1 Introduction

Blockchains, most recognized as the technology behind cryptocurrencies [1], have found a wide application area, including supply chain management [1], electronic copyrights [2], digital voting [3], used car trading [4], and real estate management [5]. Technically, blockchains are tamper-resistant and tamper-evident distributed databases. They enable transparency and traceability of data transactions and thus ensure a trusted data management system for users [1].

Our research revealed that blockchain applications are widely adopted in the health domain. For instance, the Dentacoin application [6] uses a blockchain network to associate dental clinics and patients for various services. MediBloc [7] and SRCoin [8] utilize blockchain technology to process and manage health data for healthcare professionals, patients, and researchers in a secure way. Medishares [9] provides a blockchain platform that serves as a decentralized marketplace for insurance management. It enables participants to create insurance in smart contracts; thus, anyone in the insurance circle can securely access insurance-related data. The AI Doctor platform [10] is a decentralized artificial intelligence virtual doctor application. In this application, users who provide their health data are rewarded with AIDOC tokens on the blockchain platform. Later on, the AIDOC tokens can be used as credentials for health insurance at preferential rates. Patients’ data stored on the platform can also be used by various organizations such as pharmaceuticals, medical institutions, and AI companies for drug development and clinical research. MedicalChain [11, 12] is another blockchain-based application running at various hospitals across the UK which enables patients to control their medical data. It provides a platform that enables users to give conditional data access to stakeholders such as doctors, laboratories, hospitals, pharmacists, and health insurers. Patientory [13] is a HIPAA [14] compliant platform that allows patients, clinicians, and healthcare organizations to securely access and transfer sensitive health information while providing actionable insights to improve health outcomes.

According to the 2020 OECD policy brief, adopting blockchain on a national scale is rare; however, there are government initiatives toward deploying blockchain in the health domain [15]. In 2016, the Estonian government commenced a project to use blockchain technology to find new and innovative ways to secure the health records of its 1.3 million residents. Estonia has become the first country to use blockchain in the health domain on a national scale [2]. Malta is the other country that deploys blockchain in the health domain at the national scale. A blockchain application called Dwarna manages the “dynamic consent” of citizens on biospecimens for research studies [16].

Compared with blockchain applications developed for non-safety critical domains, the health domain software applications need to be audited and verified by regulatory bodies and be ensured that they fulfill the regulatory standards.

Different regulations may be applied depending on the region in which healthcare applications are marketed. The Food and Drug Administration (FDA) [17] and the Medical Device Directives [18] regulate the healthcare applications marketed in the USA and the European Union, respectively. Currently, no policy or guideline is published to regulate blockchain usage in the healthcare system/software development by regulatory bodies to the best of our knowledge. However, recently, the FDA unveiled the Technology Modernization Action Plan [19], claiming that blockchain technology is on their radar for addressing health domain-related challenges. Additionally, IBM, Walmart, and Merck have been selected for an FDA pilot program that will use blockchain technology to improve the security of drug supply chains [20]. These latest developments indicate that regulatory bodies will remain focused on blockchain research in the health domain.

The use of blockchain in healthcare solutions has gained significant momentum in recent years. Various blockchain-oriented applications and research have been released by practitioners and researchers. We believe that this is an appropriate time to explore the recent advancements and provide a synthesis of what practitioners and researchers think about the potential usage areas of blockchain in the health domain and the extent to which blockchain technology provides a solution to the field. We also aim to lead future research directions by exploring the new challenges that blockchain introduces to the domain.

With these purposes, we present a multivocal literature review (MLR) of blockchain adoption in the health domain in this study. MLR approach provides a methodology to systematically review both formal literature and gray literature. Luxembourg states that “gray literature is produced on all levels of government, academics, business and industry in print and electronic formats, but which is not controlled by commercial publishers, i.e., where publishing is not the primary activity of the producing body” [21, 22]. Thus, both theoretical and practitioner (e.g., developers, designers, and quality engineers) viewpoints could be reflected in the study with the MLR approach.

In this study, we aim to answer the following research questions:

  • RQ1: What are the blockchain application areas in the health domain and what is the motivation behind it?

  • RQ2: What are the challenges of developing health software?

  • RQ3: To what extent does blockchain technology contribute resolving existing software development challenges in the health domain?

  • RQ4: Does blockchain introduce new challenges to software development in the health domain?

  • RQ5: What are existing solution suggestions to the blockchain-related challenges in the health domain?

In 2021, we conducted a systematic literature review (SLR) [23] with the same purposes. The SLR included 27 formal literature resources from 2016 to 2020. On the other hand, the MLR consists of 78 formal literature and 23 Gy literature resources. Among 78 formal literature studies, only 25 exist in the previous SLR we performed. This increase indicates how rapidly the blockchain health applications domain is growing. The number of new papers that passed the quality assessment criteria and were added to the paper pool has tripled in a single year.

There are other review studies focusing on potential application areas of blockchain in the health domain [24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39], but they do not cover and discuss the challenges faced by practitioners and associated solutions in depth.

This MLR study contributes to the literature with a comprehensive review and analysis of the.

state-of-the-art blockchain research from the health domain perspective. It acknowledges all current application areas of blockchain technology in the health domain. It also explores existing health domain challenges in a detailed way. Although most of these challenges are not new (e.g., being unable to specify falsified drugs and prevent modification of clinical trial data), the COVID-19 pandemic signified their impacts on people’s lives and the environment.

In the post-pandemic world, companies are exploring ways to run their businesses more effectively and make their businesses resilient against minor, major, and unexpected events. Blockchain, one of the emerging technologies, could support businesses in being resilient.

Therefore, from the managerial perspective, this study shows the potential usage areas of blockchain in the health domain and supports strategic planning and decision-making processes. Policymakers could also use this extensive study to understand the existing blockchain adoption challenges and make better plans to address problems. Finally, this MLR study sheds light on significant challenges from a theoretical perspective and leads future research.

The rest of the paper is structured as follows: Sect. 2 provides a background on blockchain technology and health domain software. Section 3 explains our research methodology while performing the MLR. Section 4 presents results and discussion. Finally, in Sect. 5, we draw conclusions and present suggestions for further work.

2 Background

2.1 2.1 Blockchain technology overview

National Institute of Standards and Technology (NIST) defines blockchain as a tamper-resistant and tamper-evident digital ledger implemented in a distributed manner, generally without a central authority [1]. Data can only be added to a blockchain after several approvals, and once added, it cannot be deleted or modified without the approval of network participants.

The basis of blockchain technology emerged in the late 1980s and early 1990s. Leslie Lamport developed the Paxos protocol in 1989 [40], which provides a consensus model to reach an agreement on an unreliable network. Haber and Stornetta proposed a method for the secure timestamping of digital documents in 1990 [41]. This method enables signing documents digitally so that none of the signed documents are changed [41]. Satoshi Nakamoto published an electronic cash system in 2008. The study includes the concepts of consensus models among distrusted participants and digital signatures [42]. In 2009, the Bitcoin cryptocurrency blockchain network was established [1]. Several different cryptocurrencies followed Bitcoin, such as Ethereum [43].

