Repurposing Geographic Information Systems for Routine Hospital Infection Control

Abstract

Purpose:

This targeted review and commentary discusses the potential role of geographic information systems (GIS) for infection control within the hospital system. The paper provides a brief overview of the role of GIS in public health and reviews current work applying these methods to the hospital setting. Finally, we outline the potential opportunities and challenges for adapting GIS for use in the hospital setting for infection prevention.

Design/methodology/approach:

A targeted literature review is used to illustrate current use of GIS in the hospital setting. The discussion of complexity was compiled using the nonadoption, abandonment, scale-up, spread, and sustainability (NASSS) framework. Challenges and opportunities were then extracted from this exercise by the authors.

Findings:

There are multiple challenges to implementation of a Hospital GIS for infection prevention, mainly involving the domains of technology, organization, and adaptation. Use of a transdisciplinary approach can address many of these challenges. More research, specifically prospective, reproducible clinical trials, needs to be done to better assess the potential impact and effectiveness of a Hospital GIS in real world settings.

Originality/value:

This review highlights a powerful but rarely used tool for infection prevention within the hospital. Given the importance of reducing hospital acquired infection rates, it is vital to identify relevant methods from other fields that could be translated to the field of hospital epidemiology.

1. Purpose

An estimated 1 in 31 hospitalized patients has a hospital-acquired infection (HAI) at any given time (1). Many of these infections are caused by Clostridium difficile and multi-drug resistant organisms (MDROs) including methicillin resistant Staphylococcus aureus (MRSA) and resistant Gram-negative infections. These organisms can be transmitted among patients, between health care workers and patients, and between the hospital environment and patients. As concern rises about extremely drug resistant pathogens and there continue to be reports of high-profile outbreaks within hospitals, it is of vital importance that health care workers have appropriate tools to track infections and identify high risk areas of the hospital so these can be targeted for interventions.

There have been recent improvements in HAI rates. Reports show a decrease in rates of surgical site infections (SSIs) and catheter associated urinary tract infections (CAUTIs). Despite this, the incidence of transmissible infections such as Clostridium difficile infection (CDI) has not declined over time (2). This may be due to the more complex nature of transmissible infections within the hospital environment. The primary prevention strategy for CAUTI is removing the catheter when no longer needed. SSI is more complex, involving patient factors as well as surgical factors; however, process measures (e.g., appropriate perioperative antibiotics and glycemic control) can reduce risk (3). In comparison, CDI and MRDO prevention requires consideration of hospital-wide antibiotic use, environmental contamination within the hospital, person to person transmission within the hospital as well as in the community, and patient level risk factors (4). In this complex environment, hospital infection preventionists monitor CDIs and MDROs by reacting to line-lists of cases and looking at trends over time. There is rarely an opportunity to integrate all the potential risk factors into one system to determine what intervention might be most appropriate. In addition, especially in large hospital systems, the work of infection preventionists is divided into regions, so there is rarely an opportunity to see transmission across the whole medical center.

Epidemiologists have addressed the issue of complex interacting factors in disease transmission with many methods, but one of the most powerful is geographic information systems (GIS). In the most basic sense, a GIS is a system that stores, analyses, and displays geographic data (5). The recognition that geographic data is important in investigating infectious disease transmission dates back to John Snow’s work with cholera in the mid 1800s (6). So, when commercial, computer-based GIS applications came into use in the 1970s, (7) they were quickly and broadly adopted for studying infectious disease transmission. The World Health Organization as well as the Centers for Disease Control and Prevention (CDC) host numerous public-facing GIS applications. GIS have been become indispensable in studying infectious disease spread and transmission in the field of public health.

Despite this, it is rare to see these methods used within the hospital setting to study infection transmission. For the purpose of this discussion, we conceptualize the structure of a theoretical Hospital GIS using the model of current interactive dashboards from major public health agencies such as the CDC’s Fluview (8). Using this as a model, a Hospital GIS would include floor plans of the hospital, location and timing of HAIs, characteristics of patients that might affect susceptibility to infection (e.g., antibiotic use), characteristics of the rooms that might affect transmission risk (e.g., size, occupancy), and statistical models that could identify areas of high risk. These would be combined and displayed in an interactive application for infection preventionists to use for routine surveillance, identification of areas to target interventions (e.g., additional environmental cleaning), and outbreak investigation (see Figure 1). For example, an infection preventionist might log into the system to see if there are any geographical clusters of infections. Space-time clusters are identified using statistical models run daily on electronic health record (EHR) data. If clusters exist, the infection preventionists could use the application to initiate an investigation into whether there is evidence of in-hospital transmission. The current workflow at many institutions includes chart review on the case-patients to try to establish a connection between the cases. This can be time-consuming and burdensome. Using the geographic layout of the hospital present in a Hospital GIS application, the infection preventionist could investigate transmission between multiple patients at the same time (e.g., were they in the same unit at overlapping time periods, or were they on the same service?), without having to open separate charts. Depending on the sophistication of the tool, they might be able to track shared equipment or health care provider movements. The tool could also be used for infections that can be caused by lapses in infection control processes or structural issues (CAUTI, SSI). It would provide a geographic distribution of these cases, potentially highlighting areas that may not be adhering to standard recommendations. See Figure 1 for an overview of the proposed structure of a Hospital GIS.

