A systematic search was conducted by 4 independent researchers from the University of Udine, Italy, for the selection, data extraction, and functional evaluation of antimicrobial resistance apps available on the Apple App Store and Google Play Store. The Italian and English terms for antimicrobial resistance, antibiotic resistance, antimicrobial stewardship, antibiotic stewardship, antibiotic guide, antibiotic therapy and antimicrobial therapy were used as keywords in both stores. All apps that were found with at least 1 keyword mentioned above were included in the study. The choice to start the systematic search using keywords was necessary because there are too many healthcare apps and it was essential to reduce the action area. The following exclusion criteria applied to the search: content in languages other than Italian or English, games, leisure apps, photo and video apps, paid apps, paid content, advertising, apps that require registration, apps that only contain references to scientific articles, and apps that are not focused on health care. Paid apps and paid content were excluded as the focus of the research was to find apps that support health care professionals in the implementation of AMS and are, therefore, accessible to all health care professionals regardless of their financial means or affiliation with a public or private institution. Pathology specific apps were excluded as the aim of the study was to find a generic app that covers all diseases and systems. A set of basic information was collected for all apps found. This included the name of the app, operating system (iOS or Android), availability of additional paid content, size (MB), number of downloads, position in the download ranking, user rating, copyright, privacy policy, possible presence of European medical device (CE) labelling, number of languages, store category, recipients, date of first release, last update, and responsibility of support. A total of 4 researchers conducted the search independently between June 15, 2022, and October 18, 2022, (2 researchers worked with iOS and 2 researchers worked with Android). All the contradictory information collected was discussed until a consensus was reached. Then the results were merged and app deduplication was performed.

A first draft of a checklist was created, listing the app features that were considered ideal for the app users. The main topics (domains) to be evaluated were related to the content of the app, notes and records, social support, and technical features. Individual questions were formulated for each domain based on the scientific literature [2,7,12] and the authors’ previous experience in the field of mHealth apps for postnatal care [18] and for antenatal and postnatal care [19]. Possible answers were yes (y), no (n), or partially (p). The first draft of the checklist was commented on by a group of experts involved in AMR at different levels, including infectious disease specialists, microbiologists, clinicians, public health and infection risk management specialists, pharmacologists, and veterinarians. The valuable multidisciplinary collaboration between these professionals [20] helped to refine the assessment tool for the final draft of the 86-item checklist, which covers 8 domains, namely: pathogens and etiological agents, diagnostic and therapeutic support, AMR, dashboard function, AMS, notes and records, network, and technical features. Table 1 shows the final version of the checklist.

This study did not include human subjects research (no human subjects experimentation or intervention was conducted) and so did not require institutional review board approval.

First, 115 apps were identified: 31 apps for Android and 84 apps for iOS. By applying the exclusion criteria, 31 apps were excluded: 16 for Android and 15 for iOS. The remaining 84 eligible apps (15 for Android and 69 for iOS) were downloaded, installed, and further analyzed using the same criteria, excluding 57 apps: 48 for iOS and 9 for Android. The remaining 27 apps (6 for Android and 21 for iOS) were selected for in-depth analysis. Based on additional inclusion and exclusion criteria, as outlined in Figure 1, the flowchart illustrates the app selection process and outcomes.

After the entire selection process, 27 apps were evaluated using the checklist questions, of which 6 were available for Android (22%), 21 (78%) were available for iOS, and 1 was available in both stores. In total, 12 of the 27 apps were associated or affiliated with health care institutions or other international organizations. As shown in Table 2, considering all the potential items that could be fulfilled by the apps for each domain, of the 27 apps selected, diagnosis and therapy support (90/513, 37%) and app technical characteristics (187/810, 23%) were the most frequently fulfilled domains, followed by AMS (13/162, 8%), pathogens and etiological agents (8/216, 4%), notes and records (5/162, 3%), network (4/189, 2%), AMR (2/162, 1%), and dashboard function (1/108, 1%).

