Responsibility-focused CBT for OCD

1. Introduction

Version followed: Salkovskis's cognitive-behavioural (responsibility) model and treatment as set out in Salkovskis (1999), "Understanding and treating obsessive-compulsive disorder". I also drew on the Oxford Guide by Bream, Challacombe, Palmer & Salkovskis (2017), but only its appendix list was visible, not the text. Treatment sits inside NICE CG31 stepped care. Gaps are filled from the international expert consensus on specialty CBT for OCD (Sookman et al., 2021), and those points are labelled as secondary.

What it is. An individual, formulation-driven CBT for OCD. It combines Beckian cognitive therapy with exposure and response prevention (ERP). Here ERP is delivered as a set of behavioural experiments that test the patient's beliefs. [source: Salkovskis 1999; NICE CG31]

How change happens. Almost everyone has unwanted intrusive thoughts, images, doubts and urges, and their content overlaps with obsessions. OCD develops when a person reads an intrusion as meaning they are responsible for causing or preventing harm to themselves or others. Responsibility here means believing you have pivotal power over a crucial bad outcome, either in the real world or morally. That reading drives distress. It also narrows attention onto the intrusions and makes the thoughts more accessible. And it drives neutralising: compulsions, avoidance, reassurance-seeking, thought suppression and impossibly strict criteria for when a task is finished. These responses then increase the intrusions, the sense of threat and the sense of responsibility. [source: Salkovskis 1999]

Therapy does not try to get rid of intrusions or argue with their content. It helps the patient build and test a less threatening explanation: the problem is worry about harm, not harm itself. It then helps the patient drop the safety-seeking that keeps that worry going. Salkovskis states the core clinical task as helping sensitive people stop trying so hard to be certain they have caused no harm. [source: Salkovskis 1999]

2. Framework

Format. Individual outpatient sessions, which can be weekly or time-intensive. Sessions are "relatively long", and recordings of them are given to the patient as homework. Home visits are a useful addition. [source: Salkovskis 1999] Within NICE stepped care:

• Mild impairment, or patient preference: low-intensity CBT including ERP, up to 10 therapist hours. This can be guided self-help, telephone or group.
• Mild impairment where low intensity has failed, or moderate impairment: more intensive CBT including ERP (more than 10 therapist hours) or an SSRI.
• Severe impairment: CBT combined with an SSRI.
• Housebound patients: consider treatment at home, or by telephone if home treatment is not possible.

[source: NICE CG31] Time-intensive delivery (daily or near-daily sessions over a few weeks) gave results comparable to weekly CBT in the Salkovskis group's case series. [source: Oldfield, Salkovskis & Taylor 2011, abstract only] The total number of sessions is not given in the primary texts [uncertain]. In practice, plan for more than 10 therapist hours for moderate OCD and review progress at three months. [synthesis]

Phases. Salkovskis (1999) describes these as overlapping tasks, not timed stages. The ordering and the move-on criteria below are my synthesis of his sequence.

Goals. Set short-term goals (reachable within 2–4 sessions), medium-term goals (by the end of therapy) and long-term goals (over several years). "No more intrusions" is ruled out as a goal. The two overarching aims are:

• seeing obsessional thoughts as irrelevant to action and not needing to be controlled;
• changing how intrusions are interpreted.

[source: Salkovskis 1999]

3. Guiding principles

[source: Salkovskis 1999 unless marked]

• Collaboration and feeling understood. The patient should leave each session feeling the therapist has grasped their experience from the inside. Practically, check your understanding aloud before you start any change work.
• Normalise. Present intrusions as universal and as a sign of the patient's values, not of danger. Practically, offer normalising information early and come back to it often.
• Guided discovery, not persuasion. Use questions so the patient reaches conclusions for themselves. Practically, ask more than you tell.
• Weave discussion and experiments together. When a discussion shows that some information is missing, design an experiment to collect it. Practically, finish most belief discussions with "how could we find out?"
• Never dispute the content of obsessions or reassure. Arguing about whether the harm will happen is itself reassurance and feeds the cycle. Practically, put forward an alternative explanation and test it instead.
• "Don't trust me, test it out for yourself." Credibility comes from the patient's own evidence, not the therapist's authority. Practically, every key claim becomes an experiment.
• Do not promise that feared disasters won't happen. A promise is reassurance and cannot be kept. Practically, frame change in terms of the cost of carrying on as now.
• The target is the appraisal and the neutralising, not the number of intrusions. Intrusions often fall as a side effect once they stop seeming important. Practically, don't measure success by thought frequency.
• Watch for cognitive work becoming neutralising. Thought records and pie charts can turn into reassurance rituals. Practically, ask whether a technique is being used to feel certain.
• Hand responsibility to the patient step by step. Practically, exposure tasks go from therapist-modelled, to done by the patient unmodelled, to planned and run by the patient.
• The therapist must know the model thoroughly. Practically, be able to draw and explain the vicious circle without notes.
• The therapist does what they ask the patient to do. Be ready to model, or go beyond, any experiment, including deliberately "anti-OCD" actions. [source: Sookman et al. 2021]
• Involve the family without making them co-therapists of the OCD. Help relatives reduce their part in compulsions, avoidance and reassurance, and do it with sensitivity. [source: NICE CG31; Sookman et al. 2021]
• Relapse prevention is built in. Towards the end, explain how the principles apply to future symptoms. [source: NICE CG31]

4. Intervention inventory

The interventions are grouped by phase. Unless marked otherwise, the purpose of each one comes from Salkovskis (1999). The detailed in-session steps fill in his descriptions and are labelled [synthesis] where they go beyond what he says. All example language is my own (illustrative) unless it appears in quotation marks with a citation.