Figure 1:

Conceptual model of a Hospital GIS (EHR: electronic health record; CAD: computer-aided drafting; GIS: geographic information systems).

Some initial steps have been taken to develop a Hospital GIS by our research team and others. As such, the purpose of this paper is to review the current state of spatial analysis for analyzing infections within the hospital, evaluate the potential opportunities and challenges for implementation of a GIS tool for infection control, and outline a course for future work in this area. Because of the concern that new health care technologies often fail to be widely adopted, we use a framework to try to better assess the complexities of adopting a Hospital GIS.

2. Design/methodology/approach

This paper includes a targeted literature review, an assessment of the complexity of implementing a hospital-based GIS, and a summary of the opportunities and challenges involved. The goal of the targeted literature review was not to systematically assess the literature, but instead to identify recent relevant work in this area to engage a discussion on the potential benefits and challenges of this technology. As such, the review targets the use of GIS or spatial modeling for HAIs in the medical literature. Search terms used were “hospital GIS”, (“GIS” and [“hospital acquired infection” or “hospital infection control” or “outbreak”]), (“geographic information systems” or “spatial”) and “hospital” and “infection”. Both PubMed and Google Scholar were searched for relevant articles. Additional articles were identified from references within these articles. Articles were screened by title. They were included only if they were published from 2000 through 2018 and included original research on spatial modeling or geographic visualization of infection related events. Reviews were excluded.

The complexity of Hospital GIS implementation is assessed using the nonadoption, abandonment, scale-up, spread, and sustainability (NASSS) framework (9). This framework was published by Greenhalgh and colleagues in 2017, using a systematic review and case studies of health care technology implementation. The NASSS framework was chosen because it focuses on outcomes that are common in health information technology (IT) such as sustainability and adaptation over time. In addition, the authors of the NASSS specifically state that it was not created as a guide for implementation, but to help with conversations about potential challenges and ideas, which is a goal of this paper. A recent paper assessed the implementation of a telemonitoring intervention used the NASSS framework. The authors used the framework to structure the discussion of challenges and concluded it would have been particularly useful prior to implementation (10).

The NASSS framework has seven domains: the illness, the technology, the value proposition, the adopter system, the organization, the wider context, and embedding and adaptation over time. Within each of the seven domains of the framework, we estimated whether implementation based on that domain would be simple, complicated, or complex. Based on the framework, a simple classification implies there are few components or issues and they are predictable and straightforward. Complicated implies there are many components or issues that interact, but in a predictable way. A complex classification is chosen if issues are unpredictable, dynamic, and multiple. Based on the theoretical model of a Hospital GIS outlined above, both authors assessed each domain and came to consensus on the final classification. When a domain was dependent on a specific organization’s characteristics, we discussed likely barriers and challenges, but did not make a specific assessment

The specific challenges to implementation were compiled by the authors based on the results of the NASSS assessment and their experience working with hospital stakeholders, GIS, and new health care technologies.

3. Findings

3.1. Targeted Literature Review

We developed our search terms to be broad, and over 600 articles resulted from our search in PubMed. The Google scholar search resulted in over 1,000 results. Not surprisingly, many articles reported on traditional GIS outside of the hospital system. One recently published article known to the authors (Murray) did not show up in our searches; however, it was included as it fit the inclusion criteria. As such, our search resulted in four articles, which we discuss below.

3.1.1. Using GIS for infection prevention

A study by Kho and colleagues from 2006 describes a prototype dashboard that used animated GIS to capture staff movement and the placement of patients with MRSA (11). The prototype was implemented for three months. During the study period, the visualizations in the prototype showed inappropriate placement of patients with MRSA in shared rooms, as well as an episode in which one health care worker checked vital signs on nine patients within 33 minutes, raising a concern for inadequate hand hygiene between patients. The authors concluded this was a potentially useful tool in order to track lapses in infection prevention processes, but outlined usability issues related to the time it took to review images.

