Infrastructure and Capacity Building for Semantic

70

Building Capacity for Health Informatics in the Future F. Lau et al. (Eds.) © 2017 The authors and IOS Press. This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0). doi:10.3233/978-1-61499-742-9-70

Infrastructure and Capacity Building for Semantic Interoperability in Healthcare in the Netherlands a

Ronald CORNET a,b,1 Department of Medical Informatics, Academic Medical Center – University of Amsterdam, The Netherlands b Department of Biomedical Engineering, Linköping University, Sweden

Abstract. Over 15 years, a broad spectrum of activities was undertaken to realize a health IT infrastructure in the Netherlands. In this paper we reflect on the history, challenges, accomplishments, changes, and the way forward. It shows that the infrastructure depends on technical, legal, and semantic aspects, which are frequently reciprocally related. It also highlights the fact that the role of health professionals and of patients is increasingly considered as a crucial element. Keywords. Electronic health records, National & international strategies and policy

1. Introduction Nowadays, many countries are in the process of furthering the use and interoperability of electronic health records (EHRs), personal health records and patient portals. Wellknown are the Meaningful Use program in the United States, and the NHS National Program for IT (NPfIT) in England. Both are examples of large-scale national programs, in which a substantive budget is allocated for the realization of interoperable EHRs. Both programs differ substantially, as the Meaningful Use program aimed at an increased use of EHRs in the first phase, and of (re)use of data in the first and second phase, whereas the NHS NPfIT aimed at interoperability at a national level of existing EHR systems. In the Netherlands, an approach was taken which differs from both the USA and England. Over 15 years, a development was realized in which the emphasis has shifted from technical interoperability to semantic interoperability, and in which the drivers have not primarily been monetary incentives, but rather stakeholder awareness, education and involvement. In this paper, we present an analysis of the various phases, and an outlook to the future.

1

Corresponding Author. Department of Medical Informatics, Academic Medical Center – University of Amsterdam, P.O. Box 22700, 1100 DE Amsterdam, The Netherlands; E-mail: [email protected].

R. Cornet / Infrastructure and Capacity Building for Semantic Interoperability

71

2. Technical Interoperability In 2002, the AORTA infrastructure was conceived. AORTA is the Dutch national infrastructure for the exchange of data between healthcare providers [1]. It consists of a so-called national switchpoint, based on a messaging mechanism using HL7 version 3. The switchpoint does not hold any clinical patient data, but contains references to systems that do, and facilitates the retrieval of patient data from those systems. To be able to exchange patient data, a unique patient identifier needed to be introduced. Furthermore, it is crucial that only authorized users have access to these systems and to the switchpoint. This requires unique identification of care providers, for which badges and badge readers have been introduced. Finally, as a switchpoint with decentralized storage requires the connected systems to have a very high availability, only systems that are accredited as “Qualified Healthcare Information Systems” are allowed to connect. Late 2016, the switchpoint is used by about 92% of care providers, exchanging 150,000 messages each day for two main scenarios: medication information and an electronic locum record 2. 2.1. Policies & Politics Whereas there were plenty of technical challenges, the main barriers for implementation of the infrastructure were non-technical. The first barrier was the use of a unique patient identifier. The decision was made to use the social-fiscal number that had been in use since 1988. However, it took until 2007 before legislation was in place to use the then-called citizen service number for other than social-fiscal purposes, and in 2008 it was allowed to use it in healthcare. Late in 2008, the use of the infrastructure officially commenced, in anticipation of legislation pertaining to use of the citizen service number for electronic exchange of health information. This legislation did not pass, and in 2011 the Dutch government decided that the national infrastructure needed to be cancelled. It was then handed over to the “Association of Healthcare providers for Care Communications”, which was formed to enable continuation of the infrastructure in a regionalized way and without compulsory participation by care providers [2]. A final important barrier was the initial choice for patient opt-out, rather than optin. After rejection of the national switchpoint, it was decided that patients should explicitly opt-in for the system, otherwise their information could not be exchanged. By the end of 2016, 10.6 million, i.e., 62% of Dutch citizens have opted in2.

