In 2011, the proponent of Australian Health Level Seven (HL7) standards, Grahame Grieve, proposed an interoperability approach called Resources for Healthcare (RFH) as a new standard for better interoperability in digital health. Technically, RFH has been designed for web technology, and the resource is based on extensible markup language (XML) with an HTTP-based representational state transfer (REST)ful protocol and a distinct URL for each resource. The RFH standard was renamed Fast Health Interoperability Resources (FHIR) with extension of previous HL7 specifications (ie, HL7 version 2 and version 3) with consideration of modern web technologies [1].

The main idea behind FHIR was to build a set of resources and develop HTTP-based REST application programming interfaces (APIs) to access and use these resources. FHIR uses components called resources to access and perform operations on patient health data at the granular level. This feature makes FHIR a unique standard from all other standards because it was not available in all previous versions of HL7 (v2, v3) or the HL7 clinical document architecture (CDA).

The basic building blocks of FHIR are the so-called resources, a generic definition of common health care concepts (eg, patient, observation, practitioner, device, condition). FHIR uses JavaScript object notation and XML structures for data exchange and resources serialization. FHIR does not only support a RESTful to exchange resources but also manages and documents an interoperability paradigm.

Since the first day of its introduction, FHIR has gained popularity and has been increasingly adopted by the health care industry. In 2018, six large technology companies, including Microsoft, IBM, Amazon, and Google, pledged to remove barriers for health care interoperability and signed a letter that explicitly mentions FHIR as an emerging standard for the exchange of health data [2]. With incorporation of Substitutable Medical Applications Reusable Technologies (SMART), a platform for interoperable apps [3], FHIR can be expected to attract even more attraction in digital health in the future. Using FHIR for the exchange of medical data can provide potential benefits in a large number of domains, including mobile health apps, electronic health records (EHRs), precision medicine, wearable devices, big data analytics, and clinical decision support.

The main objective of FHIR is to reduce implementation complexity without losing information integrity. Moreover, this new standard combines the advantages of the previous HL7 (v2, v3, and CDA) standards and is expected to overcome their limitations. FHIR allows the developers to develop standardized browser applications that enable the user to access clinical data from any health care system regardless of the operating systems and devices that a health care system uses. For example, a user runs an application on the browser and will access data from a health care system using any device, whether it is running on a desktop, smartphone, Windows, Android, or Linux operating system. Figure 1 represents the general architecture of FHIR [4].

The goal of this study was to gain a deeper understanding of the FHIR standard, and to review the use and adoption of the standard in current health care applications and organizations. This study can assist researchers and experts in understanding the FHIR architecture, design, implementation, resources, challenges, mapping, and adoption in health care informatics. Additionally, this systematic review identifies the key topics discussed in the context of FHIR in the literature.

A resource is the smallest discrete concept that can be maintained independently and is the smallest possible unit of a transaction [5]. Thus, a resource is a known identity providing meaningful data. Each resource has clear boundaries and differs from all other resources. A resource should be described in sufficient detail to define and support the medical data exchange that is involved in the process. According to the latest FHIR version (R4), the FHIR community has defined more than 150 resources to date [6]. These resources are divided into five major categories: (1) Administrative: location, organization, device, patient, group; (2) Clinical: CarePan, diagnostics, medication, allergy, family history; (3) Financial: billing, payment, support; (4) Infrastructure: conformance, document, message profile; and (5) Workflow: encounter, scheduling, order.

FHIR is the latest standard; however, to date, there has been no comprehensive systematic literature review performed in this area. Therefore, a systematic literature review was performed in this study to provide a broad view of FHIR, and to address various challenges, applications, and goals of FHIR highlighted in research in this field.

Owing to its dynamic characteristics, FHIR is gaining popularity rapidly. It is expected that FHIR will soon become an icon for clinical information exchange in the health care sector. However, it also faces numerous challenges, which is the main motivation that inspired us to perform this systematic literature review. Despite its importance in health care research, there is no comprehensive review of the literature in the field.

