations. ICSI’s regional focus on the entirestate of Minnesota is a model for similar effortsin other regions. Rarely, however, have otherregionally focused improvement efforts adoptedICSI’s high bar—requiring serious improvementwork as a condition for membership.

IHI’s 100,000 Lives CampaignIn December 2004, the Institute for HealthcareImprovement (IHI) launched the 100,000 LivesCampaign, a national initiative to reduce pre-ventable deaths for hospitalized patients.32 OnJune 14, 2006, Don Berwick, IHI’s President and CEO, announced that the campaign hadprevented an estimated 122,300 deaths in 18months.33 While some dispute the precise num-bers, the campaign had unquestionably beenhighly successful.

Fewer than 20 IHI staff led the 100,000 LivesCampaign. How did such a small number ofpeople create improvement in more than 3,000hospitals, accounting for 75 percent of acute carebeds in the United States? The experience of the100,000 Lives Campaign is an important exam-ple of spreading improvement widely and rapidly.


IHI developed partnerships with many organiza-tions to achieve its goals. Nationally, the cam-paign was supported by the Association ofAmerican Medical Colleges, American MedicalAssociation, American Nurses Association,Center for Medicare and Medicaid Services,Centers for Disease Control and Prevention, The Leapfrog Group, Kaiser Permanente, theVeterans Health Administration, UniversityHealth System Consortium, Joint Commissionon Accreditation of Healthcare Organizations,and others. In addition, IHI worked with collec-tions of regional organizations called “nodes.”These nodes aggregated organizations in moststates, often including state hospital associations,quality improvement organizations (QIOs), stateand local medical societies, state nursing associa-tions, business groups, unions, and large hospitalsystems. IHI also applied the principle that someorganizations are early adopters of change, andcalled upon early adopter hospitals to serve asmentor hospitals, assisting other hospitals in theirgeographic areas to make improvements. Thesuccess of the campaign, then, depended on activating many organizations to partner in the effort.

Another strength of the campaign was its sim-ple, focused, and persuasive improvement goals.In the terminology of The Tipping Point, themessages were “sticky.” The campaign targetedsix changes for which research evidence demon-strated improved outcomes: rapid responseteams, care for patients with acute myocardialinfarction, medication reconciliation, preventingcentral line infections, preventing surgical siteinfections, and preventing ventilator-associatedpneumonia. IHI provided practical tools thathospitals could use to quickly implementchanges in these six areas. Hospitals were free to use these tools however they wished.

Some nodes drew upon regional collaboratives toassist in the improvement effort. Others organ-ized monthly conference calls or educational sessions among participating hospitals. Regional

newsletters were another spread tactic. The nodessupported the campaign by raising funds, assist-ing hospitals in implementing the changes, andensuring that data was regularly reported to IHI.Some nodes are continuing to work on region-wide improvement even after the initial phase ofthe campaign has been completed.

The 100,000 Lives Campaign is quite distinctfrom the typical collaborative. Collaborativesbring together front-line teams to test and imple-ment improvements. In contrast, the 100,000Lives Campaign worked through top leadership.Collaboratives generally apply the Model forImprovement—testing changes through PDSAcycles—to discover what works. The 100,000Lives Campaign focused on improvement thathad already been tested; the campaign leadersknew what worked and disseminated this knowl-edge broadly. Hospitals did not necessarilyunderstand the Model for Improvement or thePDSA concept. A collaborative is actually a pre-spread methodology, using front-line championsto discover how to make improvements. The100,000 Lives Campaign, in contrast, was aspread methodology focused on six areas forwhich the improvements were well known andtested. The campaign recognized that spreadshould not take place until the improvementmethods have been carefully tested and areproven solid.

Because the 100,000 Lives Campaign was solarge, the IHI leadership could not work directlywith each organization; many of the details weredelegated to nodes and mentor hospitals. IHI’srole was to inspire and organize each region andto provide simple “how-to” materials to all par-ticipants. In a sense, the campaign was more top-down than the traditional collaborative; ratherthan encouraging teams to find their way, IHIdecided on the goals and told the participatinghospitals how to attain them. The message wassticky, the product was simple, and the goal—saving lives—was universally accepted.



THE EXAMPLES CITED IN THIS REPORT PROVIDEreasons to be optimistic that improvement can move fromsmall pilot sites to entire organizations, or can be successfullyimplemented organization-wide from the outset under certainconditions. But this optimism must be tempered. The organi-zations described here were chosen because they demonstratesuccess, but they are also few and far between. The conceptthat improvement will spread by itself once a tipping point hasbeen reached is often not true for health care organizations. In many cases, “islands of improvement” remain isolated, orrevert over time to former practices. Nonetheless, these exam-ples, plus the literature of innovation spread, can provide someuseful lessons about the ways entire organizations can imple-ment improvements in health care.

