Skip to content
  • CBT
  • treatment planning

CBT homework: how to design it, review it in session and record whether it was done (and whether it helped)

Gesell Team13 min read

A between-session assignment should be designed with the client, reviewed at the start of the next session, and recorded as more than a yes or a no. The first two are what the manual for Young and Beck’s competence rating scale asks of the therapist; the third follows from the compliance evidence, which holds wherever you practise. In the meta-analysis by Kazantzis and colleagues (2016), how much of the homework clients did, its quantity, was associated with outcome at the end of treatment at an effect size of g = 0.79, across 15 studies and 1,537 clients. How well they did it, its quality, meaning whether the skill was actually acquired, was associated at g = 0.78, but in only 3 studies and 417 clients, with a confidence interval running from 0.03 to 1.53. Quantity is well established; quality is promising and thin. Record both.

Quantity is established; quality is promising

Kazantzis’s group pooled 17 CBT studies (2,312 clients) published after calls to research homework quality. At follow-up the pattern shifts: quantity, g = 0.51 (7 studies, 1,291 clients); quality, g = 1.07, with an interval from 0.06 to 2.08 and, again, 3 studies with 417 clients. An interval that spans “almost nothing” to “very large” cannot tell you how much quality weighs; it does tell you that, by a narrow margin, it is distinguishable from zero. The authors write in the conditional: if confirmed, not measuring both will have been “an important omission”.

Mausbach and colleagues (2010) point the same way by another route. Across 23 studies and 2,183 participants, the relationship between homework compliance and outcome was r = .26 (95% CI .19 to .33), robust across target symptoms. But it depended on how compliance was measured: studies using Likert ratings found r = .31; studies counting the percentage of assignments completed, r = .17. The authors call the finding “difficult to explain” and offer a reading: a graded rating “might inadvertently reflect quality and quantity”, while a percentage reflects quantity only. A client who practised once, well and usefully, and another who practised daily without understanding the point, tie on a “done” checkbox.

Two limits. These are associations: that clients who do more homework improve more does not prove the homework caused the improvement. And who rates, and when, matters. Mausbach found higher correlations when compliance was rated retrospectively (r = .36, in only 3 studies) than session by session (r = .19), and offers bias as a possible explanation: clients who did well may be remembered as more compliant. The practical rule follows: record compliance at each session, not from memory at the end of treatment.

What national guidelines say, and what they leave out

Mexico is the worked example. The Mexican Social Security Institute’s clinical practice guideline for adult anxiety disorders, IMSS-392-10, includes homework in its definition of CBT. On page 25:

«La Terapia Cognitivo-Conductual (TCC) se caracteriza por ser un método activo y directivo, en ella paciente y terapeuta (profesional experto) trabajan en forma conjunta y estructurada, con tareas fuera de sesión.»

(CBT is characterised as an active, directive method in which client and therapist work together in a structured way, with tasks outside the session.)

That is the only time the word appears in that sense in its 60 pages. The guideline does not say how to design the task, when to review it or what to record. Its monitoring and follow-up section (4.5) sets a schedule only for medication: assessment in the first 2 weeks and again at 4, 6 and 12 weeks. This is not an error: the guideline dates from September 2010 (updated August 2011); it names the component and stops there. Check what your own national guidance says. England’s NICE depression guideline (NG222) is a useful contrast: its treatment tables frame homework as something the client must be willing to do (“The person will need to be willing to complete homework assignments”, in the rows for CBT and behavioural activation) and say nothing about how it is reviewed. What both leave out is set down in the manual for a scale that rates cognitive therapists’ competence.

Designing it with the client: four pieces

Young and Beck’s Cognitive Therapy Scale (CTRS) gives item 11 to homework. Its rating manual asks for assignments that are “custom-tailored”, that ideally follow from the problems discussed in the session, that are “clear and very specific”, and that are written “in duplicate (one copy for the patient and one copy for the therapist)”. Design has four pieces; the first two come from the manual:

  1. The rationale. The manual says the therapist “must stress the importance of homework” and suggests doing so by explaining the benefits of each assignment in detail; its collaboration item gives the reason: when the patient sees the link between an assignment and the solution to their problem, “it is more likely that the patient will participate conscientiously.”
  2. The anticipated obstacle. The manual suggests having the client visualize carrying out the assignment “to identify any obstacles that might arise”, and asking whether it sounds useful, seems manageable and is clear.
  3. When and where. Not from the CTRS but from research on if-then plans (implementation intentions): in Toli, Webb and Hardy’s meta-analysis (2016), with clinical and analogue samples, forming them had a large effect on goal attainment (d = 0.99; 28 studies, 1,636 participants, one outlying effect excluded). These were goals in general, not CBT homework; it still supports fixing a time and a place.
  4. The plan goal it serves. Nor does this one come from the CTRS, which speaks of the session’s problems. A task that does not hang from a goal in the treatment plan cannot be evaluated: you do not know what would count as success.

As therapy progresses, the manual adds, the client should play an increasing role in suggesting and designing assignments. The difference, in a fictional case:

Piece Undesigned task Designed task
What “Do thought records” “Note two sharp drops in mood: situation, emotion from 0 to 100, and the exact sentence you thought”
Rationale — “To see which thought arrives just before; we work with it on Thursday”
When and where “This week” “If my mood drops at the office, I type the sentence into my phone; that night after dinner I copy it to the sheet”
Obstacle — “‘I won’t know what to write’: the sentence word for word is enough, no analysis”
Goal — “Goal 2: identify the thoughts linked to drops in mood”
Quality criterion — “The thought is written as a sentence (‘they’re going to fire me’), not as an emotion (‘sad’)”

The last row lets you review quality without improvising.

Reviewing it at the top of the next session

CTRS item 1, the agenda, places the review. The agenda usually opens with a brief summary of the week that includes “discussion and feedback regarding homework”, and the manual notes that, done properly, “the agenda can usually be set within five minutes.” Item 11 explains why the review cannot be skipped:

“Unless the therapist routinely reviews homework assigned from the previous week, the patient may come to believe that there is no need to complete the assignments carefully. Near the beginning of each session, the therapist and patient should discuss each assignment, and the therapist should summarize conclusions derived or progress made.”

Reviewing is not asking “did you do it?”. Four questions, in this order, cover quantity, quality, learning and obstacles:

“How much of what we agreed did you get to? Show me one of the times: how did you do it, step by step? What did you notice that you didn’t know last week? What got in the way?”

The second question is the quality question. For a thought record: whether the client separated situation from emotion and wrote the thought as a sentence, or filled the sheet from memory the night before the appointment. If the skill went wrong, practise it there and then, with one of their own examples, before agreeing the next assignment.

What to record: four data points, not a checkbox

We did not locate a Spanish-language source that teaches recording completion alone, so what follows corrects no one; it comes from the evidence above. If quality carries weight of its own and a completion percentage erases it, the record for each reviewed assignment needs four data points:

  1. Quantity. Done, partly done or not done, and how much of what was agreed: “2 of 3 records”.
  2. Quality. Whether the skill was carried out as taught, with the evidence: “separates situation and emotion; thoughts appear as sentences”.
  3. Learning. What the client concluded, in their own words and in quotation marks: “not every drop comes from work”.
  4. Decision. Keep, adjust or drop, and why.

Write it in the same session where you review it, not from memory later.

When it comes back undone

An undone assignment is clinical data, not a failing. The tools are the design tools, used in reverse:

  1. First, the design. Was it clear, manageable and useful to this client? If not, the assignment failed, not the client.
  2. Then, the obstacle. Practical (no time, lost sheet) or cognitive (“what’s the point, it won’t work”, “I’ll do it wrong”). The second kind is session material: an automatic thought about therapy itself.
  3. Then, the adjustment. Smaller, at another time or place, with the first step done in session.

“That’s fine; I’m more interested in what stopped you than in the sheet. Let’s go back to the moment you were going to do it: where were you, what happened, what did you tell yourself? […] What smaller version would you actually do this week? Let’s do the first one here, now.”

