Impacts of FHIR on Data Quality and Decision Making in Healthcare
In the last two decades, health information technologies have enhanced accessibility and reduced costs of care but with significant data quality and decision-making challenges. Electronic health records (EHRs) enable hospitals, practitioners, and patients to report, store, and exchange information easily and accurately, unlike paper-based hospital management systems. These data processes can be done in a wide range of formats and structures from different devices, including computers, electronic medical equipment, and smartphones. This diversity, although makes care services more accessible, introduces incompatibility problems that impede interoperability of the inherent technologies. Earlier efforts to address this issue, such as High Level Seven (HL7) version 2 (v2), version 3 (v3), and clinical document architecture (CDA), led to loosely standardized solutions with restricted scope, a situation that resulted in proliferation of so many clinical data structures and application programming interface (API) architectures. HL7 latest version, Fast Health Interoperability Resources (FHIR), eliminates this disenfranchised implementation of EHRs by standardizing access, storage, and communication between different medical apps, users, and systems. The high level of interoperability increases not only data quality, but also the speed of clinical decision-making.
In a healthcare context, data quality refers to the ability of a particular set or source of information to adequately meet the needs of the expected application or utilization. According to Kaloyanova et al., data is fit for purpose if it is complete, consistent, accurate, current, recoverable, portable, understandable, and traceable (156). Most importantly, the information and underlying processes should comply with relevant regulations and laws regarding security, privacy, and confidentiality, especially the Health Insurance Portability and Accountability Act (HIPPA). Such data is available to the intended user at the right time with the required content in the correct format; it does not require further processing. It enables nurses and other practitioners to make informed decisions faster. Due to its enhanced interoperability standards, FHIR can facilitate realization of the data quality and decision-making objectives since it gives EHR users the capability to access, compile, and analyze disparate data sources using different apps and devices.
FHIR
FHIR is a standard that specifies how health information is exchanged across different EHRs and medical apps. The interoperability is achieved by defining the architecture of the exchangeable packets of health data into over 150 resources. The latter, which make the FHIR foundation, reflect the typical terms that are common in healthcare settings, including observation, practitioner, disease, location, institution, payment, document, and scheduling. The resources are classified into five: administrative, clinical, infrastructure, financial, and workflow (Ayaz et al., “The Fast Health Interoperability Resources” 3). The simplification enables software vendors, app developers, and hospitals to follow a consistent approach when creating EHR system architecture. This standardization eases interaction or compatibility of disparate healthcare technologies irrespective of the format of the stored data. In other words, dissimilar data sets and formats can be exchanged across varying apps or EHRs as long as the data is organized in acceptable FHIR resources.
FHIR’s RESTful API and SMART methodologies enable the enhanced cross-platform interaction. Unlike the predecessors that allow text-based (HL7 v2 and v3) or document-based interaction between users’ apps and devices with EHRs, HL7 FHIR supports access and utilization of health information systems through online representational state transfer (REST)ful APIs (Pavão et al. 1250). The latter (see fig. 1) is an architecture that guides the communication in the web. It makes the client and the server independent of each other, meaning that both web agents can interact with and understand each other even if the client is unaware of actions and identity of the server. Without a stringent method to authenticate and integrate clients, this RESTful API approach can expose EHRs to security and incompatibility risks. FHIR solves this challenge through SMART (Substitutable Medical Applications and Reusable Technologies) framework (Stoldt and Weber 193). The latter (see fig. 2) authenticates connections using OAuth and OpenID Connect, besides providing vendors, developers, users, and institutions with the freedom to try different types of applications and find the most suitable for the specific needs. Accordingly, many companies, including Microsoft and Amazon, and hospitals are nowadays implementing their EHR products as SMART on FHIR systems.