Blockchain networks are categorized according to network participants’ “permission” requirements [1]. If a participant can publish a new block to the network without getting permission from any authority, it is called a “permissionless” blockchain network. Publishing blocks in a permissionless network requires consensus of the majority of the participants. The network is called “permissioned” when only certain participants are allowed to publish blocks. In permissioned blockchain networks, data blocks must be approved by authorized nodes in a network.

Blocks are the records that comprise blockchain networks. Each block has its own hash, data, and the preceding block’s hash. This hash is a unique value created from block data by a cryptographic hash function [1]. Block data includes a list of transactions and ledger events within that block. If a single data is changed in a block, the associated hash also changes. Therefore, attempts to modify data can be detected easily [1]. Figure 1 was adapted from [1] and shows the generic chain of blocks.

Fig. 1
figure 1

Chain of Blocks based on Ref [1]

Full size image

The interactions among participants of the blockchain are represented by transactions. These transactions are validated by nodes, each of which is an individual system within a blockchain network. “Full nodes” store the entire blockchain and validate the transactions, whereas “lightweight nodes” do not store a full copy of data and usually pass it to the full nodes to be processed [1].

Consensus models provide agreement among non-trusted parties by ensuring the validity and authenticity of blocks without a central authority [1]. Network participants agree that a transaction is valid through consensus models.

Asymmetric-key cryptography is used in blockchain technology to create a trusting relationship between network participants. Asymmetric-key cryptography [1] offers a method for verifying transaction integrity and authenticity. Cryptographic hash is used to compare large datasets and verify that data has not been tampered with by creating a unique small fingerprint for all data [3, 4]

A smart contract, a concept first proposed in 1994 by Nick Szabo [5], is a piece of code on the blockchain network [1] that emerges to minimize the need for trusted intermediaries. If a series of pre-defined conditions are fulfilled, a smart contract executes itself, and the execution results are recorded in the blockchain. Smart contracts are deployed on blockchain networks using cryptographically signed transactions; therefore, they are tamper-resistant as well [1].

Participants in certain blockchain networks utilize wallets, specific software for public and private key management and calculating digital assets [1].

As the health domain is a multi-stakeholder structure, blockchain applications in this domain often require interoperability. The NIST defines blockchain interoperability as “a composition of distinguishable blockchain systems, each representing a unique distributed data ledger, where atomic transaction execution may span multiple heterogeneous blockchain systems, and where data recorded in one blockchain are reachable, verifiable, and referable by another possibly foreign transaction in a semantically compatible manner.” As this definition implies, information assets will be exchanged and used between different ledgers, and this exchange needs to be verifiable. Although the NIST defines interoperability as the communication between heterogeneous blockchain systems, Belchior et al.’s [44] definition covers homogeneous blockchain systems as well. In both cases, transactions require a trusted third party to ensure the correctness of underlying protocols [44]. It should be noted that complete interoperability may not be achieved due to reliability issues among different blockchain networks [45]. During a transaction that spans multiple ledgers, a source blockchain issues transactions against a target blockchain, and an exchange occurs between the source node and the target node. Decentralization is achieved when participants elect a source node from the source blockchain and a target node from the target blockchain [44].

2.2 Health Domain Software Overview

Health domain software must comply with the regulatory requirements defined by international standards and health regulatory agencies. Figure 2, adapted from Heidenreich [46], categorizes health domain software in terms of the regulatory compliance requirements.

Fig. 2
figure 2

Health Domain Software Categories and Standards—adapted from Ref [46]

Full size image

As shown in Fig. 2, health domain software is categorized into health software and medical software. Medical software covers both the software embedded in medical devices and software that serve as a medical device (SaMD). On the other hand, health software includes SaMDs and other health software applications. Medical device software (MDS) processes, analyzes, or creates medical data [47]. Various devices, including EKG monitors, insulin pumps, and medical imaging devices, are run with the medical device software applications. Software as a medical device (SaMD) is used for medical purposes and is not necessarily a part of hardware [48]. SaMDs may be used on desktop computers, tablets, smartphones, and smartwatches to aid healthcare professionals and patients in treatment planning, medical image viewing, heart-rate monitoring, and drug dosage calculations. Other health software, on the other hand, is stand-alone health applications that are executed on hardware [46], which can be exemplified as healthcare data management and clinical software applications.

These different types of software are subject to various regulations. While health software applications have to comply with the ISO/IEC 82,304 standard [49], medical software applications require compliance with the IEC 62,304 standard [50]. The ISO/IEC 82,304:2016 Health Software standard defines the processes and practices for guiding healthcare software development. Furthermore, it refers to the safety and security of health software that runs on general-purpose IT platforms, such as smartphones and tablets. IEC 62,304:2006 specifies the requirements to be followed while developing medical device software.

Besides the aforementioned two primary standards, the following standards are applied in the health domain:

  • ISO 14971:2009 [51]: the international risk management standard for the production of medical device

  • IEC 80,002–1:2009 [52]: the standard guiding the application of the ISO 14971 standard for medical device software

  • IEC 80,002–3:2014 [53]: the process reference model for medical devices that defines software life cycle processes based on IEC 62,304:2006

While ISO 14971:2009 is an international risk management standard for medical device production, IEC 80,002–1:2009 guides the application of the ISO 14971 standard for medical device software. IEC 80,002–3:2014 is a process reference model for medical devices that defines software life cycle processes based on IEC 62,304:2006.

Software applications must be audited by different regulatory agencies depending on the region in which the application is marketed. The Food and Drug Administration (FDA) [17] in the USA and the Medical Device Directives [18] in the European Union are responsible for conducting the regulatory audits. Above, a detailed explanation is given for the standards valid in the EU. The FDA also provides clear guidelines such as developing clinical decision support software [54], software as a medical device [55], wireless medical devices [56], and medical device software [57].

As mentioned above, several standards and guidelines exist on various components of the health domain system and software development. Blockchain technology is on the radar of the FDA [19, 20].

3 Research methodology

Considering the purpose of our study, both contextual information and evidence from the industrial community [58] would be significant. We specified that many sources indicate high practitioner interest in the topic. Multivocal literature review (MLR) suits our purposes very well, as it is used for addressing research questions by using both formal literature (FL) and gray literature (GL) sources [21]. We systematically applied Garousi et al.’s MLR guideline [21] to perform the MLR process in this study. We used a Google Drive spreadsheet to collectively extract and analyze the raw data. The source pool and extracted data are publicly available at [59].

The overview of the MLR process is given in Fig. 3.

Fig. 3
figure 3

The Phases and Steps of the MLR process we applied

Full size image

The definition of the “white,” “gray,” and “black” literature differs among studies. For instance, books are a part of the gray literature, according to [60]. On the other hand, they are considered part of the white literature, according to [61]. Compiling these studies ([60] and [61]), we agreed on the spectrum classification given in Table 1. To classify the sources based on their types in the gray literature, we used an existing model from Garousi et al.’s guideline [58]. We revised the examples of the second-tier category, i.e., preprints, e-prints, technical reports, lectures, and datasets.