Leadership. Champions may initiate improvement, but itdepends on top-level organizational leaders to create an institu-tional culture ready to accept change, and to spearhead thespread of particular improvements. Only senior leadership hasboth the authority and the breadth of vision to do so.

Senior leaders can spread improvement through a number ofdifferent strategies. They can mandate changes, use variousforms of persuasion, offer financial incentives such as internalP4P programs, and, if sufficiently adept as messengers, caninspire people throughout the organization. Purely top-downapproaches are common in large organizations, often begin-ning with a pilot program and then mandating its spread tothe entire organization. This approach usually works best if dif-ferent sites in the organization can customize the innovation tofit their particular needs, and if many people in the organiza-tion have received thorough training on the innovation.

Improvement collaboratives with front-line teams are only onestep in an improvement process. Organization-wide changerequires commitment at the top. If spread is the work of seniorleaders, they need to understand the issues and believe inthem. Begging for senior leadership support doesn’t work; thesenior leaders need to be educated about the improvement sothat they see its value.

Champions. Innovation champions are often thought of asthe engines of change within organizations. But a smallimprovement team led by champions is unlikely to spread

IV. Conclusion

improvement throughout a large-scale organiza-tion. If the champions are also front-line careproviders, they will have neither the time nor theauthority to propagate improvements organiza-tion-wide. The proper role of these champion-ledteams is to experiment with and adjust innova-tions, and to serve as messengers to explain newideas to potential adopters. To the extent thatsenior leaders need to believe in an improvementand understand the issues surrounding it, cham-pions can be invaluable in the leadership educa-tion process.

Enthusiastic champions are essential to the initia-tion of improvement work. But champions arequite different from the average front-line care-giver; they often create success by working 18hours a day. Over-reliance on champions candestine an innovation for failure. It creates unre-alistic expectations about what can be achieved ina certain time period by front-line providers.

Front-line caregivers. In general, good ideascome from the front lines of an organizationbecause people there know how things work (ordon’t work). Moreover, their ideas about innova-tions are crucial because they are the people whowill be implementing them.

Champions may have a highly developed desireto innovate, but most people do not. Thus,arguments and messages that appeal to innova-tors or early adopters may not be sufficient forthe majority of front-line workers. Instead,arguments need to speak directly to their dailyexperience: for example, the innovation willmake your work easier, will improve publiclyreported performance measures, or will increasepay-for-performance bonuses. In this context,Rogers’ division of the population—earlyadopters, the majority, and laggards—may becounterproductive. Resistance to innovation bypeople working daily to care for patients meansthat their ideas need to be heard and addressed,not dismissed as heel-dragging.

Macrosystem vs. microsystem change. A balanceis needed between macrosystem reform andmicrosystem improvement. Changing health careorganizations one by one will take decades orlonger. A thoughtful improvement-fostering pay-ment change by Medicare and large private insur-ers could stimulate improvement in every healthcare organization in the country overnight. Onthe other hand, provider organizations need theknowledge and will to harness a financing changeto create real improvement (rather than trying togame the system).

To sum up the lessons that can be taken from thetheoretical and practical work discussed in thisreport, innovations in health care can spreadmost easily if:

1. Innovations fit with the culture and values ofthe majority of potential adopters. Physicians,for example, may have a culture of doing agood job for patients but also a culture ofphysician autonomy in decision-making. Attimes, improvement requires confrontingrather than accommodating the prevailing cul-ture. Spreading improvement in such circum-stances is a significantly greater challenge.

2. Innovations can be clearly shown to improvemeasured performance for several groups ofpatients in several sites.

3. Top leadership actively promotes (and some-times mandates) not only adoption of thechange but also acceptance of the ideas under-lying it, while allowing individual sites withinthe organization to customize implementationas long as measured performance improves.

4. New ideas are framed and presented by indi-vidual messengers in a way that makes them“sticky” for the majority of people in theorganization.

5. New ideas make day-to-day work easier forthe potential adopters.

6. A financial “business case” for the innovationcan be made for the organization that is toadopt it.


1. Rogers, E.M. Diffusion of Innovations, 4th ed. (NewYork:The Free Press, 1995).

2. Institute of Medicine. Crossing the Quality Chasm: ANew Health System for the 21st Century (Washington,D.C.: National Academy Press, 2001).

3. Saydah S.H., J. Fradkin, and C.C. Cowie. 2004. “PoorControl of Risk Factors for Vascular Disease amongAdults with Previously Diagnosed Diabetes.” JAMA2004; 291:335-342.

4. Plsek, P., Spreading Good Ideas for Better Health Care: APractical Toolkit; Tools, Perspectives, and Information forHealth Care Providers. (Irving, Texas: 2000) (cited Oct.17, 2005) (www.tachc.org/HDC/Tools/Organization/Docs/Spreading_Good_Ideas_for_Better HealthCare.pdf; site discontinued).

