Sepsis, defined as “life-threatening organ dysfunction caused by a dysregulated host response to infection” [1], is estimated to affect 50 million people each year worldwide, of which more than 40% are among children younger than 5 years [2]. Despite advances in vaccines, antibiotics, and acute care, sepsis remains the leading cause of death from infection [1]. About 20% of patients with sepsis die, with global estimates recording 11 million deaths due to sepsis in 2017 alone [2]. Surviving sepsis is associated with increased mortality across an individual’s lifespan and significant reductions in quality of life, including higher rates of chronic illness, physical disability, cognitive impairment, and mental health issues [3-9]. Additionally, sepsis treatment is extremely expensive [10]. In the United States it has been listed as the most expensive condition in US hospitals (>US $20 billion annually) [11]. The World Health Organization has declared sepsis a global medical emergency and adopted a resolution in 2017 to reduce the global burden of sepsis by improving sepsis diagnosis, treatment, and management [12].

Early detection of sepsis allows for prompt treatment, which is associated with reduced mortality and lower costs [13,14]. The 2016 Surviving Sepsis Campaign (SSC) guidelines, a set of clinical guidelines designed by a panel of international sepsis experts, strongly recommend treatment begin immediately, with the administration of intravenous antimicrobials within 1 hour of sepsis or septic shock recognition [15]. To enhance the effectiveness of rapid treatment, it is critical that septic patients are identified as early as possible [15-17]. Interventions such as regular monitoring of vital signs and elevated lactate levels can aid early recognition [18]. However, studies show that delays in both disease diagnosis and treatment are not uncommon in hospitals [19-21]. One of the main barriers to early sepsis diagnosis is the lack of effective diagnostic tools, further compounded by the fact that sepsis is a heterogeneous and enigmatic syndrome with no diagnostic gold standard [22]. Consequently, clinicians face a challenge in differentiating sepsis from other acute conditions with similar signs or symptoms.

Sepsis detection and recognition pathways can play an important role in facilitating early sepsis diagnosis and initiation of treatment. A number of sepsis-risk warning tools have been developed, for example, the Quick Sepsis-Related Organ Failure Assessment, the National Early Warning Score in the United Kingdom, and the Adult Sepsis Pathway in Australia [23-25]. Many hospitals currently rely upon paper-based sepsis recognition tools, which are susceptible to transcription and interpretation errors and highly reliant upon vigilant and timely patient monitoring by clinicians. In contrast, appropriately designed automated systems have the potential to decrease delays and increase the accuracy of sepsis detection [26].

Given the difficulties associated with timely sepsis recognition, health care institutions are increasingly leveraging clinical data captured in electronic health records (EHRs), which have been rolled out extensively in recent years around the world. EHR-based computerized clinical decision support (CCDS) systems, built into hospital electronic systems, present a great opportunity to facilitate sepsis early detection and prompt treatment. CCDS systems automate sepsis-risk warning tools to alert clinicians to the possible presence of sepsis [16,27], while reducing the mental load on clinicians and nurses [28,29]. Following an alert, a protocol is followed that usually involves a patient being evaluated by senior medical staff to confirm diagnosis and initiate the appropriate sepsis treatment [27].

Over the past 10 years, two distinct CCDS approaches to sepsis identification have emerged: knowledge-based electronic CCDS following predefined rules of an established diagnosis pathway, and nonknowledge-based CCDS utilizing artificial intelligence and machine learning techniques [28,30]. Our primary focus in this scoping review is the use of knowledge-based electronic CCDSs in sepsis detection.

Implementation of sepsis CCDS in hospital clinical information systems is a novel and rapidly expanding area [28]. Furthermore, the use of technological innovations in health care is rife with complexity [31]. This is particularly true for the use and evaluation of sepsis CCDS in real-world clinical settings. Sepsis itself is a complex and multifaceted condition, and various clinical criteria for sepsis early detection have been developed over the years and implemented in an evolving range of sepsis CCDS systems [32]. These systems have been evaluated in numerous clinical settings in hospitals, such as emergency departments, intensive care units (ICUs), and various wards [32,33]. In addition, there is extensive heterogeneity in the design of studies evaluating the effectiveness of these CCDS systems, such as differing outcome measures (eg, mortality, cost, and clinical workflow) and evaluation methods [28,32]. We intend to explore the complexity present in the use and evaluation of sepsis CCDS by systematically mapping the breadth of the literature in a scoping review to identify knowledge gaps and inform future research. The research question we have formulated for this review is as follows: What is the evidence base for the use of knowledge-based CCDS systems in hospitals for sepsis early detection and how have they been evaluated?

In this scoping review, we will aim to (1) scope the study contexts, designs, and research methods employed, (2) summarize study outcomes investigated and outcome measures utilized, and (3) map out the range of CCDS designs and implementation features, such as sepsis clinical criteria and related sepsis care and management protocols.

The methodology for conducting scoping reviews presented by the Joanna Briggs Institute (JBI) Reviewer’s Manual [34] and the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) [35] will be used and adapted as guides. In particular, the five-stage scoping review framework presented by Arksey and O’Malley [36] will be followed as recommended by JBI [35]: (1) identifying the research question, (2) identifying relevant studies, (3) selecting studies, (4) charting the data, and (5) summarizing and reporting the results.

A search for current reviews and protocols on this topic was undertaken on MEDLINE, the JBI database, and Google Scholar, confirming the absence of scoping reviews in this research area. Only one similar scoping review protocol was identified [37]; however, the authors are focusing only on the use of machine learning or artificial intelligence–based CCDS, whereas our review is instead focused on only knowledge-based (ie, nonmachine learning) CCDS.

Ethical approval or consent to participate is not required for this protocol and scoping review. The data will be extracted from published articles, and no individual information will be included.

The increasing digitization of health care has promoted the use of CCDS systems in hospitals for sepsis early detection and treatment [27]. However, there is poor consensus on the effectiveness of these systems in improving the health outcomes of patients with sepsis [28]. This can be partly attributed to the complex and heterogenic nature of studies investigating this topic [16,28]. Additionally, as the field is rapidly emerging, there are increasingly varied methods of CCDS evaluation seen in the literature. In this paper, we have presented a protocol for a scoping review based on well-established methodology, as explained in the Methods section. A strength of our review, and of scoping reviews in general, is their broad search strategy and eligibility criteria, which will allow us to scope the breadth of the field more comprehensively. To our knowledge, this scoping review is the first review to comprehensively map the breadth of the literature investigating the use of CCDS for the early detection of sepsis in hospitals. Mapping the literature will provide a broad outline of the current studies within the field, promoting a more comprehensive understanding of current research efforts. We will, therefore, be better able to identify research gaps, which in turn can inform future studies and clinical practice.

Due to feasibility and time constraints, we limited our search to studies published in English, or that had English translations readily available. Therefore, we must acknowledge that our scoping review may miss some studies published only in non-English languages.

The review will provide a comprehensive summary on the use of knowledge-based CCDS systems in the early detection of sepsis in hospitals, providing researchers, clinicians, policy makers, and developers with a relevant and important evidence base. Our findings will highlight research gaps and shortcomings in existing evaluations and implementations of sepsis CCDS and, hence, guide future research efforts. The results will be shared with key stakeholders in clinical care, research, policy, and patient advocacy to inform clinical practice and policy.