There were five objectives of this study. The first objective was to profoundly investigate the literature related to FHIR and EHR to explore their multiple challenges in the health care domain and give a comprehensive summary of these issues. The second objective was to identify FHIR applications, goals, challenges, and their roles in the health care domain. The third aim was to address different models of FHIR implementation. Fourth, we addressed different existing and emerging challenges of electronic health implementation to provide the readers with up-to-date information about the different types of hurdles faced by health care project implementations. Finally, this review offers useful suggestions and recommendations about the solutions to these issues faced by health care stakeholders.

This systematic literature review was conducted through the following steps: (1) establishing the research questions to be investigated; (2) identification of digital libraries to be explored and establishing the search strategy; (3) setting the criteria for selection of relevant articles; (4) setting the quality assessment criteria to select the best articles for this study; and (5) data extraction to address the research questions from the selected articles.

According to Kitchenham et al [7], research questions are the most crucial part of any systematic literature review. Therefore, we have to set questions related to the focus fields, which are FHIR and EHR. We formulated specific research questions to identify the objectives in terms of problems, challenges, solutions, and goals. Our research questions identify the mentioned domain broadly and cover almost every aspect of the field, which is essential for the research purpose. Table 1 summarizes the research questions and their corresponding objectives.

After establishing the research questions, the next step was to search for articles to collect the required data. To perform a proper systematic literature review, an appropriate search is essential to define the scope and search keywords, which are the fundamental concepts of our research questions for retrieving accurate results.

There is a possibility that the search method may not identify some relevant studies. Therefore, to establish an optimized search string, Kitchenham et al [7] suggest breaking down the research questions into individual facets called research units, which include all of their associated acronyms, synonyms, abbreviations, related words, and alternative spellings combined using Boolean operators (AND, OR) for the construction of keyword phrases.

Finally, we obtained and used the search string shown in Textbox 1 to retrieve the relevant articles.

After performing the search queries, a total of 8144 articles from all five major digital databases were retrieved from the initial search. After thoroughly checking the web profiles of authors and their networks, 37 new articles were added with a snowballing procedure. From the 8181 retrieved articles, we first applied the duplication criteria, and then set the inclusion and exclusion criteria described in Table 2. Therefore, we first excluded all of the articles found in multiple databases. After removal of duplicates, 1514 articles remained. In the second phase, we discarded articles published in non-English journals/proceedings, resulting in 1442 articles for further screening. In the third phase, we excluded articles that were not primary studies such as reviews and survey papers. Finally, 892 articles remained for further screening.

In the fourth phase, we analyzed the remaining articles on the basis of their title, abstract, and keywords, and the number dropped to 278. In the final phase, after reading and analyzing the full text of the articles, we selected 80 articles from the list to be included in the systematic review. Table 3 shows the results of the different phases of the selection process, Table 4 presents the articles chosen for our study, and Table 5 provides the geographic information of the publications.

As shown in Table 4, the distribution of the articles was 59% and 41% for journal articles and conference proceedings, respectively. The conferences represented are the main international conferences on health care or health care informatics, whereas the journals represent the world-class reputable journals in the field of computer science and health care. In terms of geography, as shown in Table 5, the number of publications related to FHIR published by researchers in the United States was the highest among represented countries. This indicates that the research in the field is quite active in the United States, which may become a factor that pushes the adoption of the standard in the country and in the rest of the world.

The 80 selected articles are arranged based in their primary subject categories in Table 6.

Once the articles were selected, we arranged them by ascending order of publication year. We then considered the attributes of the articles, including author names, article title, venue of publication (eg, journal article, conference proceeding), and publisher name. The complete list of the selected articles and their attributes is depicted in Table 7.

According to Kitchenham et al [7], it is essential to select and assess the best articles for every literature review, and the quality of selected studies must be verified before inclusion in the study. We evaluated the selected articles with regard to research quality, related work, purposes of research, the obtained result, the methodology used, literature review, current and future objectives, conclusion, publication repository, and other factors. We evaluated the quality of each article according to the protocol defined by Roehrs [85] as displayed in Textbox 2.