Fig. 1. Implementation of a RESTful API framework (“What is REST?” Codecademy, n.d., www.codecademy.com/article/what-is-rest)
Fig. 2. Illustration of the working principle of SMART on FHIR network (“SMART on FHIR: Introduction.” Smile Digital Health, n.d., smilecdr.com/docs/smart/smart_on_fhir_introduction.html)
Microsoft Azure API for FHIR. Azure API for FHIR is a cloud-based API provided by Microsoft as a Platform-for-a Service (PaaS). It enables anyone to create a RESTful API service to access and utilize resources from any EHR or clinical database that conforms to FHIR standards (see fig. 3). Through the PaaS, a patient, a physician, or a nurse can perform such RESTful functions as conditional creation, deletion, and update of resources as well as recovery of files deleted from a FHIR database (“What is Azure API for FHIR?”). Other capabilities include secure processing of clinical data in compliance with HIPPA, data analytics based on artificial intelligence, and bulk access to FHIR-enabled server but with role-based access control privileges. Microsoft has also recently introduced Azure Health Data Services and FHIR Server for Azure (“What is Azure Health Data Services?”) Both apps are PaaS, with the former integrating other standards than FHIR and the latter being targeted at users interested in a provisioned, scalable cloud-based database. These FHIR APIs provide healthcare organizations and other clinical data users with web-based tools to access and interact with any EHR.
Fig. 3. Illustration of access control in Azure API for FHIR (“Microsoft Entra Identity Configuration”)
Amazon HealthLake. Amazon introduced HealthLake in 2021, a cloud-based platform that enables healthcare stakeholders to access clinical data securely. Recently, the firm integrated three key capabilities into the product: “SMART on FHIR authorization, Patient Access API, and FHIR Bulk Data Access API” (T. Syed et al.). As a SMART on FHIR application, HealthLake can enable hospitals and app developers as well as integrators to control EHR and database access (see fig. 4). On the other hand, the Patient Access API functionality can give users the ability to build mobile and web-based apps that can be deployed as clients to exchange data with FHIR-compliant databases and servers. The bulk functionality is designed to facilitate transfer of large volumes of information across multiple EHRs, devices, and users. However, these features are still in development stage; they are yet to be released to the healthcare market.
Fig. 4. The working principle of AWS HealthLake FHIR system (T. Syed et al.)
Data Quality
Consistency. The most prominent feature of data quality in FHIR-based healthcare information technologies regards concordance. Kaloyanova et al. define consistency as the extent to which characteristics of data from one system agree with attributes of similar information from a different technology (156). FHIR promotes data uniformity through interoperable architecture of EHRs or databases and RESTful APIs. According to Stoldt and Weber, a compliant EHR relies on specified profiles to classify objects or data packets (193). In other words, compatible datasets are organized into five key resources, each defined and sized to reflect the unique conditions of the target hospital or healthcare environment. For example, the financial resource in FHIR databases comprises payment, insurance, statements, support, and invoices (Ayaz et al., “The Fast Health Interoperability Resources” 3). It can include all these details for a large hospital, but two or three entries for a doctor’s office. The profiles make similar resources consistently structured, implying that a user accessing the resource from anywhere, any device, or any EHR receives the same results.
If widely implemented in the healthcare sector, FHIR can enhance data quality by making information stored in clinical information systems uniform across platforms and hospitals. CoxHealth and the Gravity Project efforts demonstrate this benefit (“Real-Life FHIR Implementations”). In the Gravity Project, one of the accelerator plans under the HL7 FHIR framework, a group of health and social care workers are involved in creating resources to be used in integrating social determinants of health into SMART on FHIR. The outcome of this collaborative work will be standards that developers, vendors, and IT staff in medical facilities can utilize to design and deploy industry-wide interoperable apps. By contrast, CoxHealth is already implementing the SMART on FHIR RESTful API approach in form of VisualDx. Through this app, clinicians are able to access and retrieve consistent patient details from the hospital’s Cerner EHR. Both operations demonstrate that FHIR is a dependable technology that can provide the seamless interoperability currently missing in EHRs, medical databases, mobile and electronic health apps, wearable devices, and crossplatform systems.
Completeness. The integrated nature of FHIR EHRs can foster data completeness. From a quality perspective, the latter describes the extent to which clinical data contains all expected characteristics. RESTful APIs and SMART frameworks embedded on FHIR-compliant hospital systems can guarantee this completeness. The statelessness of REST protocol implies that information from client and caregiver-facing app can interact with EHRs or information warehouses simultaneously and seamlessly. Moreover, further processing of the information, such as transfer from one device to another and formatting, are unrequired. Accordingly, resources can be integrated into the FHIR databases as captured from the point of service (Wesley et al. 2223). This capability differs from current electronic medical records (EMRs) in which incompatibility forces healthcare professionals to manipulate data being transferred from one technology to another. The information handling creates risks of omissions and commissions, affecting the quality of the information. Thus, FHIR is capable of making resources more complete compared with existing clinical data management systems.