A. Assessment and engagement

4.1 Recent-episode analysis

• Purpose: Find the specific intrusion and what it meant to the patient, and map the cycle that maintains it. Use this in the first sessions and again for each new symptom area. [source: Salkovskis 1999]
• Steps:
  1. Choose a recent, specific episode, ideally from the last few days.
  2. Slow it down moment by moment. Pin down the trigger and the exact intrusion: its words, image or urge.
  3. Ask what the intrusion meant to them at that moment: what they thought might happen, and what their part in it would be.
  4. Record the emotional reactions and the behavioural reactions: rituals, checking, avoidance, reassurance-seeking, suppression and mental neutralising.
  5. Trace the knock-on effects: mood, more thinking about it, attempts to suppress it, asking others for reassurance, and the longer-term consequences.
  6. Repeat with two or three other episodes to find what they share and where they differ.
  [source: Salkovskis 1999]
• Example language (illustrative): "Take me back to Tuesday at the front door. What was the very first thing that went through your mind? And in that moment, what did it mean to you that you'd had that thought?"
• Watch for: The patient may give only general summaries ("I just check everything"); bring them back to one episode. The patient may skip the appraisal and jump straight from the trigger to the ritual. In that case, see 4.4.

4.2 Contrast-episode questioning

• Purpose: Show that it is the appraisal, not the intrusion, that drives distress. [source: Salkovskis 1999]
• Steps: Ask about times when the same intrusion came up and caused little or no distress. Then compare what the patient made of it on those occasions: mood, setting, who else was present, whether someone else was responsible. [source: Salkovskis 1999] Add these differences to the formulation as evidence for Theory B. [synthesis]
• Example language (illustrative): "Have there been times when the thought about the gas came into your head and it didn't bother you much? What was different about how you took it then?"
• Watch for: Common contrasts are being away from home, someone else being in charge, or being in a good mood. Each one points to appraisal or responsibility as the deciding factor. [synthesis]

4.3 Helping a reluctant patient disclose intrusions

• Purpose: Get past shame and fear about saying intrusions aloud. This is common with sexual, violent or blasphemous content. [source: Salkovskis 1999]
• Steps:
  1. Normalise how hard it is to say these things.
  2. Ask what the patient fears would happen if they said the thought, for example that it would make it more likely, or that the therapist would judge or report them.
  3. Give anonymous examples of similar intrusions that other clients have described.
  [source: Salkovskis 1999] Separately, explain the limits of confidentiality accurately. Make clear that intrusive thoughts the person does not want, and is distressed by, are different from intentions. [synthesis]
• Example language (illustrative): "Many people I see have thoughts they find so shocking they've never told anyone. What do you worry might happen if you told me yours?"
• Watch for: Fear of saying the thought is itself a responsibility appraisal ("saying it could make it happen"). Note it and later test it (see 4.21).

4.4 Traffic-light analogy and the withholding experiment (revealing hidden appraisals)

• Purpose: Bring out appraisals when neutralising has become so automatic that the patient "just does it". [source: Salkovskis 1999]
• Steps:
  1. Ask why they stop at red lights. Most people don't think of danger; it is a habit.
  2. Ask what would go through their mind if the brakes failed at the lights. The danger beliefs are still there, just not in awareness.
  3. Suggest that checking or washing may work the same way.
  4. Set up a small experiment: hold back the ritual once and notice the thoughts that come up.
  For non-drivers, use crossing at a pedestrian crossing instead. [source: Salkovskis 1999]
• Example language (illustrative): "You don't think about crashing every time you stop at a red light, but the reason for stopping is still there. Shall we see what goes through your mind if, just once, you leave the check?"
• Watch for: Appraisals that come up in the withholding moment ("if I don't, it'll be my fault"). Write them down word for word for the formulation.

4.5 Home visit

• Purpose: See the full extent of avoidance and rituals in the patient's own setting, which patients often under-report in clinic. [source: Salkovskis 1999] Later in treatment, home visits also help carry gains over into the home. [source: Sookman et al. 2021]
• Steps (synthesis):
  1. Agree the purpose and the boundaries in advance.
  2. Ask the patient to show you the rooms and routines where OCD operates.
  3. Note hidden avoidance, such as rooms not used or objects not touched.
  4. Note how relatives accommodate the rituals.
  5. Write up the findings for the formulation.
• Example language (illustrative): "When I come round, it would help if you showed me the places OCD has taken over, including anywhere you've stopped using."
• Watch for: Consider the ethics of seeing patients outside the office and record your rationale. [source: Sookman et al. 2021]

4.6 Standardised measures

• Purpose: Plan treatment and track progress. [source: Sookman et al. 2021]
• Steps:
  • At baseline and at reviews, use a severity measure (Y-BOCS), a self-report symptom inventory (OCI) and an insight scale (BABS). [source: Sookman et al. 2021]
  • The Oxford Guide appendices include the OCI and a Responsibility Attitudes/Assumptions measure. [source: Bream et al. 2017, appendix listing only] Use the responsibility measure to pick out assumptions to work on directly. [source: Salkovskis 1999]
  • These are licensed instruments: score them according to their manuals. The items are not reproduced here.
• Watch for: Poor insight on the BABS predicts a poorer outcome. Tell poor insight apart from delusional belief. [source: Sookman et al. 2021]

4.7 Goal setting

• Purpose: Set goals that are concrete and reachable, and that do not depend on intrusions disappearing. [source: Salkovskis 1999]
• Steps: Agree short-term goals (2–4 sessions), medium-term goals (end of therapy) and long-term goals (years). Rule out "never having the thoughts". Phrase goals as things the patient will do (for example, cooking for the family, or leaving the house after one check). [source: Salkovskis 1999; the behavioural wording is synthesis]
• Example language (illustrative): "Since everyone gets odd thoughts, let's aim for what you'd be doing if the thoughts no longer bossed you around."

