SERIES: eHealth in primary care. Part 5: A critical appraisal of five widely used eHealth applications for primary care – opportunities and challenges

Abstract Background Given the pressure on modern healthcare systems, eHealth can offer valuable opportunities. However, understanding the potential and challenges of eHealth in daily practice can be challenging for many general practitioners (GPs) and their staff. Objectives To critically appraise five widely used eHealth applications, in relation to safe, evidence-based and high-quality eHealth. Using these applications as examples, we aim to increase understanding of eHealth among GPs and highlight the opportunities and challenges presented by eHealth. Discussion eHealth applications can support patients while increasing efficiency for GPs. A three-way division (inform, monitor, track; interaction; data utilisation) characterises many eHealth applications, with an increasing degree of complexity depending on the domain. All applications provide information and some have extra functionalities that promote interaction, while data analysis and artificial intelligence may be applied to support or (fully) automate care processes. Applications in the inform domain are relatively easy to use and implement but their impact on clinical outcomes may be limited. More demanding applications, in terms of privacy and ethical aspects, are found in the data utilisation domain and may potentially have a more significant impact on care processes and patient outcomes. When selecting and implementing eHealth applications, we recommend that GPs remain critical regarding preconditions on safe, evidence-based and high-quality eHealth, particularly in the case of more complex applications in the data utilisation domain.


Introduction eHealth in primary care
Healthcare worldwide faces a heavier and increasingly complex workload due to an ageing population and a rise in multimorbidity [1]. Primary care bears the full brunt of this pressure because it is usually the first port-of-call for the general population [2]. This increasing pressure creates a necessity for a more efficient organisation and better distribution of care [3].
Inherent to this change are general practitioners (GPs) and their staff who have to deal with changing patient-provider relationships. Tailored care that focuses on a patient's autonomy, self-management and self-efficacy is becoming increasingly important [4]. The COVID-19 pandemic further underlines the need for a transformation of care, as continuity of care should be secured even in times of limited access to face-to-face care [5]. eHealth has been proposed as a valuable tool to support healthcare and ensure universal health coverage [1]. According to Shaw et al. (Figure 1) [6], eHealth has three main functions: (1) 'Inform, monitor and track' to observe and study health parameters; (2) 'Interaction' to support communication; and (3) 'Data utilisation' to collect, manage, and use health data [1]. In daily practice, eHealth applications ranging from mobile phone applications to telemonitoring systems often encompass multiple domains.
In the first four articles of this series on eHealth, we addressed issues related to the successful uptake and use of eHealth in primary care [1,[7][8][9]. These issues were described from various perspectives ranging from ethical aspects to education and implementation, including the six preconditions that we view as pivotal to ensure safe, evidence-based and high-quality eHealth (Box 1). However, for many GPs, it may still be challenging to maintain an up-to-date overview of the opportunities and challenges pertaining to the use of eHealth in daily practice. Therefore, this opinion paper aims to apply the six preconditions for high-quality eHealth, which are described in detail in Part 1 of the SERIES [1], to critically appraise specific and tangible eHealth applications currently in use in primary care. We intend to increase GPs' understanding of eHealth and provide insight into the opportunities and challenges concerning the use of eHealth. Using examples illustrating the potential of eHealth, we hope to inspire and encourage GPs, while at the same time remaining reflective and critical regarding eHealth applications in daily practice.

Contextual factors
Country-specific aspects Adoption of eHealth differs among and within countries and is influenced by several factors, such as the details of national policy. Regarding the six preconditions (cocreation, blended care, individualisation, globalisation, evidence and ethics), there is a lack of consistent national monitoring or thorough evaluation of eHealth internationally [10]. In existing literature, several countries are mentioned repeatedly because of well-established digital health infrastructure. Denmark is considered to be an international front runner because of the combination of an advanced high-quality digital health system and an excellent nationwide Digital Health Strategy that focuses on co-creation, blended care, individualisation and ethics [11]. A comprehensive digital system enables Danish GPs, healthcare facilities, nursing homes and other facilities to cooperate better and decrease inequalities. In Australia, a national digital strategy focuses on leveraging co-creation and blended care in a digitally-savvy society, combined with a clear digital health vision that places the individual patient at the centre of care [12].
Several countries, including New Zealand, the Nordic countries and Israel, are relatively advanced in adopting eHealth and have a sound digital strategy in place that details preconditions for co-creation and blended care [12]. In New Zealand, regional systems integrate data from several primary care providers (i.e. community nurses, pharmacists and GPs) and hospitals, forming a broadly accessible data structure (depending on access rights) [13]. In the Netherlands, most primary care practices offer eHealth, although the level of use varies considerably and co-creation with the patient is still rare [14]. In Canada and the United States, GPs are increasingly adopting eHealth, including telemedicine for online consultation, a trend further boosted by COVID-19 [15]. The challenge now is to find a responsible approach to maintaining the use of eHealth as the COVID-19 pandemic recedes.

