At month-end, a dialysis centre’s schedule reports 412 appointments. Its signed session register reports 405 completed records. The billing export contains 409 lines. None of those totals says where the seven-session difference came from.
A useful software demo starts with that disagreement and works backward. It should prove which visit was planned, which session was documented, which downstream event was authorised, and which exception still has an owner.
Takeaways
- Reconcile recurring appointments to session records before reviewing charges.
- Keep slot, station, machine, approved order, and completed session as separate objects.
- Treat every unexplained difference as an owned exception, not an adjusted total.
- Leave treatment orders and clinical decisions with authorised professionals.
This is a procurement exercise, not a dialysis protocol, clinical standard, scheme guide, or legal opinion. Nephrology, nursing, technical, biomedical, quality, billing, privacy, and compliance owners must define the centre’s safe workflow.
Ask for a month-end evidence dossier
Create a fictional month with two patients, three stations, two machines, and ten planned visits. Include one missed appointment, one rescheduled visit, one machine change, one corrected session entry, and one session completed during a short outage. Use no real patient information and no treatment instructions.
Ask the vendor to export five linked lists at the same cut-off time:
| List | Minimum proof |
|---|---|
| appointment register | stable visit ID, pattern, exception, final status |
| station and machine history | identity, availability source, change history |
| session register | visit link, approved author, event time, sign-off, corrections |
| consumables ledger | source session, item reference, quantity, reversal state |
| billing export | source event, authorisation state, cancellation or correction |
The MoHFW PMNDP facility manual documents programme screens for patient registration, machine details, slot management, and dialysis history. The PMNDP portal introduction describes that public programme and its IT platform. These are useful public examples of distinct operational records, not a specification for every private centre.
Reject an export that shows only totals. Each row needs an identity that can be traced across lists without overwriting the original evidence.
Keep the six identities apart
A recurring appointment is not a completed session. A station is not a machine. An approved order is not an observation, and a billing line is not clinical proof.
Pick one fictional visit and ask the vendor to display all six identities at once: patient, appointment, station, machine, approved order, and session. Change the machine assignment. The appointment pattern should remain intact, the old assignment should stay in history, and the completed record should point to the equipment actually recorded under the centre’s approved process.
Now correct one synthetic observation. The correction should add an author, reason, and time while retaining the prior entry. It must not silently revise the approved order or another visit in the recurring series.
The Clinical Establishments portal publishes haemodialysis guidance. Qualified facility owners decide which current guidance, permissions, and local rules apply. Software should preserve authorised records; it should not prescribe, clear a patient, alter treatment, or declare a session safe.
Backtrace the seven-session difference
Return to the deliberately inconsistent month-end totals. Reconcile from the session register outward rather than editing the smaller number until it matches the larger one.
The seven differences in the fictional dataset might include:
- a missed appointment that never became a session;
- a rescheduled visit whose old slot remained open;
- a completed session awaiting authorised sign-off;
- a signed session whose billing event is still queued;
- a cancelled billing line retained for history;
- an outage record awaiting reconciliation;
- a duplicate downstream import rejected by the destination system.
For each difference, record the source row, present state, expected next state, owner, and ageing. If the product labels all seven as merely pending, ask how staff can distinguish normal work from a broken interface or an unauthorised change.
This approach also tests reporting integrity. A dashboard count should be reproducible from the underlying rows at the same cut-off, with documented treatment of cancellations, missed visits, unsigned records, corrections, and downtime entries.
Reconcile stock and billing last
Once the appointment-to-session relationship is explainable, follow two downstream paths. Trace a fictional consumable event to the authorised inventory record. Separately trace a billable event to the billing queue. Correct the source event and observe whether each destination updates, reverses, or creates an exception.
The hospital inventory guide covers stock controls. The hospital billing guide covers the finance lane. A dialysis workflow can pass scoped evidence to those systems without becoming their authoritative ledger.
Do not let finance data repair clinical history. A charge does not prove that a session occurred, while a clinical entry does not by itself authorise a price. The reconciliation pack should show the source and status of both.
Repeat the dossier after downtime
Take the test environment offline before one fictional visit. Use the centre’s approved downtime form, record only the approved minimum, and restore the software. Enter or reconcile the event with its original event time, entry time, author, and source. Then attempt to import it twice.
Regenerate the five month-end lists. The dossier should now explain the recovered visit, retain the downtime source, reject or flag the duplicate, and leave no unexplained change to the recurring pattern.
Patient Square does not publish a dedicated dialysis module and does not claim live ABDM connectivity, scheme integration, or dialysis-device integration. Do not infer appointment scheduling, machine links, treatment records, consumables, or billing functions from Hospital Copilot. Require the quoted module, deployment, interfaces, support, data flow, exclusions, and acceptance evidence in writing.