4.8 Cost-benefit discussion

• Purpose: Build motivation by making the price of OCD and its safety behaviours explicit, without promising that the feared outcome won't happen. [source: Salkovskis 1999]
• Steps:
  1. List everything OCD costs: time, relationships, work, money, mood, and the freedoms given up.
  2. Ask what keeping the safety behaviours costs.
  3. Ask how much they would give to be free of the problem. Salkovskis's question: "How much would you pay to get rid of your obsessional problem?"
  [source: Salkovskis 1999]
• Watch for: The patient may want to know the feared event is impossible before they commit. Decline to guarantee it, and go back to the costs of carrying on as now.

4.9 Theory A versus Theory B

• Purpose: The central engagement and formulation tool. It sets two explanations side by side so they can be tested. [source: Salkovskis 1999; Sookman et al. 2021] The Oxford Guide includes a blank Theory A/B sheet. [source: Bream et al. 2017, appendix listing]
• Steps:
  1. Write Theory A in the patient's own words: the problem is real danger or badness (for example, "I am dangerous" or "I have contaminated things").
  2. Write Theory B next to it: the problem is being very worried about danger or badness. Salkovskis's wording: "Maybe you are not dangerous, but are very worried about being dangerous."
  3. Work out together what each theory implies you should do. Theory A implies washing, checking and neutralising to prevent harm. Theory B implies dealing with the worry, which leads to exposure and response prevention.
  4. Collect evidence from past and present for each theory, and weigh what would follow from acting on each one. [source: Sookman et al. 2021]
  5. Ask how much effort has gone into treating the problem as Theory A, and how far it has helped. Usually the answer is "very little".
  6. Agree to test Theory B with experiments.
  [source: Salkovskis 1999 unless marked]
• Example language: "Don't trust me, test it out for yourself" (Salkovskis 1999). Illustrative: "You've given Theory A years of full effort. Would you be willing to give Theory B a fair test for a few months?"
• Watch for: Ask for belief ratings in each theory and re-rate them over time. The patient may treat Theory B as reassurance ("so you're saying I'm definitely safe"). Correct this: Theory B is a hypothesis to test, not a guarantee. [synthesis]

4.10 Time-limited wholehearted trial

• Purpose: Turn ambivalence into a commitment the patient can manage. [source: Salkovskis 1999]
• Steps:
  1. Ask whether they would "give it a wholehearted try for three months, then review it with me" (Salkovskis 1999).
  2. Ask how they think change should start. Patients often suggest their own first steps; use those.
  [source: Salkovskis 1999]
• Watch for: Set the review date in the calendar at the start. [synthesis]

4.11 Addressing fear of over-correcting

• Purpose: Deal with the fear that dropping OCD will make the patient careless or dangerous. [source: Salkovskis 1999]
• Steps:
  1. Ask which is easier: becoming less obsessional, or more obsessional.
  2. Point out that becoming obsessional again would be easy.
  3. Offer to help them re-obsess if therapy ever had an unacceptable effect.
  [source: Salkovskis 1999]
• Example language (illustrative): "If you ever became too relaxed about the oven, I'm sure we could get the worry back within a week. I'll help you do it if you want."
• Watch for: A smile or a laugh usually means the patient has taken the point. Persistent fear of becoming careless is a responsibility belief to target with the pie chart (4.20) and experiments.

B. Formulation and normalising

4.12 Shared vicious-circle formulation

• Purpose: Give a personal map of how the problem keeps itself going. [source: Salkovskis 1999; Sookman et al. 2021]
• Steps:
  1. Using the patient's own episode, draw: trigger → intrusion → appraisal of responsibility for harm.
  2. From the appraisal, draw arrows to each output: distress (anxiety or low mood), neutralising and rituals, avoidance, reassurance-seeking, suppression, selective attention, and the thought becoming more accessible.
  3. Draw arrows back from each output to more intrusions and a stronger sense of threat and responsibility.
  4. Ask the patient to explain the loop back to you.
  [source: model components in Salkovskis 1999; the drawing sequence is synthesis]
• Example language (illustrative): "Let's see what happens after you check. In the moment it calms things down. What happens to the doubt by the evening?"
• Watch for: The diagram should use the patient's words. Each arrow that goes back to the start is a target for an experiment.

4.13 Normalising: who has which obsessions

• Purpose: Show that intrusions are upsetting because they clash with the person's values. They are not evidence of hidden wishes. [source: Salkovskis 1999]
• Steps:
  1. Explore why different people are troubled by different intrusions. For example, a caring parent is horrified by thoughts of harming the baby; a devout person is horrified by blasphemous thoughts.
  2. Compare intrusions with ordinary worry.
  [source: Salkovskis 1999]
• Example language (illustrative): "Why do you think a thought about harming your child bothers you so much more than it would someone who didn't care about children?"