Table 1 Spectrum of the “white,” “gray,” and “black” literature
Full size table

3.1 Planning the MLR

We decided to carry out this MLR by observing the substantial increase in research in this area after conducting the SLR in 2020. We explored the existing literature review studies before proceeding with the research. At this stage, we also revisited the research questions given in Sect. 1. Below, we briefly discuss to what degree our research questions were covered in the previous SLR studies on developing healthcare applications using blockchain technology. The papers that do not follow a systematic literature review approach were excluded.

We list the previous SLR studies along with their publication dates, the number of the papers included in each of these studies, and the years covered in Table 2. We also added the columns for our research questions stated in Sect. 1 to highlight the similarities and differences of these SLRs with our study.

Table 2 Systematic literature review studies on blockchain in the health domain
Full size table

We specified that while the vast majority of the SLR studies in Table 2 addressed the potential application areas of blockchain in the health domain (RQ1), none of them explored the motivation for adopting blockchain in this field in detail. Thirteen SLR studies [24, 25, 28, 30,31,32,33,34,35, 38, 39, 62, 63] explore the inherited challenges in health software development that blockchain addresses (RQ2). In addition, these 13 studies discussed the role of blockchain technology in resolving these inherited software development challenges. However, they do not include technical details or explanations for solving these challenges.

Eleven studies [24,25,26,27,28,29, 32, 33, 36, 39, 62, 63] explored the challenges of developing health applications using blockchain technology (RQ4). Among these 11 studies, only Agbo et al. [26] and Al Mamun et al. [39] provided information on possible answers to these blockchain-related problems (RQ5).

Flangovan et al. [37] focus on the application areas of blockchain in the health domain and the reasons for its use. Adere [38] analyzes trends and highlights the potential benefits of blockchain deployment in IoT and healthcare. Al Mamun et al. [39] show the distribution of blockchain types and platforms adopted by the reviewed articles. Jadhav and Deshmukh [62] highlight the challenges faced and the potential benefits of blockchain technology in the healthcare supply chain. Attaran [63] identifies blockchain application areas for pharmaceuticals, challenges, and opportunities for implementing blockchain technology in healthcare. These studies mainly provide an overview of the use of blockchain in the health domain. On the other hand, our MLR study contains detailed information on constraints and solution proposals with information from various sources.

Different from the studies mentioned above, our research includes:

(i) the most recent studies in the literature searched until October 2021 (78 formal pieces of literature).

(ii) gray literature sources (selected and extracted using a systematic approach) considering the experiences of practitioners in the implementation of blockchain. The MLR includes a wide variety of gray literature resources such as videos, white papers, Web pages, Q/A sites, and formal literature.

(iii) This is the first MLR performed systematically in blockchain health domain. Fang et al. [36] included gray literature in the systematic literature review; however, the study includes fewer sources with less variety (white papers, reports). It focuses on the current trends of the blockchain, key design decisions, current limitations encountered, and future directions. However, the study only discusses these topics in the personal health records field. In this MLR, we explore the use of blockchain in the health domain from a broader perspective covering medical supply chain management, clinical trials, precision medicine, remote patient monitoring, IoT use with blockchain, electronic health records, public health surveillance, and health insurance.

3.2 Conducting the MLR

In this section, we provide information on the stages of evaluation process, selection of formal literature and gray literature sources, and data extraction process.

3.2.1 Formal literature

Evaluation process and selection of the publications. We performed the search using the search string given in Table 3 on the IEEE Xplore and ACM libraries, Google Scholar, PubMed databases, and Research Rabbit tool. The initial search was completed on October 15, 2021.

Table 3 Search string used for formal and gray literature
Full size table

To present a broader scope, we used the following inclusion criteria:

(i) the papers that share practitioner/researcher experiences;

(ii) the papers that present blockchain solutions in the health domain;

(iii) the papers that describe software development challenges encountered in the health domain;

(iv) the papers that meet the quality criteria presented in Table 3;

(v) the papers that are written in English and accessible.

We excluded secondary studies to eliminate the duplication of the findings.

The initial search yielded 4076 results. The first evaluation was performed based on the titles and abstracts of the publications. After the 1st evaluation, the paper pool was reduced to 178 papers. The second evaluation was conducted by reading the papers thoroughly. After that, we expanded the paper pool by performing snowballing on these publications and including papers related to the subject. The search process was stopped when the results reached theoretical saturation, which meant that no new concepts emerged. As a result, 78 formal pieces of literature were added to the review process.

Table 4 shows the number of publications found in online libraries and the outputs of each evaluation process.

Table 4 Results of evaluation process
Full size table

Quality assessment. To determine the quality of the papers in the formal literature, we used the following assessment criteria. Q1, Q2, and Q3 in Table 5 belong to Höst and Runeson's quality checklist [64], and we defined Q4.

Table 5 Quality assessment questions
Full size table

All authors of this paper took place in the quality assessment process. For the 178 papers that passed the first evaluation, we used three-level indicators to address the quality assessment questions:

(i) Level 0 when the criterion was addressed very poorly or not at all.

(ii) Level 1 when the criterion was addressed partially.

(iii) Level 2 when the publication successfully fulfilled the criterion.

The studies with a four-star rating or higher were included in the review.

Data extraction. All authors used a collective spreadsheet to extract the data from the papers and review the data. We retrieve bibliometric data (i.e., publication type and year, number of citations) and the information required to answer the research questions.

3.2.2 Gray literature

Evaluation process and selection of the sources We performed the search on a general Web search engine (i.e., Google), a specialized database (i.e., YouTube), and a social question–answer website (i.e., Stackoverflow) using search string given in Table 3.

The searches, which were performed between August 23, 2021, and October 1, 2021, yielded 572 results in total. We applied the following inclusion criteria to present a better scoped research:

(i) The sources that share experiences of practitioners;

(ii) The sources presenting blockchain solutions in the health domain;

(iii) The sources describing encountered challenges in software development in the health domain;

(iv) The sources that conform to specified quality criteria given in Table 4;

(v) The sources that belong to first- or second-tier outlet types;

(vi) The number of views higher than 1000 (valid only for videos);

(vii) Written in English and accessible sources.

The initial source elimination was performed based on the source titles. In the second evaluation, we examined the full content of the source based on the inclusion criteria given above. As a result, 23 sources from gray literature were included in the review process. We finalized the search process when the findings reached theoretical saturation, which means no new concepts emerged [64].

The number of sources found and the results of each evaluation are given in Table 6.

Table 6 Results of evaluation process
Full size table

Quality assessment. We applied the following criteria to assess the quality of the sources in the gray literature given in Table 7. We retrieved Q1 and Q2 from Garousi et al.’s quality checklist [58] and Q3 from Höst and Runeson's quality checklist [64] and added Q4 to these questions.

Table 7 Quality assessment questions
Full size table

All three authors of this paper were involved in the quality assessment process. We answered the quality assessment questions for 141 sources that passed the first evaluation according to the three-level indicator presented in Sect. 3.2.1.