5. Greenhalgh, T., G. Robert, F. MacFarlane, P. Bate,O.Kyriakidou. 2004. “Diffusion of Innovations inService Organizations: Systematic Review andRecommendations.” Milbank Quarterly 2004; 82:581-629.

6. Ibid.

7. Gladwell, M. The Tipping Point: How Little Things CanMake a Big Difference (Boston: Little Brown andCompany, 2002).

8. Berwick, D.M. 2003. “Disseminating Innovations inHealth Care,” JAMA 2003; 289:1969-1975.

9. Massoud, M.R., G.A. Nielsen, K. Nolan, M.W. Schal,C. Sevin. A Framework for Spread: From LocalImprovements to System-Wide Change. IHI InnovationSeries white paper. (Cambridge, Mass. Institute forHealthcare Improvement; 2006) (www.ihi.org/IHI/Results/WhitePapers/AFrameworkforSpreadWhitePaper.htm)

10. Fraser, S.W. Spreading Changes. Institute for HealthcareImprovement (Cambridge, Mass.) (www.ihi.org/IHI/Topics/Improvement/SpreadingChanges/)

11. Daniel, D.M., J. Norman, C. Davis, and others. 2004.“A State-level Application of the Chronic IllnessBreakthrough Series: Results from Two Collaboratives onDiabetes inWashington State.” Jt Comm J Qual Safety2004; 30:69-79.

12. Mittman B.S. 2004. “Creating the Evidence Base forQuality Improvement Collaboratives.” Annals of InternalMedicine 2004; 140:897-901.

13. Fraser, S.W. Accelerating the Spread of Good Practice: AWorkbook for Health Care (Chichester, UK: KingshamPress, 2004).

14. Fraser, S.W. Spreading Changes Expert Host. Institute forHealthcare Improvement (Cambridge, Mass.)(www.ihi.org/IHI/Topics/Improvement/SpreadingChanges/SpreadingChangesExpertHost.htm)

15. Fraser, S.W., “Undressing the Elephant: Why GoodPractice Doesn’t Spread in Healthcare,” Lulu.com, 2007.

16. Oliver, A. 2007. “The Veterans Health Administration:an American Success Story?” Milbank Quarterly2007;85:5-35.

17. Arnst, C. “The Best Medical Care in the U.S.: HowVeterans’ Affairs Transformed Itself—and What It Meansfor the Rest of Us.” Business Week, July 17, 2006.

18. Glabman, M. 2007. “Health Plans Can Learn fromVHA Turnaround.” Managed Care 2007; 16(2):26-38.

19. Asch, S.M., E.A. McGlynn, M. Hogan, R.A. Hayward,P. Shekelle, L. Rubenstein, J. Keesey, J. Adams, and E.A.Kerr. 2004. “Comparison of Quality of Care for Patientsin the Veterans Health Administration and Patients in aNational Sample,” Annals of Internal Medicine2004;141(12): 938–45.

20.Arnst, C. “The Best Medical Care in the U.S.: HowVeterans’ Affairs Transformed Itself—and What It Meansfor the Rest of Us.” Business Week, July 17, 2006.

21. Ibid.

22.Kupersmith J., J. Francis, E. Kerr, and others.“Advancing Evidence-based Care for Diabetes: Lessonsfrom the Veterans Health Administration,” Health AffairsWeb Exclusive, January 26, 2007.

23. Kerr E., R.B. Gerzoff, S.L. Krein, and others. 2004.“Diabetes Care Quality in the Veterans Affairs HealthCare System and Commercial Managed Care: theTRIAD Study.” Annals of Internal Medicine 2004;141:272-281.

24.Nolan K., M.W. Schall, F. Erb, T. Nolan. 2005. “Using aFramework for Spread.” Commission J Qual Safety 2005;31:339-346.

25. Ibid.

26.Institute for Healthcare Improvement. PursuingPerfection. (Cambridge, Mass.) (www.ihi.org/IHI/Programs/StrategicInitiatives/PursuingPerfection.htm)

27. www.cincinnatichildrens.org/about/measures/hpce/

28. Britto, M.T., J.M. Anderson, W.M. Kent, and others.2006.“Cincinnati Children’s Hospital Medical Center:Transforming Care for Children and Families.” Jt Comm J Qual Safety 2006;32:541-548.


Endnotes

29. Mandel K., G. Pandzik, M. McKibben, and others.2006. “Improving Asthma Care: A Model for Partneringwith Pediatric Practices,” presentation at the NationalInitiative for Children’s Healthcare Quality AnnualMeeting, March 17, 2006.

30.www.ihi.org, click on “Topics, Improvement,Improvement Methods, Improvement Stories.”

31. www.icsi.org

32.Berwick D.M., D.R. Calkins, C.J. McCannon, A. D.Hackbarth. 2006. “The 100,000 Lives Campaign,”JAMA 2006; 295:324-327.

33. www.ihi.org/IHI/About/InTheNews/PressReleases2006.htm


the science of spread: how innovations in care become the norm

Documents