Mandatory reporting map

A proof-of-concept view of NHS mandatory reporting requirements.

This page shows the kind of clear overview a Business & Performance Business Partner would want: what is due, who owns it, where the risks are, and whether each return is properly assured.

It also shows how agentic AI could help identify, organise and maintain reporting requirements from public sources.

AI can help with the admin, structure and checking. It cannot replace the named owner, the information lead, or the formal sign-off process.

The same approach could also support other complex projects. An agent could help bring together information from project plans, documents, guidance, meeting notes and approved systems such as Azure DevOps or Teamwork. It could then suggest updates, flag gaps and keep actions aligned.

In a live setting, this would need clear access controls and human approval. The agent would not make changes on its own. It would help organise the work so people can make better decisions.

Demonstration caveat: This is a personal demonstration site. It is not an official Dorset HealthCare register. It is not complete and has not been operationally validated. It does not contain confidential information, unpublished internal documents or patient-identifiable information. Owners, due dates, assurance statuses and local source labels are illustrative sample data for demonstration only. Public reference links point to published NHS sources where available. In a live setting, every return would need to be checked against local standard operating procedures, agreed owners, source systems and formal governance routes.

What this register maps

This demonstration register shows:

return name
owner or team
reporting frequency
demo due date
assurance status
risk level
key issues or notes
public reference link

More detail could sit behind the table, including the recipient, mandate, source system, confidence level, local process notes and escalation route. Additional fields are held in data/mandatory_returns_register.csv.

In a live version, this could support routine assurance meetings, handovers, risk reviews and planning conversations.

How agentic AI could help

An AI agent could support this work by helping to:

search public specifications
extract reporting requirements
organise returns by owner, frequency, risk and due date
flag missing links, gaps or unclear ownership
compare local notes against national guidance
maintain a change log when specifications are updated
prepare first-draft assurance notes for human review
answer definition questions using approved sources
suggest updates to project plans or action logs for approval

This would reduce admin-heavy work and make the reporting position easier to manage.

However, the agent would not approve anything. Human owners would still check the definitions, validate the figures, confirm the local process and sign off the return.

How this page was created using agentic AI

This page is a worked example of a human-directed AI workflow.

A Cursor agent was asked to identify common NHS mandatory and statutory returns relevant to a mental health and community trust. It used public NHS sources and kept only returns with a reliable public reference link.

The agent then helped organise the information into a register format that a Business & Performance Business Partner could use: return name, owner, frequency, due date, assurance status, risk and issues.

The local fields are demonstration data. Public sources do not provide local owners, local assurance status, internal source systems or local sign-off routes.

In a real Trust setting, those fields would need to be confirmed by named owners against local processes, validated systems and agreed governance routes.

This is the key point: AI can help draft and maintain the structure, but people must own the local truth.

How to read the table

The table is not intended to be a full official register. It is designed to show the kind of questions a Business & Performance Business Partner should be able to answer quickly:

What returns are due soon?
Which returns carry the highest risk?
Which returns need validation?
Which teams own each return?
Which returns rely on manual work?
Which public guidance supports the return?
Where are the known gaps or caveats?

A view like this helps turn a long list of reporting duties into a manageable assurance tool.

It also shows a wider point: agentic AI is most useful when it helps turn scattered information into clear, governed, actionable work.

Filter and search

Mandatory returns register

Return name	Owner / team	Frequency	Next due	Assurance	Risk	Issues / notes	Reference
MHSDS	Performance & BI	Monthly	2026-04-15	In production	Medium	Validate against MHSDS guidance; coding QA pending	Public reference
CSDS	Performance & BI	Monthly	2026-04-15	In production	Medium	Community metrics require coding QA each month	Public reference
NHS Talking Therapies	MH Performance Team	Monthly	2026-04-10	In production	Low	Recovery target tracking in place	Public reference
ECDS / urgent care	Urgent Care BI	Monthly	2026-04-20	Needs validation	High	ECDS scope needs local confirmation	Public reference
KH03 bed occupancy	Bed Management / BI	Quarterly	2026-04-30	In production	Low	KH03 reconciled with bed state each quarter	Public reference
RTT	Performance & BI	Monthly	2026-04-15	Needs validation	High	RTT incomplete for some pathways	Public reference
DM01 audiology/diagnostics	Diagnostics PMO	Monthly	2026-04-12	Manual process	Medium	Audiology DM01 partially manual	Public reference
CYP eating disorder waits	CYP Performance	Monthly	2026-04-08	In production	High	National ED access standard — high scrutiny	Public reference
KO41a complaints	Complaints Team	Quarterly	2026-04-30	In production	Low	KO41a aligned with complaints policy	Public reference
FFT	Patient Experience	Monthly	2026-04-05	In production	Low	FFT response rates monitored	Public reference
ERIC	Estates & Facilities	Annual	2026-05-31	Under review	Low	ERIC annual cycle — estates data lag possible	Public reference
DSPT	IG / IT Security	Annual	2026-05-31	In production	Medium	DSPT annual assessment	Public reference
CQC notifications	Quality & Governance	Ad hoc	Ad hoc	Manual process	High	CQC notifications require clinical sign-off	Public reference
WRES / WDES / EDS	Workforce / OD	Annual	2026-05-31	In production	Low	Workforce data from the Electronic Staff Record (ESR) — extract delay possible	Public reference

Plain-English glossary of the return codes in this table

MHSDS — Mental Health Services Data Set: The national record of mental health care activity submitted each month.
CSDS — Community Services Data Set: The equivalent national record for community health services for children, young people and adults.
NHS Talking Therapies (formerly IAPT): Monthly data on talking-therapy services for anxiety and depression, including recovery rates.
ECDS — Emergency Care Data Set: The dataset behind urgent and emergency care (A&E) activity and waiting times.
KH03 — bed availability and occupancy: A return on how many beds are available and how full they are, submitted quarterly.
RTT — Referral to Treatment: Waiting-time data measuring how long patients wait from referral to starting treatment.
DM01 — diagnostic waiting times: Monthly data on waits for diagnostic tests such as audiology and imaging.
CYP eating disorder waits: Children and young people's (CYP) waiting times against the national eating-disorder access standard.
KO41a — written complaints: A return summarising the volume and handling of formal written complaints.
FFT — Friends and Family Test: Patient feedback on whether people would recommend a service.
ERIC — Estates Returns Information Collection: An annual return on buildings, estates and facilities.
DSPT — Data Security and Protection Toolkit: An annual assessment showing the organisation meets data-security standards.
CQC — Care Quality Commission notifications: Alerts the NHS regulator must be told about certain events, often needing clinical sign-off.
WRES / WDES / EDS — workforce equality standards: Annual returns on race and disability equality (WRES, WDES) and the wider Equality Delivery System (EDS).