1 Introduction

Pressure injuries (PIs) are manifestations of sustained pressure loading causing damage to skin, tissue, muscles, and bone structures; associated with a constellation of risk factors that include both physiologic (e.g., microcirculation) as well as non-physiologic (e.g., age, limited sensation/mobility and urinary incontinence) characteristics affecting an individual [1].

PIs continue to be an important quality and safety issue for health systems both in Australia/New Zealand and across the globe [2-4]. The costs of these injuries to individuals are well documented in terms of increased financial burden and reduced quality of life [2, 5, 6]. For healthcare providers, PIs have a significant monetary impact and promote negative perceptions of the quality of health services. Generally, the development of PIs in healthcare is associated with increased costs for wound treatment, financial penalties on the provider and in some cases, legal action or litigation [7].

Despite increasing awareness of the aetiology of and socioeconomic impacts attributed to PIs, most remediation/prevention strategies rely on manual interventions that further burden already time-poor healthcare workers [8]. We now live in a time where machine learning/artificial intelligence, sensors and other technologies continue to evolve in capabilities as well as affording enhanced mainstream accessibility [9]. Today, many opportunities exist for the creation of solutions that use one or multiple technological components to address (and ultimately prevent) the incidence of PIs. However, to date there has not been any publicly available research that maps the available solutions that aim to not only manage, but prevent PIs from developing in individuals; especially those who are most vulnerable due to their mobility status, skin integrity, age or other factors. These ‘eHealth’ interventions can be broadly defined as:

“An emerging field in the intersection of medical informatics, public health and business, referring to health services and information delivered or enhanced through the Internet and related technologies. In a broader sense, the term characterises not only a technical development, but also a state-of-mind, a way of thinking, an attitude, and a commitment for networked, global thinking, to improve healthcare locally, regionally, and worldwide by using information and communication technology” [10].

New methodological tools such as Greenhalgh et al.'s Nonadoption, Abandonment, Scale-up, Spread and Sustainability (NASSS) framework enable researchers to evaluate any specific technological intervention for its potential challenges/utility in large-scale, sustained adoption as well as to understand the root cause(s) of previous attempts meeting limited success [11]. Understanding previous attempts at utilising eHealth technologies for PI prevention is essential to chart future steps towards a more comprehensive and effective solution to the persistent issue of PIs [12]. Therefore, the aim of this scoping review was to map the literature pertaining to the use of eHealth interventions to prevent PI in populations at risk of the complication, and describe the interventions encountered with the help of the NASSS framework.

2 Methods

Following the principles outlined by Arksey and O'Malley, the 5 standardised stages of a scoping review from specifying the research question to final summary, synthesis and reporting were structured using a teams-based approach [13, 14].

This scoping review was necessary to identify and synthesise the range of evidence within the field of eHealth interventions that focus on preventing PI; providing an early assessment of the available literature and whether the quantity of relevant articles could warrant a more detailed systematic review while simultaneously identifying gaps that can prevent redundant research from being performed inadvertently [14].

Covidence, a web-based collaboration software platform that streamlines the production of systematic and other literature reviews, was utilised to support the conduct of the review process [15].

Studies included in this review adhered to the following criteria:

One author conducted the literature search and extracted the title and abstract data into Covidence (PT). Three authors then screened the titles and abstracts for inclusion according to the criteria (PT, AR and AK). If differences arose, a third party judged the abstract for inclusion or exclusion (PT, AR or AK). Full texts of relevant studies were then obtained, and two authors screened these for relevance, with differences judged by a third party (PT, AR or AK). A data extraction template was formulated, and two reviewers independent of each other completed the data extraction (AH, AR, ER and AK). Authors of papers that contained missing data or unclear information were contacted.

3 Search Strategy

Databases included CINAHL, Medline, ScienceDirect, and the Cochrane library. Search terms were derived for Medline and adjusted for other databases accordingly. A full log of search terms used in this study is available in Appendix S4. No language or date limiter was applied, with studies from the inception of each database until the 29th of January 2024 included. Individual reference lists from suitable articles were also manually searched for relevant sources appropriate for the aims of this study.

