Web-Based Interactive Tool to Identify Facilities at Risk of Receiving Patients With Multidrug-Resistant Organisms

Rany Octaria; Allison Chan; Hannah Wolford; Rose Devasia; Troy D. Moon; Yuwei Zhu; Rachel B. Slayton; Marion A. Kainer

Disclosures

Emerging Infectious Diseases. 2020;26(9):2046-2053. 

In This Article

Discussion

TDH has used this interactive tool to improve statewide awareness of the importance of interfacility connectedness, particularly during outbreaks and for containment responses of novel MDROs. We designed the tool as a web-based application for real-time, easy access with internet browsers from computer desktops or handheld devices. This flexibility ensures public health staff can access the application to identify at-risk facilities in a variety of settings, such as when in the field performing point prevalence surveys or during routine office work. The application has helped epidemiologists and infection preventionists prioritize communication during public health containment responses.

We have demonstrated that a facility's ego network can accurately predict the facilities patients visit after discharge from the index facility during an outbreak.[6] The TDH HAI/AR team used the application during a particular multifacility outbreak that had evidence of MDRO interfacility and intrafacility transmission. The tool allowed us to identify facilities that frequently received patients from the 2 ego facilities involved in the outbreak. TDH alerted these downstream facilities, which led them to consider admission screening for incoming patients from the 2 ego facilities. TDH plans to continue to use this tool during similar outbreaks. Facility transmission warrants public health action to alert downstream facilities to consider admission screening or enhanced contact precautions for patients admitted from the ego facility.

In addition, the TDH HAI/AR team introduced and demonstrated the use of this application to infection preventionists at hospitals and nursing homes through a variety of webinars and in-person presentations across the state. TDH received requests from facility infection preventionists for line lists of downstream facilities because they were planning containment efforts and wanted to understand which facilities receive the most patients from their facilities. TDH did not provide hospital infection preventionists access to the application but fulfilled requests by emailing exported line lists as Excel documents. Information on downstream facilities can help inform which facilities to target for relationship development and likely will assist with communication during patient transfers.

The TDH HAI/AR team performs targeted infection control assessments as part of a MDRO prevention strategy. These assessments, conducted by TDH infection preventionists, are nonregulatory, consultative, on-site healthcare facility visits to identify gaps in infection prevention specific to a targeted pathogen or area of concern. Our web-based Shiny application was and will continue to be used to identify highly connected facilities in each EMS region and across the state. Prior studies found a correlation between the incidence of MDROs in a healthcare facility and the facility level of connectedness, measured by weighted in-degree and out-degree.[6,8] Identification of highly connected facilities is valuable because it enables us to perform preemptive targeted infection control assessments before the introduction and spread of MDROs. Public health staff can assist by ensuring adequate infection prevention practices are in place at highly connected facilities where a potential for catalyzing interfacility transmission exists.

Our patient-sharing network has several strengths. Access to the hospital discharge data and granular patient identifiers enabled us to conduct person matching with high-level identifiers. We were able to use a robust method to match patients from populations with any insurance coverage for statewide data in HDDS. Moreover, the use of 2 complementary datasets established a highly inclusive picture of a facility's ego network. With both the HDDS and CMS datasets, ACHs, CAHs, IRFs, LTACHs, and SNFs could be included in our analyses each time an ego facility is evaluated. Previously published patient-sharing network analyses were constructed by using partial data that included only direct patient transfers; a subset of patient population, such as CMS beneficiaries; hospitals; or county-level data.[6,8,12,16]

The inclusion of SNFs was critical for analysis because LTCFs are a key component to a hospital's patient sharing network.[17] Although not all types of LTCFs were included our network, the inclusion of SNFs represent the facilities carrying a considerable burden of MDRO infections. Point prevalence analysis of MDS data found that MDRO infections were found in 4.2% of nursing home residents in the United States.[18] Colonization with MDROs were found to be more common among nursing home residents.[19,20] Smaller cohort studies showed gram-negative bacteria was found in 39% of nursing home residents and MRSA was found in 42%.[19,21] Thus, communication with LTCFs is crucial for outbreak management and prevention activities.

An additional strength of the application is its built-in flexibility, which allows the user to tailor the colonization period for specific organisms. The inclusion of 365 days as the longest transfer period for indirect transfers reflects the documented colonization period of CRE in the community,[22] but users can change this parameter to account for MDROs with shorter colonization periods. The application also can display facilities connected only through direct transfers or through 30-day indirect transfers, which might reflect the colonization period of different MDROs more closely.

One limitation is the construction of 2 separate networks with the HDDS and CMS datasets. Ideally, the tool would include 1 large network with all facilities, but the construction of separate HDDS and CMS networks was required because of the differences in facility aggregation. Although both networks include ACHs, the unique number varies in each because of the difference in aggregation. Aggregating facility-level transfer data together might result in loss of information about some granular patient-sharing patterns in HDDS. One CMS certification number from the datasets can represent a group of tertiary hospitals within the same organization, creating a challenge to merge these data with the HDDS database. More recent CMS datasets include ZIP code information and the CMS certification number. We hope to use this additional datapoint in future analyses while exploring facility aggregation and standardization strategies for our databases.

We will continue to develop and improve the application with the addition of upstream facilities. We will update the network data and facility characteristics for the application annually with the most recent HDDS and CMS data. We also plan to develop models to outline the risk for transmissions based on their relative position in the network. We are working to merge the HDDS and CMS network data by standardizing facility identifications for a unified patient sharing network that provides a more complete picture of the patient population. Finally, we plan to expand the availability of this web-based platform to other public health departments by developing a feature to allow for external data uploads so health department staff can visualize their regional patient transfer networks. Access to information on patient-sharing networks would assist public health departments in mitigating MDRO transmission in their jurisdictions.

processing....