Measuring integration and influence among health professionals. A social network approach

Introduction

Efforts to improve communication, interactions and cooperation between individuals and organizations, lead to better quality outcomes in a business context (Balkundi & Harrison, 2006; Choi, Kim, & Lee, 2010; Kilduff & Tsai, 2003; Krackhardt & Kilduff, 2008). In the healthcare context, integrated care is one such strategy (Berry & Mirabito, 2010; Ehrlich, Kendall, Muenchberger, & Armstrong, 2009; Johanson, 2010; Koshiyama, Ogawa, Tanaka, & Tanaka, 2010). Most frequently, where this kind of strategy has been applied in developed countries such as New Zealand, integrated care has been applied to managing long term conditions; in particular, care for people with diabetes (Letford & Ashton, 2010; Rea et al., 2010; Reid, 2007; Smith, 2009). Integrated care for diabetes aims to improve timely detection, treatment and metabolic control for patients with diabetes. It is grounded in the assumption that good communication and collaboration between general practitioners, nurses and specialists leads to better patient care and outcomes within and across the primary and secondary care interface (Browne, Kingston, Grdisa, & Markle-Reid, 2007; Foy et al., 2010; McCormick & Boyd, 1994). Therefore, measuring how well integrated is an inter-professional network and describing influence within members of the network merge as indicators of good communication and collaboration among health professionals (Ahgren & Axelsson, 2005; Fattorea, Frosini, Salvatorec, & Tozzid, 2009).

A systematic review on how measure integration within integrated care had been proposed a set of criteria for measurement methods to guide future research and, also claims for practical and relatively easy methods for non-scientific users (Strandberg-Larsen & Krasnik, 2009). This paper proposes the social network approach to measure the level of integration and influence within inter-networks at the primary and secondary care interface. This perspective is important for both researchers and practitioners, but for managers responsible for implementation could become a tool for monitoring and maintenance of integrated care. The basic theoretical elements of Social network Theory will be outlined, also described previous experience implementing this approach in the health sector, and a practical application of this approach in the context of the primary and secondary care interface is described and analysed.

BACKGROUND

RESEARCH DESIGN AND METHODS

RESULTS

Fifty people responded to the social network survey; one was incomplete leaving 49 who contributed to further analyses. The study had 49 participants in the social network survey from 50 invitations submitted, with a gender distribution of 65% female and 35% male. The distribution of the participants by profession and sector is shown in table 2.

A total of 18 health practices from primary care were represented by at least one participant. Figure 1 shows network A and represents how heath workers were connected within and between primary and secondary care. The greater size of the spheres and a more central position on the network represents greater relative influence. The “star” graph suggests that health workers in positions more central of the network exert more dense interactions and influence over the rest of the inter-professional network. In the middle of the graph were a dense group of specialist nurses and endocrinologists (red spheres) interacting for the purpose of providing support and advice to general practitioners and nurses (blue spheres). Density of interactions within this network is 15.2, which is relatively low.

• Download figure
• Open in new tab

Figure 1: Network A, all primary care and secondary care interactions

Blue: Primary care sector & Red: Secondary care

Figure 2: network B map. Interactions between Primary and secondary care by profession and within primary care (omitting interactions within secondary care).2 shows the distribution of the network at the interface between primary and secondary care after excluding interactions solely within secondary care.

• Download figure
• Open in new tab

Figure 2: network B map. Interactions between Primary and secondary care by profession and within primary care (omitting interactions within secondary care).

Red: specialist, Blue: GP and purple and Green: DSN.

This figure shows the importance and centrality of three secondary care professionals (SCS003, SCS009 and SCN010) in this network, and contrasts with network A in that DNSs are more peripheral. This graph illustrates the relatively higher centrality of some five GPs (GD00628, GD00332, GD00514, GD00326 and GD00926) who relatively were more connected with both secondary care professionals and with primary care colleagues.

The density of the interactions in network B has reduced to 3.7%, as the interactions between secondary care’s health professionals were eliminated. Table 2: Social Network Measures. Network B interface3 shows the social network measurements at the interface’s network. The overall results are consistent with the map shown in Figure 2: network B map. Interactions between Primary and secondary care by profession and within primary care (omitting interactions within secondary care).2. Three secondary care health professionals possess the highest degree and betweenness in both Networks A and B. Because closeness centrality is one measure of strength of interaction between GPs and secondary care professionals, it appears that professionals GD0047, GD0630 and GD0723 have weakest relationships with secondary care. However, high eigenvectors shows that two GPs (GD0047 and GD0630) are strongly connected through leaders in the network.