Data extraction. As mentioned in the formal literature section, we used a spreadsheet to collectively manage the data extraction process. We downloaded the videos to a local computer to extract audio data and transform them into text using the pyTranscriber application [65]. We have added the bibliometric data of the sources and all the information needed to answer the research questions in this document.

4 Results and discussion

In this section, first, we provide the bibliometric overview of the sources included in the MLR and continue with the discussion of research questions.

In the final MLR source pool, there are 78 formal literature and 23 Gy literature sources. Out of the 78 papers, 53 were published in journals and 25 were published in conference proceedings. Figure 4 shows the publication years of the papers included and the publication dates of gray literature sources. Although Satoshi Nakamoto introduced blockchain in 2008, no publications on its use in the health domain were found until 2016. The number of publications raised gradually in 2018, 2019, and 2020, as shown in the following graph. The numbers appear to be decreasing in 2021. As the year had not yet been over when the last search was conducted on October 1, 2021, we anticipate that interest in this field will continue to grow.

Fig. 4
figure 4

The Publication Years of the Sources Included

Full size image

Figure 5 shows the publishing trend of blockchain technology in the health domain from 2016 to 2021. The number of studies on the topic shows an upward trend after 2017 and a downward trend after 2020. When we analyzed the publication types, we saw that early studies focused more on the concept, framework, and model development, and piloting implementation studies gained a rapid momentum as of 2019.

Fig. 5
figure 5

Evolution of Publications in Blockchain Technology in the Health Domain

Full size image

Figure 6 shows the frequencies of the most published venues of formal literature studies. There are two venues with the highest number in terms of frequency: Journal of Medical Systems and IEEE Access. Journal of Medical Internet Research venue follows them with three studies. International Journal of Environmental Research and Public Health, International Conference on Open and Big Data, IEEE Globecom Workshops (GC Wkshps), Blockchain in Healthcare Today, Sensors, and AMIA Annual Symposium Proceedings venues follow them with two FL studies.

Fig. 6
figure 6

The Frequencies of most Published Formal Literature Venues

Full size image

In Fig. 7a, b, we present the distribution of formal (white) literature sources’ citation numbers and YouTube video view numbers to show the interest in the field. It was specified that 770 new papers were published in the domain within just one year. As mentioned in Sect. 1, the number of the new papers added to the MLR study has tripled in a single year. Thirty-six percentage of the publications have low citation numbers (less than ten citations per publication), mainly because these papers have been published in recent years (in 2020 and 2021). Considering that the latest publication date is 2016, the citation numbers of the remaining papers in the paper pool are pretty high. Eighty percentage of the YouTube videos included in the MLR source pool also have high viewing numbers above 2000.

Fig. 7
figure 7

a Citation Numbers of the FL Sources,

Full size image

In the next part, we present the results of the MLR, i.e., the answers to the research questions based on the analyzed formal and gray literature sources.

4.1 RQ1. What are the blockchain application areas in the health domain, and what is the motivation behind it?

Table 8 highlights the primary blockchain application areas in the health domain, the rationale for adopting this technology in these areas, and examples of blockchain-oriented solutions presented in our resource pool. The use of blockchain technology for electronic health and medical record management is mentioned in most publications and GL sources (47 FL and 16 GL sources). The second most prevalent application area is Remote Patient Monitoring/Internet of Medical Things, with 12 formal literature sources (namely papers). Other application areas are medicine supply chain management (7 FL and 4 GL sources), clinical trials (7 FL), precision medicine (3 FL), blockchain in strengthening public health surveillance (1 FL and 1 GL sources), and health insurance (1 FL and 1 GL sources).

Table 8 Blockchain technology in the health domain
Full size table

We present the summary of this information in Fig. 8, including the application areas and the number of resources in those areas.

Fig. 8
figure 8

The Application Areas of Blockchain and the Number of Sourcesb. Video View Numbers of the GL Sources

Full size image

4.2 RQ2. What are the challenges of developing health software? RQ3. To what extent does blockchain technology contribute resolving these software development challenges in the health domain?

We have grouped the challenges and associated solution suggestions of the MLR sources under four main headings: 1) meeting regulatory requirements and public health surveillance, 2) security and protection of privacy, 3) ensuring interoperability, and 4) preventing waste of resources.

4.3 (1) Meeting regulatory requirements and public health surveillance

Challenge 1.1 Falsified drugs are one of the most serious threats to the pharmaceutical industry [66] 53 55 57 58 128 130. The World Health Organization (WHO) highlights that one in ten medicines produced in developing countries is substandard or falsified and has serious adverse effects on human lives [67]. To prevent the distribution of falsified medicines, regulatory agencies require to monitor the product supply chain before and during distribution [68, 69].

Solution 1.1. It is possible to detect data anomalies, unauthorized data insertions, and missing raw materials, identify authorized medication vendors/manufacturers, and store medical information by using blockchain technology and smart contract-based structures after data is inserted. Blockchain technology, which confirms and authenticates transactions, allows not only the system members but also drugs to be tracked throughout the medical supply chain [68,69,70,71,72,73,74].

The primary structure used in blockchain for this problem is smart contracts. Smart contracts provide traceability by tracking information on the blockchain in the form of transparent, immutable, and timestamped blocks by recording medicines, their active ingredients, and distribution data [66, 73, 74]. An account could be allocated to FDA and is notified by a smart contract when a transaction (i.e., the production, transportation, or receipt of medicine) occurs in a supply chain. The FDA account, as an oracle, could verify all transactions. When all other accounts attempt to upload a file, they automatically publish a session key encrypted with the FDA public key. Sylim et al. suggest that when unregistered products are entered into the medical supply chain, discrepancies in specific data points (e.g., dosage, ingredients) could be detected [68].

Additionally, a permissioned blockchain could be used to enable only trusted parties to join the network and to push data to the blockchain [70]. These mechanisms aim to prevent data manipulation at the entry of data into the chain to some degree. However, blockchain cannot handle if data is manipulated at the source. For instance, if a permissioned network participant manipulates the content of drugs and records the manipulated data into the chain, the blockchain cannot notice the manipulation. However, this manipulation can be tracked down to the source once detected elsewhere.

Challenge 1.2. During the COVID-19 pandemic, healthcare industry leaders point out that reviewing suppliers can be time-consuming. Given the ongoing demand for materials, healthcare organizations need to make quick decisions to locate and verify new vendors [75].

Solution 1.2 These days, IBM leverages blockchain technology to help address medical supply chain shortages due to the COVID-19 pandemic. The company has launched a blockchain network called, Rapid Supplier Connect to help government agencies and healthcare organizations identify alternative supplies and equipment vendors more quickly. By creating their supply chains and adding non-traditional suppliers to their networks, they had a sufficient stock of equipment and materials. The blockchain network also helps identify existing supplies and excess unused inventory, allowing hospitals to make them available to others and route supplies where they are most needed [75].

