Atlas›Domain IX · Biomedical, neuro & digital›Guide 095
Guide IX-095

Mental-health and wellbeing apps

A 23 · L 72 · M 517 interventions in the Trinity explorer
Client cardCompare
[source: …] where a claim comes from[uncertain] not confirmed in a primary source[synthesis] the writer's reconstruction

1. Introduction

Version followed: the American Psychiatric Association (APA) App Evaluation Model as published on the APA App Advisor pages (six steps: Background, Access, Privacy and Security, Clinical Foundation, Usability, Integration toward patient-centred goals; expert-panel update of April 2025 adding health equity, social factors and AI considerations), the 105-item objective version behind the M-Health Index and Navigation Database (MIND) (Lagan et al., 2020), and, for UK services, the NHS England Digital Technology Assessment Criteria (DTAC; updated form introduced 24 February 2026) and the NICE Evidence Standards Framework (ECD7; last major update August 2022). Status checked as of October 2026 per those pages.

Mental-health and wellbeing apps are not one therapy but a large, fast-changing market of self-help, mood-tracking, skills-practice, peer-support, telehealth and chatbot products. This guide is therefore a framework for the psychotherapist's own work with apps: finding out what clients already use, choosing apps jointly with an explicit evaluation method, protecting the client's data and safety, and folding an app into the treatment plan as homework, practice or monitoring. The core idea of change is that an app adds nothing by itself; any benefit comes when a well-chosen, safe tool extends practice and self-observation between sessions inside a therapeutic relationship and a clear plan [synthesis]. APA frames app choice as a decision made with the individual patient, not a pass/fail verdict on a product, and states that static ratings are of limited use because apps change often [source: APA, The App Evaluation Model; APA, Mental Health Apps hub]. NAMI's article by the MIND developers describes apps as a complement to care, not a replacement [source: Lagan & Torous, NAMI blog, 2021]. An app is never a substitute for assessment, therapy, medical review or crisis care that a client needs [synthesis].

Meditation and mindfulness apps: the contemplative practices themselves are covered in Domain XIII (contemplative); use this guide only for the evaluation, privacy, safety and integration steps that apply to any app.

2. Framework

This is a framework, not a manualised treatment; no session count is specified in the sources. The phases below organise the APA model and the UK assurance tools into a clinical sequence [synthesis; evaluation content from source: APA App Evaluation Model].

Phase 0: Digital history (first or second session, 5–10 minutes). Goal: learn which apps, wearables, online services and AI tools the client already uses for mood, sleep, sex, substances, relationships or support, what they share there and what they get from it. Move on when you know whether any current tool is helpful, harmful or a risk to privacy or safety [synthesis].

Phase 1: Fit and goal (one session segment). Goal: decide whether an app has a role at all. Name the clinical target (e.g. between-session skills practice, sleep diary, mood tracking, psychoeducation) and check access, literacy, language, disability, cost and the client's preference for non-digital options. Move on only if an app would serve a defined treatment goal and the client wants one [synthesis; access questions from source: APA model Step 2].

Phase 2: Evaluation and shortlisting (between sessions for the therapist, 10–20 minutes in session with the client). Goal: apply the APA model's six steps (or its quicker screener to rule out poor fits), use MIND to filter candidates, and in NHS settings check DTAC and local digital-health-library status. Privacy and Security is the step APA treats as among the most critical; an app with no clear privacy policy may be inappropriate for sensitive health information [source: APA App Evaluation Model]. Move on when you and the client have one or two candidates whose gaps you both understand [source: APA, no minimum number of levels required].

Phase 3: Informed trial (2–4 weeks) [synthesis]. Goal: agree what the app is for, how often to use it, what data it collects, what it does and does not do in a crisis, and when you will review it. Record the agreement. Move on at the planned review.

Phase 4: Integration (ongoing sessions). Goal: use the app as homework, rehearsal or monitoring; review its output in session; link it to the formulation. Continue while it helps and the client engages [synthesis; integration questions from source: APA model Step 6].

