Association of Visitor Contact Precautions With Estimated Hospital-Onset Clostridioides difficile Infection Rates in Acute Care Hospitals

This simulation study estimates the association between visitor contact precautions and the rate of hospital-onset C difficile infection in an acute care hospital.

Coding and debugging of the Java model was completed in Eclipse IDE v4. 8  To determine stability of results, HO-CDI per 10000 patient days and colonizations per 1000 admitted patients were examined as a function of number of replications, as seen in eTable 5. As the Java model showed a similar stability at 5000 replications to the NetLogo model, we concluded that 5000 replications were sufficient to produce stable results.

Existence of association between VCPs and HO-CDI reduction
In the interest of finding what conditions exist in our model, if any, where VCPs are associated with a greater than 1% reduction in HO-CDI rates, we conducted further experiments using large parameter changes. Of primary interest is if there exists a threshold at of visitor length of stay or rate of contact at which VCPs are associated with HO-CDI reductions. In our model, all probabilities related to visitor C. difficile acquisition or deposition are of the form: = probability of acquisition or deposition = proportion of environment/patient contaminated with C. difficile = transfer probability of C. difficile between patient/environment and visitor = rate of contact between visitor and environment/patient = length of stay with patient/in environment As visitor length of stay and visitor acquisition/deposition rates increase, the probability of transfer rapidly approaches 1. Once the exponential power is on the order of -10, increasing visitor length of stay or rate of contact more do little to affect the probability of exposure.
Therefore, there is little need to investigate lengths of stay or rates of contact that result in a power less than -10. We conducted an additional experiment where we multiplied the rates of contact between visitors and the environment and between visitors and patients by 10, thereby bringing the probability of transmission and deposition near to 1. . As the difference between these average rates is less than one percent, we are confident that increasing the rate of contact or length of stay any further would not lead to a greater association between VCPs and HO-CDI.
We believe it would be of interest to know if there is any condition where VCPs may be associated with a larger reduction in HO-CDI. We found that under extreme conditions where the only possible transmission pathway was through visitors contaminating the common room, transmission to and from visitors was guaranteed, and surveillance testing was adequately high, VCPs could have a larger association with HO-CDI. Note that surveillance testing represents the scenario that patients are screened for C. difficile at admission and then are put in isolation therefore VCPs will be implemented. Experiments varying the surveillance testing parameter are shown in eTable 6. Averages and confidence intervals were calculated from 5000 replications.
Surveillance likely played an important role in the experiments described in eTable 6 because our model assumes that VCPs may only be used when interacting with patients known to be infected or colonized with C. difficile. Therefore, increasing surveillance and testing leads to a greater number of known colonizations and infections and thus increased VCPs use.
However, even when all patients entering the hospital are tested (i.e., surveillance compliance of 100%), the association of VCPs and HO-CDI is still relatively small.
Under baseline conditions, simply expanding use of VCPs to all patients is not significantly associated with a decrease in HO-CDI, as described in eTable 7.