4 Data Extraction

For ease of analysis and brevity of content, the seven preceding questions were considered the most relevant for this paper's discussion of included studies to evaluate the specific contexts and factors influencing the adoption of a given technology (See Table 2).

All of the 19 NASSS framework questions were answered during the data extraction stage and the full set of questions/domains was completed as per design. Appendix S2 lists the remaining questions that were less relevant for this paper's analysis with a brief rationale of why each one was omitted, while Appendix S3 shows the captured responses for every question within the NASSS framework in addition to the consensus-derived rationale for grading an intervention as recorded.

All data extracted are presented in tables before the evidence is further synthesised and summarised. The tables used to present findings in this paper mirror the raw data from the Covidence data extraction template. Where data (e.g., mean age) was unavailable during data extraction, these are denoted separately.

5 Data Summary, Synthesis and Reporting of Results

Following data extraction in Covidence, an overall narrative was constructed and reported in this paper. The findings from both individual studies and the larger patterns that emerged across the studies allowed the original intent of this paper (i.e., to outline the various eHealth interventions that can be encountered in the literature) to be met with better clarity through a new classification system that describes the types of technology commonly encountered during the literature search. The final synthesis of the extracted data into more discrete themes/categories was formed through consensus between AR and AA, which also contributes to the different perspectives that were included in the discussion section of this paper. The broader team's input was then used to cross-check the various elements of results and discussion that are included in this paper.

6 Results

A PRISMA flowchart was generated at the conclusion of data extraction and presented in Figure 1. From the initial search yielding n = 876 studies, article de-duplication and data screening resulted in n = 27 studies included for discussion within this paper. Of the n = 27 studies that were under review, Table 1 shows that 48%(n = 13) of included studies were composed of studies originating in the USA and approximately 15% (n = 4) of studies reported on studies conducted in Switzerland, with all other studies representing other countries including Canada, Israel, Argentina, South Korea, the Netherlands, Austria, and Brazil. The included studies had a publication date range from 1997 to 2023 and included diverse study designs encompassing experimental trials, qualitative designs, mixed-methods, cohort studies, and randomised control trials (including secondary analyses).

The NASSS framework formed a part of our data extraction process by providing validated questions that articulate different aspects of a given intervention to critique its long-term viability as a solution to a given issue of interest. After mapping NASSS domains to each included study (See Table 2), approximately 67% of included articles noted complicated or complex requirements on patients (i.e., patients are expected to perform/adhere to routine tasks to complete the intervention) while more than 75% of included articles involve the use of technologies that have a wide scope for adaptation and co-evolution over time.

The types of interventions encountered in this review required a simpler way of grouping similar interventions to aid in understanding the variety of eHealth technologies reported in the literature. To achieve this, we provide an overview in Table 3 that details the specific and/or overarching intervention types as well as technical features grouped from the interventions observed during data extraction. Thematic findings from Table 3 were used to help classify the eHealth interventions (i.e., software/communications-focused, clinical support-focused or bespoke software/hardware fusions). For example, a study investigating the use of a video telephone device to prevent PIs in 1999 carries a similar purpose to a study of telemedicine via video link to support PI management and prevention in 2014; thus, both of these interventions could be clustered together based on their emphasis on creating said video link to help prevent PI incidence or development.