Challenge 1.3. All clinical trials must make their methodology and findings available to regulatory agencies; however, more than half of the trials fail to do so [76]. Furthermore, a recent study [77] found that clinical trials are highly vulnerable to data manipulation. Subject registration, trial registration data, and clinical measures might be subject to manipulation [76].

Solution 1.3. The clinical trial life cycle includes trial registration, recruitment of subjects, regulatory approval, data entry, compliance with the trial protocol, amendments to a clinical trial protocol, patient monitoring for giving or withdrawing their informed consent, and reporting of adverse events stages. Blockchain prevents backward data manipulation during the clinical trial life cycle if data is added to the blockchain at every stage when the data is created instead of bulk data entry at the end of a trial [76]. Although blockchain does not guarantee correct data entry, a staged data entry process would reduce data manipulation risk. Smart contracts could be used to promote transparency in reporting clinical trial data by capturing data that might be intended to be manipulated [76, 78,79,80,81].

Additionally, smart contracts mandate a stage-by-stage data entry during the clinical trial process. This way, during a clinical trial, the intermediary stages can be traced, and the results can be disclosed to regulatory agencies without any missing data.

Challenge 1.4. Clinical researchers or clinicians may be fraudulent or careless and record misleading or incorrect data into case report forms (CRFs), including all data of patients participating in a clinical trial [82]. A common healthcare fraud involves perpetrators who provide false or exaggerated diagnosis data for patients so that fraudulent insurance claims can be submitted for extra payment [11]. According to the FBI report: “The total cost of insurance fraud is estimated to be more than $40 billion per year.” in the USA [83].

Solution 1.4. One method to solve this issue is to encourage clinical researchers and clinicians to provide raw data to the blockchain as early as possible. Incorrect data entry cannot be prevented by using blockchain; however, adding raw data early into the chain would eliminate data tampering to adapt to new situations. Later on, statistical analyses can be applied to verify the data [82] to detect discrepancies that signal data manipulation at the source.

Challenge 1.5. The Health Insurance Portability and Accountability Act (HIPAA) establishes standards for healthcare-related electronic transactions [14]. The HIPAA Privacy Rule requires protecting the privacy of health information. Therefore, patient information should be stored anonymously to prevent the identification of patients [84, 85].

Solution 1.5. Blockchain offers a partial solution with pseudo-anonymity where the user is anonymous, but their account identifiers are not [86]. The pseudo-anonym structure allows patients to hide their identities with alphanumeric addresses and yet to authenticate their identity when needed [73, 85, 87,88,89,90,91,92]. Pseudo-anonymity is still considered personal data [93]. Although privacy cannot be fully ensured with pseudo-anonymized data, it constitutes a partial solution for data privacy. Smart contracts enable regulation of the access control policy and achieve HIPAA compliance [94].

4.4 (2) Ensuring security and privacy

Health data is a tempting target for criminals due to its potential economic value [11, 15, 70, 71, 86, 92, 95]. Therefore, health data storage and transmission processes must be performed in a reliable and secure way [85, 87, 94, 96,97,98,99,100,101,102,103,104,105,106].

[15, 89, 91, 107,108,109,110,111,112]. According to the Trustwave report: “a healthcare data record may be valued at up to $250 per record on the black market compared to $5.40 for the next highest value record (a payment card)” [113]. The number patient records compromised in 2020 exceeded 40 million according to the incidents reported to the USA government [114]. Criminals may attack the healthcare system and threaten confidentiality, integrity, and availability of patients’ personal health information.

Challenge 2.1. Intruders may tamper with or delete patients’ data, thereby benefiting insurance companies or hiding medical malpractices (e.g., delayed diagnosis and misdiagnosis) [111, 115]. Intruders may also tamper with the cold-chain shipping data in medicine supply chains when the essential information is stored in centralized databases [69].

Solution 2.1.a. In blockchain, all data and transactions are digitally signed which enables prevention of unauthorized access to network [70, 85, 89, 98, 99, 101, 104, 116,117,118,119,120,121]. It employs asymmetric cryptography to authenticate users and safeguard data, thus enabling confidentiality among the participants of a system [116, 122,123,124,125].

Solution 2.1.b. Each block in a chain keeps permanent logs of data transmissions [79, 84, 85, 87, 102, 103]. As data are timestamped in a blockchain, data manipulations can be recognized [121, 126]. Blockchain technology ensures transaction security; however, it does not offer a specific advantage in preventing data theft. Using zero trust principles in a blockchain could enhance the overall security. While blockchain ensures transaction security, zero trust policies, including data encryption, would improve access management and user authentication [95].

Solution 2.1.c Nodes have full access to ledgers; however, users are only allowed to perform activities based on their role and can only access the files they own or have permission to view [91, 95, 127]. Smart contracts can be used to assign the network users to different roles with associated functions and privileges [102, 128, 129] and generate immutable logs of transactions [15, 97, 110, 124, 130, 131].

Solution 2.1.d. A wrapper layer integration mechanism can be used between the cloud-based EHR management systems and public blockchain networks to develop tamper-proof health record management systems [115].

Solution 2.1.e. Because of the sensitivity of health data, permissioned blockchains (e.g., Hyperledger Fabric) could be used to enhance privacy [2, 11, 72, 73, 79, 97, 111, 125, 131,132,133]. In a permissioned network, participants are included in a system in a controlled manner. Thus, confidentiality in a network is met [131].

Challenge 2.2. When health records are stored in centralized databases, it becomes a necessity to rely on single authorities that may not effectively protect the data against internal and external attacks [11, 73, 125, 134]. For example, in disease surveillance, authorities and independent agents must record sensitive information in centralized information systems. But, centralized data control systems are subject to single point of failure problem and do not provide data transparency [121].

Solution 2.2. Blockchain’s decentralized nature enables elimination of single point of failures. If a node fails or is compromised in a chain, the failure does not cause the entire system to stop. Therefore, it has a more robust structure and resilient to cyber-attacks [74, 79, 81, 85, 89, 90, 92, 95, 98, 107, 121, 122, 127, 131, 135, 136]. Patients’ health data are stored on the servers of various healthcare providers in blockchain-based applications; thus, a single failure does not affect all locations to stop working simultaneously [74, 118]. It can be said that BC-based systems are robust against data loss or data corruption [73, 121] and eliminate the need to rely on central authorities [74, 95, 126, 134].

Challenge 2.3. Healthcare system users (e.g., patients) should have control over on their own data due to its sensitivity [70, 85, 89, 101, 110, 112, 118, 123, 135, 137,138,139,140,141,142]. Data ownership management is a major challenge for healthcare systems currently [136]. Cernian et al. mention that there is no platform to monitor patient traceability throughout the entire healthcare chain [143]. The risk of intermediaries or intruders accessing the patients’ reports without their consent remains valid. Such information may be exploited by insurance vendors and other third parties [144]. Patients should be the ones to decide with whom their health data will be shared, neither third parties nor institutions.