Phase 5: Review, change or stop. Goal: decide jointly to continue, switch or stop; on stopping, help the client export what they want, delete account data where possible and remove permissions [synthesis; opt-out and deletion questions from source: APA model Step 3]. Re-evaluate any app after major updates, ownership changes or news of a data incident, since APA notes apps change often [source: APA Mental Health Apps hub].

Regulatory and assurance landscape (as of October 2026, per the pages listed in Sources).

  • United States: APA notes that the FDA has taken a largely hands-off approach to regulating mental health apps [source: APA Mental Health Apps hub]. Products marketed as prescription digital therapeutics have separate FDA pathways [uncertain: not verified for individual products]; product status changes (e.g. Pear Therapeutics' bankruptcy; Woebot Health's closure of its consumer app) [uncertain: not verified]. Many wellness apps and direct-to-consumer services are not covered by HIPAA; the FTC has acted against companies over health-data sharing (see Section 4.5) [source: FTC 2023; The Markup 2024].
  • England: DTAC is NHS England's baseline assessment for buying digital health technology, covering clinical safety (DCB0129 documentation), data protection, technical security, interoperability, and usability and accessibility. It is used alongside, not instead of, medical device regulation. An updated form was introduced on 24 February 2026, with the old form not to be used from 6 April 2026; the hosting website was being retired, with content redirected to digital.nhs.uk from April 2026 [source: NHS England Transformation Directorate DTAC page; NHS Innovation Service news, 4 March 2026]. DTAC is a commissioning tool; an individual therapist uses it by asking whether an NHS-offered product has passed local DTAC assurance [synthesis].
  • NICE Evidence Standards Framework (ECD7) sets the evidence commissioners should expect, by evidence tier: Tier A (system impact), Tier B (understanding and communicating) and Tier C (interventions); the August 2022 update added AI and adaptive-algorithm requirements and aligned classifications with regulation [source: NICE ECD7 update information]. Apps that deliver treatment or management of a mental-health condition would generally sit in Tier C [uncertain: classification table not verified].
  • ORCHA (Organisation for the Review of Care and Health Apps) is a private evaluation company founded by NHS clinicians that reviews apps across clinical or professional assurance, data and privacy, and usability and accessibility, and supplies app libraries to NHS organisations; its published threshold for meeting its quality bar is a score above 65% [source: ORCHA mental health apps rating map, 2019, modified 2023]. ORCHA scores are one company's assessment, not a regulatory approval [synthesis].

3. Guiding principles

  • Care first, app second. Practical meaning: never offer an app instead of assessment, therapy, medication review or crisis care the client needs; say so to the client [source: Lagan & Torous 2021; synthesis].
  • Shared, informed decision. Practical meaning: choose with the client, using the APA steps as a conversation, not a verdict; APA calls app choice a personal decision based on many factors [source: APA App Evaluation Model].
  • No pass/fail; know the gaps. Practical meaning: an app meeting few levels may still be reasonable, and one meeting all may not fit this client; write down the gaps you accepted [source: APA App Evaluation Model].
  • Privacy is clinical. Practical meaning: treat health-data disclosure as a risk to the client's safety, employment, insurance and relationships, and check it before content [source: APA model Step 3; FTC BetterHelp 2023; synthesis].
  • Do not trust store ratings or 'evidence-based' labels alone. Practical meaning: star ratings and downloads do not show clinical usefulness; look for actual evidence and face validity by trying the app yourself [source: Lagan & Torous 2021; APA model Step 4].
  • Try it before you suggest it. Practical meaning: install and use any app you will recommend, including its crisis pathway and data settings [source: APA model Steps 4–5 recommend trying the app; synthesis].
  • Safety does not live in the app. Practical meaning: the client's safety plan, crisis numbers and human contacts must exist outside the app; ask what the app does when risk is disclosed [source: APA model Step 3, question on responding to clinical safety concerns; synthesis].
  • Clarity about who watches the data. Practical meaning: tell the client plainly whether you will see app data, when, and that you do not monitor it in real time unless a service explicitly provides that [synthesis].
  • Equity and inclusion. Practical meaning: check offline use, operating system, accessibility, language, reading level, cost and intended users; always have a non-digital equivalent [source: APA model Step 2; April 2025 update on health equity].
  • Age-appropriate use. Practical meaning: check the app's stated intended users and age terms, and involve parents or carers according to the young person's competence and local law [source: APA model Step 2 on intended users; synthesis].
  • Re-evaluate over time. Practical meaning: review apps at set intervals and after updates, sale of the company or data incidents [source: APA Mental Health Apps hub].
  • Stay within role. Practical meaning: a psychotherapist does not prescribe a prescription digital therapeutic or advise on medication via an app; route such questions to the prescriber [synthesis].
  • Document. Practical meaning: note the app, version or date, the evaluation, the agreed purpose, the consent conversation and review date [synthesis].