The proposed quality criteria scores were assessed for each selected article. Although the majority of the selected articles did not fully satisfy all six criteria for evaluation, they complied with at least four out of the six quality assessment criteria listed in Textbox 2.

All of the assessed articles clearly presented their research purpose, literature review, and were supported by a research methodology, bibliographical references, or models/architectural proposals. Based on this quality assessment, we did not exclude any articles from the corpus; this assessment only evaluates whether the articles have a satisfactory structure.

This systematic literature review successfully identified both qualitative and quantitative sets of studies that enable obtaining a clear view of the FHIR standard in health care in the past 8 years, starting from the selected number of articles. Some of the most relevant studies in the field are highlighted according to systematic selection criteria. We identified the main topics associated with the use of FHIR in digital health. Many articles dealt with topics related to FHIR implementations and use resources, as well as data migration data models. As expected, various application categories such as mobile apps, SMART on FHIR applications, and research applications were the main topics associated with FHIR. Multiple challenges in FHIR adoption and implementation were also highlighted in the included articles. Interestingly, only a small number of relevant articles addressed FHIR goals.

At the beginning of this study, we planned to identify some common aspects in this field by answering some fundamental research questions. Hence, we established six research questions to address the objectives, goals, applications, and challenges of FHIR emerging in recent years. As a result, we can propose a taxonomy of the literature, and identify gaps to be further investigated on existing challenges and issues related to use of the FHIR standard in recent years. We also identify other common and related aspects with respect to interoperability, privacy, authorization (access control), data type, and testing and validation tools. For example, interoperability between the provider and hospital systems poses additional barriers to effective data sharing. In addition, various testing and validation tools are used to improve server compliance with the FHIR specification. Moreover, FHIR specifications define different data types to access and process the FHIR resources element.

Various FHIR-related studies aim to address FHIR implementation challenges such as data migration and cross-institutional sharing of clinical data in the clinical environment [60,71]. The main findings that are presented in these reviews and some other related studies include the importance of realizing EHR data interoperability via adoption of FHIR by health care providers. This adoption might be essential for the improvement of health care services with respect to health data sharing, integration, and availability. Furthermore, use of the FHIR standard in the health care sector may enhance the chance of adoption of smart technologies in the health care domain, such as smartphones, mobile health apps, tablets, smart watches, fitness trackers, and any other future innovations [90]. Furthermore, use of artificial intelligence technologies and data sciences will also be dominant in implementing FHIR-based applications.

FHIR is viewed as the latest standard purely operating on resources, which are used for data storage, migration, and processing among multiple health care providers. The resources-based structure of FHIR is declared as distinct from other standards and is considered to be its main advantage. FHIR has several advantages that range from being a flexible standard, minimal implementation complexity, ability to display the patient history in a single document, granular data access, and avoiding message variability with the use of RESTful APIs.

FHIR is considered to be a unique pathway that can offer a solution to the interoperability issues of clinical data. Nevertheless, various studies indicate that FHIR also faces numerous challenges such as implementation, adoption, maintenance, mapping, and standard complexity. The RESTful API that makes FHIR unique from other standards also faces its own challenges in accessing sensitive health care data stored in the cloud environment [36].

Numerous studies have shown that several applications used in different domains are taking advantage of FHIR, including mobile apps, SMART on FHIR applications, research applications, electronic records and medical applications, graphic/image applications, HAPI FHIR application, and web-based applications.

To the best of our knowledge, this is the first comprehensive systematic literature review that focuses on FHIR-based EHR. There are some systematic literature reviews available in the FHIR domain; however, we found that the existing reviews are unable to explain the FHIR standard in detail. The FHIR standard is very rich, and therefore research in this domain is equally diverse with focus in various directions. Readers are interested in searching for articles that do not only introduce the FHIR standard but also explain various aspects of the standard in detail. For example, Lehne et al [91] only reviewed articles related to a general introduction of FHIR, without providing in-depth analysis on either FHIR or articles mentioning FHIR. In particular, the titles or identification of included articles and more comprehensive details of the included articles are missing. Based on a thorough reading, we concluded that this previous review could not fully introduce and address various aspects of FHIR, such as challenges, applications, goals, mapping, and implementation models. Moreover, individual aspects of the reviewed articles were not explained adequately. Therefore, it is not possible to find a corresponding article when interested in a particular topic. Further, the authors focused on articles published between 2002 and 2018, although the FHIR concept was only introduced in 2011; thus, this was a mixed review of EHR and FHIR with little focus on FHIR itself. Lastly, the authors included only 15 references in the review, which is not sufficient for a systematic literature survey.

