Externalities of Inter-facility Patient Sharing: Diffusion of Healthcare Acquired Infections
However, hospitals commonly share patients with other hospitals via direct transfers and indirect sharing via readmissions. This patient sharing between hospitals may increase the connectivity of hospitals and may serve as a potential means for the dissemination of healthcare-associated infections. The movement of patients between hospitals serves to connect hospitals in a way that may only weakly correlate with spatial distances, and the strength of these connections may be much stronger than those simply resulting from drawing from the same catchment population. Simulations studies support a model suggesting dissemination of healthcare-associated infections via patient transfers [1]. To date, there is little empirical work focusing on measuring the potential for spreading healthcare-associated infections via patient transfers. These patient transfers may undermine local hospital-specific efforts to control healthcare-associated infections. Here we focus on C. difficile, a common healthcare associated infection.
We used the Healthcare Costs and Utilization Project State Inpatient Database for Arizona for the years 2003-2007 to reconstruct the network of patient transfers between hospitals. We identified cases of C. difficile infection and computed the incidence for each hospital during each of the 20 quarters. We estimated a mixed model with centered log incidence by quarter as the dependent variable. Fixed effects in the model included log median length of stay, log fraction of patients over the age 65, year and quarter of the year and a random intercept. This model is “network naive" and lacks any variable measuring what is occurring elsewhere in the network. We fit a comparison model that was “network aware." The network awareness was included by adding the weighted log incidence at hospitals from which transfers originated to the model described above.
Model fit was compared using AIC. We found a 10.4 unit decrease in AIC between the network naive and network aware models. The model estimates that a 1% increase in incidence of C. difficile at the source hospitals is associated with a 0.023% (95% CI: 0.010, 0.036) increase in the expected incidence at the receiving hospital. While this is a relatively small increase in incidence, for a large hospital or system of hospitals, the marginal cost of these additional cases in aggregate may be substantial. Additionally, other infectious diseases, such as influenza, can spread over these same connections. A hospital’s network-specific risk for healthcare-associated infections has implications for the effectiveness and equity of reimbursement-based incentives for controlling healthcare-associated infections.
References
[1] B. Y. Lee, S. M. Bartsch, K. F. Wong, A. Singh, T. R. Avery, D. S. Kim, S. T. Brown, C. R. Murphy, S. L. Yilmaz, M. A. Potter, et al., The importance of nursing homes in the spread of methicillin-resistant staphylococcus aureus (MRSA) among hospitals," Medical Care, vol. 51, no. 3, pp. 205-215, 2013.