Fig. 5. Model of RESTful patient-facing App (Wesley et al. 2222)
Fig. 5 illustrates how FHIR can be deployed to improve data quality between hospital feedback system and EHR. Instead of using paper-based or online-based questionnaires and then transferring the data to the EHR, a RESTful PRO app can be developed based on SMART on FHIR framework to automatically interact with the EHR. To participate in the survey, the client can use a portable digital assistant (PDA) provided by the hospital and preinstalled with the service or can download the app and install in their personal smartphone (Wesley et al. 2223). On accessing the app (process A), the individual can be requested to input their details (B), which can be subsequently authenticated by FHIR server through OAuth and OpenID Connect protocols (C). Once authorized, the patient can interact with the FHIR-compatible data hub.
The verified app fetches the patient information from the hub and the first question from the EAC (D). Based on computer adaptive testing, EAC administers all the questions. In process E, the finished survey is stored in the data hub, from which a caregiver can access the feedback from the hospital EHR (F). The practitioner only needs to query the EHR from their workstation; the EHR communicates (F) with the data hub to pull and serve (H) the practitioner with the patient feedback in real time (Wesley et al. 2223). Through this exchange, the client’s responses are available to the caregiver on the EHR as entered in the PRO app. This seamless interoperability across the app, data hub, EAC, and EHR with minimal human intervention can safeguard data completeness when used in feedback systems, enhancing quality of the information.
Currency. FHIR is based on APIs acting as mediators between client and server-side apps, an approach that ensures that users have access to up-to-date data. Currency is a quality dimension that measures how timely is the information given the setting (Kaloyanova et al. 156). RESTful APIs provide the means to update EHRs in real-time with resources’ information from every stakeholder, including patients, insurers, doctors, nurses, and hospital administration. These operators can interact automatically with SMART on FHIR EHRs, implying that the details created are continuously integrated into the central databases. An entity that queries the data warehouse through a compliant technology is able to receive and view the most current information about the resources of interest. FHIR enhances timelines by giving every customer the methods to easily, properly, and securely keep common records updated.
The autonomous exchange across the health information technologies eradicates human or system aggregators or processors, which delay update of records and impede immediate access of clinical or administrative data. This feature of information quality is apparent in 1upHealth, a SMART on FHIR platform that provides healthcare institutions with a solution to provide real-time, accurate, and complete data. It enables the organizations to allow all stakeholders, including patients, to update EHRs without unnecessary interventions. The product consists of modules for key hospital resources, which are automatically integrated into the 1upHealth to provide end users with current information (1upHealth). It is also cloud-based, implying that individuals can share relevant details from anywhere at any time. As demonstrated in 1upHealth, the shared approach supported by the RESTful architecture makes FHIR capable of guaranteeing up-to-date health information at all times.
Correctness. Through the SMART methodology, FHIR can make data exchanged and stored across disparate health systems more accurate. Correctness indicates whether data is a true reflection of the clinical concept or term in semantics and syntax. Medical information is composed of plain language, numbers, clinical terms, and abbreviations. Various EHRs use inconsistent formats and structures, with diverse amount of details for each data type. Practitioners are expected to enter these details while providing care, including in critical or emergency services. Risks of wrong information, misplaced details, omitted entries, and improper transcriptions are high in such settings. FHIR can alleviate this problem by enabling hospitals to try different technologies until they find the right fit for every need without losing interoperability of systems. A study by Shah found that FHIR enhances data accuracy by minimizing errors during data entry or eliminating the need to transfer information from one EHR to another manually (3). Compared to conventional data management technologies, FHIR is better in fostering quality of medical information.