Comparing the two Network figures shows three important elements. First, the structure of network A is a result of the great density of interaction between health professionals in secondary care and between DNSs. Second, endocrinologist are the most central or important actors in both networks, followed by DNSs. The social network questionnaire asked if an interaction with a particular health professional influences his/her about how diabetic patients are diagnosed and treated, and also if this interaction provides current and valuable information that influences decisions about how to care patients with diabetes, which means that the network structure suggested the direction of the influence and advice across the interface. Third, there are a range of differences in centrality measures within the GPs, some of them appear well and strong connected within the network but some appear relatively isolated.

Table 4 illustrates the distribution of the network densities by frequency of interaction in relation to their profession. Each value represents the calculated density of interaction obtained when a group of actors interact in a particular frequency. In general there is a higher frequency of interaction within secondary care (0.75) than in primary care (0.03) on issues related to diabetes. Endocrinologists interact overall more about diabetes than the rest of the network followed by DNSs. Endocrinologists have more dense interaction weekly while DNSs do on a daily basis. There is a more frequent communication between nurses (PNs and DNSs), higher than communication with GPs with secondary care. As in this study frequency of interaction represents advice regarding how to treat patients with diabetes, the data reveal that GPs seeks such advice from secondary care rather than from other GPs. The frequency of interaction for general practitioners with secondary care is weekly and higher in contrast to the frequency of interaction between GPs which is weekly or monthly. Interaction between GPs and DNSs were more frequent than between GPs and endocrinologists. Physicians (GP and endocrinologist) interactions can be less than monthly, while the interactions between GPs and DNSs can be weekly or even daily.

Table 5 shows that interactions between endocrinologists, followed by interactions within secondary care, were ascribed with high quality (84% and 53% respectively). In contrast, interactions within GPs were considered high quality in just 1%. Therefore, if GPs seek advice, they were more likely to value discussion with professionals in secondary care than colleagues at the same practice or elsewhere in primary care. However, the quality of the interaction between GPs and DNSs was ranked higher than interactions with endocrinologist, and 1% of the density of the interactions was ranked low quality. That means there are factors to affect the quality of the interactions between GP and endocrinologist related to current and value of the information. For DNSs and PNs ascribe higher quality to the information they share than do GPs when they interact with secondary care.

Table 5. Perceived value of interaction about diabetes care at the primary and secondary interface6 shows the most perceived value of the interactions is hold within secondary care (endocrinologist and specialist nurses). In contrast with the low perceived value of the interactions within primary care.

Although GPs considered interactions with endocrinologist with high perceived value there is still a percentage of the interactions ranked with low perceived value. Therefore GPs considered the information from endocrinologist is current and valuable (high quality) but in some cases does not help to take decisions on how treat patient with diabetes.

DISCUSSION

The social network perspective applied to the primary and secondary care interface is one of the contributions of this study. Defining two Networks proved a useful way to confirm that the centrality of secondary care and its influence on primary care is real rather than an artefact of the high density of interactions between endocrinologist and DNSs. The selected social network measures constitute measures of integration within the current health professional’s network and, anticipate information to analyse potential issues for knowledge transfer and advice. It has been suggested that the small size of networks in some chronic care programs could be a critical barrier to patient care because of the limited resources and capacities that each actor has for building and maintaining ties (Scott & Carrington, 2011), allowing access to information from valuable collaborators (Burt, 2009), and also the introduction of new practices and behaviours reinforcement (Valente & Fujimoto, 2010). In regard to the centrality measures, in-degree may be especially important as it reflects the influence and importance of individuals. Diabetes care at the interface between primary/secondary care recognizes actor’s influence based on their specialized knowledge and expertise (opinion leaders).

Some GPs in this study showed 100% density of interaction with his/her colleagues within the general practice, and also has many interactions and connections with secondary care professionals. Those GPs could share information with important players in secondary care as well as with his/her colleagues in primary care, such that otherwise-isolated health professionals could indirectly receive the benefits of interaction with secondary care. This represents a structural hole as two separate clusters (general practice and secondary care professionals) possess non-redundant information, but a mutual connection to allow flow of information. Some health professionals in primary care appear to be disconnected from the influential actors and their specialised knowledge. However, selecting the interface network showed that some of these GPs apparently isolated had interactions with other actors that were connected with influential actors. This finding is very important in terms of access to information and knowledge.

