Couples and family therapy documentation: one chart or several?

You have been seeing a couple for eight months. They separate. One of them writes: she wants “her record” for the divorce. What you can answer was decided months earlier, when you opened the file. One chart or several? This guide separates what is required from what is defensible. It is documentation guidance, not legal advice; record-keeping rules differ by country and state, so verify the ones that bind you.

Clinical-record rules are written for one patient at a time

Record standards inherit their architecture from medicine, where who the patient is was never in doubt. Mexico’s clinical-record standard, NOM-004 is a clean specimen: it binds private consulting rooms (numeral 2), its numeral 4.4 defines the clinical record as “the single set of information and personal data of one patient,” and its 4.7 defines a patient as “the direct beneficiary of medical care.” Individual, top to bottom — read your own jurisdiction’s rule with that in mind.

What such a rule does not say weighs as much. Nothing in it contemplates treating more than one person at once, and national professional codes name couple and family work without saying how it is filed — Mexico’s FENAPSIME code, for instance, defines intervention at the “individual, couple, family, group” level yet speaks only of a “psychological record of the user” (article 15), singular. You are given an architecture, not a rule. Anyone who says the law prescribes a filing model for couples is selling a certainty that does not exist.

Who is the client: a decision you record, not a fact you discover

The most precise guidance in print is the APA Ethics Code. Standard 10.02(a) asks that, when you agree to provide services to several people who have a relationship — “such as spouses, significant others, or parents and children” — you take reasonable steps to clarify at the outset “which of the individuals are clients/patients” and “the relationship the psychologist will have with each person.” Standard 10.03 asks less of group work: describe the roles, responsibilities and limits of confidentiality at the outset, without naming the client.

Note the verb: clarify. It is a decision you make, discuss and write down, and it settles whose objectives appear in the treatment plan with verifiable goals, who can authorize a release, and which file each note lives in.

“Frame agreed in session 1: the contracted service is couple work; each person owns their own information. One chart is opened per person and conjoint session notes are filed in both. Consent signed by each.”

That conversation belongs to informed consent, and that is where the question is first raised and left open.

Three filing models, and what each one costs

None is mandated: the record rule is written around a chart per patient, but says nothing about whether a conjoint note may also live in a shared process folder.

Model	How it is filed	What it costs
One chart per person	The conjoint note goes in both	Duplicate entry; forces separable writing
Per person + shared folder	Relationship material kept apart	Defining what lives where
One shared chart	A single file for the client unit	Every future release is a separation

The APA Record Keeping Guidelines — adopted in 2007, expired by their own terms in 2017, never replaced, and still the most recent thing the APA has published — are the only source that puts the question directly. Guideline 11, Multiple Client Records, reads: “The psychologist carefully considers documentation procedures when conducting couple, family, or group therapy […].” It allows both exits — “if a couple or family is the identified client, then one might keep a single record” — and draws the consequence: on later requests “it will be necessary to release only the portions relevant to the party covered by the release,” so “the psychologist may choose to keep separate records on each participant from the outset.”

The per-person model is the most congruent with how record rules are drafted and the only one that survives a partial access request cleanly. Public documentation manuals read the same way: San Mateo County’s treats family therapy as an activity “directed at improving the client’s functioning and at which the client is present” — the client, not the family, is the unit. The shared chart is defensible when the couple genuinely is the identified client and you accept that cost; opening one with no written release plan is a standing entry on the list of clinical documentation mistakes.

Secrets: the policy you pick decides the chart’s structure

The AAMFT Code of Ethics, whose revision took effect on 1 January 2026, has the cleanest formulation in print. Standard 2.2 bars disclosing information outside the treatment context without written authorization from each individual competent to give it and — the part that matters here — provides that the therapist “must not reveal any individual’s confidences to others in the client unit without the prior written permission of that individual.”

The documentation consequence almost nobody plans for: a no-secrets policy and a secrets-held policy do not just produce different conversations, they produce charts with different structures.

No secrets. Whatever arrives by call, message or individual session can enter the conjoint session; it all fits in the conjoint note, and whoever spoke knew that beforehand.
Secrets held. You need a place where that material exists without becoming visible to the other in a release. In the per-person model it already exists: the chart of the person who said it. In the shared chart it does not, and creating it after the disclosure is improvising.

Either is defensible; not having chosen is not. Write the policy into the consent and the chart’s frame section.

Write the conjoint note so it can be separated

Today’s note decides whether tomorrow you can release one person’s part alone. An undifferentiated narrative — “worked the escalation pattern; both acknowledge their share” — cannot be separated without rewriting it; one that attributes each item to whoever contributed it can.