General practitioner-specific aspects
The speed of eHealth implementation depends for a large part on the willingness of physicians. While some healthcare providers can be considered early adopters and are open to using eHealth in daily practice [16], others may be more reluctant to change their routines. The expectations of general practitioners concerning the impact of eHealth on workload are sometimes conflicting, with one qualitative study stating that the majority of general practitioners thought that eHealth would reduce their workload while another large survey in the same country found that only 10% had similar expectations [17,18].
Other factors influencing the use and adoption of eHealth in daily practice relate to healthcare providers' preferences, e.g. seeing patients face-to-face and perceptions about the (added) value of e-consultations or telemonitoring in relation to improving care and relationships with patients [19]. Legislative aspects may also play a role, as there can be uncertainty about professional responsibilities [7]. Furthermore, healthcare providers appear less inclined to use eHealth if they are worried about their job security or cost aspects, citing issues such as the extra costs of eHealth applications without compensation or the loss of revenue if the costs but not the benefits are felt at the practice level [19,20]. Differing preferences may also be linked to a lack of knowledge about and attitudes towards eHealth's opportunities and challenges.

Aspects concerning the complexity of eHealth applications
The complexity of an eHealth application, ranging from less complex applications in the inform domain to more complex applications in the data utilisation domain, together with the required level of care reorganisation, directly affects adoption rates [20]. Using a relatively simple eHealth application to replace a specific routine, e.g. tracking the lifestyles of patients with chronic diseases with an app rather than a paper diary, requires different abilities compared to applications that radically change work processes, e.g. 24/7 telemonitoring of patients with diabetes rather than seeing them on a quarterly basis. The lessons learned from the first four papers in the eHealth series are relevant and important to all applications, especially in the case of more complex applications [1,[7][8][9]. Furthermore, patient factors are related to the complexity of applications and need to be considered. For example, a fair degree of (health) literacy and digital competencies are required, especially for complex applications [1]. Other factors influencing the use of eHealth include age, income and education [21]. Iterative co-creation processes with relevant target groups are essential during the development and implementation of eHealth applications.

eHealth applications
To support GPs interested in adopting eHealth, we describe five widely-used eHealth applications. Applications were eligible for inclusion in the paper if they were relevant to primary care and currently in wide use. As it was not our aim to provide a comprehensive overview of all existing applications, we selected only five applications. To cover all the domains detailed by Shaw et al., we set at least one application per domain, with the applications serving as examples to illustrate the opportunities and challenges of eHealth in primary care and as aids to help define critical aspects of concern to GPs. Using these examples, we elaborate on the preconditions for safe, evidence-based and high-quality eHealth (Box 1) and the five examples highlight how these preconditions ultimately influence the adoption and utilisation of eHealth applications.
The five included applications were selected pragmatically, based on consensus and brainstorm sessions with the co-authors. One of the authors (ET) checked the literature, websites and the application itself to extract information regarding the six preconditions for all applications. MK also reviewed the literature, websites and applications to validate the results. The next step was validation of the results by the whole author group and discussion to form the final opinion and interpretation of the findings.
The tables presented below summarise the five commonly used eHealth applications. Table 1 provides a description of the relevant domain(s) in relation to the Shaw classification. Table 2 describes preconditions for safe, evidence-based and high-quality eHealth.