What to avoid: skipping the review, for the reason the manual gives; scolding; or reassigning the same task unchanged, which does nothing about what stopped it. In the note, write what happened, not an adjective about the client: “not done; obstacle: no time alone; reduced to one record, on arriving home” can be reviewed; “poorly engaged” cannot. And if the assignments for one goal keep coming back undone, the question is no longer about the assignment but about the goal, or about the client’s state: if it coincides with worsening scores or with hopelessness that reaches beyond the assignment, reassess symptoms and risk before redesigning it.

Where it goes in the note and the plan

The agreed assignment goes in the plan section of the note for the session where it is set; its review goes in the next session’s note. The worked example in the BIRP notes guide does it in one line, under interventions: “Reviewed the week’s episode log with the client (previous homework).” The client’s response to that review (quantity, quality and learning) goes under R; the decision to keep, adjust or drop the assignment goes under P, with the new assignment.

Mexico’s record standard, NOM-004-SSA3-2012, never mentions homework by name, so locate it in the standard’s own language: it asks every progress note for «tratamiento e indicaciones médicas» (treatment and medical instructions, numeral 6.2.6), read through the adaptation numeral 5.16 makes for psychological care; what NOM-004 requires of a psychologist works through that adaptation clause by clause.

The plan holds not the assignments but the goals they serve; what rises to the plan is the pattern. At each review, what has accumulated per goal tells you whether the goal is moving, whether the assignment is the wrong one, or whether the goal was too big.

The worksheet that arrives as a phone photo

The client messages you a photo of the completed sheet. Does it go in the record? That depends on your jurisdiction’s record rules, and this is documentation practice, not legal advice. Numeral 4.4 of Mexico’s NOM-004 shows the reasoning; it defines the clinical record this way:

«Expediente clínico, al conjunto único de información y datos personales de un paciente, […] el cual, consta de documentos escritos, gráficos, imagenológicos, electrónicos, […] y de cualquier otra índole, en los cuales, el personal de salud deberá hacer los registros, anotaciones, en su caso, constancias y certificaciones correspondientes a su intervención en la atención médica del paciente […].»

(The clinical record is the single set of a patient’s information and personal data, made up of written, graphic, imaging, electronic or any other kind of documents, in which health personnel must make the entries, annotations and certifications corresponding to their intervention in the patient’s medical care.)

Format is no barrier: 4.4 admits «electrónicos» documents and those «de cualquier otra índole» (of any other kind), and a photo fits. But the one who records, in that definition, is «el personal de salud» (health personnel), and what they record is their own intervention. What the record needs is your note of the review. Filing the client’s sheet is optional: numeral 5.18 allows adding documents «que se consideren necesarios para complementar la información» (considered necessary to complement the information). Once filed, it is part of the record and the retention period of numeral 5.4 applies. What you want to avoid is the only copy living in a chat thread; see WhatsApp is not a clinical record.

Why there is no list of 50 assignments here

The CTRS manual itself carries a list: twelve “typical homework assignments”, from the Daily Record of Dysfunctional Thoughts to trying out new behaviours. But it sets the list after asking that the assignment be tailored and, ideally, follow from the session’s work. Picked from a catalogue, with no conceptualisation behind it, an assignment fails item 11’s first requirement.

Each technique’s assignments, with their design, are in their own guides: building a graded exposure hierarchy and the automatic thought record.

If you use Gesell, note one limit: your client cannot upload the completed sheet. Marking the assignment partly done or not done, with your review, is yours to do.

References

Share this article

Keep homework from getting lost between sessions

Assign the worksheet as homework in Gesell: your client sees it in their portal, marks it done, and leaves you a note before the next session.

Try gesell.ai

Have a question about your practice?

How to document a case, which format suits you, or how Gesell fits your practice: write to us and we will answer.

Write to us

About the author

Gesell Team

Clinical and product content written by the gesell.ai team together with certified clinical psychologists.

← Back to the blog