4.14 Normalising: how common intrusions are

• Purpose: Correct the belief that having the thought is abnormal or meaningful. [source: Salkovskis 1999]
• Steps:
  1. Summarise the research: around 90% or more of people report unwanted intrusions with content similar to obsessions.
  2. Invite a thought experiment: what would it be like to have to plan every thought in advance?
  [source: Salkovskis 1999] A survey variant, in which the patient asks trusted others about their own intrusions, is widely used in this tradition. It is not in the sources. [uncertain] If you use it, agree the questions beforehand, and make sure it is gathering evidence and not seeking reassurance. [synthesis]
• Watch for: "Yes, but mine are different or more frequent." Move back to how the thoughts are appraised, not how often they come.

4.15 Normalising: what intrusions are for

• Purpose: Present intrusions as a by-product of a creative, problem-solving mind. [source: Salkovskis 1999]
• Steps:
  1. Discuss how unbidden thoughts help problem-solving and creativity.
  2. Ask when a violent or alarming thought might even be useful (for example, noticing a hazard).
  [source: Salkovskis 1999]
• Example language (illustrative): "If your mind couldn't throw up ideas you hadn't chosen, could you ever have a new idea?"

C. Belief change: neutralising and responsibility

4.16 Metaphors for counterproductive strategies

• Purpose: Make it intuitive that neutralising helps briefly and makes things worse over time. [source: Salkovskis 1999]
• Steps: Offer one metaphor that fits the patient, then ask how it applies to their own rituals. [source: Salkovskis 1999]
  • Digging deeper to get out of a hole.
  • Fighting a fire with petrol.
  • The bully. Each time you give in, the demands get bigger. The way out is to challenge the demands, not to look for safety from them.
• Example language (illustrative): "Every time you give OCD what it asks for, does it ask for less next time, or more?"
• Watch for: The patient starts using the metaphor themselves between sessions. That is a sign the idea has landed. [synthesis]

4.17 Explaining spontaneous decay

• Purpose: Teach that urges and discomfort fade by themselves when the person does not act on them. This is the basis for response prevention. [source: Salkovskis 1999]
• Steps:
  1. Describe the finding that the urge to neutralise drops over time without a ritual.
  2. Draw the expected curve: up, then down without the ritual.
  3. Contrast it with the short-lived relief a ritual gives, which is followed by the urge coming back.
  4. Turn this into a prediction to test in exposure.
  [source: Salkovskis 1999; the curve-drawing is synthesis]
• Watch for: Patients who don't believe it. Treat their doubt as a hypothesis and test it in the next experiment.

4.18 Thought-suppression experiment

• Purpose: Show that trying to control intrusions makes them more frequent. [source: Salkovskis 1999]
• Steps:
  1. Week 1: the patient records intrusions each day (a frequency count) and rates daily how hard they tried to suppress them.
  2. Week 2: on set alternate days (Salkovskis's example is Monday, Wednesday, Friday and Sunday) they actively suppress. On the other days they only record.
  3. Plot daily frequency on a graph, marking the two kinds of day.
  4. Review the graph together. Typically suppression days show more intrusions.
  [source: Salkovskis 1999]
• Example language (illustrative): "Let's find out whether pushing the thoughts away is helping. On some days push as hard as you like, on others just count. Then we'll compare."
• Watch for: If there's no difference, explore whether "just record" days involved covert suppression. Either way, the result undermines the idea that control works. [synthesis]

4.19 Stopping-criteria experiment

• Purpose: Target the rule that a ritual can only stop when the patient feels certain or "just right". That state can't be reached and keeps rituals going. [source: Salkovskis 1999]
• Steps: Salkovskis names an alternating-treatments design but gives no schedule. [uncertain] The following is a practical version [synthesis]:
  1. On alternate days or occasions, end the ritual either when it "feels right" or at a fixed, ordinary criterion agreed beforehand (for example, one wash, or one look).
  2. Rate discomfort straight after and an hour later, and count how often the doubt comes back.
  3. Compare the two kinds of day.
• Example language (illustrative): "How do people without OCD know when they've finished washing? Let's try using their rule on some days and yours on others."
• Watch for: A fixed criterion that turns into a new ritual. Vary it, and keep moving it towards ordinary practice.

4.20 Responsibility pie chart

• Purpose: Show that responsibility for a feared outcome is shared and comes in degrees ("shades of grey"). It is not all or nothing. [source: Salkovskis 1999]
• Steps:
  1. Name the specific feared outcome (for example, a house fire).
  2. The patient lists every factor that could contribute to it: their own action or inaction, and then all the others (other people, faulty equipment, manufacturers, chance and so on). Take time to make the list complete.
  3. Starting at the bottom of the list, the patient gives each factor a slice of the pie, so their own share is decided last.
  4. The therapist provides structure and prompts but does not reassure. The aim is not to prove the patient blameless.
  5. Compare the final slice with the patient's initial rating of their own responsibility.
  [source: Salkovskis 1999]
• Example language (illustrative): "Before we look at your part, let's list everything else that would have to happen for a fire to start."
• Watch for:
  • Salkovskis notes that proving no influence at all is almost impossible, and that going over things in detail can increase doubt. Keep the exercise brief and about the principle.
  • The patient using the chart repeatedly to feel certain. That is neutralising.

