A hospital EMR is not a long note form. It is the clinical source of truth that must survive departments, shifts, interfaces, corrections, downtime, and discharge.
Many buying errors come from collapsing three layers: the clinical record, the hospital’s operational system, and external exchange. Separate them before scoring products.
Decision summary
- Define the clinical record before counting modules.
- Treat ADT, CPOE, eMAR, LIS, RIS, PACS, pharmacy, billing, and interfaces as acceptance-tested requirements.
- A FHIR or HL7 format claim does not prove a working interface or write-back.
- ABDM adds consent-based exchange; it does not replace the hospital record.
- Hospital Copilot is demo-priced and supports either a complete Patient Square HIS/EHR deployment or an alongside-existing-HIS/EHR path.
Define the clinical record before the product
Write the minimum longitudinal record as data, not department names:
| Record element | Questions |
|---|---|
| Patient identity | How are identifiers issued, matched, merged, unmerged, and audited? |
| Encounters | How are OPD, emergency, day care, admission, transfer, and discharge linked? |
| Problems and diagnoses | Which terminology is used? Who can correct or retire an item? |
| Allergies and alerts | How are source, certainty, reaction, correction, and override recorded? |
| Medications | How are history, prescription, verification, administration, hold, stop, return, and reconciliation separated? |
| Orders and results | How are order, specimen, status, result, amendment, acknowledgment, and cancellation linked? |
| Imaging | Where are order, worklist, study, images, report, addendum, and critical-result acknowledgment stored? |
| Notes and documents | How are author, signer, time, version, amendment, and attachment represented? |
| Procedures and devices | How are consent, performer, implant, batch, observation, and follow-up captured? |
| Discharge and continuity | What is available to the patient and next care team, and in which form? |
| Consent and access | How are purpose, scope, expiry, withdrawal, role, emergency access, and audit recorded? |
If a demonstration skips identity, correction, or provenance, it has skipped the hard part.
Keep system boundaries explicit
| Layer | Typical responsibility | Boundary question |
|---|---|---|
| EMR or EHR | Clinical record and clinical workflow | Is this the source of truth for the named data? |
| HIS or HMS | Registration, ADT, beds, billing, pharmacy, inventory, payer, finance, operations | Which operational modules are in the same product and which are connected? |
| LIS | Specimen, analyzer, result, validation | Which system owns order status and corrected results? |
| RIS and PACS | Imaging workflow, reports, images | How are identity, study, report, and image linked? |
| Pharmacy | Formulary, stock, dispensing, returns | Does a clinical prescription become a verified dispense request? |
| ABDM exchange | Consent-based health-information exchange | Which records are shared, under what consent, and what remains local? |
| Patient-facing channel | Access, appointments, messages, payments, consent | What is shown, and which system owns the source data? |
A single vendor may provide several layers. That does not remove the need to name the source of truth and failure owner for each handoff.
Named routes worth validating
These are not a best-to-worst ranking. They show why the buyer must first decide whether it needs programme software, an open-source operating model, a certified product shortlist, or a scoped commercial deployment.
| Route | What the first-party source establishes | What the hospital still has to prove |
|---|---|---|
| NIC e-Hospital | Its official module list separates Clinic, LIS, and RIS functions inside a government-hospital system | Eligibility, rollout, exact version, module availability, migration, support, and interfaces |
| C-DAC eSushrut | C-DAC describes it as a hospital-management system used in government health settings | Access route, deployment owner, current scope, hosting, support, and exit |
| Bahmni | Its project site describes an open-source hospital and clinic system | Implementation partner, clinical configuration, infrastructure, security, recovery, interfaces, and support |
| NABH software directory | NABH operates separate HIS and EMR product-certification standards and a directory | Certification version, facility fit, excluded workflows, implementation quality, and contract |
| Patient Square Hospital Copilot | Demo-priced with complete Patient Square HIS/EHR and existing-HIS/EHR deployment paths | Required modules, interfaces, migration, acceptance tests, and the chosen source of truth |
A named product gets onto the evaluation sheet only when its access route and evidence fit the facility. A logo on a standards slide does not complete the workflow test.
MoHFW standards turn “digital” into specific questions
India’s 2016 EHR Standards cover capture, storage, retrieval, exchange, analytics, privacy, security, and semantic and technical interoperability. Use them to ask concrete questions:
- Which patient, provider, facility, encounter, diagnosis, medication, observation, and document identifiers are stored?
- Which terminology and value-set versions are used?
- Can the system preserve authorship, digital signatures, timestamps, amendments, and audit trails?
- Which exchange formats are produced, consumed, validated, and retained?
- How are access control, authentication, consent, encryption, backup, and recovery tested?
- How is a complete copy supplied to the patient or another provider?
Do not accept “standards compliant” as the answer. Ask for the conformance statement, sample data, validation result, and exception list.
ABDM adds exchange, not a central chart
ABDM’s FAQ says records remain where providers create and store them. The architecture is federated. Patients can grant granular and time-limited consent for a provider to access shared records and can revoke that consent.
That creates two separate requirements:
- The hospital must keep a complete, safe local record and remain operational when external exchange is unavailable.
- The hospital must perform the consent-based ABDM transactions it has actually implemented.
Test provenance, consent status, duplicates, superseded documents, expiry of access, and local availability during exchange failure. Patient Square’s ABDM integration remains roadmap-only, so it should not receive points for a live workflow.
Authorship and amendments cannot be afterthoughts
For every clinical entry, ask:
- Who authored it?
- Who reviewed and signed it?
- When did each event occur?
- Was text imported, dictated, templated, or machine drafted?
- What changed after signing?
- Can the original and amendment both be reconstructed?
- Who viewed, printed, downloaded, or exported it?
The clinician remains responsible for reviewing and signing Patient Square drafts. A structured draft is not a final record until the hospital’s signing and filing workflow completes.
Interface acceptance ladder
Score each required interface at its achieved level:
| Level | Evidence |
|---|---|
| Displayed | A user can view the other system |
| Exported | A file or message can be produced |
| Imported | The destination accepts the file or message |
| Matched | Patient, encounter, order, and author identities are reconciled |
| Workflow-complete | Acknowledgment, retry, correction, cancellation, and exception handling work |
| Operationally monitored | Queues, alerts, support ownership, downtime, recovery, and reconciliation are tested |
“FHIR” can describe a format or API. It does not, by itself, prove the higher levels.
A ten-encounter trial
Use synthetic data and run ten varied encounters before committing:
- new OPD patient;
- returning patient with a possible duplicate;
- emergency-to-admission transition;
- inter-ward transfer;
- lab order with a corrected result;
- imaging order with report addendum;
- medication allergy and documented override;
- discharge with medication reconciliation;
- downtime entry and recovery; and
- patient or receiving-provider export.
For each encounter, record completion, errors, duplicate entry, elapsed staff time, source system, final source of truth, audit evidence, and recovery path.
Patient Square’s scoped hospital role
Hospital Copilot is a separate, demo-priced offering and does not inherit Practice Copilot rates. A hospital can deploy the complete Patient Square HIS/EHR or retain its existing HIS/EHR and define Hospital Copilot workflows during discovery. Neither path promises automatic filing or write-back into a named third-party system.
Patient Square’s AI Medical Scribe drafts a structured SOAP note, ICD-10 suggestions, and a prescription draft for clinician review. The product lists EHR-ready export (PDF · HL7 · FHIR). That exact phrase is an export claim, not certified interoperability or native enterprise write-back.
Required ADT, CPOE, eMAR, LIS, RIS, PACS, pharmacy, billing, and interface workflows must be scoped and acceptance-tested for the selected deployment.