3. Semantic Interoperability In 2007, the Dutch government embarked upon the journey towards semantic interoperability by joining the newly formed International Health Terminology Standards Development Organization (IHTSDO). The AORTA infrastructure was mainly technical, and the lack of semantics enabled only exchange of a limited (freetext) data set, which aimed at and required human interpretation. Hence this step created the perspective of a larger data set and interpretation of data by both humans 2

www.computable.nl/artikel/achtergrond/magazine/5840508/5215853/hoe-staat-het-met-het-lsp.html


72

and machines alike. However, as Dutch healthcare is as decentralized as the switchpoint infrastructure, there was no clear implementation path at hand. To realize capacity building in semantics in healthcare, the Dutch National Release Center Nictiz started introductory courses in SNOMED CT, to bring people up to speed. Whereas there was a perceived knowledge of SNOMED, this was largely based on previous versions. Dutch pathologists use PALGA, a thesaurus based on SNOMED II. This made people think that SNOMED CT still was a multi-axial system, and that there were issues with availability and currency. The introductory course focused on the why and how of medical terminology, i.e., the need of a concept-based system, the distinction of a statistical classification and a reference ontology, and the basic structure and contents of SNOMED CT. Until now, the course has run about 30 times, attracting among others healthcare professionals, healthcare IT staff, and consultants. In addition, more in-depth conferences and meetings were organized, such as a national healthcare terminology day, and technical courses with international experts. Through these activities, hundreds of people got a basic understanding or more of the terminology landscape in general, and more specifically in SNOMED CT, also leading to attention in Dutch medical journals [3,4]. One challenge in using SNOMED CT is that in practice Dutch terms need to be used, whereas there hasn’t been a Dutch translation of SNOMED CT. Rather than doing a translation upfront, it was decided to work on the development of interface terminologies. Inspired by two hospitals that strived at implementing an EHR and wanting to be able to exchange patients and their information, Dutch Hospital Data (DHD) started development of the so-called thesauri for diagnoses and procedures. In 2010, development of the diagnoses thesaurus commenced, and late 2011 first thoughts were given to the procedures thesaurus. The first full version of the diagnoses thesaurus was published in 2015; the procedures thesaurus is still under development.

4. National Program for “Facilitating Clinical Documentation at the Point of Care” As mentioned, in the decentralized setting of the Dutch healthcare system, it is hard to realize a change from (legacy) systems with proprietary terminologies to standardsbased systems. Early adopters have little benefit of having an interoperable system, so there needs to be a critical mass for it to be beneficial. To get the field moving, the 8 university medical centers (UMCs) joined forces and described a vision for 2020 [5]. The vision consisted of 4 elements: x Patient-oriented, continuous care x Documentation of care is part of the care process and fits the workflow of care providers x Capture once, use many times x International standards The vision was adopted by the Dutch department of health, and a national program started in 2014, aiming at a 3-tier strategy: x x x

Increase quality and usefulness of healthcare information Optimize care processes Demonstrate improved outcomes of care


73

The strategy outlines the phases of the program, and has a step-wise approach, focusing on the foundational layers first, as these are sine non to realize the third step. The strategy also indicates that the program is not about technology, but about the people providing the care, capturing the data, and ultimately the patients. The people-driven strategy led to an approach in which capacity building is an important aspect. It focuses on the following aspects: x x x x x x

User-friendly EHRs Data capture not only by clinicians, also nursing specialists, medical coders and patients Train care professionals in documentation of care Be open on purposes of data use Patient access to data Support patients in interpreting data

The national program was initially undertaken by the 8 UMCs, but as of 2017 the general hospitals are also participating.