Solution 2.3. Blockchain technology allows patients to take ownership of their health data and share data without violating patient rights. Patients could retain control over every transaction in a blockchain-based system by accessing their health data using an asymmetric encryption algorithm. Only the trusted parties authorized by patients may access the data [2, 11, 73, 74, 85, 87, 91, 96, 101, 107, 118, 124, 130, 134, 135, 137, 141, 142, 144,145,146,147,148]. As an example [11], when a patient wants to grant access to a doctor to view their health data, the patient encrypts their health data with a symmetric key in an external database. This symmetric key is sent to the doctor using the asymmetric key encryption. The doctor uses their private key to access the symmetric key of the patient and decrypts the health data using this symmetric key [11]. This secure sharing among doctors and patients eliminates the intermediary parties [106]. When the patient wants to revoke the access grant to their health data, the data is encrypted with a new symmetric key. Patients could also track how many times their records have been accessed by using smart contracts [74, 112] and whether a change was made on their records, and the owner of the change [124]. Thus, blockchain-based health applications enable the storage and sharing of health data in a patient-centric manner.

Challenge 2.4. Patients’ health data may not always be recorded electronically in healthcare systems. This issue affects the quality of health care. There is a need to establish a patient-reporting mechanism to improve the quality of care [94, 129, 149].

Solution 2.4. Blockchain gives patients the right to report their health records on ledgers [94]. This functionality creates opportunities to use health records for medical research with permission [118].

Challenge 2.5. IoT devices used for remote patient monitoring are especially vulnerable to cyber-attacks and data theft [130, 136, 150]. Hackers can take complete control of wearable IoT devices and misuse them. For instance, Johnson & Johnson had previously warned patients about the vulnerability of one of the insulin pumps that the hackers could exploit to overdose the patients [136]. In blockchain and IoT integrated systems, IoT devices may subject cyber-attacks.

Solution 2.5. Blockchain cannot be a solution to prevent the attacks or temper resistant to these attacks. However, audit trails in blockchain allow tracking who made the changes and when the changes were made [151]. Cryptographic hash functions create immutable audit trails and guarantee that the most recent version of the record is always used [148]. When a patient’s report residing in the blockchain network needs to be updated, a new report is generated with the reference of the original report and uploaded to the blockchain. This reference enables the updated reports to be identified [121].

4.5 (3) Ensuring interoperability

Challenge 3.1. Regulations on data transfer among healthcare providers are not well defined. In addition, interoperability of systems is another issue when effective and coordinated data sharing is concerned [66, 74, 101, 119, 139, 152,153,154]. Therefore, there is a need to develop a secure and efficient data-sharing mechanism for highly sensitive health information among the stakeholders of healthcare systems [155].

Solution 3.1.a. When a nationwide blockchain application covering all the health stakeholders is deployed, there is no need to transfer data among health providers. As blockchain enables the patients to control their health data, patients could give access rights to health providers directly. As a result, it is possible to avoid undefined procedures and interoperability issues in sharing data among healthcare providers [151].

Solution 3.1.b. When a single ledger is developed to include all stakeholders in a healthcare system, there is no need to manually transfer added or modified patient data from one system to another [11, 134]. Agreements among patients, government, providers, and insurance companies can be stored via smart contracts. Thus, interoperability would not be a concern. Additionally, data format requirements can be defined on blockchain to record all information correctly. In this way, the problem of preventing data sharing due to inadequate information can be reduced or eliminated [154].

Challenge 3.2. Patient mobility requires cross-border exchange of patient data, which causes difficulties in complying with different countries’ privacy and data protection standards [156].

Solution 3.2. Different data privacy, security, and sharing policies need to be addressed in designing blockchains and smart contracts considering the patient mobility fact [138, 156]. In addition, we suggest developing a structure that allows each country participating in the network to implement specific policies for the protection and control of health-related data.

Challenge 3.3. Patel states that infrastructures used for cross-site medical imaging data transfers require relying on third-party intermediaries [90]. However, ensuring the trust among the relevant stakeholders is highly difficult.

Solution 3.3. A blockchain application dedicated to corresponding stakeholders could be a solution for cross-domain image sharing. Blockchain framework eliminates the third-party access to protect health information [90].

4.6 (4) Preventing waste of health resources

Challenge 4.1. In clinical research, manual processing may be required to capture, manage, and report data [79, 80], as the patient data is collected as bio-samples, questionnaires, and laboratory results. The clinical research forms and questionnaires are usually paper based. Such a manual intervention for data management and maintenance increases the cost of clinical studies. Backup and the data recovery time for such systems are also high [79, 80, 89]. In addition, clinical trial data have to be stored confidentially and securely for audits and potential future studies [136].

Solution 4.1. Blockchain-based data management frameworks may reduce the administrative burden and the time and effort to ensure data integrity and confidentiality in clinical trials [79]. In blockchain health systems, medical data is recorded continuously. Additionally, since previous clinical studies’ trial data are encrypted and stored in the blockchain in a distributed manner, they remain unchanged and would be available for future studies [136].

Challenge 4.2 Patients often experience burden in remembering their medication history or carrying physical copies of their medication records [152]. Patient reports, tests, and medical treatments generated by various doctors are managed independently [11, 12, 99, 142, 144, 157]. Many health institutions, doctors, and laboratories have their own database and manages their own information [99], without the intervention of patients. This situation affects the prevention and treatment of diseases for the population due to misinformation about a patient, potential information loss, or data leakage, which may imply an immediate risk to individuals and increase public health costs [124]. For instance, doctors might prescribe a medicine to patients that they are allergic to, as they cannot access the patients’ medical history [158]. On the other hand, patients may be unaware of their medical reports, as they are not provided with complete documentation [117, 144].

Solutions 4.2. Electronic health records that are securely published on blockchain-based health applications with patients’ consents would address the problem given in Challenge 4.2. Using a decentralized ledger system, health professionals could update and query medication histories of patients after getting patients’ approval [152]. Thus, doctors and other healthcare providers can reach patients’ health data [148, 157] and perform transactions such as adding scans and laboratory results [12].

Challenge 4.3. The patients’ consent is essential for using their medical records for various purposes. However, most people give consent using paper forms, and they do not have control over it. Healthcare organizations are also having difficulties in dealing with the patients’ consent. Patients give consent and may want to withdraw it later. There is a need to allow healthcare organizations to manage patients’ consent [159].

Solution 4.3. Individuals’/Patients’ consent could be stored in blockchain and shared by the participating parties in an immutable way [159]. This is a way to provide individuals to have control over their data. Rather than one-time-only consent models, this dynamic structure allows individuals to override their consent terms in time as a new block [16].

Challenge 4.4. A patient’s medical history or patient’s informed consent must be available at the required time [81, 110]. In current systems, maintaining a medical history to meet this criterion is costly [99], time-consuming, and labor-intensive.

Solution 4.4. The health records stored on the blockchain network are permanent and are replicated across multiple nodes [121]. This ensures that all patient data is available at the required moment and required place [160].

Challenge 4.5. Regulations require insurance companies to maintain directories that contain up-to-date demographic information about doctors and other healthcare providers. Maintaining its index for each insurer is time-consuming and expensive. Claim and payment processing may be delayed if the information in these directories is inaccurate. Roughly $2.1 billion is spent annually to track and maintain provider data across the healthcare system in the USA. A review completed by the Centers for Medicare and Medicaid Services (CMS) found that 52% of listed provider directory locations had at least one inaccuracy [161].