In the domain of pathogens and etiological agents, only 5 apps (19%) provided information on the mechanism of action of drugs, 2 apps (7%) provided information on the main characteristics of microorganisms, and 1 app provided information on the characteristics of aerobic and anaerobic microorganisms. In the domain of diagnostic and therapeutic support, 21 (78%) apps assisted the user in identifying the drug for targeted therapy against a specific microorganism, 18 (67%) apps provided information on the dosage to be prescribed, 17 apps (63%) indicated the route of administration for the drugs indicated for therapy, and 15 apps (56%) assisted the user in selecting the drug for empirical therapy according to the patient’s clinical condition, and 15 apps (56%) considered the clinical variables related to the site of infection and patient characteristics in their therapy suggestions. In contrast, most apps (n=23, 85%) did not support the user in reading the antibiogram correctly. In the domain of AMR, hardly any app fulfilled the required characteristics, with the exception of one app that provided information on the evolution and types of AMR and one app that provided prevalence data on AMR in Italy, while no app provided prevalence data at European and Italian level.

In terms of dashboard functions, only 1 app responded to one of the items by providing regional data on AMR profiles of isolates; national and international data were not provided to the user. In relation to AMS, 8 apps (30%) provided information on the latest available guidelines, but little other information was provided on this topic. In particular, no app provided information on the possible circulation of the same pathogen between humans, animals, and the environment and did not inform the user about the role of the veterinary component in AMR and AMS from a One Health perspective. In terms of the network domain, no app was connected to the Italian postmarketing surveillance network for reporting possible adverse effects or to the infectious disease reporting system. No app offered opportunities to interact with other users or experts.

Regarding technical features, all apps (100%) had intuitive and predictable navigation patterns and offered content in text mode, the content of most apps (26/27, 96%) was available to users without payment, and 81% of the apps (22/27) could be used completely offline. However, none of the apps provided a real-time map of the AMR in relation to the reference country or located the user to provide more detailed information. In addition, none of the apps interacted with the hospital or local medical management software, nor did they offer backup, restore, or download functions. In addition, no app offered multilingual support or was certified as a medical device under Italian law, and only 11 apps (41%) claimed scientific responsibility for the content provided. Multimedia Appendix 1 shows the detailed results of the app analysis.

Multimedia Appendix 2 shows the detailed results of the in-depth analysis of each of the 27 apps.

Considering the full list of items, no app met all the desirable criteria and the highest level of fulfilment of the desirable items identified by the checklist was 36%, which was only achieved by 2 apps. All other apps achieved fulfilment rates between 9% and 26%.

Most of the apps analyzed did not have most of the desirable features, with only two of them achieving a 36% fulfilment level. Most of these apps lacked the provision of content on pathogen characteristics, diagnostic and therapeutic support, AMR, and AMS. The dashboard function was generally lacking and did not allow users to view data on local, national, and international AMR prevalence. The mHealth solutions did not support a network system, which meant that organizational and technical support between professionals could not be provided. One of the most likely reasons why these apps did not meet the criteria selected by the expert group could be that they were designed and developed for other purposes and were not aimed at supporting health care professionals in AMS. Not having a dashboard function is missing an opportunity both at the national and international level, because having local data available could be useful to prescribers.

We are aware that we identified many features and that probably no app could have responded positively to all the items. Nevertheless, to reach our purpose to fighting AMR, we inserted all the elements we considered useful for the good functioning of an app in order to verify the distance between apps available for free and our ideal standards. Starting from this study, the next step we are planning is to create an app that fulfills most of our items.

