A patient discharged from a Bengaluru hospital arrives at her regular physician a week later carrying a folder. Inside is a discharge summary as a printed PDF, two lab reports from a different chain, and a photo of a prescription on her phone. Her physician spends the first eight minutes of a fifteen minute consultation reading paper. None of that information is queryable, none of it will reach the next provider, and if the patient changes cities the folder is the record.
This is the problem interoperability standards exist to solve, and India now has both the technical standard and the national plumbing in place.
What FHIR Gives You, and What It Does Not
FHIR structures clinical information as discrete resources: a Patient, an Observation for a lab result, a MedicationRequest for a prescription, an Encounter for a visit. Each is addressable over a REST API with a defined schema. The practical consequence is that a haemoglobin value stops being a number inside a PDF and becomes a typed field with units, a reference range, a timestamp and a coded identifier.
What FHIR does not give you is agreement on meaning. Two systems can both be FHIR-compliant and still disagree on how a diagnosis is coded or which units a result uses. That is why terminology matters as much as transport. India’s National Resource Centre for EHR Standards mandates SNOMED CT for clinical terms and LOINC for lab observations, with ICD-10 for morbidity coding. Skipping the terminology layer produces systems that exchange data successfully and understand none of it.
ABDM Is the National Layer
The Ayushman Bharat Digital Mission provides the identity and discovery infrastructure that FHIR alone does not. A few components matter for anyone building clinical software:
- ABHA number and address. A 14 digit health identifier plus a human-readable address that links a patient’s records across providers without those records being centrally stored.
- Health Facility and Healthcare Professional registries. Verified directories so a request for records comes from a known, credentialled source.
- Health Information Exchange and Consent Manager. The consent artefact is the core idea. Records stay with the provider that created them, and a patient grants time-bound, purpose-bound access rather than handing over a permanent copy.
- Federated by design. There is no national record database. Software must be able to fetch on demand rather than assume local possession of a patient’s history.
For a clinic system, ABDM readiness means implementing the linking flow, honouring consent expiry, and being able to serve a FHIR bundle when a legitimate request arrives.
Workflow Is Where Adoption Is Won or Lost
Standards compliance does not make a doctor use a system. Structured data entry costs clinician time, and if capturing a coded diagnosis takes four extra clicks compared with typing free text, free text wins every single day.
The implementations that succeed reduce that cost. Smart defaults from the patient’s history, specialty-specific templates, voice input for notes, and coded fields that autocomplete from the first three characters. A useful test: a routine follow-up consultation should be documentable in under 90 seconds of clinician interaction. If it is not, the structured data will be incomplete regardless of how correct the FHIR mapping is.
At Invexa, we approach clinical software as a workflow problem wrapped in a standards problem, because a compliant record that nobody fills in properly helps no one.