4. Intervention inventory

The sources supply evaluation questions; the in-session procedures, wording and monitoring cues below are illustrative clinical synthesis built on them unless a source is named [synthesis]. Example language is illustrative throughout.

4.1 Taking a digital history

  • Purpose / when: At intake and at reviews, to find tools already in use, including risky ones (unvetted forums, general-purpose AI chatbots, tracking apps shared with a partner) [synthesis].
  • Steps: (1) Ask openly which apps, sites, wearables or AI tools the client uses for mood, sleep, anxiety, substance use, eating, relationships or loneliness. (2) For each: what it does, how often, what they enter, who can see it, whether it costs money. (3) Ask what it gives them and what, if anything, it makes worse (checking, comparison, sleep loss). (4) Ask whether anyone else has access to the device or account. (5) Note tools to keep, review or reconsider.
  • Example language (illustrative): "Lots of people use apps or online tools for how they feel. What do you use at the moment, even things that don't seem like 'mental health' apps?" "Does anyone else have access to your phone or that account?"
  • Watch for: Shame about chatbot or forum use; a partner or family member able to read entries (coercive control, safety risk); excessive tracking feeding anxiety or eating-disorder behaviour; paid subscriptions the client cannot afford. Working if the client describes use candidly and you can name each tool's role.

4.2 Deciding whether an app has a role

  • Purpose / when: Before suggesting any app; to tie it to the formulation and avoid apps replacing needed care [source: Lagan & Torous 2021, apps as complement; synthesis].
  • Steps: (1) Name the treatment goal the app would serve (practice of a skill taught in session, a diary, psychoeducation, sleep routine, peer support). (2) Ask whether a paper, calendar or in-session alternative would do the same. (3) Check the client's preference and capacity (see 4.9). (4) Check that the level of care is right: if the client needs more intensive treatment, a medical review or crisis care, arrange that first. (5) Agree whether to proceed to evaluation.
  • Example language (illustrative): "An app could help you practise this between our sessions, but it's an add-on to our work, not a replacement. Would you like that, or would a paper record suit you better?"
  • Watch for: Pressure (from an employer, insurer or waitlist) to use an app instead of therapy; the client hoping an app will make sessions unnecessary while risk is high. Working if the app has a stated purpose both of you can name.

