Part 1 traced pre-EHR hospital systems from billing to order entry. This second article covers the period from "year one of the EHR" in 1999, when electronic storage of medical records was formally permitted, to the 2010s, when EHRs spread to a majority of general hospitals. It organizes how policy, technology, and hospital practice moved to establish the EHR.
Figures such as adoption rates cited here are approximate values based on published sources such as the MHLW Survey of Medical Institutions. Always verify the latest values and detailed breakdowns against primary sources.
1999, "year one of the EHR": the three principles of electronic storage
In April 1999, the then Ministry of Health and Welfare issued the notice "On Storage of Medical Records on Electronic Media," formally permitting electronic storage under certain conditions. The conditions it set are the "three principles of electronic storage" inherited to this day: authenticity (assurance that records have not been tampered with), legibility (the ability to display and output content readably when needed), and preservation (retaining records for the statutory retention period).
The notice's significance lay in removing legal uncertainty. Until then, hospitals creating records electronically still had to print and store them on paper. The notice made a "paperless hospital" legally possible, prompting vendors to seriously develop EHR products and hospitals to seriously consider adoption. This is why the year is called "year one of the EHR."
Pioneering efforts predated the notice. Kameda Medical Center in Chiba, which put a self-developed EHR into operation in 1995, is well known as Japan's first full-scale EHR. Such cases were exceptional, however, and it was only after the 1999 notice that most hospitals came to see the EHR as a realistic option.
- April 1999: ministry notice formally permits electronic storage of records
- Three principles: authenticity, legibility, preservation
- Paperless hospitals become legally possible; the market begins to move
- 1995: Kameda Medical Center's self-developed EHR as a precedent
The 2001 Grand Design: national adoption targets
In December 2001, the MHLW formulated the "Grand Design for Informatization in Health and Medical Care," setting concrete numerical targets: by FY2006, EHRs in 60% or more of hospitals with 400 or more beds and 60% or more of all clinics. This was the first time the national government explicitly made the EHR a pillar of health policy, and it became the starting point for subsequent subsidies and fee-schedule measures.
The target was not met. In the mid-2000s, EHR adoption remained around 10% of hospitals overall. Causes included high implementation costs, concern about physicians' entry burden, vendor-specific specifications and lack of interoperability, and difficulty seeing returns commensurate with cost from a management perspective. The shortfall exposed that EHR adoption was not merely a technical issue but one of cost burden and incentive design.
Related institutions built during this period, however, steadily laid the groundwork. A 2002 notice addressed external storage of records; the e-Document Act, effective 2005, made it possible to scan paper documents for electronic storage. Also in 2005, the first edition of the "Guidelines for the Security Management of Medical Information Systems" was formulated, translating the three principles into concrete practical requirements.
- 2001 Grand Design: 60% of 400+ bed hospitals and clinics by FY2006
- Target missed; mid-2000s adoption around 10% of hospitals
- Barriers: cost, entry burden, lack of interoperability, unclear returns
- 2002 external storage notice; 2005 e-Document Act and first security guidelines
DPC and electronic claims: data transforms hospital management
In 2003, DPC/PDPS, a per-diem bundled payment by diagnosis group, was introduced for specified-function hospitals. Since inpatient care is paid at bundled points per diagnosis group, hospitals had to grasp their care by group and manage length of stay and resource use. DPC hospitals were also required to submit clinical performance data (Form 1, EF files, etc.) to the government.
This submission requirement fundamentally changed hospital information systems. Producing accurate DPC data requires diagnoses, surgeries, procedures, and medications to be recorded electronically in structured form. As DPC hospitals grew in number, structuring data through EHRs and order entry became a management necessity. The practice of "DPC analysis" — analyzing DPC data for management improvement — was also born in this period.
Electronic claims advanced in parallel. A 2006 MHLW ordinance set the policy of making online claims submission mandatory in principle by FY2011, and paper claims rapidly disappeared. With the spread of electronic claims processing, integration between billing systems and EHR/order entry became tighter, and immediate reflection of clinical entries in claims data became standard.
