NABH accreditation software checklist: what the standard actually asks your system to produce

NABH publishes no software checklist. Here is the derived one: eleven things your system should produce unprepared, and why evidence gets reconstructed.

Six weeks before an assessment, a quality team starts collecting. Someone pulls hand-hygiene audit sheets from the infection control nurse. Someone else reconstructs the CAPA register from email. A consultant is asked to sign a set of records dated four months ago. Nobody thinks this is right, and everybody does it.

If you have searched for an NABH accreditation software checklist, the honest answer is that NABH does not publish one. The standard does not name a product, a module or a vendor. It describes what your hospital must do and what it must be able to show, and leaves the question of how entirely to you. So the only useful checklist is a derived one: the list of things an assessor will ask for, turned into a test of whether your system can produce them on the day, without anyone preparing.

That is the list below. It is written for the medical director or superintendent who will be walking the assessor through the wards, because that is the person who finds out, in real time, whether the evidence exists.

Why the evidence is always reconstructed

The mechanism is not laziness. It is a structural mismatch between what hospital software was built to record and what accreditation asks you to demonstrate.

Most hospital systems are transaction systems. They exist to register a patient, post a charge, dispense a drug, raise a bill. They record that a thing happened, at the level of detail required to bill for it. Accreditation asks a different question: not whether the thing happened, but whether it happened the way your own policy said it would, whether the exception was noticed, what was done about it, and whether the same exception recurred.

The NABH hospital accreditation standards are organised as standards broken into objective elements, and it is the objective element that the assessor actually tests. An objective element is rarely satisfied by a document. It is satisfied by a record — dated, attributed, unedited, and consistent with the twenty other records around it. Documents describe intent. Records describe what occurred. Assessors have been reading both for twenty years and can tell them apart in about four minutes.

So the gap opens in three predictable places:

• Evidence lives outside the system. Infection surveillance in a spreadsheet. CAPA in email. Risk register in a document on a shared drive. Mortality review minutes in a folder. Each is defensible on its own. Together they mean that no single query can answer "what happened in this hospital last quarter".
• Records can be edited without a trail. If a clinical note can be changed after the fact and nothing records that it was changed, then every note in the hospital is weaker evidence, including the honest ones.
• Compliance is a project, not a by-product. When evidence is assembled for an event, it degrades the moment the event passes. Twelve months later the quality team starts again.

The derived checklist: eleven things your system should produce unprepared

Take these one at a time, and give the team no notice. If it takes more than a few minutes, or it takes a person rather than a query, you have found where the reconstruction work will fall.

• A named prescriber on every prescription, with statutory registration. Not a login. The registered identity, on the document, at the time it was written.
• Generic-name prescribing behaviour, measurable. Not a circular reminding consultants of the rule. A number, by department, over a period.
• The complete record of one patient, end to end. Registration, assessment, consent, orders, administration, notes, discharge summary. One record, one timeline, no stitching across four systems.
• Consent, with what it covered and when it was taken. Including consent for procedures and for use of personal data.
• Records released within the statutory window. A live view of pending requests and their clocks, not a register maintained by hand.
• The CAPA register, open items with owners and due dates. Including items that are overdue, visible rather than quietly closed.
• The risk register, as of today. Not as of the last review meeting.
• Healthcare-associated infection surveillance, by unit, by device, over time. Denominators included.
• Antimicrobial stewardship activity. Restricted-antibiotic use, approvals, de-escalation.
• Incident reports and their investigation trail. Including near misses, which is the part that tells an assessor whether reporting is actually safe in your hospital.
• Quality indicators with their raw denominators. An indicator you cannot drill into is an indicator you cannot defend.

The current standard document is public. Read the NABH Hospital Accreditation Standard, 6th Edition against your own system rather than against a vendor's compliance datasheet, and mark each objective element as produced automatically, produced on request, or assembled by a person. The third category is your real workload.

What to fix before you buy anything

Most of the distance can be closed without a procurement.

Stop maintaining parallel registers. If the incident register exists both in the system and in a spreadsheet, one of them is wrong and the assessor will find the one you did not expect. Pick the system of record per register and retire the other, even if the system version is uglier.

Make one indicator honest end to end. Choose a single quality indicator and trace it from raw event to reported number without manual adjustment. Whatever breaks in that trace will break in all the others.

Run an internal assessment with no preparation window. The rehearsed mock assessment tests whether your team can prepare. The unannounced one tests whether the system produces. Only the second is predictive.

Audit the edit trail before someone else does. Ask whether a clinical note, a drug administration entry or a vital sign can be changed after the fact, and whether the change is visible. If you do not know, that is the answer.

Ask your vendor for the evidence query, not the feature. "Show me the screen" is a demonstration. "Show me the export, for last quarter, with denominators" is a test.

Where Ospia fits

Ospia HOS is an AI-native hospital operating system rather than a hospital management suite, and the difference matters here for one reason: one patient record across the hospital's domains means the evidence is a query rather than an assembly job. Ospia's compliance position is that NABH evidence should be a by-product of daily work — CAPA, a live risk register, healthcare-associated infection surveillance and antimicrobial stewardship maintained as operations run, not reconstructed before an assessment.

Some of the statutory behaviour is enforced at write time rather than audited afterwards. NMC rules on generic-name prescribing, prescriber statutory identity on every document, serialized certificates and the 72-hour record-release clock are applied when the record is created, and the clock escalates before it lapses. ABDM and ABHA are listed as supported, meaning implemented and demonstrable; on the compliance page "by design" means the platform produces the required behaviour or evidence as part of normal operation. Registration is ABHA-ready, and every use of personal data traces to recorded consent held in a consent ledger that is consulted before use.

Background agents watch the same surface continuously — among the agents running in the platform are quality and infection-control watchers — and every agent action produces a replayable ledger entry, an event trail and a reversal path, at every level of the autonomy dial from manual to autonomous-within-policy. The dial is set per task and held by the hospital, not by us.

What we will not tell you is how much faster this made anyone's assessment. Hospitals run on Ospia HOS today, and we publish names, numbers and measured outcomes only as each hospital completes hypercare and agrees to be named. The case-study slots on our site are unfilled placeholders describing what will be measured. If a vendor shows you an accreditation outcome figure, ask which hospital, which cycle, and who measured it.

Accreditation is also a workload question, and the shape of that workload differs by hospital type — a multi-speciality hospital carries a different evidence burden across departments than a single-speciality unit does. And if you are comparing platforms on this dimension specifically, our Ospia versus Napier comparison sets out where each is stronger, including where they are not us.

No system makes a hospital compliant. A system can only decide whether your compliance evidence is a record or a reconstruction.

If you want to see the evidence surface rather than a slide about it, ask for an executive demonstration, and bring three objective elements you expect to be difficult.