We chose to provide data about the evaluation of each of the 27 apps in Multimedia Appendix 1. This makes it possible for readers to analyze apps also using subgroups of items. Ensuring an audit and feedback mechanism for prescribers by bringing together relevant information can indeed promote changes in practice, as is done in many clinical contexts [21] through quality improvement programs [22]. Therefore, the audit and feedback mechanism for prescribers and the opportunity for benchmarking should be supported to enhance their role in practice [23]. In addition, the recent pandemic has revolutionized the previous approach of health care professionals and researchers seeking access to organizational and institutional sources of information. This process toward greater transparency, accountability, and democratization of health data began with the introduction of the free availability of Italian COVID-19 surveillance data in 2020 [24]. The use of technology is not only changing the way people and professionals communicate with each other, but is also providing innovative ways to monitor the health and well-being of patients and the planet, and giving health care professionals better access to information for a more holistic view of health issues. Digital technologies are emerging as an important resource for health care and public health [15], but much remains to be done to translate evidence-based, validated information into clinical practice through medically certified devices. Probably the most recent, scientifically pervasive, and currently ground-breaking topic that is receiving massive and increasing investment in the public sector, but even more so in the private sector, is AI. In health care, AI is already being used to some extent for chatbots (software applications or web interfaces, designed to have textual or spoken conversations) and triage systems to assess the need for doctor visits. It is also sometimes used in diagnostics, where it achieves accuracy comparable to that of experts, in predictive modelling, pharmaceutical research, and epidemiological studies [25]. As for its use in antimicrobial devices, ChatGPT (a comprehensive language model developed by OpenAI) appears to have access to sufficient training data (although it does not have access to specific medical databases), but has shortcomings in consistency, reasoning, and situational awareness that could jeopardize patient health. missing important aspects in scenarios of increasing complexity [26]. Despite these advances, problems remain regarding the robustness of studies and algorithmic bias. Furthermore, there are currently no specific regulations for AI-assisted care. If AI systems prove to be sufficiently reliable, they could provide significant support to health care providers. Although the use of AI in clinical practice is still limited, initial studies show promising results, suggesting that external validation and clinical trials are likely to increase in the near future [27]. Most importantly, local, regional, and national institutions and professionals need real-time data at the local, regional, and national level to make the best clinical and organizational decisions regarding antimicrobial stewardship. Therefore, the availability of features to train algorithms with real AMR and AMS data without the risk of data breaches is critical to the pursuit of this goal.

The Italian National Plan to Combat Antimicrobial Resistance (Piano Nazionale di Contrasto dell’Antimicrobico-Resistenza, PNCAR) recommends a One Health approach to contain AMR [12]. The apps analyzed in this study lack information on the circulation of zoonotic AMR pathogens between humans and animals and on the role of the veterinary component and the environment in the AMR and AMS problem, thus missing the opportunity to consider the broader One Health perspective. To effectively contribute to global health security by supporting prevention, early detection and warning, response and recovery for emerging and endemic communicable diseases, the One Health approach requires cross-sectoral integration and collaboration to efficiently and effectively bring together information, capacity and expertise to protect people, animals and our environment. On this basis, the tripartite collaboration between the Food and Agriculture Organization (FAO), the World Organisation for Animal Health (OIE) and the WHO has recently reaffirmed its commitment to cross-sectoral collaboration on disease risks at the human-animal-environment interface [28], which is also supported by many other global and local stakeholders [29]. Given the high economic risks of disease epidemics and their far-reaching societal and economic impacts, cross-sectoral One Health approaches are critical to improving prevention, detection, response, and recovery capacity [30]. In line with this vision, any action to address climate change can in turn benefit the global burden of disease, migration, conflict over natural resources and political instability—all of which are closely linked to the economic, environmental, and social determinants of health. In Europe, for example, climate change has been observed to lead to changes in rates of communicable and noncommunicable diseases due to temperature changes, increase the risk of outbreaks and exacerbations of clinical diseases (especially cardiovascular diseases), influence extreme meteorological events (eg, rainfall, floods, and hurricanes), cause sea level rise, and alter the distribution of plant species and arthropod vectors, as well as the quality of air and food. Nonetheless, climate change also has an impact on food security, as an increase in global temperature is associated with an increase in global demand for food [31].