4.21 Behavioural experiments on beliefs and on the maintaining cycle

• Purpose: Test predictions that come from Theory A against those from Theory B. Show that reassurance and rituals keep the problem going. [source: Salkovskis 1999; Sookman et al. 2021] The Oxford Guide includes a blank behavioural experiment form. [source: Bream et al. 2017, appendix listing]
• Steps:
  1. Pin down a specific belief and rate how strongly the patient holds it.
  2. Write down what Theory A predicts will happen, and what Theory B predicts.
  3. Design a task that is approachable and that will tell the two apart (for example, a day asking for reassurance versus a day not asking, then comparing anxiety and urges by the evening).
  4. Agree what result would count as support for each theory.
  5. Run the task, in session or as homework.
  6. Review: what was expected, what happened, and what it means.
  7. Re-rate the belief.
  [source: Salkovskis 1999; Sookman et al. 2021; the step order is synthesis] Fear of saying or writing an intrusion can be tested the same way. [synthesis]
• Example language (illustrative): "If Theory A is right, what should happen when you don't ask your partner? If Theory B is right, what should happen?"
• Watch for: Covert shifts of responsibility during experiments, for example "you told me to, so it's your fault". Find them and discuss them openly. [source: Sookman et al. 2021]

4.22 Reversal question

• Purpose: Expose how implausible the feared causal link is by turning it the other way round. [source: Salkovskis 1999]
• Steps: Ask whether doing the opposite of the ritual on purpose could bring about the feared outcome. Salkovskis's example asks whether a washer could kill someone by deliberately not washing their hands. Then discuss what the answer says about the link between ritual and harm. [source: Salkovskis 1999]
• Example language (illustrative): "If you wanted to cause a fire by not checking the cooker, how confident would you be that it would work?"
• Watch for: Keep a light, curious tone. Done mockingly, this can feel dismissive. [synthesis]

4.23 Mood-induction demonstration

• Purpose: If the patient doesn't see how low mood affects their thinking, show it directly. [source: Salkovskis 1999, referring to Clark's procedure]
• Steps: Use a brief standard mood-induction exercise in session. Then compare how threatening an intrusion seems before and after. [source: Salkovskis 1999 names the procedure; the steps are synthesis] [uncertain: procedure details not given in the source]
• Watch for: Bring mood back up before the session ends.

4.24 Modified dysfunctional thought record

• Purpose: A written record of intrusion, appraisal, feeling, neutralising and alternative appraisal, adapted for OCD. [source: Salkovskis 1999]
• Steps (synthesis):
  1. Columns: trigger and intrusion; meaning or appraisal (with belief rating); feelings; urge or neutralising; alternative appraisal (Theory B view); re-rating.
  2. Use it to spot patterns, not to argue each thought away.
• Watch for: Records that become long or repetitive, or that the patient must do "right". These are a reassurance ritual. Shorten them or stop. [source: Salkovskis 1999 warns of this]

4.25 Downward arrow and two-column technique

• Purpose: Get to the core feared meaning of the intrusion and of having it. [source: Salkovskis 1999]
• Steps:
  1. Downward arrow: starting from the appraisal, ask repeatedly what it would mean if it were true, until you reach the core feared meaning (for example, "I'd be a monster" or "I'd be to blame for ever").
  2. Two-column technique: set the obsessional appraisal of the intrusion's occurrence and content against an alternative appraisal.
  [source: Salkovskis 1999 names these tailored to appraisals of occurrence and content; the procedural detail is synthesis]
• Example language (illustrative): "And if you had left the door unlocked and someone got in, what would that say about you?"
• Watch for: Feeling responsible "forever" or feeling unforgivable. These moral appraisals often drive the most rigid rituals.

4.26 Working directly on responsibility assumptions

• Purpose: Change general rules such as "not preventing harm is as bad as causing it" or "if I think of a danger I must act on it". [source: Salkovskis 1999]
• Steps:
  1. Pick out assumptions from questionnaire responses (Responsibility Attitudes measure) and from the formulation.
  2. Write each assumption out.
  3. Look at its costs and its double standards, for example whether the patient would judge a friend the same way. [synthesis]
  4. Test predictions that follow from the assumption.
  [source: Salkovskis 1999 for using the measure; the steps are synthesis]
• Watch for: High-responsibility life transitions such as parenthood or promotion. These can reactivate the assumptions. [source: Sookman et al. 2021]

4.27 "What's bad about that?" (Wells's variant)

• Purpose: Ask about the safety behaviour itself to bring out the costs of being obsessional. [source: Salkovskis 1999, citing Wells]
• Steps: When the patient says "I've got to check", ask what's bad about that. Keep following the answers to bring out costs such as lost time, exhaustion and strain on the family. [source: Salkovskis 1999]
• Example language: "What's bad about that?" (Salkovskis 1999, attributing to Wells).