Plausibility. FHIR enhances plausibility, validity, or relevance through standardization of data structures. R. Syed et al. observe that information is believable if it is granular and meaningful to the specific context (7). FHIR accomplishes these objectives by providing a framework to record, transfer, and store data. The latter is indexed as resources, which are then granulized into separate attributes. As previously stated, the resources entail the typical concepts used in clinical settings, such as patient, observation, and administration. Within each profile, the attributes are organized in granular format, reflecting the context of the provider. For example, the patient resource can include clinical identification number, name, social security, ethnicity, date of birth, allergies, age, gender, income, and location of residence. Healthcare organizations, software vendors, and app developers can adjust and scale these objects to make the resource unique to the target environment according to the applicable regulations and laws (“8.1 Resource Patient”). Through FHIR, EHRs and wearable devices as well as mobile and electronic health services can be configured to be relevant to the medical context, whether based on type of disease, patient, or care.
Decision-Making
FHR improves the speed of decision-making by fostering real automation and interoperability of healthcare systems. Medical information technologies have revolutionized access, delivery, and management of care. For example, wearable devices and mobile health apps enable individuals to consume clinical information and monitor health status in real-time. On the other hand, hospitals have access to EMRs, videoconferencing, imaging, and clinical decision support (CDS) systems to process and record protect health information (PHI) of their clients. Even though these innovations signify an improved way to offer quality, patient-centered services, they are mainly proprietary; they are poorly interoperable and incompatible (Thiess et al. 2). This situation makes communication across the technologies or facilities manual, redundant, and tedious, which slows the process of taking clinical actions. FHIR solves these challenges to make the planning faster by eliminating manual data processes, harmonizing access and transfer standards of hospital electronic systems, speeding analytics, and enhancing efficiency of evidence-based practice.
Automation of Data Processes. FHIR eradicates or minimizes manual input and transfer of health data into EHRs and other clinical systems. In spite of deployment of advanced information technologies in healthcare, data is mostly entered into EMRs manually. At the hospital, practitioners transcribe patient narratives and type clinical observations, laboratory details, diagnosis, and treatment instructions into computer systems. In other cases, referrals are done by telephone or fax, implying that typing of the patient details is repeated at the receiving facility’s HER (Odisho et al. 406). Even within the same institution, data is physically captured into different information systems at each unit, department, or ward. These manual tasks are time-consuming as the details have to be carefully recorded and verified before the client is released to the next point of service. For instance, if a laboratory technologist finds missing instruction or ineligible note, they have to call and confirm the right details with the physician requesting the tests. These physical processes cause delays in diagnostic and therapeutic decisions, compromising patient outcomes.
FHIR automates data entry and exchange processes, making the communication between different stakeholders, devices, and systems seamless. The RESTful API approach allows systems to be designed for every data source, and the technologies are then combined through SMART to create an interoperable FHIR-compliant system. For instance, in a clinical research, Odisho et al. build and tested a cloud, SMART on FHIR hospital referral system that integrated diverse EMRs and standards (see fig. 6). They found that the FHIR effectively eliminates the physical “processing of faxed referrals, looking up the patient in the EHR, entry of demographic information, updating insurance and coverage, creating the referral, finding patient information, calling the patient, and logging the referral” (408). Clinicians do not have to wait for client or treatment information to be typed, transferred between computer-based forms, or verified. Instead, the data is captured or updated at the source and interchanged with the successive network points in a readily accessible and understandable structure. The efficiency in data processing enables practitioners to make decisions faster than in conventional EHR or paper-based environments.
Fig. 6. An automated hospital referral system based on FHIR standards (Odisho et al. 408)
Data Access and Transfer Standardization. As an interoperability standard, FHIR provides a framework to harmonize data from disparate medical systems and inform efficient decision-making. Currently, the health sector is characterized by many proprietary technologies based on differing standards that hardly communicate with each other across devices. Every institution or practitioner adopts the most suitable for their context; even departments or units in a large hospital implement varying EHRs. The diversity slows planning, treatment, or administrative actions in clinical practice because of the high level of efforts required to convert and make data compatible across the technologies. FHIR, such as AWS HealthLake, 1upHealth, and Azure API for FHIR, creates a hub through which these traditional EHRs can seamlessly interact with one another as well as with FHIR-compliant EMRs and other health information technologies. For example, AWS HealthLake allows users to store and utilize information from SDoH, claims processing, laboratory systems, EHRs, and diverse devices without reprocessing (T. Syed et al.). This capability accelerates the pace at which physicians and nursing professionals can make informed plans, enhancing care delivery and outcomes.