Measures to improve the appropriateness of antimicrobial prescribing are a key component of AMS programs [10,32], as it is widely recognized that inappropriate use of antimicrobials increases the risk of treatment failures and adverse events in individual patients and accelerates the selection and transmission of antimicrobial-resistant pathogens in health care facilities and populations [33]. Nonetheless, the awareness and knowledge of prescribing physicians on these issues still needs to be improved and strengthened. To achieve this goal, much needs to be done, from preparation to ongoing education of health care professionals, to reinforce the message and support the behaviors. To promote a culture of optimal antibiotic use, the WHO published guidance in 2022 based on the AWaRe classification, which was introduced to support measures to improve antibiotic management. Antibiotics are categorized into 3 groups (Aware, Watch, and Reserve) according to their clinical importance and the risk that their use promotes the development of resistance. The guidelines contain evidence-based information on the selection, formulation, dosing, and duration of administration of antibiotics and defines clinical situations in which antibiotics are not recommended based on the best scientific evidence [34]. The WHO’s “Firstline” app, which is only partially available to unregistered users, is also moving in this direction. This app provides various national and some local guidelines on different types of infections and some of their antibiotic treatment for adults and children, and its content is also available online [35]. In Italy, for example, “Firstline” is temporarily available for the pilot phase of the Enforcing Surveillance of Antimicrobial Resistance and Antibiotic Use to Drive Stewardship (ENSURE project) in Padua or for the users of the Integrated University Hospital of Verona after authentication. However, as this is a first attempt to address the problem, more advanced solutions are being sought to match guidelines with data and prescribing behavior to reverse the general trend. Antibiogo (Médecins Sans Frontières Foundation) is a mobile app, which was developed to help laboratory workers without expertise in low-resource settings measure and evaluate antibiotic susceptibility tests so that clinicians can prescribe the right antibiotics to their patients [36].

The European Centre for Disease Prevention and Control (ECDC) has defined core competencies for professionals in the field of infection control and hospital hygiene. These include minimum requirements common to all professionals in this field, ranging from leadership and management of infection prevention and control programs, microbiology and surveillance, infection prevention and control in clinical practice, education and training, quality, patient safety and occupational health [37]. Another issue to consider in relation to this topic is the sale of over-the-counter (OTC) antibiotics. This is seen as a significant problem, particularly in regions of the world (ie, low- and middle-income countries) where prevalence is quite high. In these areas, it is not enough just to influence prescribers to tackle the problem of antimicrobial resistance; other measures must also be taken to control the use of antimicrobials. For this reason, the WHO has launched an electronic consultation in 2023 involving key stakeholders such as politicians, pharmacist associations, and civil society to identify possible actions for a global campaign to phase out over-the-counter antibiotics [38]. Despite these considerations and the awareness that there is still much to be done in this area, we believe that this study was able to provide an initial insight into the information that already exists and what could be done to further support health care professionals in combating AMR. For example, providing an institutionally validated app with content tailored to the individual needs of the health care professional and setting would help to tailor prescribing decisions to the context and tailor support to the prescriber’s expertise. For example, the app could differentiate between primary care providers (general practitioners, family medicine, pediatrics, and internal medicine), acute care providers (inpatient or hospitalists), and infectious disease specialists to provide the most needed information.

This study has several limitations. First, the exclusion criteria of this study limited the systematic search for the apps available in the Apple App Store and Google Play Store, specifically the requirement that the app must be available for free and without registration to be included in the in-depth analysis. These criteria aimed to evaluate the apps available to all health care professionals, regardless of their geographical or institutional affiliation. During the screening phase and the eligibility phase excluded apps were overall 22 of 115 (19%). Although aware that this share is relevant, we opted to exclude them anyway, because of the consideration that was more important that the contents were freely available to health care professionals, without the obligation of payment, in order to ensure equity for all economic levels. However, by choosing to include both Italian- and English-language apps, apps were selected that primarily target the educational level and professional profile of prescribing physicians. Second, we cannot know whether presenting the same checklist to other larger groups of experts would confirm all the items on the checklist we used to test the apps. Therefore, further future research along these lines may be useful to confirm, model, refine, and integrate domains and items. Third, the availability of the apps in stores at the time of the systematic search and a limited app validation panel may have limited this study. The search strategy, which involved the use of keywords, was probably unable to find some apps containing AMR modules. However, our goal was to find an app totally and not partially focused on AMR. Furthermore, this study does not value the length of therapy based on the organism and site of infection, surely an AMS domain. It could be a parameter to include in future works.

Finally, the actual effectiveness of the apps was not evaluated as this was not our focus. Nonetheless, this initial evaluation could form the basis for future work exploring this aspect in more depth.

This systematic search for apps available to Italian health care professionals to support antimicrobial prescribing in line with AMS principles shows that there is still much room for improvement in the development of apps and more generally in mHealth solutions to combat antimicrobial resistance. There is currently no user-friendly and useful app that meets all the requirements identified by the expert group; such a tool would support the engagement of prescribers in the fight against AMR.