- 2003 DPC introduction; data submission mandates drive structured records
- DPC analysis emerges; data use becomes a management issue
- 2006 policy for mandatory online claims → transition by FY2011
- Immediate reflection of clinical entry in claims becomes standard
Standardization efforts: SS-MIX and exchange standards
As EHRs spread, the problem of vendor-specific data formats preventing exchange between hospitals and departments became apparent. In response, SS-MIX (Standardized Structured Medical Information eXchange) began in 2006 as an MHLW project, establishing a specification for outputting and storing patient demographics, prescriptions, and test results in a standard folder structure using HL7. It evolved into SS-MIX2 in 2012.
SS-MIX has served as "standardized storage" from which data can be extracted in a common format regardless of EHR vendor, underpinning regional networks, secondary data use, and migration at system replacement. However, limited data items and vendor-specific implementation details kept it from achieving full interoperability. This limitation leads to standardization via HL7 FHIR in the 2020s.
For exchange standards, HL7 v2 was used for messages such as tests and prescriptions and DICOM for images, while Japan-specific code systems such as JLAC10 (laboratory tests), HOT (pharmaceuticals), and ICD-10-based disease name masters were developed as standard masters. From 2010, the MHLW has recognized these successively as "MHLW standards." Standardization is unglamorous work, but its accumulation became the foundation for later information exchange.
- 2006 SS-MIX, 2012 SS-MIX2: vendor-neutral standardized storage
- HL7 v2, DICOM, JLAC10, HOT, disease masters, and other standards established
- From 2010: MHLW standards recognition system
- Limitations lead to HL7 FHIR standardization in the 2020s
2010s: accelerating adoption, regional networks, and lessons from the earthquake
In the 2010s, EHR adoption accelerated steadily. According to the MHLW Survey of Medical Institutions, adoption in general hospitals rose from about 14% in 2008 to about 22% in 2011, 34% in 2014, 47% in 2017, and about 57% in 2020, passing the majority mark. Among hospitals with 400 or more beds it exceeded 90% by 2020, making the EHR virtually standard equipment in large hospitals. Adoption lagged in hospitals under 200 beds, and the gap by size became clear.
Regional medical information networks in which local institutions share patient information were also built across the country. Following precedents such as Nagasaki's "Ajisai Net" (launched 2004), many regions built networks in the 2010s with national subsidies, sharing clinical information via SS-MIX2 storage. Challenges such as stagnant participation, low patient enrollment, and securing operating funds also became apparent.
The Great East Japan Earthquake of March 2011 left heavy lessons for hospital information systems. Many institutions lost both servers and paper charts to the tsunami, driving home the importance of backing up clinical information and business continuity planning. The experience spurred interest in off-site data center backups and cloud-based systems. The 2010 revision of the external storage notice, which permitted storage at private data centers, also provided institutional support for the move to the cloud.
- General hospital EHR adoption: ~14% (2008) → ~57% (2020); 90%+ for 400+ beds
- Adoption lagged under 200 beds; the size gap became visible
- Nationwide regional networks and their operating challenges
- 2011 earthquake → rising interest in backup, BCP, and cloud
Problems that took shape during the adoption era
By the late 2010s, EHRs were nearly taken for granted in large hospitals. New problems, however, took shape during adoption. First, vendor lock-in. Because EHRs are tightly coupled with departmental systems, migrating to another vendor at replacement time incurs large costs for data migration and reconnection, creating a structure of repeated same-vendor renewals.
Second, growing documentation burden. While EHRs made sharing and searching easy, entries to satisfy billing requirements and various evaluation items increased, lengthening the time physicians and nurses spend facing screens. The criticism that clinicians "look at the computer, not the patient" became widespread. Third, accumulated data was underused: despite vast records, mechanisms to feed them back into care and management improvement were limited.
These problems — lock-in and slow standardization, documentation burden, and underuse of data — are precisely the themes that medical DX policy since 2018 and next-generation AI-embedded EHRs seek to address. The final Part 3 covers the medical DX roadmap, standardization and the national medical information platform, cybersecurity, and the shift toward AI-native EHRs.
- Vendor lock-in: tight coupling with departmental systems raises migration costs
- Documentation burden: entries for billing and evaluation weigh on clinicians
- Underused data: accumulated but not fed back
- These frame the agenda for medical DX and AI-native EHRs
RELATED ARTICLES