Solution 4.5. Solution 4.5. Administrative costs and data quality can be improved by sharing healthcare provider data and sharing changes of different parties on a blockchain. This feature enables identifying data inaccuracies within healthcare provider data [161].

In Table 9, we provide the FL and GL sources that highlight the blockchain-related solutions given above.

Table 9 Titles of blockchain powered solutions and the sources
Full size table

4.7 RQ4. Does blockchain introduce new challenges to software development in the health domain? RQ5. What are existing solution suggestions to the blockchain-related challenges in the health domain?

We compiled the blockchain-related problems and solutions below.

Challenge 1. Data cannot be altered or deleted after storing it in a blockchain [74]. However, according to health data protection laws, data is required to be deleted when a patient requests it [127, 138, 159] or to be changed (e.g., if a manufacturer enters incorrect information about a medicine) [127].

Solution 1. Storing the actual health data in an external storage and its hash value in the blockchain could be a solution to this problem [15, 84, 102, 131, 138, 141, 147]. The details of this topic are discussed more in-depth in Solution 9 as part of this paper. By not having the data itself on the blockchain, we could delete the data when requested by its owner. Hash values whose data has been deleted would remain in the blockchain.

Challenge 2. Heterogeneous data (e.g., X-rays, images and ECG signal data) are heavily used in the health domain [162], and size of health data can be pretty large [66, 146]. Along with the increase in data size, we need to deal with storage issues [124, 163] and mining costs [107].

Solution 2. Storing the original large-scale data in an external storage and keeping its hash in a blockchain would resolve dealing with large-sized data [84, 102, 131, 138, 141, 147] without compromising the tamper-resistant nature of blockchain. The hashes of large data embedded in a digitally signed transaction are added to blockchain by consensus. When the hash for the data in the external storage matches the hash in the blockchain, the origin and timestamp of the data can be verified. Furthermore, when the data in the external storage changes, the hash of the data also changes, and thus, data manipulations could be detected.

However, there is a deletion risk of external data when the data is stored off-chain. The rollup technology could be an alternative to this solution. Rollups move the computation off-chain but retain some data per transaction on-chain. It also provides a solution to the storage problem as the amount of data published on the chain is the minimum amount required to validate the rollups transaction [164] locally.

Challenge 3. Blockchain poses performance [91, 101, 117, 127, 130, 133, 146] and scalability challenges [66, 92, 110]. As a performance issue, the read latency increases with the growth of ledgers [94, 122]. Public blockchains suffer from scalability issues due to ledger replication in all network participants and consensus mechanisms [165] and the need for significant computing power and storage space required on each node [92]. As each node repeats the same process for mining the next block, it is impossible to perform parallel executions in a blockchain, which reduces system efficiency, and therefore may cause bandwidth and response time problems.

Solution 3.a. Architectural design decisions such as using consensus models impact blockchain system performance [127]. The choice of consensus algorithms affects both scalability and computing performance. For example, Practical Byzantine Fault Tolerance (PBFT) consensus algorithm is not that scalable but offers superior performance than Proof of Work (PoW) consensus algorithm [91, 92]. Instead of the Proof of Work (PoW), the Delegated Proof of Stake (DPOS) algorithm could be employed in medical blockchain to avoid high energy costs, as DPOS does not require competition over discovering the blocks and is, therefore, more efficient [89]. Another design decision is to make the blockchain public, consortium or fully private [166]. Consortium blockchain networks with trusted nodes may be preferred if high performance is expected from an application. They have much higher execution and processing efficiency (35,000 transactions per second) and higher computing power than public blockchain solutions [66].

Solution 3.b. Dividing the network into small groups, called shards, could be used to address scalability issues [165]. Hyperledger Fabric supports multiple channels, each maintaining a separate ledger and smart contract [79]. Transactions can be processed in parallel while running consensus within each shard with a subset of blockchain nodes. Although this technique could help solve scalability issues, the communication overhead between shards can degrade network performance. Minimization of cross-shard communication is possible by creating complete shards based on “the need to participate” nodes per patient [165]. We should also note that currently there is active research ongoing on sharding features in Ethereum as of July 2022 [127, 167].

Solution 3.c. The efficiency of the blockchain is highly dependent on the coding of smart contracts. A smart contract which is coded properly (e.g., reduced external data storage access) could be executed in a quick and efficient manner [127].

Solution 3.d. Standardizing the data to be stored and exchanged on a blockchain could be a solution to achieve better performance and efficiency. Ledgers could align and define data’s type, size, and format. Restricting access to the blockchain network also helps standardize the data [66].

Challenge 4. Data providers may not have a culture of handing over the control of the data [151]. Furthermore, not everyone is capable of managing their personal health data. Studies conducted by the Connected Health Cities Programme and Wellcome Trust Fund have shown that most citizens are not interested in managing their data [147]. On the other hand, a vast majority of the population is unfamiliar with blockchain technology. If patients lose their private keys, the associated resources become inaccessible to these patients and this issue requires recovery solutions outside of blockchain to re-establish ownership of the patient [90]. To fully obtain the potential benefits of blockchain in the health domain, all parties involved in a health system need to be part of blockchain-based solutions [82, 99]. Some stakeholders may be reluctant to join the network for fear of losing their competitive advantage [66].

Solution 4. Solutions to these problems have not yet been proposed.

Challenge 5. Blockchain development imposes certain constraints on the development processes. Challenge 5.1 A smart contract code cannot be changed once it is added to a network. When a change request is received, a new contract needs to be deployed (upgrading) [168, 169].

Solution 5.1 While this issue reminds us running a waterfall-like plan-driven development process for smart contract development, it does not guarantee error-free features. However, this problem can be addressed by establishing new software design principles to assist the development of high-quality smart contracts.

Challenge 5.2 Smart contracts need to be tested in a production environment as part of the development process. Testing smart contract in the production environment in the public blockchain includes an execution fee (called gas price in Ethereum). This fee varies depending on the operations performed by the smart contract. Calculating the cost of executing a smart contract on a blockchain network may be challenging, particularly for large-scale projects with complex coding [170].

Solution 5.2 There is no solution addressing this problem in our resource pool. Smart contract testing is free in test networks. Before deploying smart contracts to production environment, testing practices need to be applied in the test network to reduce the costs.

Challenge 5.3. Gas costs are proportional to the number of stored data and operations (i.e., memory and storage access) in smart contracts. Storage access needs increase gas cost dramatically [127].

Solution 5.3. There are tools (e.g., Remix IDE) that could estimate execution and transaction costs and helps adjust these costs [127]. Additionally, this problem may be solved by establishing cost-efficient smart contract programming practices and a software development life cycle specific to blockchain-based applications to manage the development process better.

Challenge 6. There is a trade-off between transparency and confidentiality. Blockchain is intended to increase trust and enable transparency by sharing health data. Access control imposes limits on data sharing and provides a level of confidentiality [105].

