“Us and Them”: a social network analysis of physicians’ professional networks and their attitudes towards EBM

Prior literature has widely documented that there is a significant association between the propensity of physicians to use Evidence-Based Medicine (EBM) in their practice and the structural characteristics of their professional networks [1, 2]. In particular, this stream of research has shown that the network characteristics of professional relationships among clinicians are important predictors in explaining their different orientation towards EBM [2‐4].

Although EBM has been widely considered as an individual attitude, its actual impact within organizations strongly relies on its pervasiveness and widespread diffusion at the organizational level [5]. If physicians do practice EBM individually, the risk is that barriers to the effective implementation of innovative clinical solutions are not translated “from the bench to the bed” of the patient [6, 7]. These difficulties are often due to social constraints and barriers which some elitists may establish against other non-elitarian members within organizations [8], as well as the resistance of other clinicians who have a different behavioral orientation [9, 10].

Despite its general importance, this topic has been seldom analyzed on empirical grounds in healthcare organizations. The aim of the present paper is to fill this gap by exploring and testing whether physicians’ self-reported frequency of EBM adoption is related to the network position they hold in the overall web of collaborative relationships established within healthcare organizations, where they routinely visit and treat their patients.

Data regarding a community of hospital physicians staffed in one of the biggest Italian healthcare organizations were collected and used in the present study. Social network analysis was firstly performed to identify structurally important physicians in the network. Specifically, we derived a core-periphery structure of the overall inter-physician network, distinguishing the dense cohesive core of the professional network from the sparse, unconnected periphery. Then, a new class of network centrality indicators, overall called Hubs and Authorities centrality, were employed to capture the structural prominence that physicians exhibited in the network. Finally, we explored whether their self-reported frequency of EBM adoption predicted the degree of coreness and structural importance that individual doctors assume within the observed network.

Social networks research has provided ample evidence that individuals’ attitudes and other personal characteristics influence the shape of their social networks as well as the position they assume in the overall web of relationships [11‐13]. On the basis of previous work developed in this field [14], we assume that the propensity towards EBM is a relatively stable physician individual characteristic, which in turn influences his/her network position within organizations. We hypothesize that there is an association between the physicians’ propensity to use EBM and their degree of coreness within organizations, taking other relevant individual and organizational characteristics into consideration.

The overall sample is made up of 297 physicians. Table 1 shows the main characteristics of sampled individuals. They are, on average, 47 years old (SD 8.01) and are mostly men. The number of years they have accumulated in the INHS is, on average, 16.01 (SD 8.01). Years of experience that have been accumulated within the organization is, instead, 10.95 (SD 7.94). Only 51 physicians are clinical managers. As for EBM adoption, the majority (71.5%) reported adopting EBM frequently, followed by physicians declaring to adopt EBM very frequently (14.49%), occasionally (12.56%) and never (1.45%). Overall, almost 86% of sampled physicians reported to adopt EBM frequently or very frequently.

Figure 1 illustrates the network of collaborative relationships among sampled physicians. The circle (node) represents physician and the link (edge) represents an existing collaborative tie among node pairs. Physicians' locations in Figure 1 were determined using a spring-embedding heuristic, multidimensional scaling algorithm, with proximity indicating the extent to which two clinicians were connected directly and indirectly through mutual colleagues [21].

Table 2 shows the pairwise correlations among variables. The inspection of coefficients reveals a strong and positive association between age, tenure in the INHS, and organizational tenure. Tenure in the INHS is, in turn, positively associated to the variable making distinction of whether the clinician has a managerial role or not. Network coreness and network authority variables showed to be moderately associated with both the propensity to adopt EBM and geographical proximity of physicians, albeit with a different sign.

Tables 3 and 4 show the OLS (Ordinary Least Squares) regression results. Stata 10 was used to perform the regression analysis.

Table 3, in particular, presents three different models that we built to explore the clinician’s coreness within the professional network. Model M1 contains only the EBM adoption variable, and it should be considered as a null model against which the explanatory power of the subsequent models can be compared to. Model M2 includes only control variables. Model M3 is the full model incorporating all explanatory variables.

Model M1 in Table 3 shows that there is a significant association between the EBM variable and the network position that physicians hold in their professional network. In particular, it was found that a negative association exists between the physicians’ attitudes towards EBM and their degree of coreness (β = -0.004; p < 0.05). The regression results also documented that, among all of the structural and characteristic variables included, the coreness of individual physicians in their professional network was associated with Managerial Role, Geographical Proximity and the variables reflecting the clinician’s affiliation to hospital structures and directorates. In particular, Managerial role (β = 0.011; p < 0.05) and Geographical Proximity (β = 0.001; p < 0.01) were positively associated with the coreness of professionals within the organization. Physicians in Department #2 were more likely to exhibit a higher coreness than those in Department #1 (β = 0.020; p < 0.01), which is the baseline category of the model. Physicians in hospital facilities #2 (β = -0.056; p < 0.01), #3 (β = -0.045; p < 0.01), #4 (β = -0.029; p < 0.01), #5 (β = 0.026; p < 0.01) and #6 (β = -0.017; p < 0.10) exhibited a lower coreness score than those working in hospital #1.

Table 4 presents all models exploring the association between the EBM adoption variable and the Network Authority variable. Our model building follows, similarly to what previously presented, a stepwise approach. According to this logic, Model M1 contains only the EBM adoption variable, Model M2 includes only control variables, and Model M3 is the full model that incorporates all explanatory variables.