Similarly, another systematic literature review [90] only explained the general concept and current status of FHIR, whereas core issues such as challenges, goals, and application implementation were not discussed. Important articles that discuss the FHIR resources used in various application implementations were also not included in this review. Although this previous review analyzed some articles in the FHIR domain, the list of articles was not provided or explained properly. Therefore, it is quite difficult for readers to search for articles related to specific information of interest, such as FHIR applications, goals, challenges, and used resources. This information is the core requirement for readers interested in this field. Thus, we concluded that the existing reviews only introduced the FHIR standard without performing a comprehensive analysis of the current state of the field.

In this work, we deeply explored the literature and identified articles that not only mention the FHIR standard but also discuss its major aspects such as core challenges, applications, goals, mapping, used resources, and implementation models. In addition, we highlighted every article along with their references and addressed aspects such as those mentioned above. We searched existing databases for articles on the FHIR standard published between 2012 and 2019, and then included every article that discussed or mentioned even a single aspect of the FHIR standard. This approach provides a convenient resource for readers to easily search articles of interest in the literature.

FHIR is a new standard in the health care domain. It is still in the early stages of development and evaluation, and consequently faces numerous obstacles. We believe that these obstacles might eventually be overcome, thereby opening a new roadmap to solving the problem of data interoperability in the health care sector, which is in line with the findings of the literature review and remains the main objective of our research.

This study was limited to aspects related only to the FHIR standard rather than the general health care concept. In this sense, the literature review focused exclusively on articles addressing FHIR concepts. This work sought to answer research questions proposed for providing an outline of the current literature related to FHIR without specifically assessing any computer system that refers to FHIR use. In addition, our search focused on articles published in various scientific journals related to health care and computer science within a limited time frame. This investigation was limited to articles selected from journals/conferences through implementations of standard steps of the systematic literature review methodology. We focused on scientific articles and did not address commercial or more technological approaches or solutions.

This study provides a systematic literature review regarding the FHIR EHR standard, with the main objective of identifying and discussing the main issues, challenges, goals, and possible benefits from adoption of the FHIR standard in the health care sector. We have explored the FHIR-related literature and investigated articles associated with the FHIR standard in health care information systems. We identified various data models, methods/techniques used in FHIR implementation, FHIR beneficiary applications, and resources used in FHIR implementation. Various data mapping techniques/approaches, key challenges, and primary goals of FHIR were also explored. We observed that FHIR studies mainly focus on clinical data interoperability and portability issues between health care information systems.

The FHIR standard is capable of providing an optimized solution for medical data exchange between two systems and will establish data-sharing trust among health care providers. Furthermore, the FIHR standard is identical in terms of the support of smart technologies such as smartphones, tablets, mobile health apps, smart watches, and fitness trackers, which could solve numerous health care problems that were not possible for the previous standards (ie, HL7 v2, v3 and CDA). Based on this thorough investigation of the literature, we recommend the FHIR standard as a future suitable solution for addressing the health care interoperability problem. Nevertheless, FHIR itself faces some challenges such as implementation, standard complexity, and adoption, among others. Therefore, further research is required to address these challenges.

This review on the standard, purpose, and applications of FHIR will provide readers with a more comprehensive view and understanding of FHIR. This review should also help researchers and health care information technology professionals to access FHIR-associated information in the research community and to assess its impact on digital health. Lastly, this work can provide a roadmap, and suggest possible directions for future research and development in the FHIR domain.