A hospital can be entirely paperless and still be unable to answer a basic question: what happened to this patient at another facility last month? Digitisation and connection are different problems. One is about replacing paper inside an organisation. The other is about moving meaning between organisations that never agreed on how to describe anything.

Across Asia, that second problem is moving from a strategic aspiration to a legal obligation. Singapore has legislated mandatory contribution to its national record. India has built one of the largest linked health record systems anywhere. Malaysia has funded a national electronic medical record rollout with fixed deadlines. Japan is standardising clinical document exchange on an international data standard. For hospital leaders, the practical consequence is that decisions made in the next two procurement cycles will determine whether an institution connects cleanly or spends years paying for custom integration work.

Digital is not the same as connected

Most hospital systems in the region were bought one department at a time. Laboratory, radiology, pharmacy, billing and clinical documentation each arrived with their own vendor, their own data model and their own idea of what a diagnosis code looks like. Each system works. Together they produce an institution that generates enormous quantities of data and can share almost none of it without manual effort.

This matters beyond convenience. The World Health Organization’s global digital health strategy, extended by Member States through 2027 with a successor strategy in development for 2028 to 2033, places interoperable systems at the centre of health system strengthening. WHO has also worked with HL7 International to support adoption of the FHIR standard, which underpins most modern health data exchange, and encodes its own clinical recommendations into machine-readable formats intended to travel across systems. The direction of policy is unambiguous. The question for individual institutions is sequencing.

The four layers, and where hospitals usually stall

Interoperability is commonly described in four layers. The distinction is useful because most hospitals solve the first two, declare victory, and discover the failure much later.

Layer	What it means	Typical failure in practice	How to test it
Foundational	Two systems can exchange data at all	Connection exists but is one-way, or breaks on version updates	Send a test message both directions after a vendor patch
Structural	The data arrives in a predictable format	Fields arrive intact but land in the wrong place or truncate	Compare a transmitted record field-by-field against the source
Semantic	Both systems understand the data the same way	Local drug names, free-text allergies and non-standard codes that mean nothing to the receiving system	Count how many of your clinical terms map to a recognised coding standard
Organisational	Governance, consent and workflow allow the data to be used	Data is technically available but nobody is permitted, trained or incentivised to open it	Measure how often received external records are actually viewed before a decision

Semantic failure is the expensive one. A system that transmits perfectly formatted nonsense passes most acceptance tests. Allergies recorded as free text, medications recorded under local trade names, and diagnoses recorded in a proprietary internal code set will all move successfully between systems and mean nothing on arrival. Retrofitting terminology across years of accumulated records is substantially harder than specifying it correctly at the point of purchase.

The regional picture: rules are arriving faster than readiness

Singapore

Parliament passed the Health Information Act in January 2026. It requires licensed healthcare providers and retail pharmacies to contribute key health information to the National Electronic Health Record, including allergies, vaccinations, diagnoses, medications, laboratory results, radiological images and discharge summaries. Implementation is batched, with the first tranche of providers expected to begin contribution and implement cybersecurity and data security measures by September 2027, and later groups following through 2028 and 2030.

Two details are frequently missed. Contribution applies prospectively on connection, so there is no requirement to upload historical records. And contributing is mandatory while accessing the national record is not, which means a provider can be fully compliant and still derive no clinical benefit if nobody in the organisation uses what arrives.

India

The Ayushman Bharat Digital Mission has linked more than a billion health records to over 930 million health accounts, according to a government fact sheet issued in July 2026. Linked records roughly doubled in fifteen months, and more than 450 public and private health technology solutions have integrated with the ecosystem. The identifier-plus-consent architecture is the part worth studying: rather than centralising records, it federates them and makes discovery possible.

Malaysia

The Ministry of Health has committed to completing electronic medical record implementation across primary care facilities by the end of 2027, through a cloud-based clinic management system for health clinics and a parallel system for dental clinics. In August 2026, the government raised a digitalisation allocation to RM1 billion to accelerate the rollout and expand connectivity at public hospitals and clinics. The sequencing here is notable: connectivity and record systems are being funded together, which is a recognition that a digital record in a facility with unreliable bandwidth solves very little.

