Figure 1.
Health care utilization in the Netherlands in 2004.
A) Map of the Netherlands showing the location of the university medical centers, top clinical and general hospitals. Patients were stratified based on the number of admissions in one year, and per stratum we measured: B) the number of patients and the distributions of C) the length of stay, D) time between admissions, E) number of different hospitals visited and F) the number of changes between hospitals, i.e., the number of admissions in a different hospital than the previous one.
Figure 2.
Spread of hospital-acquired infection between hospitals in absence of interventions, according to our individual-based model results using the recorded health care utilization patterns.
The thick lines show the mean and shaded areas show all runs between the 5th and 95th percentile. A) Time to encounter of the first colonized patient. B) Prevalence of colonization among admitted patients.
Figure 3.
Patients referred between hospitals in the Netherlands.
A) The reconstructed Dutch health care network based on the potential infection rate between hospitals, red squares denote university medical centers, blue triangles the top clinical and black circles the general hospitals. B) Inward (blue), outward (red) and relative inward (black) degree of connectedness per hospitals category, calculated from the Dutch medical registration. The relative indegree is the indegree divided by the total number of admissions. Solid lines show mean degree per category and the dashed line shows the overall mean degree. University medical centers take a clear central position, in the sense that they have a high degree of connectedness. The network is directional towards the UMCs as they have a higher indegree than outdegree.
Figure 4.
Impact of directionality of the hospital network on the spread of hospital acquired infections.
We created three networks with alternative directions. One with equal referral probabilities to all other hospitals (A, B & C), one exact reversion of the original network (D, E & F) and one with an exaggerated reversion of the direction (G, H & I). A, D & G show the time to encounter of the first colonized patient, B, E & H show the prevalence among admitted patients and C, F & I show the indegree (blue), outdegree (red) and relative indegree (black). Reversion of the network direction leads to a lower prevalence in university medical centers, while they are still the first to encounter the infection, showing that the relative indegree, the indegree divided by the total number of admissions, relates to the found prevalence and the high absolute indegree relates to the time to first encounter.