Turn a health transformation priority into a useful local evidence question

1. Begin with the decision a local team can actually make

The Saudi Ministry of Health’s February 2026 announcement of the 2025 Health Sector Annual Report describes prevention, quality of life and more efficient, people-centred services in the context of the Health Sector Transformation Program. That provides a reason to investigate a service problem. It does not establish the performance of a particular clinic, validate a local measure or show that a specific change caused an improvement.

For a useful first brief, name a decision and its owner: for example, whether a clinic manager needs a closer review of the referral-handling process. Describe the question before choosing a chart. A broad aim such as “demonstrate transformation success” invites selective evidence. A narrower proposed question is how much calendar time separates receipt of an eligible referral from the first completed clinical assessment at one clinic.

Reference: Saudi MOH: release of the 2025 health sector report, 15 February 2026

2. Define both ends of the proposed interval

In this teaching exercise, the proposed scope is adult, non-urgent referrals received by one fictional clinic during a specified month. The unit is a referral record. One person might have more than one referral, so the record count is not automatically a patient count. Before extraction, settle how resubmissions, transfers, cancellations and records with no assessment by the end of follow-up would be represented.

Receipt by the clinic is different from creation in a referring system. A booked appointment is different from an assessment that took place. Put the operational meaning, source field, time zone and follow-up end date beside each timestamp. If the available fields capture different events, revise the question or seek a suitable source through the institution. Renaming a column does not make the intended event observable.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

3. Draft the smallest data request that can answer the question

Prepare a field-level request as a draft, with the purpose of each item. The proposed exercise needs a controlled referral key, the relevant timestamps and the fields needed to apply the agreed eligibility rules. It does not need a name or telephone number to demonstrate interval arithmetic. Any actual data access, linkage, disclosure and retention must follow the institution’s applicable decisions; the small size of a request does not itself provide permission.

Ask the responsible data owner to explain how the fields were produced, which changes affected them and whether the extract includes all records in the proposed scope. Specify who may inspect an identifiable source if a discrepancy needs investigation. Keep that source separate from a permitted analysis file. A draft request and a list of questions are the products of this planning stage; neither is evidence that a request was sent or access was granted.

4. Work through a synthetic extract before interpreting it

The following numbers are entirely fictional. An exercise file contains 84 referral records. Of these, 70 have both timestamps in a valid order, nine lack the first-assessment timestamp and five place the recorded assessment before receipt. The categories are mutually exclusive in this example and sum to 84. No real patient information was obtained, and duplicate-record checks have not been supplied.

The count with a calculable interval is 70, or approximately 83.3% of the 84 records. This is a statement about the supplied fields, not a completion rate, a target-achievement rate or the percentage of patients who received good care. The 14 remaining records require a defined handling rule and, where permitted, source investigation. Do not replace a missing finish with zero days or silently remove a reversed timestamp.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

5. Separate an incomplete measure from a service conclusion

No interval values or acceptable threshold are supplied in this example. It therefore supports neither a median waiting time nor a proportion meeting a target. Even after resolving the definitions, report the observation window and how unfinished referrals enter the account. A summary restricted to completed assessments may describe a different group from all referrals received. The choice should follow the question, with its consequences visible.

A later comparison before and after a service change would need its own design. Referral urgency, routing, staffing, holidays and data capture might also change. Write these as questions for the proposed analysis, not explanations already established by the synthetic audit. A local change in a descriptive number would not, by itself, establish a causal effect of a national programme. Seek methodological review appropriate to the decision before choosing an analysis.

6. End with an actionable brief that names the unknowns

A practical brief can fit on one page: the local decision, the proposed population and interval, the permission status, the field audit, the unresolved questions and the next responsible person. Attach the dictionary version so that a future reader can tell what was measured. Keep national context in the introduction and local evidence in its own results area. This makes it easier to update the brief when the next piece of evidence arrives.

For this exercise, the next proposed action is to ask the designated owner to clarify the timestamps and the permitted investigation process. No clinical workflow change is justified by the counts alone. If a future service intervention is proposed, define in advance which outcome would inform the decision, who will review possible unintended effects and what evidence would cause the team to reconsider. These remain planning questions, not completed approvals or findings.

Questions for a practical final check

• The local decision is distinct from the national policy context.
• Start, finish, unit and follow-up are defined.
• The data request and permission status are explicit.
• Missing and reversed timestamps retain their own states.
• Calculable records are not presented as a clinical success rate.
• The next action has an owner and does not overstate completed work.