Hfp

SDa

b

c

d

a

A

R

R

2

A

K

H

P

C

1

InTi

T

1d

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129

journa l homepage: www. int l .e lsev ierhea l th .com/ journa ls / i jmi

ealth information exchange in small-to-medium sizedamily medicine practices: Motivators, barriers, andotential facilitators of adoption

tephen E. Rossa,∗, Lisa M. Schillinga, Douglas H. Fernaldb, Arthur J. Davidsonc,avid R. Westd

Division of General Internal Medicine, University of Colorado Denver, Aurora, CO, United StatesDepartment of Family Medicine, University of Colorado Denver, Aurora, CO, United StatesDenver Public Health, Denver, CO, United StatesDepartments of Medicine and Family Medicine, University of Colorado Denver, Aurora, CO, United States

r t i c l e i n f o

rticle history:

eceived 17 July 2009

eceived in revised form

0 November 2009

ccepted 3 December 2009

eywords:

ealth information exchange

rimary care

ase study

a b s t r a c t

Purpose: For small-to-medium sized primary care practices (those with 20 or fewer clinicians),

determine desired functions of health information exchange (HIE) and potential motivators,

barriers, and facilitators of adoption.

Methods: Case study approach with mixed quantitative and qualitative methods. Nine prac-

tices in Colorado were purposively selected. Five used paper records and four were already

participating in health information exchange.

Results: Practices particularly desired HIE functions to allow anywhere/anytime lookup of

test results and to consolidate delivery of test results. HIE-generated quality reporting was

the least desired function. Practices were motivated to adopt HIE to improve the quality and

efficiency of care, although they did not anticipate financial gains from adoption. The great-

est facilitator of HIE adoption would be technical assistance and support during and after

implementation. Financial incentives were also valued. Trust in HIE partners was a major

issue, and practices with rich professional and social networks appeared to be especially

favorable settings for HIE adoption.

Conclusions: These findings may assist policymakers in promoting adoption of HIE among

small-to-medium sized primary care practices, a major component of the US healthcare

system.

. Introduction

nteroperability will be essential to reap the full value of

ational investments in health information technology [1].he estimated annual net value of a fully interoperable health

nformation infrastructure in the US may be as high as $78

∗ Corresponding author at: Mail Stop B180, 12631 E. 17th Avenue, Roomel.: +1 303 724 2267; fax: +1 303 724 2270.

E-mail address: Steve.Ross@ucdenver.edu (S.E. Ross).386-5056/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights resoi:10.1016/j.ijmedinf.2009.12.001

© 2009 Elsevier Ireland Ltd. All rights reserved.

billion and may provide positive net value even during theimplementation period [2]. In contrast, the estimated value ofan incompletely interoperable infrastructure drops to $24 bil-

8504, PO Box 6511, Aurora, CO 80045, United States.

lion with negative value during implementation. States andfederal agencies such as the Agency for Healthcare Researchand Quality [3] have accordingly made substantial invest-ments in health information exchange (HIE).

erved.

i c a l

124 i n t e r n a t i o n a l j o u r n a l o f m e d

However, while the majority of patients receive their pri-mary care in smaller ambulatory practices, involvement ofthese practices in HIE projects has lagged that of hospitals andlarge ambulatory care settings [4]. To increase engagement ofsmaller ambulatory practices in HIE, they must be presentedwith a clear value proposition for adoption, which may dif-fer from the value proposition for larger organizations [5–9].We therefore sought to elucidate the perspectives of clinicaland administrative leaders in smaller ambulatory practicesregarding desired HIE functions, key motivators for adoptingHIE, barriers to adoption, and potential incentives for adop-tion.

2. Methods

We used a case study approach, which allowed us to gain amore comprehensive understanding of the issues given thelack of prior research in this area [10]. Through a mixture oftelephone and on-site guided discussions we collected andanalyzed mixed quantitative and qualitative data to exploremotivators, barriers, and potential incentives in depth. Iter-ative cycles of data collection and analysis were conductedbetween November 2008 and April 2009. All methods wereapproved by the Colorado Multiple Institutional Review Board.Practices were provided $1000 to compensate for their partic-ipatory effort.

2.1. Site selection

Sites were purposively selected from the SNOCAP-USApractice-based research networks in Colorado. Nine small-to-medium sized (20 or fewer clinicians) primary care practicesagreed to participate, distributed among three groups:

• COMMUNITY-HIE: Currently engaged in community-wideHIE, defined as a system that consolidates and providesmore than one category of information (e.g., laboratory testresults, radiographic results, clinical notes) from more thanone independent organization. COMMUNITY-HIE practicescould use either paper or electronic medical records.

• PAPER-CHARTS-ONLY: Using only paper records and notengaged in community-wide HIE.

• EMR-ONLY: Using an electronic medical record (EMR) andnot engaged in community-wide HIE.

We originally recruited three sites in each of the threecategories, representing a variety of geographic settings andpractice types. However, one EMR-ONLY site revealed itself tobe engaged in COMMUNITY-HIE during the site visit and waschanged to the COMMUNITY-HIE category.

2.2. Data collection

2.2.1. Topic guide

The topic guide for in-person guided discussions was devel-oped based on a review of existing literature on the potentialvalue of health information technology and HIE [11–13]. It wasorganized into the following categories:

i n f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129

(1) Practice ownership and decision making.(2) Current use of charts and information technology.(3) Current electronic and non-electronic exchange of clinical

information.(4) Scenarios of community HIE functions that might be

offered, based on services offered by existing HIE projects[4,14–16]:(a) Information lookup,(b) Information delivery,(c) Electronic prescribing,(d) Order placement for tests and referrals,(e) Report generation,(f) Electronic communication with providers and/or

patients,(g) Electronic forms for home health agencies and oxygen

services,(h) Real-time eligibility checking for insurance coverage.

(5) Motivators, barriers, and incentives for adoption of com-munity HIE.

(6) Governance of community HIE.

In addition, appendices to the on-site topic guide askeddetailed questions about practice demographics (e.g. patientvolume, distribution of patient age and gender, insurance mix,and staff roles) and information systems used by the prac-tice (e.g. key software applications, functions, vendors, andversions).

2.2.2. Guided discussions and process mappingData collection began with brief telephone interviews. Thesewere followed by half-day site visits with guided discussionsconducted by two members of the research team (DF plus SRor LS) using the topic guide. Key informants always includedthe office manager and lead physician of the practice. We alsoconsulted other staff and clinicians to clarify details of clinicprocesses. Guided discussions were documented with bothhandwritten notes and audio recordings. Immediately aftereach site visit, the two team members conducting the guideddiscussions debriefed to create a field report of key findings.The appendices to the topic guide were generally completedoutside of the guided discussion process.

Process mapping of health information workflows werealso conducted during the site visits. For laboratory and radio-graphic tests, team members documented the methods ofordering tests, tracking and receiving results, presenting themto clinicians for review, acting on physician orders, and chart-ing. The workflows used to receive and review clinical notesfrom outside physicians and hospitals were also mapped, aswere workflows used to send clinical notes to these agencies.

The notes taken during guided discussions, audio record-ings, field reports, and process maps were later reviewed andconsolidated into case reports.

2.3. Analysis

Qualitative analysis proceeded through an iterative process

allowing for investigator corroboration, triangulation, andmember checking [17]. After initial review of the case reports,the team collaboratively developed a template of codes toorganize data for further analysis. For each of the three cat-

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129 125

Table 1 – Practice characteristics.

Practice number Setting Number of clinicians Type of practice Medicalrecordssystem

Current use of HIE

Group 1: PAPER-CHARTS-ONLY1 Suburban 5 Private Paper Hospital web portal2 Urban 7 FQHC for the homeless Paper None3 Suburban 7 Private Paper Hospital web portal

Group 2: EMR-ONLY4 Suburban 1 Private EMR None5 Rural 8 Private EMR None

Group 3: COMMUNITY-HIE6 Suburban 16 Private EMRa COMMUNITY-HIE using common

EMR system within anindependent practice association

7 Urban 9 Indigent clinic Paper COMMUNITY-HIE with limitedEMR functionality

8 Urban 11 Private Paper COMMUNITY-HIE with limitedEMR functionality

9 Rural 2 Private Limited EMRb COMMUNITY-HIE with limitedEMR functionality

enco

eCtrmtctwlt

faippvia

3

3

Tm

rNolts

a EMR functions were inextricably linked with HIE in this practice.b Use of EMR in this practice was limited to documentation of clinic

other EMR functions.

gories of practices (PAPER-CHARTS-ONLY, EMR-ONLY, andOMMUNITY-HIE), three or more members of the research

eam met to code and analyze the case reports. The teameached consensus on key themes and findings regarding the

otivators, facilitators, and potential incentives for the adop-ion of HIE by smaller primary care practices. Contrasts andommonalities among and across the three groups were iden-ified, as well as questions to be explored further. Findingsere consolidated into a briefing sheet summarizing the pre-

iminary key findings and identifying potential implications ofhese findings for policymakers.

An iterative, modified Delphi method was then employedor final refinement of the analysis [18]. Each practice receivedsummary of its own practice-specific findings and the brief-

ng sheet. In recorded guided discussions by telephone, eachractice’s office manager and lead physician checked theirractice-specific findings for accuracy and were encouraged toalidate, contradict, and elaborate on the content of the brief-ng sheet. Their refinements were incorporated in the finalnalysis presented here.

. Results

.1. Description of practices

he nine practices are listed in Table 1. All practices are familyedicine practices.Ultimately, two models of COMMUNITY-HIE were rep-

esented. Practices 7–9 were members of “Quality Healthetwork” (QHN), a traditional regional health information

rganization (RHIO) in Mesa County, Colorado. QHN provides

imited EMR functionality including storage and retrieval ofest results and dictated notes, electronic prescribing, andhared medication and allergy lists. QHN does not provide

unters. The practice used the HIE’s limited EMR functionality for all

functions to place orders, to create notes, or to generate claimsor bills. During its site visit, Practice 6 revealed itself to beengaged in a second, non-traditional model of COMMUNITY-HIE. This practice is a member of an independent practiceassociation (IPA) that provides a common EMR to its mem-bers. Although at the time of the interview health informationwas not exchanged across practices (each practice’s EMR wasdescribed as a “silo”), the common EMR system still met ourpredetermined definition of COMMUNITY-HIE because it con-solidated more than one type of information (laboratory andradiology results) from more than one independent organiza-tion (reference laboratories and a local network of hospitals).

3.2. Current methods of exchanging clinicalinformation

Each practice inventoried its current methods for exchangingclinical information: the types of information exchanged, theorganizations with whom it was exchanged, and the methodsused for exchange. Although methods of requesting, receiv-ing, reviewing, acting upon, and storing clinical informationvaried among—and even within—practices, several patternswere noted. First, the bulk of clinical information exchangerelated to ordering tests and studies and receiving the resultsfrom hospitals and independent laboratories. Each hospitaland laboratory used different forms and methods for trans-ferring information. Second, it was vital and common forpractices to exchange information with clinicians in hospi-tals and specialty practices. Information from these clinicians(e.g. consultation reports and discharge summaries) arrivedinconsistently; practices often actively needed to request this

information. Practice requests for information were handledexpeditiously for the small circle of organizations with whichthe practice had close relationships (typically one to threecommunity hospitals). Outside of this circle, however, the pro-

126 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l

Table 2 – Ranking of health information exchangefunctions.

Function Mean ranking from 1(highest rank) to 5(lowest rank)

Looking up information 1.9Delivering results 2.2

Electronic prescribing 2.5Placing orders 3.8Creating reports 4.7

cess of requesting information was particularly formal andtedious, requiring the practice to obtain and fax signed autho-rizations for information to be released.

3.3. Desired functions of HIE

Practices were asked to consider what they would desirein a HIE. Universally, practices valued improved ability toreceive and review clinical information from outside the prac-tice much more than improved ability to send or make availableclinical information from inside the practice. PAPER-CHARTS-ONLY and EMR-ONLY practices anticipated little value inmaking their own practices’ data available via HIE butCOMMUNITY-HIE practices did find HIE-supported access totheir own practices’ data to be useful in practice (enablinganywhere/anytime access as described below). In prioritiz-ing which participants should be included in HIE, there wasconsensus that community hospitals and independent labo-ratories would be essential. It would also be highly desirable toinclude specialists to whom patients were referred. In priori-tizing which data types should be included in HIE, test resultswere considered most important, followed by discharge sum-maries and clinic notes.

Because HIE was something many of the practices had notactively considered previously, it was useful to discuss poten-tial HIE functions separately. We asked practices to considerand rank five core functions (Table 2). Although the resultswere not unanimous, the functions fell into two groups. Eachof the top three functions received at least two #1 rankings,while the bottom two functions received no #1 or #2 rankingsand were ranked #4 or #5 by seven of the nine practices.

Looking up information was the most valued function overall.It was defined as an HIE allowing users to look up any clinicalinformation available via the HIE (test results, clinic notes, anddischarge summaries) on a patient-by-patient basis. (Whetherthis was to be accomplished by a central repository vs. a dis-tributed query was left unspecified.) All practices anticipated(or experienced) that this would greatly improve the accessi-bility of clinical information. While PAPER-CHARTS-ONLY andEMR-ONLY clinicians considered the lookup function to beuseful for reviewing data from other practices, COMMUNITY-HIE clinicians reported that the lookup function was alsouseful for reviewing data from their own practices. Any-where/anytime access to clinical information was especiallyuseful when outside the office (for instance, when admitting

patients to the hospital).

Delivering results was valued nearly as much as the abilityto look up information. It was defined as the HIE deliveringto the practice the results of the tests it ordered in a con-

i n f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129

sistent, consolidated manner, regardless of where the resultswere generated. Delivery could be by printer/fax, electronicinbox, or directly into an EMR system, at the discretion ofthe practice. PAPER-CHARTS-ONLY practices anticipated thatmoving from multiple streams of result delivery to a sin-gle, consolidated method would likely streamline workflowsrelated to review of test results. This would be particularlyhelpful in reducing the effort needed to train new or floatingstaff members. EMR-ONLY practices liked the idea of havinga single electronic interface delivering results to their EMR,rather than multiple electronic interfaces that might conflictwith each other. COMMUNITY-HIE practices thought they ben-efitted from result delivery and valued it highly, regardless ofwhether they used EMR or not.

Electronic prescribing also received favorable rankings. It wasdefined as the HIE delivering prescriptions to pharmacies,either by electronic faxes or point-to-point electronic delivery.The HIE could provide its own standalone electronic pre-scribing function or interface with EMR prescribing functions.Perceived benefits of electronic prescribing included fasterprocessing of prescription refills and the ability to retrieve pre-scription and fulfillment histories. However, workflow issueswere significant barriers to adoption of electronic prescrib-ing in some practices. Physicians in those practices withoutEMR were reluctant to begin using computers to write pre-scriptions, particularly since computers were not availablein many examination rooms. Conversely, a practice whichwas already fully satisfied with the e-prescribing functionswithin its EMR saw little incremental value in the HIE provid-ing e-prescribing services. While some practices were awareof incentives offered by the Center for Medicare and Medi-caid Services for use of electronic prescribing at the time ofthe interviews, no practice identified these incentives as acompelling motivation to adopt this function.

Placing non-prescription orders was considerably less valued.It was defined as the HIE allowing practices to electronicallyenter orders for laboratory tests, radiographic studies, andreferrals. Some value was seen in consolidating various pro-cesses for placing orders or making referrals. Although somehypothesized that electronic ordering might be a preconditionfor the more highly valued function of “delivering test results,”this was not the case in several practices already engaged inhealth information exchange. As with electronic prescribing,workflow issues and lack of computers in examination roomswere barriers to provider order entry in some practices.

Recognizing that practices commonly received and sentorders and other forms from home health and oxygen sup-ply companies, we also asked whether practices would likethe HIE to provide electronic routing of these forms. Practicesacknowledged they might see some efficiency gains, but onlyif the forms could be received, completed, signed, and routedcompletely electronically. Practices anticipated no incremen-tal value if the forms were received electronically but had tobe printed and signed in ink.

Creating reports was the least valued function. It wasdefined as the HIE creating standard quality reports for prac-

tices. Although practices recognized the value of qualityassessments, practices were generally skeptical about an HIEgenerating reports over which they had little control of themethods, content, or use. They were concerned that these

a l i n

rppiap

tepcmaabtp

3

Sfttooatftsm

3

AHtmabii

tictssupwi

3a

Ti

i n t e r n a t i o n a l j o u r n a l o f m e d i c

eports might be of questionable validity, reflect on themoorly, and adversely affect them in contracts with healthlans. Rather than having the HIE generate standard qual-

ty reports, practices preferred the idea of the HIE providingccess to more comprehensive data sources, from which theractices could create their own quality reports.

In addition to ranking these core HIE functions, prac-ices were also asked about the value of a supplementarylectronic messaging (secure e-mail) function. While severalractices were interested in secure messaging among physi-ians, it was considered to be a low priority until a criticalass of community practices committed to regularly checking

nd responding to electronic messages. There was consider-bly less enthusiasm for HIEs to offer electronic messagingetween patients and practices, as they were concerned abouthe time and effort that would be required to respond toatient messages.

.4. Essential attributes of HIE

everal general attributes were considered to be essentialor HIE adoption. Most important was for HIE networkso deliver solid reliability and responsive service. Prac-ices recognized—some through painful experiences of theirwn—that information system failures could create severeperational problems if not resolved quickly. When problemsrose, it was essential for practices to be able to access liveechnical support quickly and directly. It was also essentialor HIE networks to have comprehensive policies and systemso address privacy, security, and data use issues. Practices pre-umed that these would be well established by the time theyade decisions about adoption.

.5. Motivations for adopting HIE

ll practices reported that they would be motivated to adoptIE to gain more uniformity in workflows (making staff

raining easier), to improve efficiency (obtaining clinical infor-ation with less effort, especially anywhere/anytime access),

nd to improve the quality of the care they provided (throughetter coordination of care and better informed decision mak-

ng). All of these were thought to be of substantial and equalmportance.

Potential efficiency gains were strong motivators evenhough practices did not anticipate that these would resultn actual monetary benefit (i.e. through increased revenue orost savings). Practices did not think HIE would improve theurnaround time or yield of their current claims processingystems. Employee positions were unlikely to be eliminatedince employees often served multiple roles. Those practicessing paper charts planned to continue routing and filingaper copies of test results for clinicians, even if the resultsere accessible in the HIE system, so there would be no sav-

ngs in space or other costs to maintain paper charts.

.6. Barriers to and facilitators of

doption—technology and workflow

he primary barrier to HIE adoption was technical: the need tonterface HIE with existing practice management systems and

f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129 127

(where employed) electronic medical records. Workflow issueswere also significant barriers. While some practices were opento reengineering workflows with HIE adoption, most practicesinstead wanted HIE to complement their existing workflows.It is notable that Practice 9, a newer practice that was formedafter QHN was already operational, was able to reap unusuallylarge efficiencies (purchasing an inexpensive, lightweight EMRfor documentation and remaining paperless) by building itsworkflows and information systems to be integrated with theHIE services from the start.

These practices were not concerned that competingsources of sources of clinical information, such as hospital-specific web portals, would pose workflow barriers toadoption. They anticipated that would use hospital-specificweb portals and HIE lookup in different contexts, and that HIEwould still have significant incremental value.

Given the primacy of technical and workflow barriers,practices thought that the most powerful facilitators for HIEadoption would be technical assistance (for both implemen-tation and maintenance) and training.

3.7. Barriers to and facilitators of adoption—financialissues

Financial issues were secondary to technical and workflowissues but still important. Some practices identified capitalcosts, such as installing and supporting new computers andupgraded networking in the practice, as a significant bar-rier to HIE adoption. Practices were not, however, concernedabout loss of revenue. Unlike adoption of EMR, practices didnot anticipate that adoption of HIE would require temporarydecreases in the number of office visits (e.g. for training). Prac-tices also were not concerned about potential decreases inrevenue related to laboratory testing (i.e. through a reductionin duplicate testing), as might occur for other stakeholderssuch as hospitals.

Practices assessed potential financial incentives for adop-tion pragmatically, and thought they were less important thanongoing technical and training assistance. Incentives fromgovernment agencies or grantors to subsidize upgrades ofhardware and software were particularly desirable. Increasedreimbursement for practices that joined HIE networks werealso welcomed and were considered appropriate if healthplans reaped efficiency gains from practice use of HIE.Nonetheless, because practices anticipated that such reim-bursement would be comparatively modest, they did notanticipate that the promise of increased reimbursementwould play a significant role in their decisions to adopt HIE.

3.8. Barriers to and facilitators of adoption—solidarityand trust

Solidarity and trust were common themes in practice inter-views, and social networks appeared to play a strong role asfacilitators of adoption. Practices in Mesa County explicitlyreported that a regional sense of solidarity promoted devel-

opment of QHN. They thought that a spirit of cooperationarose out of necessity, as being west of the Continental Divideisolated them from the resources of the major metropolitanColorado Front Range area. Their sense of community was

i c a l

128 i n t e r n a t i o n a l j o u r n a l o f m e d

reinforced in regular peer encounters, both professional (asmany practitioners met while rounding in the hospital) andsocial (frequent casual contacts common in a smaller city).In contrast, practices in metropolitan Denver reported a lackof regional solidarity and multiple barriers to cooperation,including strong competition with other practices, more con-flict with health plans, and few encounters with peers in thecommunity (especially as many clinicians no longer madehospital rounds). Notably, the metro-Denver COMMUNITY-HIErepresented by Practice 6 arose from practices already cooper-ating under an existing IPA.

Trust issues were reflected in preferences for HIE gov-ernance. Practices strongly wanted smaller ambulatorypractices to be represented equitably in governance, andthought that doctors with leadership positions in IPAs or asso-ciations of community health centers could represent theirinterests well. They wanted HIE networks to include leaderswho are clinically experienced, IT savvy, and well-respectedwithin the healthcare community. Practices also wanted com-munity hospitals to be represented. In contrast, practicesthought that involvement of health plans in HIE governancecould be a barrier to their participation. Many practices hadconflicts with health plans in the past and were concernedthat health plans might use data from the HIE to affect thepractice’s contracting adversely. If health plans sponsored HIEefforts, practices were willing to consider their representationin governance as long as data use agreements were tightlycontrolled. Attitudes towards involvement of nonprofit foun-dations, local government, and professional societies in HIEgovernance were inconsistent and largely neutral.

Practices were not concerned that patient mistrust of theHIE would be a barrier to their adoption. While some patientsmight have an initial negative reaction to having their clinicaldata exchanged, practices thought that patient attitudes couldbe managed by explaining how HIE could improve their care.COMMUNITY-HIE practices noted that their patients did notexpress concerns about the HIE, but instead appreciated andeven came to expect the higher service levels that HIE enabled(e.g. immediate transmission of prescriptions to pharmaciesand quick access to clinical results during visits).

4. Discussion

The smaller primary care practices studied varied in theirinformation technology resources and their experiences withHIE but had similar perspectives on the value HIE would pro-vide and the functionality they desired. These practices weremotivated to join HIE to improve the quality, coordination, andefficiency of the care they offer. Technical and workflow issuespresented the greatest barriers to adoption, and financialissues were also important for some practices. The most valu-able facilitators for adoption would be technical assistanceand training. Financial incentives were less powerful facili-tators. Existing social and professional communities of trustfacilitated the development of health information exchange

networks. These results are generally consistent with otherreports on HIE adoption, although some issues of relevancefor other stakeholders and contexts (e.g. loss of revenue froma reduction in redundant testing [19] and adverse compe-

i n f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129

tition with web portals [20]) were not important for thesepractices.

In framing our questions, we adopted a permissiveapproach, allowing practices to express a preference for theHIE to provide unidirectional flow of information into thepractice. We did not insist that practices consider only fullybidirectional flow of information, which would likely requireEMR adoption. Our rationale was that successfully engaging inlimited HIE without practice redesign may be a path to morecomprehensive HIE engagement in the future. Accordingly, wefound that HIE functions requiring less change in pre-existingpractice workflows (looking up information and deliveringresults) were more highly valued than those requiring work-flow changes (ordering prescriptions or tests electronically),particularly if those functions accommodated paper processes(e.g. through interfaces that facilitate printing and faxing ofresults). Given the high value of these functions, the participa-tion of hospitals and independent laboratories, the dominantproviders of test results, may provide a critical mass of partic-ipants necessary to launch a HIE network.

The primary limitation of our approach is a relatively smallsample that may limit the generalizability of our findings. Themodels of community HIE in the COMMUNITY-HIE practicesare not representative of the entire spectrum of communityHIE employed nationwide. Other limitations include selectionbias, as practices willing to participate in this study may differfrom nonparticipating practices. Also, by focusing on guideddiscussions with lead clinicians and office managers we mayhave overlooked important considerations that other clinicaland non-clinical staff could have provided. These limitationswere inherent in a case study process emphasizing depth ofdata collection and analysis rather than breadth.

These results may provide guidance for policymakers inter-ested in increasing HIE adoption in small-to-medium sizedpractices. Beyond providing technical, training, and financialassistance in HIE adoption, several other efforts may be help-ful. Since more efficient and higher quality care appears tobe a primary motivation for adopting HIE, developing andpromoting formal and informal evidence of these benefits islikely to promote adoption. Since reliability and security areessential attributes of HIE, HIE networks would benefit fromdevelopment of best practice guidelines for secure and reliableservice and performance and for transparent and enforce-able data use agreements and privacy protections. Efforts toencourage adoption of HIE in small-to-medium sized ambu-latory practices might be most successfully targeted not toarbitrary geographic regions, but instead to communities oftrust defined by existing business, professional, and socialrelationships, from which HIE adoption could diffuse morebroadly.

5. Authors’ contributions

Dr. Ross took primary responsibility for developing the studyprotocol, which was further refined in consultation with the

remaining authors (Dr. Schilling, Mr. Fernald, Dr. Davidson,and Dr. West). Data was collected by Mr. Fernald, generally inconjunction with Dr. Ross and Dr. Schilling. Primary analysis ofthe data was conducted as a team with Dr. Ross, Dr. Schilling,

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n

Summary pointsWhat was already known on the topic:

• By facilitating the flow of health information, HIE mayimprove coordination of care as well as the quality andefficiency of care.

• Adoption of HIE by small-to-medium sized primarycare practices, where the majority of ambulatoryencounters take place, is likely to be an essential ele-ment of a health information infrastructure.

• Sustainable adoption of HIE in a community requires athorough understanding of its costs and value propo-sitions for key stakeholders.

What this study added to our knowledge:

• HIE adoption may be promoted by offering functionssuch as consolidated test result delivery, test resultlookup, and electronic prescribing.

• Primary care practices look forward to HIE improvingthe quality and efficiency of the care they provide, butthey do not anticipate any financial return on invest-ment. While monetary incentives would be welcome,technical assistance and support is likely to be a morepowerful motivator for adoption.

• Practices already engaged in a network of professionaland social ties are more ready to adopt HIE. Distrust ofother potential stakeholders (such as health plans) in

Mapset

A

TRBTTam

r

7.7.09).

governance is a significant barrier to adoption.

r. Fernald, and Dr. West. During the data collection andnalysis phases, Dr. Davidson reviewed interim reports androvided additional insights. Dr. Ross took primary respon-ibility for writing the manuscript, which was reviewed anddited by all members of the team. All authors have approvedhe final manuscript.

cknowledgements

his study was funded by the Agency for Healthcareesearch and Quality (AHRQ) under Primary Care Practice-ased Research Networks Contract # HHSA290200710008.he authors wish to thank Rebecca Roper MS MPH, theask Order Leader on the contract, for her assistance anddvice throughout the study and her thoughtful review of theanuscript.

e f e r e n c e s

[1] S.L. Penfield, K.M. Anderson, M. Edmunds, M. Belanger,

Toward Health Information Liquidity: Realization of Better,More Efficient Care from the Free Flow of HealthInformation, Booz Allen Hamilton, Rockville, MD, 2009.

[2] J. Walker, E. Pan, D. Johnston, J. Adler-Milstein, D.W. Bates, B.Middleton, The value of health care information exchange

f o r m a t i c s 7 9 ( 2 0 1 0 ) 123–129 129

and interoperability, Health Aff. (Millwood) (Suppl. WebExclusives) (2005), W5-10–W5-18.

[3] Agency for Healthcare Research and Quality NationalResource Center, Available from http://healthit.ahrq.hgov(Accessed 30.4.09).

[4] eHealth Initiative, Fifth Annual Survey of HealthInformation Exchange at the State and Local Levels, eHealthInitiative, Washington, September 11, 2008.

[5] J.M. Grossman, K.L. Kushner, E.A. November, Creatingsustainable local health information exchanges: can barriersto stakeholder participation be overcome? Res. Briefs 2(February) (2008) 1–12.

[6] J.M. Marchibroda, Health information exchange policy andevaluation, J. Biomed. Inform. 40 (6 Suppl.) (2007) S11–S16.

[7] R.H. Miller, B.S. Miller, The Santa Barbara County Care DataExchange: what happened? Health Aff. (Millwood) 26 (5)(2007) w568–w580.

[8] J. Adler-Milstein, A.P. McAfee, D.W. Bates, A.K. Jha, The stateof regional health information organizations: currentactivities and financing, Health Aff. (Millwood) 27 (1) (2008)w60–w69.

[9] S.R. Simon, R. Kaushal, P.D. Cleary, C.A. Jenter, L.A. Volk, E.G.Poon, D.H. Williams, E.J. Orav, D.W. Bates, Correlates ofelectronic health record adoption in office practices: astatewide survey, in: AMIA Annu. Symp. Proc., 2006, p. 1098.

[10] R.K. Yin, Case Study Research: Design and Methods, SagePublications, Thousand Oaks, CA, 2002.

[11] J.M. Grossman, A. Gerland, M.C. Reed, C. Fahlman,Physicians’ experiences using commercial e-prescribingsystems, Health Aff. (Millwood) 26 (3) (2007) w393–w404.

[12] R.H. Miller, C. West, T.M. Brown, I. Sim, C. Ganchoff, Thevalue of electronic health records in solo or small grouppractices, Health Aff. (Millwood). 24 (5) (2005) 1127–1137.

[13] R.H. Miller, I. Sim, Physicians’ use of electronic medicalrecords: barriers and solutions, Health Aff. (Millwood) 23 (2)(2004) 116–126.

[14] C.J. McDonald, J.M. Overhage, M. Barnes, G. Schadow, L.Blevins, P.R. Dexter, B. Mamlin, The Indiana network forpatient care: a working local health informationinfrastructure. An example of a working infrastructurecollaboration that links data from five health systems andhundreds of millions of entries, Health Aff (Millwood) 24(September–October (5)) (2005) 1214–1220.

[15] J. Sutherland, Regional Health Information Organization(RHIO): Opportunities and Risks, PatientKeeper, Inc.,Newton, MA, November 2005, Available from http://www.himss.org/Content/Files/Sutherland RHIO WhitePaper.pdf(Accessed 30.4.09).

[16] J. Halamka, M. Aranow, C. Ascenzo, D. Bates, G. Debor, J.Glaser, A. Goroll, J. Stowe, M. Tripathi, G. Vineyard, Healthcare IT collaboration in Massachusetts: the experience ofcreating regional connectivity, J. Am. Med. Inform. Assoc. 12(November–December (6)) (2005) 596–601.

[17] B.F. Crabtree, W.L. Miller, Doing Qualitative Research, SagePublications, Thousand Oaks, CA, 1999.

[18] H.A. Linstone, M. Turoff (Eds.), The Delphi Method:Techniques and Applications, 2002, Available fromhttp://is.njit.edu/pubs/delphibook/ (Accessed 7.7.09).

[19] L. Burchell, Best Practices for Community HealthInformation Exchange, Center for Community HealthLeadership, 2008, Available fromhttp://www.allscriptscenter.com/NR/rdonlyres/6B8E9E8A-93BD-467D-A3BB-52B0E4DC6107/0/CCHL BPG.pdf (Accessed

[20] J.M. Grossman, T.S. Bodenheimer, K. McKenzie,Hospital–physician portals: the role of competition indriving clinical data exchange, Health Aff. (Millwood) 25 (6)(2006) 1629–1636.