Solution 6. When developing a blockchain platform, access control should only be on the identifiable data, yet a level of transparency should be allowed on the blockchain for other data types/categories [105].

Challenge 7. Blockchain keeps log of all the activities taken place in the chain. Data is stored in every block of the blockchain; so, there is no chance of losing data, but there is a possibility of creating redundant data [162].

Solution 7. The blockchain technology would require upgrades to solve the redundant data creation problem [162]. Additionally, InterPlanetary Filesystem (IPFS) is a distributed, peer-to-peer storage network which inherently allows deduplication [80]. Therefore, IPFS could be used together with blockchain to leverage deduplication feature of IPFS [171].

Challenge 8. Blockchain networks work in their unique way, leading to interoperability issues where different blockchains cannot communicate [127].

Solution 8. This problem can be avoided if a unified blockchain-based solution is used between healthcare centers. However, it will be very difficult to make them interoperable if healthcare centers decide to use different blockchain-based solutions in varying platforms [127].

Challenge 9. Many countries have strong regulations for storing or transmitting medical data [131]. Therefore, storing and transmitting personal data in a transparent medium are not allowed.

Solution 9. Rather than storing the actual data, its hash value could be stored or could be transmitted in a blockchain [15, 84, 102, 131, 133, 138, 141, 147]. Whether hashed data being considered as personal data is an ongoing debate. If it is considered as personal data, hash sharing becomes another challenge. In this regard, it is important that regulatory guidance is issued on this subject [93]. Kim et al. also mention that medical data is only kept by certified bodies in some countries including South Korea and member states of the European Union. As a solution to this problem, they suggest keeping the hash of the data in a blockchain and leaving the storage of the data only to certified medical bodies, in separate databases [131]. On the other hand, hash itself may not be secure enough, as the hash can be linked with patients, and is subject to brute force attack [172]. In such circumstances, a keyed-hash may be a better alternative, as it uses a secret key as an additional input to hashing [173].

Challenge 10. In blockchain solutions, it may be difficult to define the legal boundaries in blockchain technology components, which complicates the role of health authorities. For instance, when a new drug-related transaction is executed on a blockchain network, health authorities need to define legal obligations for the stakeholders involved in the transaction. Although there is still no definitive provision in the current laws and regulations for blockchain technology in healthcare [66], blockchain networks need to comply with the existing regulatory requirements, such as the U.S. Drug Supply Chain Security Act (DSCA) and General Data Protection Regulation (GDPR) [66].

Solution 10. Blockchain frameworks need to be developed to comply with existing regulatory frameworks. Hyperledger Fabric can be given as an example which was designed compliant with HIPAA and GDPR [174]. Additionally, the policymakers need to consider addressing issues such as block ownership and access permissions on blockchain networks.

5 Conclusion and future work

In this study, we performed a multivocal literature review to identify blockchain application areas and the motivation behind adopting blockchain in the health domain. We extracted data from both formal and gray literature sources to identify the problem and solution suggestions on this topic. On the problem side, we focused on both the challenges of developing software and adopting blockchain technology in the health domain. On the solution side, we specified blockchain technology’s contribution in resolving inherited health application development challenges and the solutions occur due to adopting blockchain technology into health software development.

We found that blockchain technology in the health domain is applied in the following fields: electronic health and medical record management, Internet of medical things/remote patient monitoring, medicine supply chain management, clinical trials, precision medicine, empowering public health surveillance, and health insurance.

Blockchain has a significant potential in contributing to the inherent health domain challenges such as ensuring regulatory requirements and ensuring public health surveillance, dealing with security, privacy, interoperability issues, and preventing waste of health resources. Blockchain technology contributes to resolving these challenges by enabling regulatory bodies to monitor medical supply chains and preventing data manipulation. The technology increases transparency in transactions, gives patients control of their health data and the right to report personal health records, and enables pseudo-anonymity by eliminating the auditing role of central authority and intermediaries.

While providing solutions to the inherent health domain challenges, blockchain introduces new challenges as well: issues of data protection, data size, performance, confidentiality, personal data management, development process, creation of redundant data, and ensuring interoperability among different blockchain parties.

We summarize solution suggestions to these challenges as follows: addressing data protection and size problems by storing health data in external storages and its hash in the blockchain; increasing performance with architectural design decisions, consensus protocol decisions, and cost-effective smart contract coding; meeting confidentiality by using permissioned blockchain type and allowing pre-selected participants to join the network; addressing development process issues by using tools for estimation of gas cost, developing new software design principles, and by creating a new software development life cycle to meet specific constraints of blockchain-based applications; and addressing interoperability among different blockchains by using unified blockchain-based solutions among healthcare centers.

The FL and GL sources’ analysis reveals that both source types highlight similar challenges and solutions. Compared with the SLR we performed in 2020 [23], we found new blockchain application areas, challenges, and solution suggestions with the MLR. We summarize the differences of the results of these studies in Table 10.

Table 10 Comparison of the SLR [23] and the MLR
Full size table

Blockchain technology introduces specific constraints to development processes in the health domain. While developing a blockchain application in the health domain, we recommend addressing the blockchain-related challenges and solution suggestions we have presented in this MLR.

5.1 Future work

We can state that there is still room for improvements in adoption of the blockchain in the health domain. The results of this study which present both theoretical and practical perspectives could be used for identifying potential research topics and issues to consider before adoption. Below, we list these potential research topics on blockchain adoption in the health domain:

  • Policymakers, blockchain researchers, and developers need to work together in developing regulatory guidelines for addressing the rules for the collaboration of network participants, data privacy, what data to share, data deletion on patient demand, and ensuring interoperability between various ledgers in blockchain networks.

  • Different data privacy, security, and sharing policies need to be addressed in designing blockchains and smart contracts considering the patient mobility across nations.

  • Use of blockchain technology in the health domain has the potential to generate massive amounts of sensitive personal health data. Blockchain technology could be used together with artificial intelligence to meaningfully process the data collected in blockchain.

  • In large-scale blockchain adoption, data grows exponentially, and this causes performance issues in data retrieval. Effective and efficient data retrieval processes suitable for such blockchain applications need to be developed.

  • Integration of blockchain technology with currently running healthcare systems (e.g., electronic health record management systems) could be explored.

  • Data in the blockchain is transparent, which causes data privacy infringement. In the health domain, sensitive patient data is stored transparently. Additional privacy mechanisms such as keyed hash need to be developed.

  • Current development life cycle models are not fully compatible with the development process of blockchain-based applications due to the structural characteristics of blockchain, such as contract immutability, necessity of testing in production environments, and test execution fees. Therefore, it is necessary to develop a new blockchain development life cycle model considering its structural characteristics and guiding an efficient and sustainable blockchain development environment.

  • Malta, Estonia, and South Korea are countries adopting blockchain at national scale in the health domain. These successful implementations at governmental levels may be inspirations for other countries.

  • We also suggest regularly continuing the MLR studies to reflect current state of the art both from the researcher and practitioner perspectives.