D. Exposure and response prevention as behavioural experiments

4.28 ERP: rationale, planning and handing over responsibility

• Purpose: ERP does four jobs in this model: it targets neutralising, tests the formulation, disconfirms feared outcomes, and wins life back from compulsions. It "follows naturally" once belief change has started. [source: Salkovskis 1999] NICE treats ERP as the core of CBT for OCD. For obsessions without overt compulsions, NICE specifies exposure to the obsessional thoughts plus prevention of mental rituals and neutralising. [source: NICE CG31]
• Consent and safety (before starting):
  • Explain the rationale clearly. Describe out-of-office and home practice in advance. Get informed consent, and tell the patient they can stop at any time. [source: Sookman et al. 2021]
  • Only use tasks within ordinary, socially acceptable everyday risk. Never ask for anything genuinely dangerous, illegal or harmful to others.
  • Before contamination tasks, check medical issues such as skin conditions and immunosuppression.
  • Use graded rather than flooding exposure unless the patient chooses otherwise. Graded exposure is generally preferred. [source: Sookman et al. 2021]
  • Postpone or adapt exposure in acute suicidality, active psychosis, or when substance misuse makes the work unsafe, and assess risk first. [synthesis]
• Steps:
  1. Identify all triggers, obsessions and compulsions, including avoidance, washing and mental compulsions. [source: Sookman et al. 2021]
  2. Build a graded list of in-vivo and imaginal exposure tasks that covers every feared situation. [source: Sookman et al. 2021]
  3. Frame each task as a test of Theory A versus Theory B, with a specific prediction written down beforehand. [source: Salkovskis 1999; Sookman et al. 2021]
  4. Therapist-modelled stage: the therapist does the task first (for example, touching the feared object without washing), then the patient does it. Be prepared to go beyond the patient's task ("anti-OCD" actions). [source: Salkovskis 1999; Sookman et al. 2021]
  5. Unmodelled stage: the patient does tasks without the therapist demonstrating first. [source: Salkovskis 1999]
  6. Patient-led stage: the patient plans, carries out and reviews their own tasks, and the therapist explains why responsibility is being handed over. [source: Salkovskis 1999]
  7. Response prevention: no rituals, reassurance or mental neutralising during or after the task. If a compulsion slips through, re-expose straight away. [source: Sookman et al. 2021]
  8. Set homework at each stage and follow it up in the next session. [source: Sookman et al. 2021]
• Example language (illustrative): "I'm going to touch the bin and then make us both a drink without washing. Then it's your turn. What does Theory A predict? What does Theory B?"
• Watch for:
  • Subtle safety behaviours that block learning: distraction, wiping, "undoing" thoughts, or touching only with fingertips. [source: Sookman et al. 2021]
  • Signs it is working: predictions not coming true, urges fading without rituals, the patient suggesting harder tasks, and Theory B ratings going up.

4.29 Not telling the therapist the details of the task

• Purpose: Stop the therapist's knowledge from becoming a source of reassurance. For example, "you knew what I did and didn't say it was dangerous". [source: Salkovskis 1999]
• Steps: Once tasks are patient-led, the patient reports how they responded (anxiety, urges, what they learned) but not the specific details of the task. Explain why. [source: Salkovskis 1999]
• Example language (illustrative): "Tell me how it went and what you learned, but not exactly what you touched. If I know, part of you will feel I've approved it."

4.30 The unlimited-reassurance offer

• Purpose: Help the patient see for themselves that reassurance lasts only briefly and works against them. Use it when reassurance-seeking carries on inside therapy, for example when the patient mentions behaviour to get a reaction. [source: Salkovskis 1999]
• Steps:
  1. Raise the reassurance-seeking openly.
  2. Offer "as much reassurance as you need", even suggesting the remaining appointments be spent only on that. The condition is that the patient promises the reassurance will last the rest of the year.
  3. The patient usually says that's impossible because reassurance lasts minutes.
  4. Ask whether reassurance-seeking is helping, then.
  [source: Salkovskis 1999]
• Watch for: Use it with warmth, not as a trap. Afterwards, agree a shared phrase that names reassurance-seeking when it happens. [synthesis]

4.31 Imaginal exposure

• Purpose: Use exposure when in-vivo exposure isn't possible or isn't safe, or when the fear is about consequences of not doing compulsions. [source: Sookman et al. 2021]
• Steps (synthesis):
  1. Write a short first-person narrative of the feared intrusion or consequence in the present tense, with no neutralising content.
  2. Read or listen to it repeatedly while preventing neutralising.
  3. Rate distress and urge before and after.
  4. Review what the patient expected and what happened.
• Watch for: Narratives that include reassuring endings or "undoing" lines. Take them out.

4.32 Loop tapes for obsessional rumination

• Purpose: ERP for intrusions with mainly covert neutralising. [source: Salkovskis 1999]
• Steps:
  1. Identify the intrusion, the responsibility appraisals and the urges to neutralise.
  2. Record the intrusion in the patient's own voice on a short continuous loop (any phone or audio loop will do today).
  3. Rate discomfort, play the recording while preventing neutralising, deal with any problems, and rate discomfort again.
  4. Repeat until the procedure is being done correctly and discomfort or urges start to fall.
  5. Homework: use charts and full written instructions, first at home and then in real-life situations.
  6. Later, natural intrusions become the cue to practise, and the patient records their appraisals.
  [source: Salkovskis 1999] Recording length and listening duration are not specified. [uncertain]
• Watch for: Covert neutralising while listening, such as counter-thoughts, praying or mental reviewing. Ask about it directly after each trial.

4.33 Further techniques named for rumination

Salkovskis (1999) lists these without procedures: imagery work; catastrophising exercises; pie charts; thought experiments on how the feared event could actually happen; pros and cons of being obsessional; cumulative-probability downward arrows. [uncertain: steps not in the source]

A practical outline of the cumulative-probability method [synthesis]:

1. List every step that would need to happen for the feared outcome to occur.
2. Estimate a realistic probability for each step.
3. Multiply them together.
4. Compare the result with the patient's original estimate.