The first type of solution class includes smartphone apps or web-based/telephone-based eHealth interventions that primarily aim to facilitate communication between the patient, their family, and the healthcare team [16-20]. This represents the largest proportion of eHealth interventions encountered in included studies (15/27). Since modern technology also allows for both synchronous and asynchronous communication, other eHealth interventions (within this class) encountered in our review investigated automated counselling and response systems as possible eHealth functions that could complement traditional hospital interventions for PI prevention [21, 22]. In the year 2000, a study featured a video camera device installed onto TV sets in the living room, thereby allowing health professionals to provide video consultations and conduct ongoing observations of a patient's health status all from the comfort of their home dwelling [23]. In 2014, other researchers trialled the use of email, telephone, and video links to support an enhanced multidisciplinary team with telehealth [24]. Other studies focused on apps providing patient education (i.e., acting as a repository or knowledge base for information that would previously only be acquired from visiting a general practitioner or attending a hospital visit). These studies focused on delivering health education via YouTube and detailing how the public accesses health information on these types of social media sites [25], computer-aided instruction to promote skin integrity/self-care for PI prevention [26, 27] and a mobile educational tool for PIs [28]. Another type of eHealth intervention in this class adopted a social-centric model by using communication features to link patient communities with healthcare professionals for collaboration in promoting better health outcomes and health literacy/empowerment through positive social pressure [29]. Lastly, we encountered eHealth interventions in this category connecting users directly to healthcare professionals in a more holistic way by combining multiple functions including smart camera, PI diary, knowledge repository, and reminder functionality [16, 17, 30].

The second type of eHealth intervention centers on knowledge and clinical support to drive a data-driven approach to optimal care. One such intervention explored the use of a simple ‘red, yellow, green’ traffic light system to alert healthcare workers to patients that are vulnerable to hospital-acquired infections [31]. Beyond this is an intervention that utilises continuous monitoring of patient pressure data to provide healthcare professionals and patients/caregivers with access to live readings of the pressure being exerted upon bed interfaces in order to visualise potentially dangerous build-up of pressure on bodily areas that are vulnerable to developing PIs [32]. Another paper investigated treatment plan support via clinical algorithms in 1997 as a system that supports nurses to develop personalised and evidence-based treatment plans for patients who have or are at-risk of developing PIs [33]. Health professional education is also a focus for eHealth apps in the form of a web-based training system [34], a guideline reference app for PI prevention [35] and a digital feedback system that tracks the healthcare team's alignment with incorporating the latest evidence-based guidelines to prevent PIs [36].

The third type of eHealth intervention seen in this review combines software with custom hardware configurations to create bespoke solutions. In contrast to the first and second eHealth classifications that usually depend on mainstream hardware owned by the end-user (e.g., a healthcare professional or patient with a smartphone), this third class of eHealth intervention utilises dedicated hardware, such as Raspberry Pi mini-computers, sensors, or bed mattress systems. For example, the Matrix 200 system was considered a novel approach to PI prevention among patients with spinal cord lesions [37]. By using a computer programmed to monitor and react to changes in weight distribution, the Matrix 200 adjusts an array of air cells on a mattress to redistribute pressure according to the sustained pressure measured at specified anatomical sites. Other researchers have also explored the development of sensors and pressure sensing mats that interface with a dedicated tablet or Raspberry Pi minicomputer for data collection [38-42]. These computer systems can be programmed to respond via alerts and reminders to encourage users vulnerable to PIs to perform protective behaviours (e.g., weight shifting, and/or otherwise avoid pressure build-up known to contribute to PI).

7 Discussion

This scoping review aimed to map current research evidence pertaining to the use of eHealth interventions to prevent PIs in vulnerable populations. A total of 27 studies were elucidated from the initial search of articles before proceeding with an analysis using the NASSS framework to inform our evaluation of each intervention. The NASSS framework will enable research teams to better plan for future intervention development by allowing for a detailed understanding of the factors needed to transition from experimental prototypes to more mainstream and scaled-up projects; thus, an eHealth intervention becomes sustainable when it can maintain its relevance over the long term. This point regarding sustainability over time not only represents the opportunity to take advantage of emerging technologies and capabilities to manifest more functionally powerful and diverse interventions, but also highlights the threat of non-adoption due to overly complicated interventions. More complicated eHealth interventions tend to be harder to implement and may also overburden patients with reminders or pressure them into medical adherence. In short, interventions with comprehensive and diverse functions can enable PI prevention through a multi-pronged approach but must be carefully designed with accessibility, ergonomics, and ease-of-use in mind for it to be successfully adopted by the greatest number of stakeholders.

