[Objectives] This study conducted systematic review over research literature on interorganizational transfer, developed integrated theoretical framework for knowledge transfer, and propose research hypothesis. Questionnaire was developed based on theoretical framework and field investigation. Research hypothesis was then tested over survey data with random graph models. Mechanism of knowledge transfer among healthcare organizations was summarized.[Data sources and methods] Past studies interpreted inter-organizational knowledge transfer with lens of organizational learning, relational view and network theories. This study integrated theoretical insights of all these three perspectives and developed a comprehensive framework for knowledge transfer among healthcare organizations, with factors including knowledge stock, absorptive capability, proximity, resource complementary, contractual relationship, external environment and network effects. Research hypothesis are proposed according to interaction between knowledge transfer and these factors. Operable research hypothesis was proposed according to factors’ impact on knowledge transfer. The questionnaire was developed based on research literature, revised in terms of pilot survey, and sent out to all 55 public hospital above second level in a city. 46 questionnaires were recycled and the response rate was 83.6%. Statistical tests were conducted using exponential random graph model(ERGM), with hospitals as nodes, knowledge transfer as links and independent variables as network statistics. Statistical significance was inferences in terms of model coefficients and their standard deviations. Test for goodness of fit was conducted to ensure the validity of statistical models. Response rates for questionnaire is 74.5%. Study samples were representative.[Results] Hospitals were active in external communication, with a certain few hospitals as subjects. Results from the statistical models indicated that knowledge stock, absorptive capability, spatial proximity, cognitive proximity and most network effects were proven valid(P < 0.001), while results for resource interdependence and social proximity was not consistent with research hypothesis, though statistically significant(P < 0.001). Other hypothesis including competitive pressure and administration were not validated by the survey model. In addition to visual comparison between networks simulated by statistical models and observed network. In-degree distribution, out-degree distribution, shared partner distribution and geodesic distance distribution was also compared. We concluded that simulated networks was consistent with actual observation. Statistical inference on research hypothesis based on this graph model was sound.[Conclusion] The result suggest that hospital mainly choose a few hospitals as stable knowledge sources. Major factors affecting choices are:(1) rich stock in knowledge;(2) similar competitive advantage in clinical medicine;(3) smaller geographical distance. In terms of the whole network, knowledge transfer was mostly asymmetric, reflecting the unequal status of hospitals. Sources were often senior or, at best, equal to recipients. Network was clearly hierarchical, which indicated lack of division of labor. For hospital managers and policy makers, disadvantaged clinical departments should be encouraged to participate in knowledge activities. Competitive advantages of each hospital should be more visible so as to encourage mutual learning and efficient division of labor. |