Main findings
The goal of this opinion paper was to provide insight into and give examples of the opportunities and Box 1. Conditions to ensure safe, evidence-based and high-quality eHealth [2] 1. Engagement of and co-creation by all stakeholders 2. Blended care: eHealth combined with regular care 3. Individualised and inclusive 4. Applicable in high-and low-resource settings 5. Evidence-based and supported by educational guidance 6. Being attentive to ethical considerations, privacy and patient safety challenges pertaining to the use of eHealth in primary care. We will discuss five specific and tangible eHealth applications currently used in primary care. Our critical appraisal of these applications confirmed the three-way division of eHealth applications, characterised by increasing degrees of complexity. While all five applications provide information, interaction can be added to expand functionalities (Liva Healthcare and SHUTi), and data analysis and artificial intelligence can optionally be used to (fully) automate care processes (Babylon and SkinVision). Preconditions on safe and high-quality eHealth are relevant for all applications. Still, following our evaluation, we now argue that this is especially the case for more complex interventions in the data utilisation domain. In our opinion, these applications possess high-risk, high-gain potential both at the patient and GP level.
To clarify, applications in the information domain, such as Thuisarts.nl, as well as some in the interaction domain, such as Liva Healthcare, can be readily recommended by any GP. Risks due to privacy or ethical issues are minimal, as no or limited data are collected and no drastic changes are made to the clinical process. While the impact on clinical outcomes of applications in the information domain may be minimal, the impact of applications in the interaction domain is likely to be much more profound.
Applications in the data utilisation domain present significantly greater risk, as ethical aspects including patient safety and privacy might be at stake if an algorithm is inaccurate or data are not safely stored.
On the other hand, the potential yield regarding patient outcomes and care processes is often high, e.g. diagnostic processes are performed at home instead of in a clinical setting, thereby increasing efficiency, or algorithms are used that outperform  Participants felt that having an initial oneoff, face-to-face meeting with the dietician or lifestyle coach was important for their future webbased interaction Effective in improving the lifestyle of patients with diabetes, e.g. more exercise and weight loss [32,33] Open to peer-reviewed scientific evaluation; programme will be improved after analysis by scientists [30] No adverse or unintended effects among the patient population Further improvement could be made by adding an ethical expert to the team, and by promoting co-creation with vulnerable and low-literacy patient groups Liva Healthcare is an easy-to-use application to improve lifestyle, with evidence-based results on clinical outcomes, high usersatisfaction and limited resistance from doctors [38].
This eHealth tool can support behavioural changes through monitoring and providing relevant feedback. Support from family and peers also matters, and long-term success depends on ability to establish positive day-to-day support [30] SHUTi Effective and practical online solution for the treatment of insomnia in general practice. Insomnia is currently not adequately treated in the primary care setting [44] due to limited knowledge of sleep treatments [39] . Hence, contrary to international clinical guidelines that recommend CBT-I as the first line treatment for insomnia, hypnotic medication remains the primary treatment modality in the majority of the patients [45]. SHUTi is considered a viable option in the treatment of insomnia [43] (continued) According to existing law SkinVision is a healthcare provider and must meet established qualitative and safety requirements. Initially, it did not comply with legal requirements and, among others, had to draw up a policy plan by October 2020 [47] False-negative outcomes might delay the detection and treatment of skin cancer False-positive outcomes may cause unnecessary stress among users or unnecessary visits to GP's The user may ignore the advice given in the smartphone app due to a lack of trust or unawareness GP's need to understand the basics of AI to deal with such algorithms in the diagnostic process The app seems easy to use.
Users need to be able to take a photo with their mobile phones There are no (scientific) publications available on usability The accuracy of algorithm-based smartphone apps for the assessment of skin cancer is debateable [35] The algorithm has a sensitivity of 95% and a specificity of 78.3% in recognising signs of skin cancer [36] SkinVision outperforms other skin cancer detection apps in terms of accuracy [35] The doctors in, for example, diagnostic processes and thus positively affect patient outcomes.

