This study used a structured questionnaire to explore the current adoption of DHTs in 5 public hospitals in Vietnam. The assessment items were developed based on the Vietnam HIT Maturity Model, which comprises 8 domains related to technology adoption and digital capabilities in hospitals [11]. The 8 domains are IT infrastructure, HIS, LIS, radiology information systems–picture archiving and communication systems (RIS-PACS), EMR, administrative and operation software, security and information safety, and nonfunctional criteria. Except for security and information safety and nonfunctional criteria, each domain includes a set of infrastructure requirements, such as having a server room, or software functional criteria, such as having laboratory test management functionality. The model also sets out further digital capabilities focusing on assisting clinical practice, such as CDSS, clinical data repositories (CDRs), and medication management; these are named extra capabilities.

Following a stage-based approach, a hospital reaches a digital maturity level in a domain if a predefined set of criteria is met. The hospital can only move to a higher digital maturity level when all criteria of the lower levels have been satisfied. The domains of IT infrastructure and HIS have 7 digital maturity levels (1-7), while the other domains have 2 levels (basic and advanced). Multimedia Appendix 1 presents the digital maturity levels and their scoring criteria based on the domains and extra capabilities.

Since the focus of this study was the adoption of digital health systems in hospitals, we developed an assessment tool based on the criteria from the IT infrastructure, HIS, LIS, RIS-PACS, EMR, and extra capabilities domains (see Multimedia Appendix 2). The administrative and operation software, security and information safety, and nonfunctional criteria domains were not included in this study as we considered these less relevant to the study’s focus despite their importance to the overall digital maturity of a health care institution.

For each domain, we gathered information through the hospitals’ IT departments to determine whether the hospitals met the component criteria or whether they planned to achieve the criteria in the next 3-5 years. Results of the assessments were discussed with the heads of the IT departments to ensure accuracy. Data collection was carried out from July 2022 to March 2023. Information about the digital maturity of each hospital was entered into Microsoft Excel for initial data cleaning and then analyzed using R software (R Foundation for Statistical Computing) [16].

Our aim was to provide a descriptive summary of what infrastructure and functionalities, or criteria, were available in each hospital across the domains rather than calculating scores for their digital maturity. The proportion of criteria that each hospital met in each domain was calculated. A radar plot was created using the fmsb package in R [17] to depict the difference in the proportions of criteria met among the hospitals by domain. Radar plots are regarded as an efficient tool to compare various groups on multiple variables, as discussed by Saary [18]. In addition to summary statistics, we further compared technology adoption between the 5hospitals by delving into the individual domains. Tile graphs were created using the ggplot2 package in R [19] to illustrate the criteria that the hospitals had satisfied within each domain. These graphs allow for the visualization of individual data points (eg, which criterion was met by a specific hospital) as opposed to visualizing the counts or proportions of hospitals that met a particular criterion, as in the case of bar charts.

Until the beginning of 2024, it was estimated that there were nearly 1500 public and over 300 private hospitals in Vietnam [20]. Five hospitals were purposively selected for this study based on previous interest in DHT innovation and participation in digital health research activities. Our approach aimed to sample hospitals where DHT innovation was likely to occur the soonest, rather than providing a comprehensive view of all government hospitals. Of the 5 hospitals that participated in this assessment, 4 (80%) were in Hanoi or Ho Chi Minh City, the 2 largest cities in Vietnam, and 1 (20%) was in a Mekong Delta (ie, rural) province. All are top-tier teaching public hospitals with a capacity of over 500 beds; 2 (40%) are tertiary hospitals that focus on a limited number of specialties, while 3 (60%) are general hospitals that provide care in a wide range of specialties. Most hospitals have between 8 and 12 IT staff per 1000 beds, except for 1 (20%) hospital that has an approximate ratio of over 50 per 1000 beds. Characteristics of the hospitals are shown in Table 1. To ensure confidentiality, we withheld information that could be used to identify the hospitals.

This project was exempted from ethical review as it constituted a service review, as defined by the Oxford Tropical Research Ethics Committee (OxTREC) [21,22]. The selected hospitals were identified from a professional network that the study’s principal investigators (authors LT and CP) were part of. The hospitals were invited to participate in the assessment through an invitation letter explaining the purpose and scope of the study. Upon approval by the hospitals’ authorities, the hospital IT directors were provided with an information sheet, the questionnaire, and an instruction document. Queries regarding the assessment, if any, were clarified by the first author (DMT).