Most of the studies we encountered focused on smartphone applications as interventions to prevent PI, usually incorporating some degree of web-based or telephony functions. This category of intervention encompasses a variety of functions allowing for rapid information sharing enabled through modern technologies, including the high-resolution cameras found in many smartphones/mobile devices today and widespread Internet accessibility. Whatsapp and similar Voice over Internet Protocol (VoIP) apps have become an attractive option for patients and the healthcare team to communicate and coordinate with one another either in real time (synchronously) or with a time lag (asynchronously); both modes have their respective advantages and disadvantages but ultimately serve the same goal of achieving better teamwork in the therapeutic alliance [20].

Fundamentally, the potential of using apps that are portable and follow the user presents opportunities for health professionals to deliver responsive education and accomplish tasks that would otherwise have been limited to interactions within the healthcare site [43]. Beyond the patient education context, a recent trend can be seen with learning apps such as Duolingo or Coursera, which provide evidence that different paradigms exist to educate/upskill the public, even in modest amounts over a longer period of time [44]. An app with a preventative focus on PI education could, in theory, follow a predefined pattern to provide information ‘just-in-time’ in response to anticipated or previously defined risk factors [45].

In today's clinical landscape, health professionals work in a high-stress environment that places high cognitive load by way of managing all facets of a patient's care [46]. With greater insight into healthcare priorities, computerised systems offer a clear benefit to help with automated triage and prevent health states from deteriorating outside the awareness of healthcare professionals [47]. Since these types of systems rely on patient charts and other internal data that are constantly being updated, it is logical that treatment plan recommendations can be generated based off previous successes and current trends.

eHealth interventions can aid healthcare teams to develop evidence-based treatment plans and provide timely education (e.g., reminders or highlighting critical information for PI risk) to healthcare professionals in response to emerging patient data or statistics. Algorithmic clinical support not only distils critical intelligence across many different areas of interest/concern, it also reduces the cognitive burden on healthcare professionals needing to balance procedural flexibility for client-centred practice with adherence to established clinical guidelines [33]. With these algorithms, guidelines and living meta-analyses/systematic review evidence can be rapidly adopted into standard practice as clinical decision or knowledge support systems can be instantly recalibrated to bring these changes to the forefront in an intuitive and user-friendly manner [31, 35].

In contrast to personal smartphones that are multi-purpose computing devices, dedicated computing platforms such as the Raspberry Pi (A single-board computer) allow for bespoke hardware and software development to take place [48]. The study of computerised mattress systems in 1999 demonstrates an interest in eHealth interventions since the dawn of the century. Such a system would have made use of rudimentary custom programming and hardware sensors to achieve its intended outcomes; a feat that could be replicated today with all the advancements and progress acquired over the years in the hands of the right development team [49]. Bespoke hardware and software fusions may also be preferred to offer a single combined solution for commercial sale or other uses. Since both components are designed to work with each other, a higher degree of functionality and reliability can be expected as opposed to ensuring compatibility when there are many permutations to the end-user's hardware or software environment.

8 Strengths and Limitations

One of the strengths of this review is that it comprehensively maps the current eHealth solutions available to prevent PIs and evaluates them using a validated framework. This review's use of the NASSS framework provides validity to our investigation as it provided the structure for our data extraction process in terms of domain-based questions when evaluating the sustainability of a given intervention. Another strength of our paper is the inclusion of a transdisciplinary team of reviewers who contributed specific expertise in the domains of wound care, spinal cord injury, community pressure care, occupational therapy, and software/hardware lifecycles. Furthermore, our search strategy was developed and cross-checked with experienced research librarians to ensure that appropriate keywords were selected.