4.3 Structured evaluation with the APA App Evaluation Model

  • Purpose / when: For every app you recommend or the client asks about; repeat after major updates [source: APA App Evaluation Model; APA hub].
  • Steps (follow the six steps in order; APA offers a full version and a shorter screener to rule out poor fits):
    1. Background: identify owner, trusted source, funding and conflicts of interest, stated purpose, whether it claims to be medical, costs, update within the last 180 days, and a feedback route.
    2. Access: offline or low-bandwidth use, operating systems, web access, accessibility features, languages, reading level, support needed, intended users (age, literacy, culture, condition).
    3. Privacy and Security: readable policy, declared data uses, handling of sensitive data and de-identification, secure log-in, opt-out and deletion, storage location, third parties receiving data, whether and how it responds to clinical safety concerns, breach plan and history, and use of AI with bias monitoring.
    4. Clinical Foundation: try the app for face validity; check accuracy and currency of content, a clinical or recovery basis, evidence (publications, user feedback), study in intended users, retention rates, user involvement in design.
    5. Usability: modes of engagement, fit with the client's priorities, customisation, intuitive design, help and support, personalisation.
    6. Integration toward patient-centred goals: individual versus care-team use, access to and export of own data, secure sharing with the care team, links to records or wearables, useful data display. (Paraphrased from APA's published questions [source: APA App Evaluation Model].) Then: (7) write a short note of strengths and accepted gaps; (8) take it to the client (4.5).
  • Example language (illustrative): "There's no perfect app. I've gone through this one against a checklist the American Psychiatric Association publishes. Here's what looks good and here's what I'd want you to know before you decide."
  • Watch for: No privacy policy or one you cannot understand (APA suggests such an app may be inappropriate for sensitive health information); medical claims without evidence; no update for over 180 days; unclear owner. Working if you can state, for each step, what you found and what remains unknown [source: APA App Evaluation Model].

4.4 Shortlisting with MIND (M-Health Index and Navigation Database)

  • Purpose / when: To find candidates quickly before a full APA review. MIND, at mindapps.org, is run by the Division of Digital Psychiatry at Beth Israel Deaconess Medical Center; its evaluations are based on the APA model [source: AHRQ Academy listing]. The underlying index is 105 objective yes/no or numeric questions across accessibility, privacy and security, clinical foundation, engagement and interoperability (Lagan et al., 2020); users filter by the criteria that matter to them, with more than 100 filters, and the database shows which apps meet them rather than labelling apps good or bad [source: Lagan et al. 2020; Lagan & Torous 2021].
  • Steps: (1) With the client, choose non-negotiables (e.g. free, works offline, no third-party data sharing, can delete data, specific language). (2) Apply those filters plus the clinical feature needed (e.g. mood tracking, CBT skills). (3) Note the date of each app's MIND rating, since apps change. (4) Take the top one to three into a full APA review (4.3) and try them yourself.
  • Example language (illustrative): "Let's pick what matters most to you, say free, private and usable offline, and see which apps fit, then I'll check the shortlist properly."
  • Watch for: Ratings older than the current app version; treating a filter match as a recommendation. Working if the shortlist reflects the client's stated priorities. Current coverage and number of rated apps [uncertain: not verified].

4.5 Joint privacy review with the client

  • Purpose / when: Before the client creates an account or enters sensitive data; whenever a client already uses an app with health information [source: APA model Step 3; synthesis].
  • Steps: (1) Open the privacy policy and settings together. (2) Find, in plain words: what is collected, why, who receives it (advertisers, analytics, 'partners'), where it is stored, how to opt out and delete. (3) Check whether questionnaires are taken before consent is asked (a pattern Mozilla reported in 2023 for some services) and whether trackers or advertising are present. (4) Turn off unnecessary permissions (location, contacts, microphone), set a strong password or biometric lock, and consider a non-identifying username or email. (5) Explain that many consumer wellness apps are not covered by health-privacy law such as HIPAA [uncertain: depends on the company and jurisdiction]. (6) Agree what the client will and will not enter.
  • Reliable reports to draw on (status as of October 2026 per the cited pages):
    • FTC v. BetterHelp: final order announced 14 July 2023; the FTC alleged BetterHelp disclosed email addresses, IP addresses and health questionnaire information to Facebook, Snapchat, Criteo and Pinterest for advertising despite privacy promises; BetterHelp agreed to pay USD 7.8 million for partial refunds; the order bans disclosing health data for advertising and personal information for retargeting, requires affirmative express consent for certain disclosures, a privacy programme, third-party deletion and a data retention schedule [source: FTC press release, July 2023].
    • FTC and Cerebral: proposed order announced April 2024; the FTC alleged that from 2019 to 2023 Cerebral sent contact details, medical histories, insurance information and prescriptions to third parties via tracking tools, including onboarding mental-health answers to Meta; Cerebral agreed to pay USD 7 million and to stop using or disclosing health information for most advertising purposes; at the time of reporting the order needed court approval [source: The Markup, 22 April 2024]. Final court status [uncertain: not verified].
    • Mozilla Privacy Not Included (2 May 2023): 19 of 32 mental-health apps reviewed carried a warning label (29 of 32 in 2022); concerns included data sharing and personalised advertising, weak password requirements, consent requested after questionnaires, and one AI companion app rated among the worst reviewed; some apps improved and about 40% got worse than the year before [source: Mozilla Foundation blog, 2023].
  • Example language (illustrative): "Before you put anything personal in, let's look at where it goes. Some well-known services have been fined for sharing users' health information with advertisers, so it's worth five minutes." "What would you not want an employer or a family member to see? Let's keep that out of the app."
  • Watch for: Shared devices or accounts, domestic abuse, immigration, custody, employment or insurance concerns that raise the stakes of disclosure; clients who stop using a helpful tool out of fear (offer safer settings or paper). Working if the client can say what the app collects and has chosen settings deliberately.