Impact on the care process
While patients seem very satisfied with applications that automatically use data, GP resistance tends to increase when they feel that their control of the care process is diminishing or if there is uncertainty surrounding the reliability and security of data [22,23]. Nevertheless, it is doubtful that all patients fully understand the risks of these types of applications and initial satisfaction will probably diminish following negative experiences, such as when an app 'gets it wrong' or after a data leak. If using an eHealth application is too disruptive to regular GP routines or if it leads to higher costs, job insecurity or loss of revenue, GPs will understandably perceive this as a barrier to using the application [19]. Moreover, the level of proficiency in the use of information technologies might also affect the use of eHealth [8]. Applications that mainly provide information to patients, such as Thuisarts.nl, are the least 'threatening' for GPs, offer multiple ways to support their care processes and understandably are also popular with GPs. Applications like SkinVision could radically change the diagnostic process, require GPs to have a basic understanding of AI and therefore have a greater impact on daily GP practice. Clearly, the degree of disruption to a GP practice will form a barrier to using an eHealth application [9]. The greater the disruption, the greater the need for reliable testing of technical accuracy and clinical effectiveness. GPs should also consider that implementing complex eHealth interventions might require the reorganisation of care [24]. If adopted, eHealth applications must be implemented and integrated into the daily work processes of GPs.

Scientific evaluation
While scientific evaluation of eHealth applications is essential to ensuring the safety and successful implementation of eHealth [25], scientific evidence concerning effectiveness, usability and accuracy is often lacking [26,27] or is of insufficient quality [28]. This is a particularly thorny issue in the data utilisation domain, where scientific evaluation is still in an early phase, meaning that critical external scientific evaluation is an ongoing process. The applications discussed here differ greatly in terms of supporting evidence on efficacy. Evidence concerning thuisarts.nl is available, but this focuses solely on healthcare usage [29], and no peer-reviewed evidence is available regarding, for example, usability or outcomes. Liva Healthcare and SHUTi have undergone more extensive testing, including their effects on patient outcomes and experiences [30][31][32][33][34]. Even though this evidence may be derived from the general population, this information is still relevant to primary care. Nonetheless, further investigation in a primary care setting would be valuable. Evidence is available regarding the accuracy and performance of SkinVision [35,36], but additional real-life studies are needed. Babylon is claimed to be effective, but there is no evidence to support this. Furthermore, it should be noted that (co)founders or employers of Thuisarts.nl, Liva Healthcare and SkinVision are coauthors of the scientific publications describing these applications. We suggest GPs remain critical towards eHealth applications in terms of available evidence and to check whether demonstrably independent experts provide effectiveness.

Take home messages
The main advantages of the applications described in this paper, considerations that often apply to eHealth in general, are the relative ease of use for many patients, the high level of satisfaction, and the higher degree of independence and 24/7 applicability. In parallel, GPs can work more efficiently because eHealth supports patient self-management, lowering the demand for physical consultations. Nevertheless, perceived barriers for large-scale implementation include dealing with complex applications or a feeling of diminishing control of the clinical process. In Part 1 of the SERIES, we concluded that scientific research on eHealth applications is essential, preferably performed continuously while incorporating different viewpoints [1]. With the current paper, it became clear that evidence is often limited, even in already widely-used applications. Although some research is available on the applications included here, it mainly focussed on one specific aspect and continuous assessment of effectiveness was lacking. Furthermore, some studies were performed by the developers rather than by independent scientists, which might have compromised reliability. Therefore, we recommend a critical stance regarding eHealth applications, which should include assessing any external scientific evaluation that may have been performed. Independent institutes could play a leading role in collecting and publishing this evidence. If eHealth applications are not validated by peerreviewed scientific research, they should be implemented only with great restraint. If an application is likely to cause disruption to a GP's routine work processes, such as the case with Babylon, we believe that effectiveness must be proven before implementation.
A limitation of this paper was the description of only five selected applications, as we wished to discuss the potential of eHealth based on specific examples. To obtain a broader overview of the availability and effectiveness of eHealth in primary care, it might be interesting to perform an umbrella review summarising published systematic reviews in this field.
Conclusion eHealth applications in primary care appear to exhibit a three-way division, with increasing degrees of complexity ranging from relatively simple applications providing information to more complex applications that use data and/or change the organisation of care. To select viable eHealth applications for implementation in primary care, six preconditions that ensure safe, evidence-based and high-quality eHealth must be met, particularly in the case of more complex applications.