By Kelly M. Pyrek
Artificial intelligence could become a busy infection preventionist (IP)'s best friend and should be embraced where feasible to help reduce the burdens associated with traditional surveillance methodologies.
As Russo, et al. (2018) remind us, "Key to preventing and controlling healthcare-associated infections (HAIs) is their identification and reporting through standardized classification, which comprises a major component of surveillance. Surveillance is a fundamental tool to successful infection prevention and control programs. The purpose of HAI surveillance is to provide quality data that can act as an effective monitoring and alert system and reduce the incidence of preventable infections. Effective surveillance systems deliver information that can be used to inform decisions."
In a review by de Bruin, et al. (2014) the researchers found some studies indicated that surveillance rules are complex and open to subjective interpretation, causing considerable variability in manual surveillance results, both within and between healthcare institutes. They say that electronic surveillance systems lack variability as they consistently apply surveillance definitions; studies that directly compared electronic and manual surveillance performance showed that electronic surveillance achieves equal or better sensitivity than manual surveillance. They note, "Driven by the increased availability of electronic patient data, electronic HAI surveillance systems use more data, making systems more sensitive yet less specific, but also allow systems to be tailored to the needs of healthcare institutes’ surveillance programs."
Studies have demonstrated that automation and the use of electronic surveillance systems (ESS) result in improved data accuracy and sensitivity when compared to traditional methods. This is particularly important with the increasing requirements for public reporting of HCAIs and their use for performance measurement associated with financial penalties. It is also claimed that ESS can assist IP staff with surveillance by decreasing the burden of data collection, resulting in significant time savings.
In their review, Freeman, et al. (2013) found an emphasis was on the linkage of electronic databases to provide automated methods for monitoring infections in specific clinical settings. The studies they examined assessed the performance of their method with traditional surveillance methodologies or a manual reference method. Where sensitivity and specificity were calculated, these varied depending on the organism or condition being surveyed and the data sources employed. They concluded that, "The implementation of electronic surveillance was found to be feasible in many settings, with several systems fully integrated into hospital information systems and routine surveillance practices. The results of this review suggest that electronic surveillance systems should be developed to maximize the efficacy of abundant electronic data sources existing within hospitals."
Despite widespread availability, Hebden notes that the adoption of ESS is slow, and suggests that this may be due to a lack of understanding the barriers to implementing ESS. Grota, et al. (2010); Halpin, et al. (2011) and Masnick, et al. (2014) report that between 23 percent and 56 percent of facilities in the U.S. have ESS.
"Certainly by now, the vast majority of both acute-care and critical-access hospitals have acquired electronic medical records (EMR) systems as a result of the Health Information Technology for Economic and Clinical Health (HITECH) Act, and the Centers for Medicare and Medicaid Services (CMS) incentive programs for hospitals to transition from paper to digital patient-care systems," confirms Gwen Borlaug, MPH, CIC, FAPIC, an infection prevention consultant and former director of the HAI Prevention Program at the Wisconsin Division of Public Health. "But although EMRs aid in conducting HAI surveillance by delivering more comprehensive patient care data faster to the IP, there is more to electronic surveillance that just the use of EMRs. Electronic surveillance systems integrate patient data from multiple databases—medical records, laboratory results, pharmacy—and apply an electronic algorithm to determine whether a case definition for a particular HAI has been met, enabling more efficient surveillance."
Borlaug continues, "Many hospitals still do not have electronic surveillance systems, however, despite their use of EMRs. Why? Because, with the exception of electronic transmission of antibiotic use and resistance data, healthcare entities do not have to implement systems with electronic surveillance capacity to receive financial incentives from CMS. Thus, the challenge for facilities is justifying the additional cost of digital health information systems that provide electronic surveillance capacity and providing information technology (IT) support for maintaining and updating electronic algorithms and automatic data importation and exportation capacity. Most electronic health systems vendors do not provide that support, resulting in additional burdens for facility IT departments as they take on these tasks. On the other hand, the availability of electronic surveillance systems provides the opportunity for healthcare facilities to evaluate the costs of current labor-intensive manual surveillance methods and explore how electronic surveillance systems can lead to more efficient HAI surveillance and ultimately enhanced HAI prevention efforts."
Russo, et al. (2018) acknowledge the demands put on IPs by surveillance activities: "Surveillance is a cyclical process encompassing recognition of an event, data collection, analysis, interpretation and dissemination. Surveillance of HAIs is a highly resource-intensive activity, with ‘traditional’ surveillance methods involving the infection prevention (IP) staff in exclusively manual data collection processes that are time-consuming, resource intensive, and generate data of variable quality. It has been reported that up to 45 percent of IP staff time is dedicated to undertaking surveillance. A recent Australian study identified that IP staff spend 36 percent of their time on surveillance. The proportion of overall surveillance time spent on data collection has not been identified. It is also suggested that the annual staffing costs of IP staff (nursing component) in Australian hospitals could be AU$100 million. This represents significant resources devoted to HAI surveillance."
