Perceived organizational connectivity was correlated with out-degree network centrality but not in-degree centrality. These results suggest that perceived organizational connectivity may serve as a partial proxy measure of formal ties objectively reported, but not received, by organizations with respect to pandemic influenza preparedness. This finding makes intuitive sense since ties that other organizations report to an organization of interest may not be recognized or reciprocated. This may explain why in-degree centrality is not statistically associated with organizational connectivity perceptions. Out-degree centrality may be biased by organization's perceptions of their own influence in a way in which in-degree scores are not.
Further understanding of these results can be gained through consideration of the concept of cognitive accuracy. Krackhardt (1990) defined cognitive accuracy as the degree to which a person's perceived networks correspond to actual networks. His work concluded that an actor's accuracy in perceiving the influence of other actors in the network was positively associated with actual influence in the network. Choi and Brower (2006) extended the concept of cognitive accuracy to collective cognitive accuracy in their research assessing local emergency management systems in Florida. Collective cognitive accuracy shows how well participants are aware of the networks in practice and which participants are influential within each network. It is calculated as a simple percentage of the responding organizations that accurately identify the influence of an organization within a network (as compared to the relationships identified through a formal network analysis). Choi and Brower's work supported the conclusion that collective cognitive accuracy has a strong positive relationship with the centrality of networks, and that when actors have a clear mental picture of the network, greater collaboration and information sharing is observed, resulting in more effective decision making and actions. Unlike the current results however, these authors noted a strong correlation with in-degree centrality. One possible explanation for this discrepancy is that the networks in Choi and Brower's work were decentralized horizontal networks with distributed power and authority structures, rather than the primarily hierarchically-linked organizations that were the subject of our work. The scope of the network in Choi and Brower's work was focused at the local level only while our work had a much larger network structure, focused at multiple jurisdictional levels. Furthermore, Carley, Lee, and Krackhardt (2001) have previously shown that distributed decentralized networks exhibit dynamic patterns which are very different than those found in hierarchical networks. These differences in structural dynamics may offer an explanation as to why in-degree centrality was significantly correlated with organizational connectivity in our work, rather than out-degree centrality. Further research is needed to examine these inter-organizational relationships in greater detail.
Using an organizational network analysis approach provides insight into the overall structure and types of relationships existing in the public health preparedness system in Alberta for pandemic influenza. This work helps address a general gap in the public health field related to organizational network analysis, identified by Luke and Harris (2007) in their work on network analysis methods and applications in public health, in that it helps provide a structural evaluation of public health systems. With the recent shift and emphasis on using a systems approach to design, study, and evaluate public health programs, the findings of this study offer a technical means of assessing the extent to which each organization in a network is linked to others and the patterns of relationships among different organizations for one specific case example on pandemic influenza. Not only do these findings provide a means of illustrating how organizations function together as a unit, but it serves to offer an effective and timely means of assessing organizational connectivity, a critical component of public health preparedness. Moreover, formally identifying organizational connections using out-degree centrality scores or using the partial proxy measure of perceived connectivity provides a valuable tool for identifying the prominence of organizations in a network. This information can aid policymakers in developing strategies for collaboration that build on the current strengths of highly connected and central organizations while enhancing the capacity of less connected and more peripheral organizations[7, 16, 17].
Analysis of variance tests revealed statistically significant differences in in-degree and out-degree centrality measures by jurisdictional level. Differences were observed among all jurisdictional level combinations for in-degree centrality as well as between provincial and each of regional, city, and town/village levels with respect to out-degree centrality. In each case, the higher jurisdictional level had a significantly higher centrality score indicating these organizations were more connected to surrounding organizations. This finding would seem to confirm the hierarchical character of the preparedness system in Alberta as organizations at higher jurisdictions receive and report greater formal connections.
Limited literature exists which assesses the role of jurisdiction on centrality, connectivity, coordination, or specific preparedness measures for public health organizations, making it challenging to provide support for the current results. Also, unique to the current work is the depth of coverage of the inter-organizational network for pandemic influenza at four jurisdictional levels. Most other studies have typically compared only two levels, for example local and state, or county and regional levels[18, 19]. In general, however, our results are supported by recent work in Florida examining emergency management networks. According to this work, smaller municipalities have primarily relied upon counties or regional network supports for disaster-related functions since higher jurisdictional levels are better prepared for managing emergencies. Networking is advantageous because links form between individuals and agencies that would work together during a public health emergency and this serves to heighten capacity and address metropolitan fragmentation issues which are often present at lower jurisdictional levels. Other U.S. research has indicated that aside from major metropolitan areas, few counties, cities, or towns have the capacity to respond to public health emergencies independently since they lack the necessary coordination or supports with other agencies. The greater interconnectedness observed at higher jurisdictional levels in the current work is important as it may enhance communication, likely decreases coordination problems during the time of an emergency, and builds cohesion that could enhance preparedness.
There are several limitations of this work worth considering. First, the study compares perceived connectivity with network centrality arising from an organization's formal ties and relationships. Although we found out-degree centrality based on formal ties correlated with perceived connectivity, it is conceivable that other dimensions of inter-organizational relationships, such as information or material-resource sharing, may or may not be correlated with perceived connectivity in the same fashion. Further research is needed to examine if perceived connectivity is more closely linked to other dimensions of inter-organizational relationships. Second, this study is of one specific provincial preparedness network. Since the conclusion of data collection for this project, Alberta has reorganized its public health delivery system, particularly at the regional level. As a result, the type of network connections described in this study may not be the same as those currently in place. Finally, this is a cross-sectional analysis and the network data are from one moment in time. Yet, the networks in which organizations are embedded are dynamic. With recent emphases on preparedness planning, the development of action plans, and the increased use of simulation exercises, networks are continually evolving to reflect the changing landscape. The Alberta sample is no exception, and while data is collected from one time period there remains value in studying the structure and composition of the Alberta preparedness network as it existed in late 2007-early 2008. This makes possible in part the further monitoring of how relationships form and become institutionalized. Future research can use such information as a starting point for designing network- or system-level surveillance systems of organizational preparedness and conduct more dynamic analyses using longitudinal data on the structure of preparedness and response networks. While this study was conducted prior to the current H1N1 outbreak, examining how this recent event has tested the public health system and the connectivity within it would be a logical next step as investigators would likely find today a very different connectivity and set of patterns in the field.