5. Capacity Building for Point-of-Care Clinical Documentation As health professionals and patients play an essential role in improving the quality of health data they need to be involved and informed. 5.1. Patient Involvement The technical and semantic infrastructure and the national program are the basis for the ambitions of the Dutch department of health. Their aim is that by 2020: 80% of chronically ill has direct access to medication info, vital functions and test results, available for mobile apps or internet applications; 75% of chronically ill and vulnerable elderly can perform self-tests, remotely monitored by care provider; and home care & support is supplemented by 24/7 telecare. For 2017-2020, a total of €105 million is budgeted to realize this aim. In order for patients to be able to enter and access medical data, they need to be able to understand the technological aspects as well as the contents of their record. Whereas technological barriers disappear by increased experience of patients with devices such as tablets and smart phones, understanding and interpreting the contents of their record remains challenging. However, positive experiences have been gained [6], and the further implementation of the semantic layer makes it possible to provide additional functionality, such as presentation of patient data in layman’s language, and linkage to authoritative sources for further information. The semantic layer will also be important for a further refinement of the record access regulations. The current generic opt-in/opt-out mechanism will over time be replaced by a system based on specific permission for data provision, in which a patient can specify which data may be provide to which care providers or categories of care providers. This can only be determined adequately if data can be categorized, so that for example information pertaining to diabetes treatment can be distinguished from information regarding a period of depression.

74


Furthermore, patients will need to be made aware of their rights and of the possibilities. There will be an important role for patient associations to support this process. 5.2. Health Professional Involvement There is increasing involvement of healthcare professionals in technological innovation. Rather than being users of technology that is imposed onto them, they become partners in the process of selecting technology, of implementation and configuration, and of improvement during use. However, these skills are not yet trained in the current medical curricula. Therefore, two information and education efforts are undertaken to raise awareness and capacity of health professionals. Within the national program, a short e-learning course is developed to inform health professionals on the potential of reusable data. It addresses changes in documentation behavior as well as changes in the information systems and terminologies being used. Make users more aware is essential, as research has shown that they feel that good documentation is important, but in practice the quality and extent to which data are structure are suboptimal [7]. In addition to the short course, a Master program in Health Informatics has been developed by the University of Amsterdam, aiming at filling the gap between the knowledge and skills taught in current medical programs, and the required capacities of the current and future generation of healthcare professionals.

6. Conclusion In 15 years, we have moved from laying the technical and legal foundations for an exchange of healthcare information to building the semantic infrastructure and realizing the capacity of health professionals and patients to exploit the potential of the infrastructure. While we are at the brink of an incremental national implementation of the semantic infrastructure, we will need to closely watch the impact it has on the actual collection, use, and quality of data, and ultimately, on quality of healthcare.

References [1] R. Spronk. AORTA, the Dutch national infrastructure, http://www.ringholm.de/docs/00980_en.htm, last accessed 2016-09-18. [2] J. Smits. The future of the national virtual EPR in the Netherlands: Changes in approaching large-scale exchange of information in Dutch Healthcare. Journal of Chain-computerisation (2013) Jun 12;4. [3] G. Gooiker, M. Wouters, R. Tollenaar, E. H. Eddens. EPD: artsen aan zet. Medisch Contact 64 (4) (2009), 165-6. [4] R. Cornet, N. F. de Keizer. Investeren in de goudmijn, Medisch Contact 11(2009), 474-475 [5] J. Hazelzet, P. Georgieva. Registratie aan de bron. Visie op documentatie en gebruik van zorggegevens 2013–2020. Nederlandse Federatie van Universitair Medische Centra (NFU), 2013 April, 2013. Report No. NFU-13.3694. [6] S. S. Woods, E. Schwartz, A. Tuepker, N. A. Press, K. M. Nazi, C. L. Turvey, W. P. Nichol. Patient experiences with full electronic access to health records and clinical notes through the My HealtheVet Personal Health Record Pilot: qualitative study. J Med Internet Res 15(3) (2013), e65. [7] E. Joukes, R. Cornet, N. F. de Keizer, M. de Bruijne. Collect once, use many times: End-users don't practice what they preach. Proceedings of MIE 2016, München (2016), Stud Health Technol Inform 228 (2016), 252-256.