NABH 6th edition: what the assessor will ask your software to prove

A NABH 6th edition documentation checklist is less useful than knowing what an assessor tests: existence, contemporaneity, traceability and closure.

Most medical directors looking for a NABH 6th edition documentation checklist already have the standard. The chapters are published. The objective elements are numbered. The quality manager has printed them and colour-coded them. That is not the gap.

The gap is the distance between a document existing and a document being believed. An assessor does not read your policy and tick a box. The assessor reads your policy, then asks a nurse on the floor what she does, then asks to see the record of the last time it was done, then asks what happened when it was not. The checklist tells you which documents to have. It does not tell you what will be done to them.

The four questions underneath every objective element

Strip away the chapter numbers and an assessment is a series of the same four questions, asked of one document after another.

• Does it exist? The easiest to pass, and the only one a binder solves.
• Was it contemporaneous? Was the entry made when the event happened, by the person who was there, or was it reconstructed later by somebody tidying a file.
• Is it traceable? Can you walk from the summary indicator back to the individual patient, the individual shift, the individual signature — and forward again, without a spreadsheet in the middle that nobody can explain the provenance of.
• Did it close? An incident raised and never resolved is worse than an incident never raised, because it is documented evidence of a system that notices and does nothing.

Hospitals lose marks on the last three. Almost never on the first.

Why the six-week scramble happens

The mechanism is not laziness. It is that in most hospitals, documentation is a separate activity from the work it documents. The work happens in one system, or on paper, or in somebody's head. The evidence is assembled afterwards, by a quality team that was not present, from sources that were not designed to be read as evidence.

That design makes the scramble inevitable. If evidence is produced by a second process, that process only runs when somebody starts it, and somebody only starts it when a date is approaching. Between assessments, the register goes stale — not because anyone decided to let it, but because nothing in the daily work was keeping it warm.

Evidence assembled in accreditation season is evidence about your file room. It is not evidence about your hospital.

It also produces a quieter cost that nobody puts in a budget: the clinical quality function spends its scarcest weeks producing paper rather than reducing risk. The hospital passes. The hospital does not necessarily improve.

What to do about it without buying anything

Several of the highest-return moves here cost nothing but discipline.

• Pick five objective elements at random each month and dry-run them. Not the ones you are proud of. Random. Ask the four questions above and see which of the last three fail. The pattern will repeat across the whole standard.
• Date-stamp honestly and let the gaps show. A register with visible gaps and a CAPA against the gaps is a stronger artefact than a register that is suspiciously complete.
• Give every CAPA an owner and a closure date at the moment it is raised, not at the moment it is reviewed. Open-ended actions are the single most common thing that fails the closure question.
• Stop maintaining parallel registers. If the infection control data lives in one workbook and the incident data in another and the indicator submission in a third, reconciliation work is being created every month by the structure itself.
• Ask your existing vendor a direct question: can this system produce the evidence for objective element X without anyone exporting anything. If the answer involves Excel, you know where your next six weeks are going.

Where Ospia fits

Ospia HOS is an AI-native Hospital Operating System rather than a hospital management system. One patient record across 31 domains, worked by a governed workforce of AI agents — each with a defined scope, permissions, KPIs and escalation path, on an autonomy dial the hospital sets per task.

The compliance surface covers FHIR R4, HL7 v2, ABDM, DPDP, NMC, NABH indicator packs and GST e-invoicing. On our compliance page, supported means implemented and demonstrable, and by design means the platform produces the required behaviour or evidence as a by-product of normal operation. NABH sits in the second category: CAPA, a live risk register, HAI surveillance and antimicrobial stewardship are intended to fall out of daily work rather than out of a pre-assessment project. NMC rules are enforced where they are cheapest to enforce — at write time, covering generic-name prescribing, prescriber statutory identity on every document, serialized certificates and the 72-hour record-release clock, escalated before it lapses.

Several members of the agent fleet exist for exactly this surface: quality, infection control, NMC compliance and biomedical watch among them. They do the monitoring that a quality team cannot do continuously, and escalate rather than decide.

Ospia HOS runs hospitals in production today. We do not publish how many, and we publish no case studies, until each hospital completes hypercare and agrees to be named. When we have measured accreditation outcomes we will publish those too, and not before. What we will show you now is the mechanism, live, against your own objective elements.

The ask is modest: bring five objective elements you expect to struggle with to an executive demonstration, and we will show you where the evidence for them would come from.