Russo, et al. (2018) conducted a systematic review on the impact of electronic surveillance software (ESS) on infection preventionist (IP) resources, finding 13 studies that demonstrated a reduction in IP staff time to undertake surveillance-related activities. The reduction proportion ranged from 12.5 percent to 98.4 percent (mean: 73.9 percent).
The researchers explain, "Of the 16 studies, estimates in the reduction in time spent on surveillance were calculated for 13. Six of these 13 studies reported a reduction in time of ≥90 percent, and all but two suggested a reduction of >50 percent. The range of reduction identified varied from 12.5 percent to 98.4 percent (mean: 73.9 percent). Two studies demonstrated a time reduction of 98.4 percent. Six studies supplied data to compare the proportion of overall time pre- and post-ES; of these, four were statistically significant.
The researchers clarify, "Half of the studies identified were conducted in the U.S., possibly reflecting a more advanced implementation of electronic medical records and automated data retrieval. Despite the availability of commercial surveillance products, it is interesting to note the common use of in-house systems. Grota, et al.’s large cross-sectional study reported that the split between in-house and commercial products was almost half, and this is similar to the findings from this review. This may be explained by an iterative in-house development process over many years within a hospital, favoring a local product over an external commercial product. Alternatively, it may reflect the heterogeneity of existing hospital information technology systems, making implementation of an off-the-shelf ESS product that requires modification at each site less attractive and less feasible.
Russo, et al. (2018) emphasize that "In an environment of public reporting, financial penalties and increasing demands on infection prevention resources, ESS has the advantage of minimizing the subjective nature of many manual surveillance activities that lead to variable accuracy, making them ideal for large scale implementation. Furthermore, combining the time benefits this review suggests that, with the recent demonstrated advantages of surveillance algorithms that can be built into ESS, the case for implementing an ESS is strengthened. Many infection prevention interventions fail due to a lack of careful implementation planning and associated resourcing. The planning for the introduction of an ESS would require a comprehensive implementation strategy to ensure uptake …Although the studies included in this review demonstrate that the use of ESS requires less IP time to undertake the same surveillance, it is important to understand that this may not equate to IP teams spending less time on surveillance overall. Time efficiencies identified through the use of ESS could potentially enhance the identification and response to possible outbreaks, or indeed extend the scope of surveillance activities, thereby improving the quality of HCAI surveillance undertaken. In an era of gradual transition to electronic medical records, it is imperative that data technology is in alignment with the requirements of IP, and there is scope for further research in this area. To be convinced of the benefits of ESS, we suggest that research is required in several areas: the effect of ESS on infection prevention resource as a primary outcome; how IP staff redirect their resources following the introduction of ESS, and whether this has any impact on infection and/or patient outcomes; and the influence of ESS on HCAI rates."
Hebden (2015) explored the case for automated infection prevention surveillance, observing that, "The slow adoption of automated technology seems surprising in light of the time-consuming nature of manual surveillance and nearly three decades having passed since the first report detailing a computer system using microbiology data to identify patients with possible HAIs. In recent years, supportive evidence for the use of automated surveillance includes a 61 percent reduction in time spent on surveillance activities, achievement of greater depth in the implementation of evidence-based infection control practices, and improvement in implementation of isolation practices."
She continues, "Proponents of automated surveillance technology cite standardization of IP workflow and consistent and accurate case finding as a solution to potentially biased manual processes. A recent qualitative analysis of the evolving role of IPs concluded that data standardization would lessen the tension that IPs experience due to expanding responsibilities that outstrip resources. The authors noted that streamlining surveillance would aid IPs in achieving a better balance with competing priorities. An equally important advantage to automated surveillance is the capture of discrete data elements from the electronic medical record that prevents the reliability problems inherent in traditional surveillance based on human interpretation."
Many practitioners still engage in “shoe leather epidemiology," and even in the age of automation, Borlaug observes, "IPs must remain skilled and knowledgeable 'detectives,' that is, they must walk the floors, see with their own eyes how patients are cared for, observe the environment of care, and evaluate infection prevention practices on an ongoing basis."