The IT directors were informed in writing that they could refuse to provide any data if they wished and that the data collected would be securely managed and only be used for the purpose of the study. Participating departments were able to review and agree on the contents of the manuscript prior to submission. No personally identifiable data were collected during the research. The research data were managed as per the Oxford University Clinical Research Unit’s data management policy. Information that can be used to identify the hospitals, such as names and the number of beds, was deidentified to ensure confidentiality. No compensation was given to the hospitals for participating in this study.

HIS, LIS, radiology information systems (RIS), PACS, and EMRs are popular information systems in health care institutions worldwide. However, their functionalities and use cases may differ to some extent depending on the national context that they are implemented in. In this section, we provide a Vietnam-oriented description of these information systems to support the interpretation of the study’s findings.

Although we were not able to collect information about 3-5-year investment plans from the 5 hospitals, we were able to collect data on all the assessment criteria from the maturity model. The radar plot in Figure 1 presents the percentage of criteria the 5 hospitals met in each domain.

In general, HIS, LIS, and RIS-PACS were implemented in all 5 hospitals, albeit with varied digital maturity levels. Two hospitals were not using any EMR system or fulfilling any extra digital capability. One hospital reported meeting all the criteria across the 6 domains. In the following sections, we describe the domains in more detail and compare the implementation strategies between the hospitals.

Figure 2 shows the IT Infrastructure criteria that were met by the 5 hospitals. Level 1 criteria were met by all 5 hospitals, and most hospitals reported meeting level 2-5 criteria.

All 5 hospitals had sufficient workstations for staff. They were connected to the internet and local area networks. All 5 hospitals’ digital systems and databases were run on dedicated servers, and 3 (60%) hospitals had their own on-premise server rooms with physical controls. The hospitals’ data were stored on storage devices that were supported by storage networks, such as network attached storage (NAS) or storage area network (SAN) systems. Backup storage systems were available in 3 (60%) of the 5 hospitals.

Almost all hospitals in this assessment adopted technologies to help reduce errors and improve patient flow. These included barcode readers, barcode printers, queue management systems, screens displaying queue orders, and electronic notice boards. Information kiosks were less common as only 3 (60%) of the 5 hospitals implemented this technology.

Level 6 and 7 criteria were less available in the hospitals. Only 1 (20%) hospital reported adopting mobile devices, such as tablets and smartphones, and a wireless local area network (LAN) to support hospital activities.

This study assessed the implementation of some key digital health systems in 5 top-tier hospitals in Vietnam. Overall, many processes related to administration, laboratory tests, and medical imaging were digitized in the hospitals through the implementation of HIS, LIS, and RIS-PACS on the basis of adequate basic IT infrastructures. The digital maturity of these systems, however, varied significantly between the hospitals, although all are large top-tier hospitals. EMR systems were implemented in 3 of the 5 hospitals. Although these systems could satisfy a wide range of data collection needs, their integration with CDRs and CDSS was sparse.

This is the first original research that examined digital health system implementation in Vietnamese hospitals since the Vietnam MoH enacted the National EMR plan and the Vietnam HIT Maturity Model in 2018 as part of the sectorwide digital transformation agenda [11,21,24]. Academic publications on adoption and implementation of hospital-based digital health systems are limited [2]. Most of the previous studies have been conducted before 2014 and focused on the readiness to transition from paper-based processes to electronic systems rather than describing DHT adoption [28,29]. Other publications have described the development and pilots of technical solutions, such as LIS, data retrieval from medical devices, and data visualization for EMRs. We could not find any research exploring the digital systems currently used in hospitals in Vietnam [30-33].

Muinga et al [34] showed that most of the digital health systems adopted in Kenyan public hospitals from 2014 to 2016 were aimed at administrative and billing purposes rather than supporting clinical tasks. Systems such as LIS, RIS, and PACS, when available, were usually stand-alone systems and lacked interoperability with other systems in the same institution [34]. Our study found that all the Vietnamese hospitals assessed had gone beyond administrative systems and had used clinical information systems that are interoperable with each other, especially with HIS. EMRs, as a crucial clinical information system, had been implemented in 3 of the 5 hospitals to capture a wide range of data, such as medical histories, clinical documents, orders, and test results. However, the adoption of EMR capabilities, such as CDRs and CDSS, was limited, even though these are large teaching hospitals at the top tier of the Vietnamese health system. This is inferior to the digital maturity status of public hospitals in Turkey, as over 98% of the Turkish hospitals surveyed reported having CDRs regardless of their size [35]. In addition, CDSS for medication orders and nonmedication orders were implemented in 71% and over 57% of Turkish public hospitals, respectively. The early use of EMRAM to guide the national digital health care transformation may be an important facilitator for this wide adoption of CDRs and CDSS in Turkish hospitals [36].

