A public-service worker receives a question about urgent support. Before answering, she checks whether another unit already holds the documents that could help. The system offers a button labelled “refer case.” It looks like ordinary coordination, yet one click may send a complete history, interview notes, a private phone number, and a hurried assessment of family circumstances. The question is not whether the receiving unit might find all of that material interesting. The question is what it actually needs to perform the specific task it has been given.
Referrals between services are often described as an obvious improvement. Sometimes they are: they prevent someone from repeating a painful account, shorten waits, and put an issue in front of the people able to solve it. That usefulness does not authorize an unlimited transfer of information. Every referral creates a new relationship, new people with access, and new opportunities for meaning to be assigned to a record. Defining minimum data is not a way to obstruct work. It prevents administrative convenience from turning every case file into an open suitcase.
Start with the decision, not the available form fields#
The familiar mistake is to build a transfer from the fields already visible in an application. When a screen contains a name, address, diagnosis, observations, and attachments, selecting everything feels efficient. A responsible sequence goes the other way. First, write the decision the receiving service must make. Next, identify the information strictly required for that decision. Then ask whether any item can be summarized, confirmed without being copied, or requested directly from the person.
The familiar mistake is to build a transfer from the fields already visible in an application.
Consider a referral to an accessibility team that must arrange an appointment. The file number, a secure contact method, the proposed date, and the person’s stated access need may be enough. There is no reason to transfer every earlier inquiry or a note about an argument with another institution. Once the purpose is phrased in testable terms—“arrange an accessible appointment before Friday”—excess becomes easier to see. Fields stop looking neutral. Each one has to make a case for travelling.
Minimum does not mean identical in every case#
A minimum set cannot be a universal list written once and forgotten. It changes with the action, the risk, and the moment. To book a call, a time window and contact preference may be sufficient. To refer a safety concern, the receiver may need a more precise context, a response channel, and the name of the person who assessed urgency. Treating those cases as interchangeable causes two kinds of harm: necessary action is delayed, or the broad circulation of sensitive material becomes normal.
Short referral cards can make that distinction operational. For each referral type, the card should name the purpose, sending and receiving services, permitted fields, fields explicitly excluded, access duration, and review route. This is not decorative paperwork. It means a colleague can ask why an old address, a photograph, or a subjective observation is included in a case and receive an answer that does not depend on one person remembering an informal rule.
Keep facts, context, and judgments apart#
A referral becomes safer when it separates three layers that are regularly mixed together. The first is operational fact: an appointment is pending, a document arrived, a deadline is approaching. The second is context provided by the person: a support need, preference, or boundary that helps the next service treat them properly. The third is interpretation: “uncooperative,” “seems confused,” or “hard to reach.” Those labels can influence the next encounter even when they are stale, partial, or rooted in a misunderstanding.
This is not an argument for removing all context. A person may explicitly ask staff to communicate a language barrier or a safe way to contact them. The difference is that the data has a clear connection to the task, is expressed precisely, and can be corrected. A vague observation is easy to repeat and rarely carries its source, date, or uncertainty with it. Before sending one, a team should be able to answer: who said this, when, for which decision is it needed, and could the person describe the situation differently?
Design a screen that helps staff say no#
Technology can support minimization or quietly defeat it. A screen with every box selected by default encourages over-sharing, especially under time pressure. A better design displays the referral purpose first, suggests a minimum package, and asks for a short reason before extra categories can be added. The aim is not to make routine work exhausting. It is to place a proportionate pause at the point where a hard-to-reverse disclosure could otherwise become habit.
The screen should also tell staff what is not moving. A plain summary—“the case identifier, requested appointment, and interpreter preference will be sent; interview notes and attachments will not”—prevents more mistakes than a lengthy policy hidden in another tool. When an additional item is genuinely essential, the reason should be recorded in ordinary language. A later review can then distinguish a justified exception from an inherited routine.
Explain the referral to the person living with it#
Minimization is stronger when the person knows a referral will occur. Notice does not require a legal handbook. It requires an explanation of which service will receive what, for what purpose, for how long, and how to request a correction. In the opening example, the worker might say: “I will send your case number and your need for an accessible appointment to the team that schedules them. I will not include notes from our conversation. If email is safer than a call, we can record that now.”
That exchange protects dignity and improves accuracy. It gives someone a chance to correct an old telephone number, limit an unsafe channel, or add a practical detail absent from the record. It can also reveal that a referral is unsuitable. Consent alone does not answer every duty of protection, but a concrete explanation reduces the imbalance. People should not discover later that parts of their lives were circulating through systems they never knew existed.
Make access temporary and traceable#
Sending less is not enough if the receiver keeps that smaller set forever. Access should expire when the task ends or require explicit renewal. A resolved appointment does not justify another service retaining years of preferences or document copies. Expiry turns purpose from a slogan into a practice: it requires a decision about whether there is still a specific reason to keep or consult the information.
Useful logs matter as well. The record should not be an unreadable technical trail. It should state who referred the case, which categories were selected, which unit received them, for what purpose, and under which exception if any. That trace protects both the person and staff. It allows an error to be reconstructed without invented stories, reveals oversized packages, and identifies which forms are encouraging needless transfers.
Test the uncomfortable situations#
Before generalizing the workflow, the team should rehearse situations that expose its limits. What happens when someone shares a phone with a person who must not learn about the appointment? What if the identifier reaches the wrong recipient? How can a note be corrected after it has been viewed? What option does a worker have when urgent assistance is needed but no category fits? Answers do not emerge by reviewing only an ideal flowchart. They emerge by listening to staff handling real cases and to people carrying the consequences of a poor referral.
A small trial can compare two paths: the current form, which attaches the whole record, and a minimum package for a single receiving service. Measure time, mistakes, requests for additional material, and corrections made by affected people. If the minimum package repeatedly prevents a necessary action, revise it with evidence. If it does not, remove a transfer that previously looked inevitable. Coordination is not measured by the amount sent; it is measured by the ability to solve a problem without expanding risk.
A practical rule for the next referral#
Before pressing send, a team can ask four questions: what decision must the receiver make, which item is indispensable, which item may be irrelevant or prejudicial, and when will access stop being needed? If there is no clear answer, the process needs a pause rather than more fields. If an urgency or safety exception exists, it should be described and reviewed afterward, not hidden inside custom.
Minimum data is not a signal of distrust between services. It is a more mature form of collaboration: each unit receives what it needs to act, accepts its part of responsibility, and does not turn a person into a portable collection of background details. Start with one frequent referral, define its smallest workable package, and ask someone outside the team to test it. That is a modest next step, but it changes the governing question from “what do we have?” to “what must we share to help without intruding?”