Hebden (2015) alludes to the tension that exists between old-school epidemiology and the use of ESS for IPs that are caught in the middle: "Although IPs are familiar with the tasks associated with traditional manual surveillance—reviewing microbiology reports and using the electronic medical record to obtain additional information for decision making—many are unfamiliar with the tasks of data retrieval and management inherent to automated systems and must adapt their workflow to a different way of doing things. In a recent report examining IPs' awareness of and engagement in health information exchange to improve public health surveillance, <20 percent of respondents with access to an electronic health record reported being involved in the design, selection, or implementation of the system. The authors concluded that these findings may limit an IP's ability to influence or utilize key information technologies to facilitate transition to paperless surveillance processes. Further, an essential but unfamiliar task when using an automated system is the need for data validation that must be performed at start-up and whenever upgrades or changes are made to the foundational databases. Validation is necessary to ensure that the received data are complete and accurate and establish trust in the system. Most programs in use currently by IPs are semiautomated surveillance systems that deliver user alerts based on large inputs of data and require additional tasks for HAI classification. When multiple IPs are using the system, it is imperative that the surveillance process be standardized to ensure that each user can easily identify which tasks need to be completed to avoid duplication of work."
Automated HAI and infectious disease surveillance does not take the place of 'shoe leather' epidemiology, but rather can help reduce the time and effort needed to conduct routine surveillance, allowing more time for the IP to investigate, observe, and interact with facility staff and colleagues, thus enhancing HAI prevention efforts and outbreak detection capacity," says Borlaug.
To that end, IPs should possess certain skills relating to surveillance for infection that don’t change, regardless of paper or digital facilitation. "IPs should have a basic knowledge of microbiology, infectious disease epidemiology, and the infectious disease process," Borlaug advises. "They must also be able to conduct systematic and reliable HAI data collection and be able to summarize, analyze, interpret, validate, and present HAI surveillance data to their colleagues in meaningful ways. And whether HAI surveillance is conducted manually or electronically, the IP must have a thorough knowledge and understanding of HAI surveillance definitions, because by either method, the IP ultimately has the last word in determining whether an HAI is present."
Hebden (2015) notes, "As described in the Association for Professionals in Infection Control and Epidemiology competency model, professional and practice standards assume that IPs will have access to information technology hardware and some degree of experience in the use of software applications. However, it is recognized that the IPs performing surveillance will have varying degrees of competency and it is proficient IPs who can integrate both manual and electronic findings for comprehensive reporting and expert IPs who can apply principles of information management to emerging technology. Before implementing the system, the experience and skill set of each IP should be assessed because one or more members of the team may be at the proficient or above competency level and could be identified as 'superusers' to assist the other team members with the workflow transition. The team also needs to discuss who has the authority to make decisions regarding system design and configuration; for example, modifications to the data feeds, revision of user alerts, and report development."
Despite competing priorities, cost-cutting and other challenges faced by front-line clinicians and IPs, surveillance remains the key intervention relating to the identification, management and prevention of outbreaks.
"In many ways, modern technology such as molecular laboratory diagnostic methods that provide rapid, nonculture-dependent results have greatly enhanced early detection and prompt response to outbreaks and incidents of infectious disease transmission in healthcare facilities," says Borlaug. "I think one of the main barriers to effective outbreak response, however, is that many IPs do not possess the tools and resources available to conduct a systematic, methodical investigation of outbreaks and incidents of infectious disease transmission in their facilities. To bridge that gap, I recommend a review of the Centers for Disease Control and Prevention web-based course, “Investigating an Outbreak," [https://www.cdc.gov/ophss/csels/dsepd/ss1978/lesson6/section2.html] which describes a step-by-step outbreak investigation process and provides tools IPs should have available to maintain outbreak response readiness."
Borlaug shares her personal experiences with surveillance as she matured in her work as an IP. "In reflecting on my early years as an infection control practitioner, as we were called back then, I realize my initial approach to HAI surveillance was to conduct it as a soloist, without discerning the important roles of the laboratorians, pharmacists, physicians, nurses, IT staff, and other colleagues in the facility," she says. "Of course, this was a big mistake, and I quickly learned that these individuals are key players in conducting HAI surveillance. In the milieu of more complex surveillance definitions and methods, reporting burdens and evolving technology, it is increasingly imperative for IPs to use a team approach to HAI surveillance, through education of healthcare personnel and providers regarding HAI surveillance processes and case definitions so that they have an understanding of how their practices (e.g., ordering of cultures, medical records documentation), affect HAI case detection. Keeping up with rapidly evolving laboratory and information management technology is also a challenge, and IPs should not hesitate to reach out to these key partners for assistance in learning and implementing current technology tools."
Borlaug continues, "Finally, as I conducted HAI surveillance data validation exercises among dozens of acute care and critical access hospitals, occasional flaws in electronic surveillance algorithms were revealed, hence implementation of electronic surveillance methods requires active engagement, oversight, and ongoing evaluation and validation by the IP to ensure the accuracy and reliability of electronic methods. And whether electronic or manual surveillance methods are employed, both internal (conducted by facility staff) and external (conducted by an outside agency such as public health departments or independent IP consultants) data validation exercises can help detect systematic surveillance errors and improve data reliability."
