1. Dispersal of antibiotic-resistant high-risk clones by hospital networks: changing the patient direction can make all the difference 
T. Donker, J. Wallinga, H. Grundmann 
Journal of Hospital Infection 
Volume 86, Issue 1, Pages (January 2014)
DOI: /j.jhin
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

2. Figure 1
English patient referral network in Hospitals are grouped into 12 healthcare collectives, indicated by different colours on the outer ring. Dots represent hospitals, and lines19 between dots depict the exchange of patients between hospitals. The strength of lines scales with the number of patients exchanged per year.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

3. Figure 2
Geographical distribution of healthcare use in England, (A) The catchment areas of the healthcare collectives, using the same colour coding as in Figure 1. (B) Distance to the border of the catchment areas. (C) Proportion of admissions to hospitals outside the collective. (D) Proportion of re-admissions between hospitals in different collectives. These maps show that patients living close to the borders of healthcare collectives switch between collectives more often than patients living further away from the borders.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

4. Figure 3
Effect of place of residence on hospital admissions. Proportions (%) of (A) admissions to hospitals outside the assigned healthcare collective and (B) re-admissions between hospitals in different healthcare collectives. Both are plotted as a function of distance to the border of the catchment area of the healthcare collective. The blue line denotes the mean proportion per km, and the black line shows the mean proportion over all admissions and re-admissions further than the given distance from the border. The red dashed line shows the exchange between collectives that does not depend on the distance to the catchment border (‘baseline’).
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

5. Figure 4
Effect of changing referral patterns on the spread of hospital-acquired pathogens, shown by the time course of mean prevalence in the hospitals. Shaded areas show all runs between the 2.5th and 97.5th percentiles. In comparison with the observed referral patterns (black), redirecting patients to hospitals in a collective they have visited previously (blue) slows down the pathogen's spread, while creating specialist centres (red) accelerates the spread. Redirecting patients to their original hospital (green) has limited effect.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

6. Figure 5
The increase in incidence is faster when patients are referred to specialist centres and slower when patients are referred to their own collectives. (A) Growth rate for hospitals in collectives other than that of the index hospital. (B) Growth rate for hospitals in the index collective. The creation of specialist centres results in the highest growth rate, while redirecting patient to their original collective has a dampening effect. All proposed interventions only have a marginal effect within the index collective, while the effect between collectives is larger.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

7. Figure 6
Model results with a lower transmission parameter show that the relative difference between the referral scenarios is robust to choices in the transmission parameter. (A) Prevalence of infected patients over time, (B) increase in incidence as a function of the incidence outside the index collective, and (C) increase in incidence as a function of the incidence inside the index collective.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions

8. Figure 7
The choice of seeding of the epidemic did not affect the model results. The original model results, seeding by colonizing 5% of the patients, are shown in black. The results from seeding by colonizing a single patient are shown in red. The shaded area and the area between the dashed lines show all runs between the 2.5th and 97.5th percentiles. During the original simulations, 6% of runs resulted in stochastic extinction of the pathogen, while the simulations based on a single colonized patient resulted in 85% extinction.
Journal of Hospital Infection  , 34-41DOI: ( /j.jhin )
Copyright © 2013 The Healthcare Infection Society Terms and Conditions