Murray and colleagues identified hospital-related risk factors for CDI and used spatio-temporal analysis without geographic visualization (12). They used EHR data from all adult hospitalizations to examine the risk of CDI if a patient was placed in a room within 24 hours of a patient with CDI. They found that patients moved through a mean of 4.2 locations and that risk of transmission varied across locations. Specifically, they found that travelling to the computed tomographic scanner within 24 hours of a patient with CDI increased risk of subsequent CDI. This finding allowed them to investigate and identify potential lapses in cleaning in this area.

Kong and colleagues used a hierarchical Bayesian spatio-temporal model to study MRSA acquisition within a tertiary hospital (13). They found weak evidence of spatial clustering of MRSA, but were able to show that a room occupied by a patient with MRSA in the prior two weeks was a predictor of MRSA acquisition in that room.

Implementation of these kinds of analysis could highlight areas of concern for investigation, so that potential causes (e.g., inadequate cleaning, contaminated devices) could be discovered.

3.1.2. Using Hospital GIS for outbreak investigation

Kistemann and colleagues used GIS to study an outbreak of Salmonella enteritidis in a hospital in Germany (14). They developed a sophisticated Hospital GIS that included ward-level information, building and infrastructural data, recorded cases, spatial analysis, visualization, and mapping. They concluded the GIS helped them to determine the source of the outbreak (contaminated vanilla pudding served at lunch) and facilitated communication of their findings. The authors suggested this approach would be more useful if the GIS infrastructure had been developed ahead of time instead of for the purpose of the outbreak investigation.

3.2. Analysis of research gaps

The medical literature has only a few reports of GIS use for infection control and hospital outbreak investigation. All the studies reviewed suggest that this method can be useful, but none report on its integration as a standard tool for infection control. The study by Kho and colleagues is the only one to develop a tool geared toward routine use by infection preventionists. None of the articles discuss the user interface in-depth or include feedback from potential users. Finally, none of the reports employ prospective use of the tool or test its impact on reducing HAIs.

There are some clear areas of opportunity based on this review. There is a need to develop a scalable framework for spatial analysis within the hospital. Most large institutions have architectural blueprints (computer-aided drafting or CAD files) for their buildings, which can be processed and imported into a GIS. However, transforming these types of files into geographic files is time-consuming and requires expertise. Likewise, transforming raw EHR data into data that can be easily analyzed using geospatial techniques requires clinical and informatics expertise. Developing a user interface that is both useful and usable is also challenging and requires understanding of usability heuristics and content expertise of the needed tasks. There is a clearly a need for transdisciplinary teams to develop these tools and address these challenges.

Our own research group has taken the first steps in developing a Hospital GIS. First, we developed a functional prototype dashboard that uses hospital floor plans to show where CDI cases have occurred (15). We also examined how EHR data can be used to model risk associated with movement of patients around the hospital. We used EHR data to examine the risk of CDI based on the number of intra-hospital transfers (16). In a retrospective study, we combined admission-discharge-transfer data, limited clinical data, and case dates and locations of hospital-acquired CDI. Using multivariate logistic regression accounting for comorbidity and length of stay, we found that each additional intra-hospital transfer increases the odds of hospital-onset CDI by approximately 7%. These kinds of models, along with those of Murray and Kong, discussed above, can be used to identify patients or areas of the hospital at higher risk of infection transmission. These areas can then be targeted for additional cleaning or closer surveillance.

3.3. Assessment of complexity using NASSS (9) framework

The assessment of complexity is illustrated in Table 1. This assessment is focused on the development and adoption of a Hospital GIS for infection control, as described above, specifically for identification of HAI high-risk areas, outbreak detection, and investigation. The users are assumed to be infection preventionists with the user interface designed for this user type.