Looking toward the future of surveillance, Hebden (2015) summarizes, "Although automated surveillance technology has been evolving for decades, adoption of these technologies is in a nascent state. The current trajectory of public reporting, continued emergence of multidrug-resistant organisms, and mandated antimicrobial stewardship initiatives will result in an increased surveillance workload for IPs. The use of traditional surveillance methods will be inefficient in meeting the demands for more data and are potentially flawed by subjective interpretation. An examination has been offered the slow adoption of automated surveillance technology from a system perspective with the inherent ambiguities that may operate within the IP work structure. Formal qualitative research is needed to assess the human factors associated with lack of acceptance of automated surveillance systems. Identification of these factors will allow the NHSN and professional organizations to offer educational programs and mentoring to the IP community that target knowledge deficits and the embedded culture that embraces the status quo. With the current focus on fully electronic surveillance systems that perform surveillance in its entirety without case review, effective use of the data will be dependent on IPs' skills and their understanding of the strengths and limitations of output from algorithmic detection models."
As Freeman, et al. (2013) observe, "Automated methods for the identification of HAI allow the consistent application of simplified definitions designed for the purpose of surveillance. ESS should be seen as an opportunity to enhance current surveillance practices. Staff involved in surveillance activities should not feel threatened by advances in this area but should recognize that these methods can reduce the burdens associated with traditional surveillance methodologies, which will only increase as the emphasis on transparency and public reporting causes increased demand for more information to be reported."
de Bruin JS, Seeling W and Schuh C. Data use and effectiveness in electronic surveillance of healthcare-associated infections in the 21st century: a systematic review. J Am Med Inform Assoc, 21; Pp. 942-951. 2014.
Freeman R, Moore LSP, García Álvarez L, Charlett A and Holmes A. Advances in electronic surveillance for healthcare-associated infections in the 21st century: a systematic review. J Hosp Infect, 84; Pp. 106-119. 2013.
Grota PG, Stone PW, Jordan S, Pogorzelska M and Larson E. Electronic surveillance systems in infection prevention: organizational support, program characteristics, and user satisfaction. Am J Infect Control, 38 (2010), pp. 509-514
Halpin H, Shortell SM, Milstein A and Vanneman M. Hospital adoption of automated surveillance technology and the implementation of infection prevention and control programs. Am J Infect Control, 39 (2011), pp. 270-276.
Hebden JN. Slow adoption of automated infection prevention surveillance: are human factors contributing? Am J Infect Control, 43 (2015), pp. 559-562
Masnick M, Morgan DJ, Wright MO, Lin MY, Pineles L and Harris AD.SHEA Research Network. Survey of infection prevention informatics use and practitioner satisfaction in US hospitals. Infect Control Hosp Epidemiol, 35 (2014), pp. 891-893
Russo PL, Shaban RZ, Macbeth D, Carter A and Mitchell GB. Impact of electronic healthcare-associated infection surveillance software on infection prevention resources: a systematic review of the literature. Journal of Hospital Infection. Vol. 99, No. 1. Pages 1-7. May 2018.
Atreja A, et al. Opportunities and challenges in utilizing electronic health records for infection surveillance, prevention, and control. Am J Infect Control, 36 (3 Suppl) (2008), pp. S37-S46.
Hebden JN, et al. Leveraging surveillance technology to benefit the practice and profession of infection control. Am J Infect Control, 36 (2008), pp. S7-S11.
Keller SC, et al. Variations in identification of healthcare-associated infections. Infect Control Hosp Epidemiol, 34 (2013), pp. 678-686.
Klompas M and Yokoe DS. Automated surveillance of health care-associated infections. Clin Infect Dis, 48 (2009), pp. 1268-1275.
Lo YS, et al. Improving the work efficiency of healthcare-associated infection surveillance using electronic medical records. Comput Methods Programs Biomed, 117 (2014), pp. 351-359.
Mitchell BG, et al. Time spent by infection control professionals undertaking healthcare associated infection surveillance: a multi-centred cross sectional study. Infect Dis Health, 21 (2015), pp. 36-40.
Perl TM and Chaiwarth R. Surveillance: an overview. Eds: E. Lautenbach, K.F. Woeltje, P.N. Malani, Practical healthcare epidemiology (3rd ed.), University of Chicago Press, London (2010), pp. 111-142.
Stone PW, et al. Staffing and structure of infection prevention and control programs. Am J Infect Control, 37 (2009), pp. 351-357.