Data Analytics. Automated data analysis is the other significant way through which FHIR enhances the decision-making speed in clinical settings. In today’s environment that lacks such standardization, healthcare workers mine high volumes of data in different formats from disparate sources to inform actions. The mining process is labor-intensive and slow because of the amount and types of information involved. With the RESTful API feature, FHIR-based platforms have data analytics module or API through which practitioners can rely on user-friendly apps to query and receive actionable feedback for use cases, such as a specific patient, medical condition, or adverse drug interactions (Thiess et al 4; Ayaz et al., “Transforming Healthcare Analytics” 4). SMART supports the API ecosystem by empowering the FHIR-enabled network to authenticate and authorize only genuine requests. Like in any other web-based search service, users only need to type their keywords into SMART on FHIR app and retrieve the relevant details. These two capabilities facilitate generation of actionable intelligence in form of texts, images, graphs, and reports, on which medical professionals can depend to deliver efficient services to patients.
Majority of cloud-based FHIR platforms offer inbuilt data analytics options. In 1upHealth, 1up Analyze is provided as one of the five modules (see fig. 7). The company describes the functionality as a structured query language (SQL) on FHIR that enables clinicians, as well as other stakeholders in the care value chain, to quickly ask FHIR-compliant EHRs hard questions and receive targeted responses about payment and medical details of a particular client. AWS HealthLake analytics employs the same technique to provide patients, developers, and practitioners with CDS dashboards (T. Syed et al.). The analytical capability saves time needed to access a database, retrieve and process the relevant information, and combine the details with those from other data warehouses for evidence that can influence a diagnostic or therapeutic decision. By typing a clinical query on SQL on FHIR platform and hitting enter, a professional is able to receive the right response and take the necessary action with little time lapse.
Fig. 7. Data analytics module in 1up FHIR ecosystem
Efficiency of Evidence-Based Practice. FHIR can make health policies and research outcomes readily available to medical professionals for not only quick, but also reliable and accurate planning. Care delivery is heavily dependent on translation of research and policy into practice. Nevertheless, like any other EHR, political and study data systems are designed as silos, compelling practitioners to spend much effort and time in meticulously finding the most suitable possible action from large amounts of evidence on best practices, regulations, and procedures (Lomotan et al. 113). This problem is exacerbated by new study findings, new laws, or amended regulations released to the market regularly. FHIR fastens the process of transforming evidence into clinical guidelines and procedures by providing a framework to connect research, policy, and practice databases. In the same way, it eases the reverse process, that is, sampling and collection of health data from EHRs in clinical research. For example, AHR CDS Connect (see fig. 8) is a FHIR-based network that facilitates seamless interchange of data between researchers, policymakers, care providers, and communities. It reduces the amount of time and activities required in making study findings usable in disease and patient management. If adopted by many small and large clinical centers, FHIR can improve the turnaround time to create actionable evidence from policies and research.
Fig. 8. The link between research electronic system and EHRs through FHIR-compliant CDS Connect (Lomotan et al. 114)
Conclusion
Through its RESTful API and SMART architecture, FHIR improves data quality and quickens decision-making in healthcare sector. It replaces most of the manual tasks with automated data processes, an innovation that guarantees accuracy and completeness of information exchanged and stored in EHRs. The specification further promotes data consistency, relevance, and currency through standardization of collection, storage, and interchange of hospital data among disparate EHRs, technologies, devices, and formats. The SMART approach gives users the capability to try different technological solutions in a sandbox and select the most relevant to the specific context. The potential use cases build from FHIR resources simplify the process of taking clinical actions. In particular, the eradication or minimization of physical tasks reduces the time spend in typing, copying, transcribing, or transferring client or hospital details from one EHR to another. Harmonized standards make all health information technologies interoperate seamlessly provided the data is structured, stored, or accessed in FHIR-compliant resources and servers. Data analytics and utilization of research results and policies are made efficient, as well. These tasks are integrated into SMART on FHIR EHRs like any other RESTful API. FHIR improves data quality and efficiency in planning by making traditional EMRs truly automatic, interoperable, and compatible.