The Vietnam HIT Maturity Model [11] that this study’s questionnaire was based on and the HIMSS EMRAM [9] share several similarities regarding assessment criteria. For example, CDRs, CDSS, closed-loop medication administration, and electronic documentation are included in both models, including the maturity stage that these systems belong to. However, EMRAM particularly focuses on EMR capabilities, in which systems such as laboratory, radiology, and pharmacy are seen as EMR ancillaries, while CDRs and CDSS are crucial components deciding the maturity of the EMR system. All the ancillary systems need to be implemented as early as in stage 1 of EMRAM. The Vietnam HIT Maturity Model, in contrast, splits the ancillaries into basic and more advanced maturity levels along the digital maturity roadmap. Systems such as CDRs and CDSS are the extra capabilities that should be met, alongside other domains, such as HIS, LIS, RIS, and PACS, so that a hospital can move up to a higher digital maturity level. Although the division of digital systems into smaller milestones may allow hospitals to better benchmark their digital maturity and plan for investments, the requirement to meet digital maturity criteria across multiple domains can be challenging for hospitals in a resource-limited setting such as Vietnam. Duncan et al [10] discussed that most published maturity models, including EMRAM, are not assessed in LMICs, questioning their applicability in cultures different from those of high-income countries, such as the United States and the United Kingdom. In addition, the implementation of EMRAM poses the risk of investing in complex systems without meeting the organization’s local needs [37].

Most hospitals participating in this assessment had a low adoption rate of EMRs, CDSS, and CDRs. The low number of complete EMR systems may be caused by the cost required to upgrade the infrastructure and a lack of the required technical capabilities to meet all the requirements in EMR regulations. It is difficult for hospitals with incomplete EMR systems and CDRs to embark on using CDSS. Integrating, implementing, and maintaining CDSS is usually complex and depends on multiple factors, such as the availability and interoperability of data, especially from EMRs; an updated knowledge base; and understanding of the hospital’s workflow [38].

The wide availability of HIS, LIS, and RIS-PACS in the hospitals participating in our study suggests that digital systems for administrative and billing purposes and simpler systems to implement tend to be prioritized. Many hospitals may choose to implement clinical documentation functions before upgrading the other components to fully comply with EMR regulation. This leads to an increasing amount of medical data that are electronically available and useful for AI training and implementation even without full EMR implementation. However, the scarcity of CDSS, CDRs, and standard-based data-sharing experience found in this assessment suggests challenges that stakeholders may face when seeking to integrate AI-enabled solutions in Vietnamese hospitals. Although these challenges should be explored in further research, we believe a lack of national frameworks and detailed guidance for meaningful and secure clinical data exchange is 1 of the key barriers. In addition, recently, Chanh et al [39] mentioned exploring system interoperability, collaboration with local stakeholders from multiple disciplines, and understanding local data-sharing policies as some of the key considerations for applying AI in Vietnam.

We examined the adoption of digital health systems using the criteria published in the Vietnam HIT Maturity Model. This is a national benchmarking tool for HIT implementation in Vietnamese public hospitals, making it familiar to the hospitals participating in this study.

Only 5 public hospitals at the top-tier level were included in this assessment, so our findings do not represent the overall picture of HIT implementation in Vietnamese hospitals. The hospitals participating in this study were ones that had previous interest in DHT implementation and digital health innovation research. However, it is likely that hospitals at low tiers and located in less-funded regions have lower digital maturity across all the domains, especially in the EMR and extra capabilities domains.

The adoption of digital health systems in this study was assessed based on technological criteria, such as the software functions that were available. Future work could address other dimensions related to the implementation of these systems, such as the organizational context and human factors [40]. The use of qualitative methods can be particularly suitable for such research questions as they can unveil the characteristics unique to technology adoption in hospitals in LMIC settings, such as Vietnam.

Several major public hospitals in Vietnam have a sound digital infrastructure in place to support fundamental administrative and clinical tasks, such as patient management, insurance claims, laboratory result management, and medical imaging inspection. Most of the hospitals have implemented EMR systems to a basic level that prioritizes data collection, management, and access. Nevertheless, the more advanced level of data management and use via CDRs and CDSS is not common. This can be seen as a barrier to the introduction of new DHTs in these hospitals. Along with the increased amount of data collected by the systems, the adoption of HL7, DICOM, and other international standards can be seen as a facilitator for new DHTs, such as AI-based CDSS, to be implemented in these hospitals.