Overall analysis confirms Zappa’s (2011) findings related to the heterogeneity in the tendency to build interactions with other colleagues (Zappa, 2011). A phenomena seeing other networks and in different contexts (Freeman, 2008; Nohria & Eccles, 1992; Scott & Carrington, 2011; Wasserman & Galaskiewicz, 1994). This variability represents differences in motivation and interest for building interactions within a social network in order to solve personal needs and is associated with the decision to adopt innovations or new knowledge (Kim & Higgins, 2007; Zappa, 2011). Some actors seek advice when they encountering an issue or opportunity to resolve individual doubts, and as a result actors able to share relevant knowledge and information have more central positions in the network (Nebus, 2006).

The current results presents an opportunity to take in account two different interpretation of density measures: Coleman (1988) states that dense networks are the optimal social structure, while Burt (2009) states that disconnected alters is the optimal strategy for a network. Coleman and others would argue that higher network density implies better access to timely information for primary care (Coleman, 1988; Gulati & Gargiulo, 1999; Rowley, Behrens, & Krackhardt, 2000). This could be particularly evident for complex knowledge, as this requires direct, frequent and intense relationship. In contrast with this position, Burt (2009) considers that density may affect the diffusion of information within the groups in a network rather than between groups in a network. As a result structural holes within the network offer one effective way to share information. This explanation could apply in this context as GPs and PNs do not work alone, they share work places and patients within a general practices. Communication and information is relatively easy to maintain within practices as there are a variety of ways and opportunities of interaction: i.e. clinical meetings, lunch time, informal corridor meetings, and virtual patient clinical data on the local system are all examples provided by participants during interviews. As a consequence of these forms of interaction, professional’s thinking may be homogeneous among health professionals within health practices, as people converge in beliefs and practices developed within the group. In fact, they could attempt to block different views of information from external sources, if they consider the information is not concordant with their perspectives or interest.

However, novel and complex knowledge comes from specialised and expert health professionals at the external setting. Secondary care represents one of the main sources to obtain this kind of knowledge and, the current results have shown low density of interaction between primary care groups and secondary care. Burt would see this as an opportunity to build effective networks. Some of the health workers in general practices do not have strong relationship with specialised people. But it does not mean they are not aware of the existence of secondary care. It means people are focussed on their professional daily life activities and they do not attend the activities from other different groups. As a result there are gaps or empty spaces within the inter-professional network at this interface. GPs have to deal with patients with a great range of diseases and complications; not all GPs or PNs are likely to be equally involved or even interested specifically in diabetes care. Fortunately, other GPs or PNs more involved in diabetes care are willing to invest time and resources to build ties with expert people and then spread and share this new knowledge with their colleagues in primary care.

Social Network Analysis could be a useful tool to measure levels of integration at the integrated care programs. Managers, policy makers and staff could use this approach for the purpose of measure the structure and dynamic of the inter-professional networks, the similar way as it were described and analysed in this work. The maps and mathematical analysis showed several social network characteristics in which this network performed and interact.

This measurement model applied in this study is limited to the functional aspects of the integration within integrated care. However, it is possible to measure the synergy of the integration or integration efficacy, in other words how well-functioning is likely to influence patient outcomes by the way health professionals interact.

The present study has a number of limitations that need to be take in account when interpreting its results. First, a cross-sectional study limits the possibility to draw conclusions that imply causality. Second, the size of the sample and the particularities of the context involve generalization issues. However transferability should be considered. Third, potential limitations implicit in the social network questionnaire should be considered as other studies that used this tool.

Conclusion

The literature suggests that the strength of the ties between people in a network facilitates knowledge transfer and the diffusion of novel information. If this is a critical issue for improving patient outcomes at the primary and secondary interface in the chronic care programs, it is also critical to identify how patterns of social network characteristics affect patient outcomes. The future of the research will focus on identifying how interactions between health professional might affect patient outcomes at this interface

Social network analysis as a method used in this study offers a graphic model and information to show how health workers are interacting across the primary and secondary care interface. The density of interactions across the interface between primary and secondary care in the context of diabetes care is relatively low compared with the level of interaction within secondary care. There are differences between how health practices interact with secondary care, who doctors interact with and who nurses interact with. This may indicate different channels that the two groups operate through in developing networks. Specialist nurses are more likely to be connected with the rest of the network, and they perceive the quality and value of these interactions as a critical factor in making decision with patients with diabetes. Recommendations arising from this empirical evidence include investigating opportunities and systems for supporting general practices to interact more frequently with secondary care in order to improve health outcomes for patients.