Practice Transfers

Overview

When a patient moves from one GP practice to another, their full clinical record (consultations, medications, referrals, etc.) transfers to the new practice. The OpenSafely pipeline handles this automatically — you do not need to do anything special when querying the data.

---

How It Works

One Row Per Patient

The pipeline maintains one row per patient globally in the patient model, identified by NHS number. This means:

• A patient who moves from Practice A to Practice B will appear once, not twice.
• All clinical data (consultations, drug records, issue records, etc.) is linked to the patient’s current active registration.
• The pseudonymised nhs_number is stable — it does not change when the patient moves practice, so longitudinal research across practice transfers is straightforward.

How the Current Practice Is Determined

When multiple registrations exist for the same NHS number, the pipeline selects the most recent active registration using this priority:

1. Most recent registration_start_datetime
2. Most recently updated record (model_updated_datetime)
3. Deterministic tie-break on patient_id

Only registrations at qualifying England GP practices with a valid NHS number are considered.

---

What You Will See

Scenario	Result in the patient model
Patient registered at one practice	Single row with that practice’s organisation
Patient transfers to a new practice	Single row updated to show the new practice
Patient’s old practice record is closed	No change — the new practice is already represented

Clinical event models (consultation, medication_drug_record, medication_issue_record, etc.) are filtered to the patient’s current winning registration, providing a unified longitudinal view of the patient’s record.

---

Impact on Analysis

Longitudinal Studies

Because nhs_number is stable across practice transfers, you can track a patient’s full clinical history regardless of how many times they have moved. All historical clinical events are accessible under the same patient_id once the new practice holds the record.

Organisation-Level Analysis

If you are analysing data at the organisation level, be aware that:

• A patient’s clinical history will appear under their current practice, not their historical practice(s).
• Patient counts by organisation reflect the current registration landscape at the point of the monthly data refresh.
• Patients who transferred mid-month will appear under their new practice at the next monthly refresh.

---

Edge Cases

Timing Between Refreshes

The pipeline processes data weekly, but consumer-facing views update monthly (last Sunday of the month). If a patient transfers mid-cycle:

• The transfer will be reflected at the next monthly refresh.
• You will never see duplicate records for the same patient — deduplication is enforced before data reaches the consumer-facing models.

Record Merges

Occasionally, practices perform manual record merges (e.g. when a patient was accidentally registered twice at the same practice). These merges happen upstream in the clinical system and flow through as a single consolidated record. The pipeline trusts the upstream source and does not perform its own clinical record merging.

Soft-Deleted Patients

When a patient’s record is soft-deleted at a practice (e.g. the patient has left and the record is marked inactive):

• If the patient has an active registration elsewhere, that registration becomes the current record.
• If no active registration remains anywhere, the patient is excluded from all models.

Patients Without a Valid NHS Number

Patients missing a valid NHS number cannot be deduplicated across practices. These records are excluded to prevent false matches.

---

Related Documentation

• NHS Number De-duplication — detailed technical explanation of the deduplication logic and eligibility criteria
• Data Freshness & Quality — update cadence and data scope information