4.6 Checking UK assurance and regulatory status

  • Purpose / when: In NHS or commissioned settings, or when a client is offered an app by a service, employer or insurer [source: NHS England DTAC page; NICE ECD7].
  • Steps: (1) Ask whether the product is on the local NHS digital-health or app library and whether the organisation holds a completed DTAC assessment for the current version. (2) Ask whether it is a regulated medical device (UKCA/CE marking) and for which intended purpose [uncertain: not verified for individual products]. (3) Check whether NICE has assessed it (e.g. early value assessment or guidance) and note the date. (4) If an ORCHA review is shown, note the score and date and remember ORCHA's above-65% bar is a company threshold. (5) In the US, check for FDA status only where a product claims to diagnose or treat; most wellness apps are not FDA-reviewed [source: APA hub on FDA's largely hands-off approach]. (6) Record 'status as of <month year> per <source>'.
  • Example language (illustrative): "This one has been through the NHS's own safety and data checks for this area, which is reassuring, but we'll still check it suits you."
  • Watch for: Treating an assurance or listing as proof the app helps this client; outdated DTAC forms (the previous form was to be retired by 6 April 2026) [source: NHS Innovation Service news, March 2026]. Working if you can state the status with a date and source.

4.7 Safety and crisis-pathway check

  • Purpose / when: Before any app is used by a client with current or past suicidality, self-harm, psychosis, mania, eating disorder or domestic abuse; and for every app with chat, peer or AI features. APA's model asks whether the app can respond to clinical safety concerns and discloses how [source: APA model Step 3].
  • Steps: (1) Find the app's crisis features yourself: what happens if a user types or selects something indicating risk; which numbers it shows and whether they are correct for the client's country; whether anyone monitors messages and when. (2) Check moderation in peer or community features. (3) Make the client's safety plan independent of the app: written plan, personal contacts, local crisis line and emergency number, and how to reach your service. (4) Tell the client clearly whether you, or anyone, will see entries and when; state that you do not monitor app data in real time unless your service has explicitly set that up. (5) Agree what the client does if the app's content upsets them. (6) For high-risk periods, consider pausing apps that encourage rumination, calorie or body tracking, or contact with unmoderated communities.
  • Example language (illustrative): "If you write in the app that you're thinking of ending your life, nobody may see it straight away. So let's be clear what you'd actually do and who you'd contact." "I'll look at your mood chart when we meet, but I won't be watching it between sessions."
  • Watch for: Wrong-country crisis numbers; chatbots responding unsafely to risk; clients using the app as their only outlet during escalating risk; tracking that feeds restriction or compulsive checking. Working if the client can describe their non-app crisis plan without prompting [synthesis].

4.8 Age limits and young people

  • Purpose / when: With anyone under 18 and with parents asking about apps. APA asks whether the app identifies its intended users by age [source: APA model Step 2].
  • Steps: (1) Check the app's terms for minimum age and whether it is designed for young people. (2) Consider the young person's competence and confidentiality: discuss with them what a parent or carer could see. (3) