Influence of information sources on the adoption of uterine fibroid embolization by interventional radiologists

Abstract

Objectives: The purpose of the research was to (1) understand the influence of information sources on the awareness and adoption of uterine fibroid embolization (UFE) by interventional radiologists in Michigan and (2) to decipher communication relations in the social network of interventional radiologists that were most conducive to the flow of information about UFE.

Methods: Diffusion of innovations theory and constructs in social network analysis formed the basis for the development of an interview guide. Thirty-two interventional radiologists in Michigan were interviewed over the phone. Chi-square statistics were employed to analyze the awareness and adoption of UFE. Factor analysis was applied to decipher important communication relations in the social network of interventional radiologists.

Results: Conferences were found to be an initial source of information, creating awareness among early adopters (P < 0.05), but other individuals were found to be influential sources in the adoption of UFE by later adopters (P < 0.05). Radiologists rarely browsed Websites for information. Work relations in everyday clinical practice were the communication relations most conducive to the flow of information about UFE. Preliminary qualitative data indicated that opinion leaders in the diffusion of UFE in Michigan were located in hospitals primarily dedicated to practice rather than in hospitals affiliated with universities.

Conclusions: Journals are important information sources for creating awareness and stimulating adoption of innovation among both early and late adopters of new procedures in interventional radiology. Conferences, however, are significantly more important for creating early awareness, while interactions with colleagues is the most important factor in stimulating use of the innovation among later adopters. Among colleagues, opinion leaders in nonacademic hospitals may be more influential than individuals in the academic community.

INTRODUCTION

Change is a rapid and expected phenomenon in medicine. Over the years, new drugs, new therapies, and new procedures have been developed. Although the medical profession welcomes change, individual physicians face considerable challenges when a medical innovation is introduced.

After receiving information about an innovation through various channels, a physician must undergo the problematic process of evaluating the innovation. In addition, today's patients are educated consumers who place high demands on physicians [1]. Thus, after hearing about the innovation, the physician, in order to avoid learning about it through a negative experience, must devote considerable thought to the characteristics and the efficacy of the innovation, as well as any complications or side effects [2].

There are some physicians who tolerate uncertainties and take risks early in the process of the spread of the innovation in their communities and decide to accept it earlier than others. There are other physicians who wait until the new idea has become a standard; as a result, it takes time for the innovation to permeate the entire community. To understand such a process, and especially to study the influence of information on the process, the current study examines the effect of information sources on the adoption of a milestone development in radiology, a medical procedure termed uterine fibroid embolization (UFE), among interventional radiologists in Michigan.

INFORMATION SOURCES

Existing studies conducted under the rubric of physicians' information needs provide insights into how physicians, in general, use information in their work. The investigations indicate that physicians seek information mainly for two purposes—either to care for their patients or to learn about new developments [3, 4]. The focus in research, however, has been the use of information sources by physicians to answer questions arising in the context of caring for their patients. It has been found that the information sources physicians commonly use to find the answers are journals, textbooks, drug reference manuals, drug texts, conferences, consultants, colleagues, and pharmaceutical representatives [5–8]. These sources are familiar, relevant, accessible, and easy to use [9–11].

The frequency with which specific sources are used varies among physicians in different disciplines: family medicine, internal medicine, and the specialties. Haug, in 1997, conducted a meta-analytic study of information sources used by physicians to synthesize results from various studies and concluded that the top three sources physicians use in clinical practice are journals, books, and colleagues [12].

The literature indicates that physicians, in fact, rely heavily on human sources for certain types of information—advice and evaluation of information found in standard sources [13–15]. Knowledgeable colleagues, in particular, are influential sources of this type of information [16, 17]. Gorman explains that “For some questions, purely descriptive medical knowledge may be sufficient, such as that found in medical textbooks and journal literature” [18]. He continues: “In other situations, however, the practitioner may require higher-order information, such as confirmation, explanation, analysis, synthesis, and ultimately judgment […]” [19] to make sense of the codified factual information. In effect, informed and experienced colleagues provide the higher-order information to other colleagues by translating the codified information for practical use.

TYPES OF INFORMATION SOUGHT

Gorman identifies five types of information sought by physicians: (1) patient data, which refers to information about individual patients; (2) population statistics, which are related to information about groups of patients; (3) medical knowledge, a more general type of knowledge that can be applied to many patients; (4) logistic information, that is, information about policies and procedures for managing care; and (5) social influences, which is knowledge of how other local area physicians perform their jobs. [20].

A different classification by Forsythe, Buchanan, Osheroff, and Miller [21] of information sought by physicians is somewhat abstract compared to Gorman's [22]. Gorman's [23] categories, however, can be mapped to the classifications developed by Forsythe and colleagues [24].

According to Forsythe and colleagues, information can be classified as (1) formal and general, (2) formal and specific, (3) informal and general, or (4) informal and specific [25]. Codified medical knowledge such as that found in textbooks is considered formal and general information. Information in medical records, for example, is also formal, but it is specific because it relates to individual patients. Undocumented side effects of a drug constitute informal and general information because this is a type of general information not found in standard sources. Undocumented information about a particular patient or a particular practitioner, for example, is considered informal but specific knowledge, as this is a type of specific information not found in standard sources.

After describing their classification, Forsythe and others conclude that the most important information in clinical practice is the informal and specific [26]. Also, informal information, both general and specific, tends to be conveyed to others through human sources. A physician could also acquire this information through experience. As noted earlier, physicians often rely on colleagues as sources, in that colleagues translate formal information found in standard sources into informal information acquired through experience and transfer it to others in the community.

Innovations, however, tend to be born in research institutions, such as universities. With their resources and a culture based on discourse, these organizations provide a prime locale for innovation. From this hub, the innovation then flows to other members of a community. To examine the diffusion, and especially the impact of information sources on the diffusion, of uterine fibroid embolization—a milestone development in the radiologist community—the current study employs two theoretical constructs: diffusion of innovations theory and social network analysis.

THE INNOVATION: UFE

Uterine fibroid embolization (UFE) is a relatively new procedure for treating uterine fibroid disease. Uterine fibroids are tumors that occur in women of reproductive age, with the prevalence of these tumors increasing with age. The traditional treatment for symptomatic uterine fibroids has been hysterectomy—a major surgical procedure performed by obstetrician/gynecologists. In hysterectomy, the uterus is entirely removed surgically, and the disease is completely cured. However, the removal of the uterus also means infertility in the women for life [27].

As a result, uterine sparing treatments have become preferred alternatives for women with the disease. One such procedure is uterine fibroid embolization—a minimally invasive uterine-sparing procedure performed by interventional radiologists. UFE is not a completely new development in interventional radiology. It is a new application of embolization of uterine arteries, which is a technique of injecting embolic particles through catheters to block blood flow. Interventional radiologists have been performing uterine artery embolization since the 1970s to control bleeding in various gynecologic conditions [28]. The new application, however, in performing UFE is the embolization of the arteries specifically to treat uterine fibroids.

Ravina and others in France first published the success of UFE in 1995 [29]. Ravina was using embolization as an adjunct procedure to control bleeding before hysterectomy and myomectomy—another surgical but uterine-sparing procedure for treating fibroids—and noticed that the embolization itself was having an effect by shrinking the fibroids. Since 1995, UFE has been rapidly integrated into practice by interventional radiologists. The major benefits of UFE for women with fibroid disease, in general, are avoiding surgical risks and shorter hospitalization; for women of childbearing age, there is also the potential for preserving fertility [30].

In addition to UFE and myomectomy, other viable uterine-sparing alternatives to hysterectomy exist: myolysis, endometrial ablation, and drug therapy. With the exception of drug therapy, which is suitable only for mild cases, the other procedures are surgical [31] and are suitable for patients with different uterine fibroid conditions [32]. In addition, each procedure presents its advantages and limitations [33]. For instance, in 1998, Ravina et al., and others reported preliminary results that indicated myomectomy, a highly invasive treatment with an adequate post-treatment fertility rate, might have higher recurrence rates than UFE [34].

Thus far, the reports of clinical trials conducted to study UFE indicate that it is safe and effective in shrinking uterine fibroids and in treating the symptoms of the disease in select patients [35–39]. Its effectiveness in preserving fertility, as mentioned above, has not yet been established. Physicians still prefer myomectomy, in fact, for women who wish to become pregnant [40, 41]. However, the potential for fertility after UFE is present as the uterus is preserved during the procedure. Long-term results of the procedure in patients, however, remain to be seen.

THE INNOVATION-DECISION PROCESS

The process by which physicians tolerate these uncertainties and incorporate a new medical procedure such as UFE into continued medical practice could be explained by the diffusion of innovations theory. The theory purports that individuals move through stages of the innovation-decision process over time before incorporating an innovation [42]. As Rogers notes, “The innovation-decision process is essentially an information-seeking and information-processing activity in which the individual is motivated to reduce uncertainty about the advantages and disadvantages of an innovation” [43].

The five stages of the innovation-decision process are (1) knowledge, (2) persuasion, (3) decision, (4) implementation, and (5) confirmation. In the first stage of the innovation-decision process, knowledge, the individual becomes aware of the innovation. The information channels or sources through which this knowledge comes might not have been actively sought. However, individuals are often predisposed to certain information sources and are likely to find out about new developments through these channels.

After knowledge of the innovation, the individual actively seeks evaluative information about the innovation. Based on this information, in the next stage, persuasion, the individual forms either a favorable or an unfavorable attitude toward the innovation. If the evaluative information is not received, the individual might not move further in the process. However, once information is received and the attitude is formed, continued efforts to evaluate the attributes of the innovation lead to the third stage—decision. Here the individual reaches a decision to either adopt or reject the innovation. In the last two stages, implementation and confirmation, the individual puts the innovation to full use and looks for reinforcement of the decision made earlier in the process to adopt the innovation. At this point, the individual might choose to reverse the original decision if unsatisfactory messages are received.

According to the theory, different types of information sources or channels are effective at the various stages of the innovation-decision process. Mass-media channels are relatively more important in creating awareness at the knowledge stage, while interpersonal channels are relatively more important in persuading and influencing decision at the later stages of the innovation-decision process. This means that individuals tend to become aware through mass media channels but will wait to reach a decision until they have heard about experiences with the innovation through interpersonal sources. Supporting this view, in a classic study of adoption of a medical drug by physicians, Coleman and others found that drug companies were important sources of information in hearing about the drug; at later stages of the process, however, physicians sought information from colleagues [44].

In the beginning of the diffusion of an innovation through the system itself, some early adopters would not have the interpersonal sources to rely on to reach a decision because there are as yet no adopters in the system. However, these venturesome early adopters do not require affirmation from interpersonal sources, and after hearing about the innovation through mass-media sources, they pass through the other stages. Meanwhile, the later adopters, who are ambivalent, wait and look to the earlier adopters for opinions based on experiences with the new procedure.

It appears that the importance of information sources at various stages of the innovation-decision process is somewhat different for early adopters and later adopters. Mass-media sources seem to be relatively more important than interpersonal channels for early adopters than for later adopters.

Guided by these constructs in the theory of diffusion of innovations, the research questions in this study are formulated as follows:

1. Is there a relationship between the use of information sources—mass versus interpersonal—and early or late awareness of UFE among radiologists?

2. Is there a relationship between the use of information sources—mass versus interpersonal—and early or late adoption of UFE among radiologists?

SOCIAL NETWORK ANALYSIS

Physicians frequently talk to each other, seeking information and advice from their peers. Experienced colleagues provide the opinion and the evaluation and validation of codified information found in formal sources [45]. To probe a communication structure such as this among interventional radiologists, this study employs social network analysis techniques. This analysis of interpersonal communication is meant to complement the analysis (based on research questions one and two) of the relative significance of mass media versus interpersonal information sources in the adoption of UFE.

Social network analysis is the study of communication relations among individuals in a social network [46]. Individuals in the network are viewed as nodes that are linked by communication relations. When information is exchanged over these communication relations, these relations become information exchange relations [47, 48]. The type of information exchanged defines the nature of the communication relation among individuals. For example, when two interventional radiologists discuss patient education, the information-exchange relation is “discussing patient education,” based on the type of information transferred, which is patient education information.

By observing the types of information exchanged among individuals and the frequency of communication, one can identify patterns of communication relations in the network—cliques consisting of individuals who communicate frequently about certain types of information [49, 50]. The premise here is that the information exchange relations in the prominent cliques in the social network of interventional radiologists would have been the conduits for the flow of information about UFE, and these same relations would also be conducive to the flow of information about other new developments.

Research questions one and two, stated earlier, are based on the theory of diffusion of innovation. The complementary research questions three and four are based on social network analysis and are formulated as follows:

• 3. Are certain types of relations in the social network of radiologists more conducive to the flow of information about UFE?

• 4. Are there opinion leaders among radiologists who were influential in the diffusion of UFE?

METHODOLOGY

A sample of 32 interventional radiologists in Michigan was interviewed. Radiologists were first randomly selected from the online physician directory of the Society of Cardiovascular and Interventional Radiology (SCVIR). These radiologists were contacted by telephone to set up interview appointments. If the receptionist was reached, messages were left for other interventional radiologists also working in the same hospital or clinic. Some of these radiologists were not listed in the SCVIR directory. Of the radiologists contacted, 52% responded during the study period. All interviews were conducted over the telephone. An interview guide designed for this study was used to conduct the interviews. Data collection began during August 2002 and continued through December 2002.

Interview guide.

As conveyed in the literature review, physicians use certain information sources to conduct their work. These information sources formed the basis for constructing a list of information sources that interventional radiologists might use in the process of adopting a new procedure. Questions (1–12b) in the interview guide (see Appendix) probed the use of information sources from this list at three stages of the innovation-decision process: knowledge, persuasion, and decision. Chi-square statistics were used to analyze the influence of different mass communication and interpersonal information sources on awareness of UFE and adoption of UFE (research questions 1 and 2). Additional information related to the adoption of UFE sought during the interview includes

• The relative advantage of UFE over other procedures for treating uterine fibroids, according to interventional radiologists (question 14),

• Whether interventional radiologists required assistance in performing the procedure for the first time once they decided to integrate it into their practice (question 13),

• Whether interventional radiologists extensively recommended UFE to other physicians after they had adopted it (questions 15a–17),

• The reasons for non-adoption of UFE, if this were the case (question 20).

As a first step in probing the social network of interventional radiologists and identifying cliques/groups that were conducive to the flow of information about UFE, a set of possible information exchange relations in the social network of interventional radiologists was developed. Information exchange relations are relations between individuals through which information is exchanged. The various types of information sought by physicians, discussed in the literature review section, were employed to guide the development of the set of possible information exchange relations in the social network of interventional radiologists for the current study.

Each participating interventional radiologist was first asked in the interview (question 18) to provide job titles and specializations of three individuals with whom he/she had discussed UFE. The participant was asked to include individuals with whom he/she conversed most frequently. Then, the communication frequency—daily, weekly, monthly, or yearly—in each possible information-exchange relation, with each of the three individuals with whom the participant had discussed UFE, was recorded in the interview guide (question 19).

Factor analysis was employed to analyze the communication frequency data and to draw the significant groupings of information exchange relations, in other words, major dimensions of interpersonal relationships in the social network of interventional radiologists that are conducive to the flow of information about UFE (research question 3). The factors, or groupings of information exchange relations resulting from factor analysis, also represent cliques, as the relations in a grouping represent communication, with similar patterns among certain interventional radiologists and certain other individuals in their social network.

An important premise here is related to the way information exchange relations are set up. These information exchange relations are not regarding UFE. Instead, these are general information exchange relations, which are possible in the social network of interventional radiologists. However, the communication frequency recorded with respect to each information exchange relation was with individuals with whom the interventional radiologist had discussed UFE. The expectation was that information exchange relations in the prominent factors, or cliques resulting from factor analysis, would be the relations that would be good for the flow of general workplace information. However, since the general discussions were with individuals with whom the interventional radiologist had discussed UFE at some point, it is safe to assume that in addition to being good for carrying general workplace information, these information exchange relations would also be conducive to the flow of information about UFE.

RESULTS

The 32 interventional radiologists participating in the study were somewhat similar regarding two demographic characteristics—gender and age. All participants were male, and 59% fell into the 36–45 age group. Table 1 provides the frequencies in the other three age groups. All interventional radiologists were affiliated with at least one hospital. Some were affiliated with additional hospitals and clinics.

Table 1 Demographics

Geographically, participants were located in different regions of Michigan. Adopters, however, were found to be concentrated in cities—Troy, Detroit, Ann Arbor, and Dearborn—and several smaller towns in southeast Michigan. Adopters were also found in cities and smaller towns in most of the other regions. There appeared to be a lack of the treatment being offered in central Michigan. Two participants, both adopters of UFE, working in other regions were also affiliated with clinics in smaller towns in central Michigan. However, it is not clear whether these adopters perform UFE in these locations. Based on data gathered, it appears that UFE is not being offered in central Michigan.

Awareness and adoption.

Table 2 displays the number of interventional radiologists in Michigan who first heard about UFE early on and the numbers for interventional radiologists who heard about it in the years after. Table 2 also provides the numbers of interventional radiologists who adopted UFE in different years after hearing about it, as the procedure was diffused through Michigan. The categorization into early knowers, late knowers, later knowers, early adopters, late adopters, and later adopters (Table 2) is based on the spread of the frequencies across the spectrum.

Table 2 Awareness and adoption of uterine fibroid embolization (UFE) by years

As can be seen, the largest number of interventional radiologists became aware of UFE in 1998, and the largest number adopted in 1999, a year after. Participants who became aware of UFE in 1998 and participants who adopted UFE in 1999 were categorized as late knowers and late adopters, respectively. Knowers and adopters before 1998 and 1999, respectively, were categorized as early, and knowers and adopters in years after 1998 and 1999, respectively, were categorized as later.

A significant relationship was found between information obtained at conferences and early awareness (Table 3). Early knowers were more likely to have found out about UFE at conferences. In addition to the traditional national and local professional conferences, meetings, such as local Angio Club meetings or fellowship review meetings, were also categorized as conferences. Radiologists cited journals as information sources creating awareness of UFE at all three times of awareness—early, late, and later. There was no significance in its impact on interventional radiologists' becoming aware at a particular time. The other mass media information sources—Websites, medical-device company literature, books, newspapers, and newsletters—rarely played a role in informing interventional radiologists about UFE.

Table 3 Information sources creating awareness of UFE

It was surprising, given the prevalence of the Internet, to find that interventional radiologists in Michigan rarely browsed the Web for information. When they did visit a Website, it was that of their professional society, Society of Cardiovascular and Interventional Radiology (SCVIR). Books, obviously, would contain information about new developments such as UFE later on, but would not serve as an important source in creating awareness. Indeed, books were not found to be sources of information of any significance in informing interventional radiologists about UFE.

Only 11% of the radiologists found out about UFE early on through an individual, for example, a colleague. The reason for the low number is that at the time there were few or no knowers/adopters of UFE in Michigan. Those interventional radiologists who found out about UFE through an individual had contacts with colleagues in hospitals in other states. As time passed, there were interventional radiologists who knew about UFE in Michigan, and others were finding out about UFE from them. Hence, a higher number of interventional radiologists (Table 3), at late awareness, had found out about UFE through individuals who were most often located in hospitals within the state.

Information obtained through an individual was influential, however, in the later adoption of UFE by interventional radiologists. A significant relationship was found between later adoption of UFE and the interventional radiologist's receiving information about UFE through an individual, such as a colleague (Table 4). Early adopters of UFE moved quickly to adoption after receiving information through mass media information sources alone, particularly journals. Since there were few or no adopters of UFE in Michigan at that time, there was little chance that they would hear about colleagues' experiences with the procedure. Thus, it was likely that their venturesome natures allowed these early adopters to bear the uncertainties involved in experimenting with a new medical procedure and adopting it.

Table 4 Influence of information sources on the adoption of UFE

Later adopters, however, waited to hear about their colleagues' experiences with the procedure. As the early adopters built knowledge based on personal experience, they shared their informal experiences with other interventional radiologists in Michigan. The earlier adopters were also spreading the information to primary care physicians—family practitioners, obstetrician/gynecologists, internists—and trainees within their hospitals through informal conversations as well as to the masses through in-house patient newsletters.

For later adopters of UFE, the informal information, especially, communicated through conversation with early adopters of UFE, appears to have been important. In addition to other interventional radiologists, whether they were colleagues within the hospital or colleagues practicing at another hospital, the later adopters also found other types of individuals as influential in their decision making regarding adoption of UFE. The patients who were contacting the interventional radiologists and inquiring about the procedure provided a major impetus in the interventional radiologist's decision to perform the procedure. Medical-device company representatives, obstetrician/gynecologists, interventional radiology trainees from other hospitals, and visiting professors from universities were also identified, but infrequently. Many later adopters and 35% of all adopters also had a colleague assist them when they performed the procedure for the first time.

Mass media information sources such as journals alone were not sufficient in convincing later adopters. Most interventional radiologists read the important journals in their discipline. As seen in Table 4, adopters at different times—early, late, and later—used journals as an information source. Conferences appear to have played an insignificant role in adoption (Table 4). Conferences were important information sources in creating awareness but were not an influential information source in adoption. Again, Websites, medical-device company literature, and books rarely played a role in the adoption of UFE, just as they did not play a role in creating awareness of UFE.

Overall, interventional radiologists who were adopters of UFE, when asked to list and rank information sources important in learning about new developments in their discipline in general, most frequently ranked conferences first, journals second, and individuals third (Table 5). They found conferences to be a forum for exchange of up-to-date information with a trusted body of peers. Journals were easy to access and a credible source that provided details such as description of a procedure and results—safety and efficacy—of a procedure. Colleagues were trusted sources for opinions based on personal experience and thus provided direct one-on-one conversation as a guide for positive/negative experiences to avoid duplication of mistakes.

Table 5 Frequency of information sources ranked 1, 2, and 3 in learning about new developments in interventional radiology

Communication network of radiologists.

Table 6 displays the results of factor analysis performed to find cliques that were conducive to the flow of information about UFE. In addition, it presents a description of network members involved and the interactions between them.

Table 6 Communication network of interventional radiologists: patterns of communication with individuals with whom interventional radiologists discussed UFE

Analysis of the communication frequency between interventional radiologists and others in their social network produced three prominent factors. Each factor includes information exchange relations loading on the factor with, at least, a correlation value of 0.70. These are the information exchange relations, possible in the social network of interventional radiologists for which the communication frequency was recorded.

The first factor, clinical practice, includes the highest number of information exchange relations loading with the highest correlation values. Information exchange relations loading on clinical practice are parts of an interventional radiologist's everyday medical practice (Table 6). All interventional radiologists are involved in clinical practice, some more than others. Full-time clinicians spend more time in medical practice than clinician/researchers. It makes sense that information exchange relations within the context of clinical practice would be important conduits for transferring general information about work. Conducting research, the second factor, includes information exchange relations related to the conduct of research. Leisure, the third important factor, includes information exchange relations related to social interaction.

All the three factors—clinical practice, conduct of research, and leisure—and the information exchange relations that constitute each of the three factors appear to have been conducive to the flow of information about UFE among interventional radiologists. However, of the three factors, the most prominent factor, and hence the factor with the information exchange relations most conducive to the flow of information about UFE, appears to have been clinical practice. Again, the information exchange relations for which the communication frequency was recorded were relations possible, generally, in the social network of interventional radiologists. These relations were not specific to discussions about UFE. The frequencies and the resulting factor loadings are based on general information exchange relations. However, as these general information exchange relations were with others in the social network with whom the interventional radiologists had discussed UFE, the premise is that some of these discussions were conducive to the transfer of information about UFE.

Each of the three prominent factors is also a clique because it includes interactions between certain interventional radiologists and others in their social network. The top section of Table 6 presents description of pairs in the social network of interventional radiologists. The representation of pairs is not an exact description of pairs within each clique; rather, it is a description of pairs within the three time periods of adoption. Some of these pairs make up each clique. As such, the description of pairs at different time periods of adoption gives some idea as to the constitution of pairing within each clique.

To obtain this data, each participant was asked to provide job titles and specializations of three individuals with whom he/she had discussed UFE. The results of this question indicate that 4 early adopters discussed UFE with 7 other interventional radiologists, 4 obstetrician/gynecologists, and 1 nurse practitioner. Seven of these individuals were in the same hospital and 5 were in another hospital, either in the same city or a different city. Most frequently, they were in hospitals in the same city.

Eleven late adopters discussed UFE with 13 other interventional radiologists, 14 obstetrician/gynecologists, 4 family practitioners, and 2 internists; 30 of these individuals were in the same hospital. Eleven later adopters discussed UFE with 22 other interventional radiologists, 9 obstetrician/gynecologists, 1 family practitioner, and 1 interventional radiology technician; 23 of these individuals were in the same hospital.

Obstetrician/gynecologists, family practitioners, and internists are physicians who are the first point of contact for the patients. They refer the patients to the interventional radiologists when UFE is determined to be the best treatment option for particular patients. This situation of referring patients is somewhat complicated, as some of the alternative procedures to UFE, such as hysterectomy and myomectomy, are performed by obstetrician/gynecologists and not by interventional radiologists. At times, interventional radiologists volunteered statements about experiencing difficulty in securing patients suitable for UFE from obstetrician/gynecologists.

Table 6 also provides the number of interactions between the interventional radiologists and these others in their social network, within each of the three prominent factors or cliques conducive to the flow of information about UFE.

Preliminary qualitative analysis.

An interesting observation related to the social network of interventional radiologists is the link between research, practice, and geography. Participants were asked to provide names of three individuals with whom they discussed UFE frequently. This was the basis of the network analysis above. The same data also provided some other interesting observations. Individuals in two hospitals were nominated frequently by their peers. The hospitals where these individuals were located were Henry Ford Hospital of Detroit and William Beaumont Hospital of Royal Oak. Although names of individuals were not requested, participants nominated two individuals, in particular, at these hospitals. Participants also volunteered statements about how beneficial it was to receive the information and assistance in performing the procedure from these two interventional radiologists. These two interventional radiologists appear to have played the role of opinion leaders in the diffusion of UFE in Michigan.

It is interesting that neither was located in universities, but both were in hospitals primarily dedicated to practice. Also interesting is the fact that the participants who were primarily clinicians, in contrast to clinician/researchers in universities, did not identify interventional radiologists from the universities in Michigan as individuals with whom they discussed UFE. However, the two hospitals where the frequently nominated opinion leaders were located were close to two large universities in Michigan.

DISCUSSION AND CONCLUSIONS

The study contributes to the understanding of the diffusion of a significant innovation in women's health. Uterine fibroid embolization appears to be a viable treatment alternative that avoids surgery while preserving the uterus. Almost all of the radiologists attested to its relative advantages over other procedures for select women with uterine fibroid disease.

The results of the study reveal radiologists' information behavior. Journals were cited as information sources both in creating awareness and providing influential information for adoption by most of the radiologists. To keep themselves informed, interventional radiologists regularly read the professional journals in their discipline. They rarely browsed Websites for information. Conferences were an important initial source of information about UFE, creating awareness in early adopters of UFE, but other individuals, particularly colleagues, were influential in the adoption of UFE by later adopters. Early adopters would find out about UFE at a conference, but the later adopters would wait to hear about early adopters' experiences with the procedure after reading the journals and learning more about it.

Of importance in the study was the influence of patients themselves. Patients were influential information sources in the adoption of UFE. This, in some ways, is related to the nature of the innovation itself. The physical and psychological effects of hysterectomy are widely discussed in the literature. The phenomenal advantages of the procedure over alternative treatments might have motivated patients to be an impetus in the radiologists' decision to perform the procedure. Several radiologists repeatedly stated that patients were calling their office to inquire about the procedure when they were still thinking about offering the medical service. This is unusual as most of the interventional radiologists' patients are referred to them by other physicians—obstetrician/gynecologists, family physicians, or internists. Seldom do patients contact the radiologists directly.

Patients were, however, absent in the social network analysis, since the network is based on individuals with whom interventional radiologists communicated frequently. Work relations, in particular, information exchange relations in the context of clinical practice, such as (a) discussing patient care, (b) receiving advice, (c) collaborating on clinical protocol, (d) discussing general care, (e) discussing patients on radiology rounds, and (f) discussing new developments, were found to have been most conducive to the flow of information about UFE among members of the interventional radiologists' social network. These are general workplace discussions, and it is only appropriate that these would be prime situations for discussions about new developments as well. The other two factors, which included information exchange relations conducive to the flow of information about UFE, were conducting research and leisure. These factors were also important and a part of the social network of interventional radiologists.

The results of the preliminary qualitative analysis are revealing. Interventional radiologists repeatedly nominated two other radiologists as adopters of UFE with whom they had conversations. It is inferred in this study that these adopters were playing the role of opinion leaders in the diffusion of UFE in Michigan. Both radiologists were located in hospitals primarily dedicated to practice and not affiliated with universities. This indicates that opinion leaders in the diffusion of UFE were situated in the practice community rather than in the practice/research community, that is, in hospitals affiliated with universities. However, the hospitals in which they worked were close to two large universities in Michigan.

Publications reporting early results of clinical trials of UFE appeared in European journals in 1995 and in American journals in 1996. There is a possibility that the opinion leaders had increased awareness of these publications because of their proximity to universities, and, once they adopted UFE, they spread the information about the procedure to other interventional radiologists. They were more involved in the diffusion of UFE than their academic counterparts. Many interventional radiologists in Michigan are affiliated with hospitals that are not university hospitals, and it appears that they waited to hear about personal experiences from other clinicians in hospitals that are not affiliated with universities either.

There were still some non-adopters of UFE in Michigan at the time of the interviews. The non-adopters of UFE cited lack of time, lack of administrative support, lack of cooperation from obstetrician/gynecologists, and doubts about the relative advantages of UFE over alternative procedures as reasons for non-adoption.