Models M1 and M3 in Table 4 document that a negative and significant association exists between the EBM adoption variable and the structural importance that physicians hold in the collaboration network (β = -0.009; p < 0.05 in M1; β = -0.019, p < 0.05 in M3). The inspection of parameters corresponding to control variables overall confirms our previous results documenting a significant association between the network centrality of clinicians and a number of other contingencies, such as their spatial proximity from colleagues (β = -0.001; p < 0.01 in M2; β = -0.001, p < 0.01 in M3), the managerial position they eventually occupy in the organization (β = 0.020; p < 0.1 in M2; β = 0.023, p < 0.05 in M3), and their belonging to specific hospitals (hospital #5, β = 0.052; p < 0.01 in M2; β = 0.048, p < 0.01 in M3) and departmental arrangements (Department #3, β = -0.047; p < 0.01 in M2; β = -0.047, p < 0.01 in M3).

EBM represents one of the most important paradigms in modern medicine [24‐26]. Clinicians and healthcare professionals in general are requested to increasingly adopt and integrate the latest available medical knowledge produced in their clinical practice [27].

In this study, we explored how the propensity towards EBM is associated with the position that professionals occupy in the overall network of collaborative ties they create within healthcare organizations. Our findings documented that there is a significant negative association between the physicians’ propensity to use EBM and the coreness they exhibit in their organization. Our analysis also indicated that the core is formed by physicians having a significantly lower propensity towards EBM than their peers located in the peripheral part of the network. Supplementary analyses were performed to capture closely the structural importance of physicians in the professional network through the use of network centrality indicators. Our findings again provided evidence for a negative association between EBM adoption and the network prominence of individual clinicians.

Although the homophily of physicians in terms of EBM adoption has been documented elsewhere [23], in this study we show that higher EBM adoption can cause the isolation of such groups of professionals. Within organizations, there is the potential risk that professionals having this kind of behavior can be viewed as elitists who may behave in contrast with the practices routinely adopted within hospitals. As prior studies have shown [28], innovators within healthcare organizations often face hard times in changing the way consolidated practices are used daily. In particular, those who are located in the center of the network are less exposed to novelty and innovative behavior since their higher interconnectedness with homophilous pairs likely increases the risk to be influenced from colleagues [3].

Our findings are also consistent with extant research demonstrating that the acquisition of new knowledge by physicians often occurs more likely through personal relations than through explicit guidelines and clinical protocols [2]. Gabbay and le May [29] have shown that physicians often use mindlines instead of guidelines, because of their tendency to discuss clinical matters with colleagues instead of relying on documentation such as articles, meta-analysis and Cochrane library. In addition, the superior propensity towards EBM of clinicians forming the periphery might reduce their risk of becoming over-embedded.

Our findings have a number of implications. First, hospital executives are encouraged to identify groups of professionals that exhibit potentially virtuous attitudes and behaviors within their organizations. Social network analysis tools and techniques appear useful in this vein. They are, in addition, encouraged to foster collaboration across groups characterized by different propensities to use EBM in daily practices. The adoption of new organizational arrangements, processes and informal occasions of meeting, would all be useful means to achieve this objective. For example, increasing collaboration might be achieved through the internal restructuring of hospital organizations. The adoption of specific types of clinical directorates or interdisciplinary and interprofessional groups is an example in this direction [15, 30]. New internal processes have to do with both organizational and professional streams of activities. Organizational processes for example concern the definition of objectives such us budget, quality standards and appropriateness, which may be targeted by administrators in order to encourage collaboration across heterogeneous groups. Finally, executives have the possibility to support the inclusion of medical leaders within organizations so that their role might be leveraged to persuade other professionals to collaborate more with EBM users [8].

Policy implications are also strong. Health systems around the world are urged to ensure coordination and integration amongst providers by fostering the collaboration of healthcare professionals belonging to different organizations. Policymakers may want to encourage healthcare administrators to implement the above mentioned actions. In addition, in this context, interorganizational cooperation may be better achieved by identifying EBM users in organizations and then leveraging the higher tendency to cooperate by the virtue of their homophily [23]. These initiatives might be, for instance, headed to foster a better continuity of care across organizations through the formation of interorganizational groups or the definition of innovative clinical pathways.

Our findings should be interpreted in light of several limitations. First, the degree of EBM adoption was self-reported in the present study. Although this is not an objective approach to studying physicians’ orientation to EBM, our approach seems to be consistent with extant research on this topic [2, 5]. The type of study design poses another limitation. Given that all data were gathered at the same time, we cannot ascertain whether the collaboration of physicians with colleagues is an antecedent or consequence of EBM adoption. Even though the cross-sectional design adopted in this research prohibits us to determine causality, it provides however that a causal link exists between EBM utilization and social collaborative relationships. We encourage future longitudinal studies to disentangle the effect of physicians’ attitudes towards EBM and their propensity to establish collaborative ties in healthcare organizations.

Our study documents that the overall network structure is made up of a dense cohesive core of physicians and a periphery made up of less connected clinicians. The social structure of this model underlies a group tightly connected physicians who interact strongly in order to exchange relevant knowledge, and a large number of less cohesive clinicians who are more likely to be connected amongst themselves than to members of the core part of the network. This result might be interpreted as a marginalization of physicians who are more prone to use EBM in their clinical practice.

This social structure may result in a fragmented organization, in which different habits and characteristics of groups of physicians likely increase the risk of conflicts and barriers for integration within hospital boundaries [31, 32].

Social network analysis tools and techniques should be increasingly adopted by policymakers and administrators in order to support integration and coordination of clinical activities in complex social systems such as healthcare organizations.