“Conjoint session. A: reports three arguments this week; verbatim: ‘I go quiet so it ends sooner.’ B: reports postponing the conversation about finances. Intervention: interrupting the demand-withdraw cycle. Response: A held the full turn; B asked for a pause after two minutes.”

It is more work today, and the only real prevention of the partial access request two sections down.

Recording the session: permission from every voice

Recording is never the default, and in a conjoint session the requirement multiplies by the people in the room. The ethical layer is APA Standard 4.03: before recording the voices or images of the individuals you serve, you obtain permission from all of them or their legal representatives. A legal layer sits underneath and varies by place: where the law treats health information as sensitive — as Mexico’s data-protection law, published in the DOF on 20 March 2025, does — processing session audio requires express written consent, revocable, from each person whose data it is, not from the couple as a block; how to word that permission belongs to informed consent in psychotherapy.

Two consequences. If one person revokes consent, the revocation may not undo processing already done, but in practice you cannot keep using a file where one voice is inseparable from the other: plan to delete the whole recording, not “their part.” And with several voices, the risk of any draft — yours or one a tool generates for you to review — is cross-attribution: one person’s disclosure recorded under the other’s name. What to capture, and with what safeguards, is in writing better session records.

“I want my record”: when two people are in the file

Back to the opening scene. Professional guidance answers first: AAMFT Standard 2.3 provides that in couple, family or group treatment the therapist gives no access to records without written authorization from each individual competent to give it, and protects other individuals identified in the record.

Data-protection law can reach a comparable result by another route. In Mexico, the right of access covers the person’s own personal data held by the controller (article 22), and article 33 lets the controller refuse, under fraction III, where “the rights of a third party would be injured,” adding the decisive detail that the refusal “may be partial.” You neither hand over the couple’s file nor stonewall: release the portion that is the requester’s own data and document the reasoning. Deadlines and identity verification belong to the privacy notice a Mexican practice publishes.

The hard part, plainly: a law may permit a partial refusal and still say nothing about how to separate a narrative note with two people’s data interwoven in the same sentence. We located no published criterion on the point. Prevention is structural.

Do not conflate two channels: a client’s access request is asked of you, a court order is not. Mexico’s record norm names judicial, prosecutorial and administrative authorities as competent to request records and says nothing about multi-person files, redaction, or grounds to resist. We found no citable authority on what a court can compel from a couple’s chart in a divorce — treat an order as a matter for your own counsel, not a records question you settle alone.

The role problem: therapist today, witness tomorrow

The role conflict, by contrast, is yours from day one. APA Standard 10.02(b) names it with the exact example: if it becomes apparent that you may be called on to perform “potentially conflicting roles (such as family therapist and then witness for one party in divorce proceedings),” you take reasonable steps to clarify and modify, or withdraw from, the role. AAMFT is more categorical — 7.5 bars forensic or expert-witness work for current or former therapy clients unless legally mandated, and 7.6 does the same for custody evaluations. The couple’s therapist is not the expert witness in their divorce — say so in the consent, not the day the subpoena arrives.

When the process splits

No source resolves this, and saying so is more useful than inventing a rule: not the record norms, not data-protection law, not the APA Ethics Code or its record-keeping guidelines, not the AAMFT code. The closest anything comes is Guideline 11 — prevention, not a rule for the day of the break-up.

What you can decide in advance and write into the consent:

The conjoint process closes like any other. Reason, status at closing and recommendations, with the discipline of a documented termination.
The individual treatment that follows is a new service. Its own frame, consent, plan and notes. The conjoint chart does not become the individual chart of whoever stayed.
Continuing with one member is not neutral. We located no guidance prohibiting or authorizing it in general; apply 10.02(b) and leave your reasoning in the file.
Retention does not shorten because the couple separated. The APA guidelines suggest that, “in the absence of a superseding requirement, psychologists may consider retaining full records until 7 years after the last date of service delivery for adults or until 3 years after a minor reaches the age of majority, whichever is later” — note the double hedge. The binding period comes from your own jurisdiction’s law, and each chart counts its own clock. Mexico’s norm, for instance, sets a minimum of five years from the last act of care.

A well-documented conjoint process is recognizable by one thing: when the uncomfortable request arrives, the answer is already in the file. It holds up better when each person’s chart has its own place, which is what gesell.ai maintains for you: intake, clinical history, treatment plan and session notes, with informed consents signed digitally and archived in the chart. You decide the filing model and the secrets policy; every note the platform drafts is one you review and approve.