However, this review also inherits limitations from the pool of papers sampled. Since most studies included in this review were conducted in the USA and Switzerland, this may have introduced a geographical bias towards preventative technologies that are developed and trialled primarily in those countries. Non-English and non-published journal articles were also omitted from inclusion. Despite this, we believe sufficient care was taken to consider studies from all countries (including reports where their abstracts were written in languages other than English) during our initial search and where possible, efforts were made to contact authors of elusive papers that were uncovered through manual hand searches of reference lists. It can be argued that the vast majority of solutions have already been represented in our report and that new innovations that may be reported in studies missed by our search strategy take the form of artificial intelligence/machine learning (AI/ML) improvements, in line with trends in the computing industry, that do not change the specific type of solution that would have been identified (e.g., an app with AI/ML is still an app irrespective of its intended setting for use).

Another limitation is that the nature of this scoping review does not account for the rigour and clinical effectiveness/efficacy of the interventions mentioned, as its main goal is to report on the different interventions that have been mentioned in the literature. Thus, any evidence for/against a specific intervention type should be precisely retrieved as part of any further research spurred by this review.

9 Local Context/Implications

After analysis of our results, we can identify several implications for future research that can be applied to the local context in Australia (and abroad more generally). Future research targeting PI prevention should take into consideration the different population and policy parameters experienced locally. Research and development of interventions should also factor in existing technological capabilities such as any national infrastructure (e.g., Australia's myHealthRecords) to improve developmental efficiencies and to avoid security and integrity pitfalls that may exist in a from-scratch solution [50].

According to the NASSS framework, an intervention that exhibits complexity in multiple domains is predicted to experience challenges in adoption over time [11]. Besides the potential demotivating factor of being unable to achieve the ‘ideal’ therapeutic trajectory of using complex interventions, designing eHealth interventions in this manner reflects the trend of self-quantification passively revealing data about health states without offering the active support/personalised recommendations to actually improve identified issues [51]. A similar phenomenon occurs in users of fitness tracking apps whereby its gamification features include leaderboards/rankings that are designed to celebrate the most active users while merely offering an ‘ideal’ for less active users to aspire towards; potentially leading to feelings of frustration and shame [52]. To overcome the perils of users losing motivation to engage with PI prevention, ethical user experience and design principles should be applied to compensate for the requisite active participation in PI prevention.

The rising accessibility to high-speed internet over the past decade may enable the use of virtual coaches (i.e., digital representations or ‘avatars’ of health professionals) that can act in ways to promote positive health behavioural change. Virtual coaching can increase access to these eHealth consultations regardless of geographical locality [53]. Furthermore, solutions today emphasise ‘open-source’ design that makes use of freely distributable software and hardware developers can assemble into bespoke configurations. This process of democratisation may enable more rapid innovation to take place while reducing barriers associated with the previously high costs of development and/or access to these options for users [54].

Today's technological advancements also usher in a new era for computer-assisted decision making based on Big Data (i.e., health informatics) and integration with large language models (e.g., ChatGPT); two different methods to optimise clinical administration workflows and enable the healthcare team to focus on the patient, enhanced with recommendations and insights from a computer database [55].

10 Conclusion

Interventions in the eHealth domain can be broadly classified into three categories that reflect the focus of their functionality. All the eHealth solutions encountered during this scoping review demonstrate varying levels of utility and effectiveness; however, no one solution fully addresses the criteria of the NASSS framework that promote long-term adoption of said solution. Overall, our findings suggest that additional research and development are warranted to harness the successful features implemented in prior works while mitigating pitfalls that could lead to underutilisation or abandonment of the intervention in question.

Acknowledgements

We acknowledge funding from the School of Health Sciences, and the College of Health Medicine and Wellbeing, University of Newcastle that has made this research paper possible (University of Newcastle, College of Health, Medicine and Well-being: CWMWB Industry Pilot Grant G2300448). Open access publishing facilitated by The University of Newcastle, as part of the Wiley - The University of Newcastle agreement via the Council of Australian University Librarians.

Ethics Statement

N/A. The authors wish to assert that this study is based on secondary analysis of papers describing interventions of interest based on inclusion/exclusion criteria discussed, and that the conduct of this study did not involve, nor expose any participants to harms, directly or indirectly.

Conflicts of Interest

The authors declare no conflicts of interest.