Elsewhere in the region

Japan’s health ministry has been standardising electronic medical record information and exchange formats around HL7 FHIR-based document specifications, covering areas such as referral letters, discharge summaries and health check-up results. In Taiwan, a study of record exchange across medical centres, regional and district hospitals and more than five thousand clinics found that laboratory reports and outpatient records accounted for the overwhelming majority of downloads, far ahead of medical images and discharge summaries. That is an observational finding rather than a rule, but it is a useful signal: when clinicians are given access to everything, they reach for a small number of document types. Connecting those first delivers most of the usable value.

Five questions to ask before the next system purchase

Vendor demonstrations rarely fail. Integrations do. These five questions surface most of the risk before a contract is signed.

1. Which standards does the system support in production, not on the roadmap? Ask for a named reference site in your jurisdiction where the standard is live, and ask to speak to its technical lead.
2. Who owns the data, and what happens on exit? Establish in writing the export format, the timeline, the cost, and whether structured coding survives the export. A migration that returns flat files has erased your semantic layer.
3. What proportion of our clinical vocabulary maps to a recognised code set? If the honest answer is unknown, that audit is the real first project, not the purchase.
4. What is the integration cost for the next three systems? A platform that is inexpensive to buy and charges for every connection will cost more over five years than a more expensive system with open interfaces.
5. Who is accountable when an exchange fails at three in the morning? Interoperability creates dependencies between organisations that have no contract with each other. Escalation paths should exist before they are needed.

What to measure, and what the numbers do not prove

Digitisation programmes tend to be reported in activity metrics: facilities connected, records linked, systems integrated, documents transmitted. These are legitimate and verifiable, and they are the right way to describe institutional progress. They should not be presented as evidence of clinical improvement.

The distinction is worth stating precisely, because it is regularly blurred in healthcare communications:

• Activity measures — facilities connected, records exchanged, uptime, integration count. Objective, auditable, and a fair description of operational achievement.
• Utilisation measures — how often external records are actually opened before a clinical decision, and by whom. Harder to collect and far more informative.
• Outcome measures — duplicate test rates, medication reconciliation errors, transfer delays. These require study design, comparison groups and time. They cannot be inferred from connection counts.

Policy documents across the region describe better continuity of care and reduced duplication as the intent of national record systems. Intent is not evidence. An institution that wants to claim clinical benefit needs to measure it directly, and should expect that measurement to be considerably less flattering than its connectivity dashboard.

Milestones, recognition and the limits of a claim

Large digitisation programmes produce genuine institutional milestones: the first hospital group in a market to connect every facility, the largest volume of records migrated in a defined period, the widest network of clinics brought onto a single system. These are the kind of achievement that can be specified, counted, dated and independently checked, which is what separates a documented milestone from a marketing adjective.

Organisations that reach an exceptional milestone of this type sometimes seek independent documentation of it, and record recognition in Asia is one route. Bodies such as Asia Record assess measurable achievements submitted by businesses, institutions and organisations, and set out publicly how proposed records are assessed and verified before any recognition is confirmed. Healthcare organisations considering it should understand exactly what such recognition establishes and what it does not. Becoming an Asia Record holder documents the scale or nature of a stated achievement. It is not a clinical endorsement, it does not substitute for accreditation, regulatory approval or medical device authorisation, and it says nothing about whether patients were better treated. Any institution that presents a scale record as proof of superior care has overstated it. Where a hospital group, healthtech company or health programme does have a milestone that is specific, measurable and evidenced, the route to apply for Asia Record recognition is a documentation exercise, and the supporting evidence is the substance of it.

A realistic twelve-month agenda

For boards and executive teams, the useful work over the next year is unglamorous:

• Establish which of your obligations are already legislated in your market and when each tranche applies to your licence categories.
• Audit your clinical terminology before buying anything further. Knowing the mapping gap is the precondition for costing everything else.
• Prioritise the two or three document types clinicians will actually use, rather than attempting complete exchange at once.
• Write exit and export terms into every contract signed from now on.
• Instrument utilisation from day one, so you can eventually tell whether connection changed behaviour.

Interoperability is rarely the project that wins internal enthusiasm. It has no demonstration screen and no obvious launch moment. But every clinical technology a hospital buys over the next decade — decision support, remote monitoring, population health analytics, anything built on machine learning — will be limited by the quality and consistency of the data underneath it. Institutions that fix the foundation now will be able to adopt those tools. Institutions that keep buying at the surface will keep paying to translate their own records back to themselves.