Table 1:

Assessment of the complexity of implementing a routine Hospital GIS interface

NASSS Domains and questions		Explanation	Overall Assessment
Domain 1: Condition or illness	What is the nature of the condition or illness?	• Simple: HAIs are well-characterized with some factors that are clearly predictable. • Complicated: There are a small number of outcomes, which could make modeling difficult.	Complicated
	What are the relevant socio-cultural factors and comorbidities?	• Simple: HAIs are not significantly affected by socio-cultural factors within the hospital. • Simple: HAIs are associated with well-described comorbidities and risk factors (e.g., antibiotic use, urinary catheter use).	Simple
Domain 2: Technology	What are the key features of the technology?	• Complicated: Key features of GIS have been developed for public health use, but they require reworking for interoperability with clinical systems.	Complicated
	What kind of knowledge does the technology bring into play?	• Simple: It displays the spatio-temporal distribution of HAIs in the hospital.	Simple
	What knowledge and/or support needed to use technology?	• Simple: The users are experts in outbreak investigation. • Complicated: The user interface would be new to users and most have not used a GIS application before. Detailed instructions and some training likely needed for optimal use.	Complicated
	What is the technology supply model?	• Simple: The software could be built using standard data models used at many institutions. • Complicated: Because of the need for institution-specific CAD files or maps, an application would likely require significant customization at a new institution.	Complicated
Domain 3: Value proposition	What is the business case for the technology?	• Simple: The hospital could have direct cost savings with reductions in HAIs. • Simple: The software could reduce the time spent on outbreak investigation.	Simple
	What is the desirability, efficacy, and cost effectiveness?	• Complicated: Unknown; this needs to be evaluated in a research study.	Complicated
Domain 4: Adopter system	What changes in staff roles, practices and identities are implied?	• Simple: This would not alter the role of the infection preventionist.	Simple
	What is expected of the patient/caregivers?	• Simple: The users for this tool would be hospital personnel working in quality and safety. Nothing would be expected of the patient or caregiver.	Simple
Domain 5: Organization	What is the organization’s capacity to innovate?	• This depends on the resources and time given to staff for innovation.	Site dependent
	How ready is the organization for this technology-supported change?	• This depends on the support for new technology among infection preventionists and leadership.	Site dependent
	How easy will the adoption and funding situation be?	• This depends on the interoperability of current information system with new software.	Site dependent
	What changes will be needed in team interactions and routines?	• This depends on the current organization of the infection prevention team. • Potential barriers may exist in changes in workflow as a novel tool is added.	Site dependent
	What work is involved in implementation and who will do it?	• This depends on the effort required to engage staff and monitor impact. • Potential barriers may exist depending on the presence of local experts in information technology and geographic information systems.	Site dependent
Domain 6: Wider context	What is the political, economic, regulatory, professional, and socio-cultural context for program rollout?	• Simple: HAI prevention is a national priority with incentives for reduction. • Simple: The spread of multi-drug resistant organisms in hospitals is a growing public concern.	Simple
Domain 7: Embedding and adaptation over time	How much scope is there for adapting and coevolving the technology over time?	• There are potential barriers to adapting if no local experts exist. • There are potential barriers to expanding the technology across institutions. • There are potential barriers to adapting the technology as other information systems change and evolve.	Site dependent
	How resilient is the organization to handling critical events and adapting to unforeseen eventualities?	• This depends on the organization’s ongoing support for the software. • There is a high risk of loss of support if the implementation is too complex.	Site dependent

3.4. Challenges to the implementation of a Hospital GIS

Based on the assessment of complexity, the challenges fall primarily in the domain of technology as well as organization and adaptation over time, although the latter two are site-dependent. Transmissible HAIs as a condition are relatively simple to identify and for the most part are not significantly affected by socio-cultural factors. These outcomes are regularly tracked within institutions and the definitions are standardized nationally. The one complicating factor for using spatial analysis with HAIs is that these are relatively rare events within the hospital. However, new Bayesian spatial methods have been developed to cope with high levels of uncertainty in statistical models based on small numbers (17).

In the technology domain, the complications are mainly due to a lack of plug-and-play GIS applications currently available for the hospital setting. New applications would have to be developed and tested. Environments and building structures are unique to every institution, so a software application would need to easily ingest CAD or other geographic files to develop the proper data visualizations. Depending on the complexity of automating this process, local GIS experts may be needed for implementation.

The value of this technology needs to be studied before formal cost analyses can be done. However, reduction in HAIs is already clearly linked to financial incentives, making the business case a clear one.

The area with the fewest barriers to implementation is the adopter system. The hospital has staff charged with infection prevention who are already using simple information systems to conduct surveillance and outbreak investigation. They are engaged and knowledgeable stakeholders. The addition of a Hospital GIS would not involve patients or caregivers directly and would not substantially change the role of the infection preventionist. Rather, it would provide a new approach to HAI investigation, which could reduce the time to detection and subsequent infection rates.