Use it once to make the point, not as a repeated check.

4.34 Consolidating learning after each experiment

• Purpose: Make the most of disconfirmation. [source: Sookman et al. 2021, extinction-learning recommendations; secondary to the Salkovskis model]
• Steps:
  • Before the task, state a specific feared outcome that can be checked.
  • Remove safety signals.
  • Vary the contexts and spread the practice out.
  • Afterwards ask: what did you expect, what happened, and how do you know?
• Watch for: Disgust-based responses can fade more slowly than fear. Expect slower change and keep repeating the tasks.

E. Family, homework and ending

4.35 Involving family and carers

• Purpose: Reduce family accommodation, which predicts poorer outcome, and harness support. [source: NICE CG31; Sookman et al. 2021]
• Steps:
  1. With consent, invite relatives to a session and share the formulation.
  2. List the ways they take part: answering reassurance questions, doing rituals for the patient, and avoiding things.
  3. Agree a gradual, jointly planned reduction, with an agreed supportive script the relative can use instead of reassuring.
  4. Where suitable and acceptable, a relative can support ERP homework. [source: NICE CG31]
  5. Address criticism or hostility, which is also linked to worse outcomes. [source: Sookman et al. 2021]
• Example language for relatives (illustrative): "I can see this is hard. We agreed I won't answer that one. I'm here with you while the urge passes."
• Watch for: Relatives who withdraw support suddenly or angrily. Reductions should be agreed with the patient, not imposed.

4.36 Session recordings as homework

• Purpose: Sessions are long and emotionally intense, so listening back helps consolidate them. [source: Salkovskis 1999]
• Steps: With consent, record each session and ask the patient to listen once before the next one, noting key points. [source: Salkovskis 1999; the "once" and note-taking are synthesis]
• Watch for: Replaying the recording for reassurance ("did the therapist really say it's fine?"). If so, limit it to one listen.

4.37 Relapse-prevention blueprint

• Purpose: Prepare the patient to be their own therapist. [source: NICE CG31; Sookman et al. 2021] The Oxford Guide includes a blank "blueprint" form. [source: Bream et al. 2017, appendix listing; contents not seen]
• Steps (synthesis built on the sources):
  1. Summarise how the problem developed and the vicious circle.
  2. Write down Theory A versus Theory B and the key evidence.
  3. List the most useful experiments.
  4. List remaining avoidance and plan ongoing "anti-OCD" practice.
  5. Name warning signs and high-responsibility transitions (new baby, new job). [source: Sookman et al. 2021]
  6. Write a plan for a lapse: go back to experiments and re-read key session notes. [source: Sookman et al. 2021]
  7. Offer booster sessions as needed. [source: Sookman et al. 2021]
• Watch for: A blueprint that reads like a safety guarantee. Keep it about tolerating uncertainty.

5. Handling problems

Each item is sourced as marked. Unmarked guidance is [synthesis].