Works Cited
1upHealth. “1up FHIR Platform” n.d., 1up.health/products/1up-fhir-platform/. Accessed 13 Feb. 2024.
“8.1 Resource Patient – Content.” HL7 FHIR Release 5, 26 Mar. 2023, www.hl7.org/fhir/patient.html. Accessed 13 Feb. 2024.
Ayaz, Muhammad, et al. “The Fast Health Interoperability Resources (FHIR) Standard: Systematic Literature Review of Implementations, Applications, Challenges, and Opportunities.” JMIR Medical Informatics, vol. 9, no. 7, 2021, pp. 1-21.
Ayaz, Muhammad, et al. “Transforming Healthcare Analytics with FHIR: A Framework for Standardizing and Analyzing Clinical Data.” Healthcare, vol. 23, 2023, pp. 1-28.
Kaloyanova, Kalinka, et al. “Addressing Data Quality in Healthcare.” CEUR Workshop Proceedings, vol. 2933, 2021, pp. 155-164.
Lomotan, Edwin, et al. “To Share is Human! Advancing Evidence into Practice through a National Repository of Interoperable Clinical Decision Support.” Applied Clinical Informatics, vol. 11, no. 1, pp. 112-121.
“Microsoft Entra Identity Configuration for Azure API for FHIR.” Microsoft, 12 Oct. 2023, learn.microsoft.com/en-us/azure/healthcare-apis/azure-api-for-fhir/azure-active-directory-identity-configuration. Accessed 13 Feb. 2024.
Odisho, Anobel, et al. “Design and Development of Referrals Automation, A SMART on FHIR Solution to Improve Patient Access to Specialty Care.” JAMIA Open, vol. 3, no. 3, 2020, pp. 405-412.
Pavão, João, et al. “The Fast Health Interoperability Resources (FHIR) Standard and Homecare, a Scoping Review.” Procedia Computer Science, vol. 29, 2023, pp. 1249-1256.
“Real-Life FHIR Implementations.” Prolifics, n.d., prolifics.com/us/resource-center/specialty-guides/fhir-guide/fhir-implementations. Accessed 13 Feb. 2024.
Shah, Wasim Fathima. “Exploring the Impact and Evolution of Fast Healthcare Interoperability Resources (FHIR) on Healthcare Systems, Personal Health Records and Data Security.” Journal of Health Education Research & Development, vol. 11, no. 3, 2023, pp. 1-5.
Stoldt, Jean-Philippe and Jens Weber. Safety Improvement for SMART on FHIR Apps with Data Quality by Contract.” Proceedings of IEEE International Conference on Software Architecture Companion, IEEE, 2020, pp. 192-195.
Syed, Tehsin, et al. “New FHIR API Capabilities on Amazon HealthLake Helps Customers Accelerate Data Exchange and Meet ONC and CMS Interoperability and Patient Access Rules.” AWS, 7 June 2023, aws.amazon.com/blogs/industries/new-fhir-api-capabilities-on-amazon-healthlake-helps-customers-accelerate-data-exchange-and-meet-onc-and-cms-interoperability-and-patient-access-rules/. Accessed 13 Feb. 2023.
Syed, Rehan, et al. “Digital Health Data Quality Issues: Systematic Review.” Journal of Medical Internet Research, vol. 25, 2023, pp. 1-23.
Thiess, Henrik, et al. “Coordinated Use of Health Level 7 Standards to Support Clinical Decision Support: Case Study with Shared Decision Making and Drug-Drug Interactions.” International Journal of Medical Informatics, vol. 162, 2022, pp. 1-17.
Wesley, Deliya, et al. “A Novel Application of SMART on FHIR Architecture for Interoperable and Scalable Integration of Patient-Reported Outcome Data with Electronic Health Records.” Journal of the American Medical Informatics Association, vol. 28, no. 10, 2021, pp. 2220-2225.
“What is Azure API for FHIR?” Microsoft, 28 Sep. 2023, learn.microsoft.com/en-us/azure/healthcare-apis/azure-api-for-fhir/overview. Accessed 13 Feb. 2024.
“What is Azure Health Data Services?” Microsoft, 1 Sep. 2023, learn.microsoft.com/en-us/azure/healthcare-apis/fhir/overview. Accessed 13 Feb. 2024.