The complexities from an organizational standpoint are difficult to assess in a generalized way as each institution has its own unique challenges. However, the health care system is generally slow to accept new technologies. The specific organizational challenges would likely include: 1) limited information technology (IT) resources available for new projects; 2) concerns about data security given the incorporation of protected health information (PHI); and 3) concerns about disrupting staff workflow with a new tool.

From a wider context perspective, HAI reduction is a clear national priority, and using a tool such as GIS – which has been well-studied and widely adopted in public health to address this challenge – is a straightforward next step.

Finally, the biggest challenge to a Hospital GIS is creating a tool that can adapt and evolve over time. Especially with the ever-evolving EHR systems, creating a tool that is resilient to these changes is a significant barrier.

There are challenges to implementation, but each of these challenges has a potential solution based on an understanding of current GIS technology. In Table 2, we list each challenge and propose a solution.

Table 2:

Challenges to Hospital GIS implementation and Proposed Solutions

Identified challenges	Proposed solutions
Rare outcomes can complicate statistical modeling	• Use spatial Bayesian methods, which have been developed for small sample sizes
No current plug-and-play GIS software for use in a hospital	• Develop and test a new application with a focus on replicability • Use novel geographic techniques (18), which can extract features directly from CAD files • Develop methods for automated CAD file processing to geographic files
Unknown value to an institution	• Develop and test a Hospital GIS application in a trial setting with a focus on:  ∘ Outcome (i.e., HAI rates)  ∘ Usability  ∘ Cost effectiveness
Data security concerns	• Avoid unnecessary PHI in the application • Use standard security precautions and maintain the application behind the institutional firewall
Limited IT resources	• Create a ready-to-use application using common data models to make implementation easier from an IT perspective
Disrupting staff workflow	• Include relevant stakeholders prior to implementation • Complete formal usability and workflow analysis testing
Adapting over time to changes in linked information systems	• Use accepted standards and data models • Use of open source software so modifications can be ongoing • Encourage good communication between infection preventionists and IT when a change is noted

3.5. Limitations

There are several limitations to this commentary and review. The literature review was not systematic and focused only on reports of GIS used to address hospital infections in the medical literature. Without formal terminology for this type of application, we may have missed relevant publications. In addition, there are a few examples present in the geography and informatics literature (19, 20), but these focus more on feasibility, methodology, and simulation rather than real-life development and implementation, on which this review is focused. It is also possible, as with much health care technology, that local efforts to develop a Hospital GIS are proceeding in the operational space without formal publications to track them.

The assessment of complexity is inherently subjective. We used a published framework to try to avoid undue bias, but it is likely that other stakeholders would have a different assessment based on their experience in other health care settings. However, this commentary is intended to start the discussion about challenges and potential opportunities, not provide a finalized assessment. Finally, we outlined a set of challenges, but these will evolve as there are more studies on this technology and experience with the tools in the hospital setting.

3.6. Future directions

There are clear avenues forward for adapting GIS for use in the hospital for infection control. First, we need to develop an application that is robust, usable, and can be used prospectively, incorporating data as it is collected and analyzing it in real time. Next, the tool will need to undergo formal evaluation, both for efficacy and usability purposes. Finally, we need to evaluate how best to implement the technologies across different institutions. Accomplishing these steps will require close collaboration of a transdisciplinary group of researchers and hospital personnel including IT, clinical informaticians, geographers, infection preventionists, and human factors researchers.

4. Originality/value

The toolbox of GIS methods has been used widely in public health since its development in the 1970s. Over the past decade, interactive applications developed by public health agencies have become ubiquitous, allowing non-geographers to better understand the distribution of diseases and risk factors across environments. Hospital epidemiologists have not had the same access to these powerful tools, and we suggest it is time for this technology to be formally adopted in the hospital environment. We believe the incorporation of routine Hospital GIS into the workflow of infection preventionists will have a profound effect on our understanding of in-hospital disease transmission and the impact of HAI risk factors within the hospital.

Biographies

Authors Biographies:

Courtney Hebert MD, MS, The Ohio State University, United States: Dr. Hebert is a practicing physician in the field of infectious diseases and an Assistant Professor in the Department of Biomedical Informatics. Her research focuses on the secondary use of electronic health record data to better treat and prevent infections.

Elisabeth Dowling Root, PhD, The Ohio State University, United States: Dr. Root is a health geographer and Associate Professor in the Department of Geography and Division of Epidemiology. Her work applies spatial statistics and Geographic Information Systems to examine socio-environmental drivers of health in hospital and community settings.