• Patient wants certainty that the feared harm won't happen before starting.
  • Don't give it. Go back to the cost-benefit discussion (4.8) and Theory A/B (4.9).
  • Name the demand for certainty as the problem itself, not a condition for treatment. [source: Salkovskis 1999]
  • Offer the three-month trial with a review (4.10).
• Patient won't disclose the content of intrusions.
  • Normalise, ask what they fear would happen if they said it, and give examples from other clients (4.3). [source: Salkovskis 1999]
  • If they still can't, work with an agreed label (for example, "the bad thought") at first, and make saying it a later experiment.
• Hidden or automatic rituals, so no appraisals come up. Use the traffic-light analogy and a withholding experiment (4.4). [source: Salkovskis 1999]
• Reassurance-seeking inside therapy. This includes questions such as "Is that normal?", mentioning behaviour to see how you react, and asking for details about risk.
  • Name it openly and use the unlimited-reassurance offer (4.30). [source: Salkovskis 1999]
  • Move the patient to describing their own responses rather than the details of tasks (4.29). [source: Salkovskis 1999]
  • Agree a shorthand such as "Is that question Theory A or Theory B?"
• Cognitive techniques becoming rituals. Signs are repeated pie charts, perfect thought records, or arguing every thought. [source: Salkovskis 1999 flags this risk] Cut back the written work, switch to experiments, and ask "What is this record for: learning or feeling sure?"
• Extreme reluctance to do experiments. [source: Sookman et al. 2021 names this as a competence; specific steps below are synthesis]
  • Break the task into smaller steps.
  • Model it yourself first, then in a more extreme form (4.28).
  • Check for unspoken catastrophic predictions and turn them into the experiment itself.
  • Go back to costs and goals.
  • Add motivational interviewing if ambivalence dominates. [source: Sookman et al. 2021]
• Covert transfer of responsibility to the therapist. For example, "I only did it because you said so."
  • Find it and challenge it. [source: Sookman et al. 2021]
  • Speed up the move to patient-planned tasks (4.28, step 6). [source: Salkovskis 1999]
  • Plan with the patient that they will choose the next task.
• Non-adherence to homework. Homework adherence matters for outcome. [source: Sookman et al. 2021]
  • Review homework at the start of every session.
  • Ask about specific obstacles: unclear instructions, a task that was too big, covert rituals, or family accommodation.
  • Agree a smaller task that the patient rates as likely to succeed, with a written plan.
  • If a ritual happened, teach re-exposure rather than giving up. [source: Sookman et al. 2021]
• Overwhelm during exposure.
  • Stay with the task at a lower step rather than ending it with escape or neutralising.
  • Re-run the prediction: "What is your mind telling you will happen right now?"
  • Ask whether the urge is fading, as spontaneous decay predicts (4.17).
  • Never use reassurance to bring distress down.
  • If a panic attack, dissociation or a risk crisis occurs, pause the exposure, use grounding, assess risk, and review the plan.
• Mental rituals carrying on during ERP. Map the covert neutralising and use loop tapes or imaginal exposure with explicit prevention of mental rituals (4.31, 4.32). [source: Salkovskis 1999; NICE CG31]
• Fear of becoming careless or dangerous without OCD.
  • Use the over-correcting question and the offer to help re-obsess (4.11). [source: Salkovskis 1999]
  • Use the pie chart and assumption work for the underlying rule (4.20, 4.26).
• Thoughts not getting less frequent. Restate that the target is the appraisal and the neutralising, not how often the thoughts come. [source: Salkovskis 1999] Check for suppression and use the suppression experiment (4.18).
• Low mood or depression.
  • Low mood makes intrusions seem more significant, and loss-themed appraisals produce depression. [source: Salkovskis 1999]
  • Use the mood-induction demonstration if the link isn't seen (4.23).
  • Assess suicide risk routinely using current standard practice.
  • If depression is severe, treat it first or alongside, and consider referral for medication in line with NICE. [source: NICE CG31 for the medication pathway]
• Poor insight or overvalued ideas.
  • Assess insight (BABS) and tell overvalued ideas apart from delusions. [source: Sookman et al. 2021]
  • Lean heavily on Theory A/B evidence-gathering and small experiments rather than discussion.
  • If the beliefs are delusional, refer for psychiatric review.
• Family accommodation or hostility undermining progress. Hold a joint session and agree a plan for withdrawing accommodation gradually (4.35). [source: NICE CG31; Sookman et al. 2021]
• Housebound patients, hoarding, or contamination fears that prevent attendance. Offer home-based treatment, or telephone CBT if home treatment isn't possible. Help with travel if impairment is significant. [source: NICE CG31]
• Poor response after a full CBT trial (more than 10 hours).
  • Arrange a multidisciplinary review, then offer CBT combined with an SSRI. [source: NICE CG31]
  • Revisit the formulation for missed neutralising or accommodation.
  • Consider a time-intensive format. [source: Oldfield et al. 2011 abstract]
• Alliance ruptures.
  • These are often triggered by the therapist declining reassurance ("you don't care whether I'm safe") or by exposure feeling too pushy.
  • Acknowledge it directly and explain again why you don't reassure.
  • Hand back choice over the pace while keeping the direction.
  • Repair is part of the treatment.
• Distressing intrusions about harming children or about sexual content.
  • Separate unwanted, distressing intrusions (ego-dystonic, with avoidance and neutralising) from intent or behaviour. Assess risk properly and record your reasoning.
  • Don't reassure the patient; Theory A/B covers this work.
  • Know your local safeguarding duties, and explain the limits of confidentiality accurately.
• Perinatal OCD. A time-intensive CBT format reduced maternal OCD symptoms in a pilot RCT from Salkovskis's group. [source: Challacombe et al. 2017, abstract only]
• Relapse at life transitions. Plan in advance for transitions that bring more responsibility and use the blueprint (4.37). [source: Sookman et al. 2021]

Sources

Sources:

• Salkovskis, P. M. (1999). Understanding and treating obsessive-compulsive disorder. Behaviour Research and Therapy, 37, S29–S52 (PDF, Dimence). Primary source for the model and most interventions.
• NICE CG31: Obsessive-compulsive disorder and body dysmorphic disorder: treatment — Recommendations
• Bream, Challacombe, Palmer & Salkovskis (2017), CBT for OCD (Oxford Guides) — appendices chapter page (OUP). Appendix list only.
• Same book — Salkovskis chapter page (OUP). Truncated abstract only.
• Sookman et al. (2021). Specialty competencies in CBT for adult OCD: international expert consensus (UCL Discovery accepted manuscript). Secondary; co-authored by Salkovskis's Oxford Guide co-authors Bream and Challacombe.
• Oldfield, Salkovskis & Taylor (2011). Time-intensive CBT for OCD: case series and matched comparison (Bath research portal). Abstract only.
• Challacombe et al. (2017). Pilot RCT of time-intensive CBT for postpartum OCD (Bath research portal). Abstract only.
• Williams, Salkovskis, Forrester & Allsopp (2002). Changes in symptoms of OCD and appraisal of responsibility during CBT (Bath research portal). Abstract only.
• HRA research summary: RCT of CBT for OCD, time-intensive and standard weekly formats. Administrative details only.

Books and papers cited (not drawn on):

• Salkovskis, P. M. (1985). Obsessional-compulsive problems: a cognitive-behavioural analysis. Behaviour Research and Therapy, 23, 571–583.
• Bream, V., Challacombe, F., Palmer, A., & Salkovskis, P. (2017). Cognitive Behaviour Therapy for Obsessive-Compulsive Disorder. Oxford University Press. This is the current manual from the originating group. The text is not drawn on, so session-by-session structure, survey methods and blueprint contents are [uncertain].
• Salkovskis, P. M., Forrester, E., & Richards, C. (1998). Cognitive-behavioural approach to understanding obsessional thinking. British Journal of Psychiatry, 173(Suppl. 35), 53–63.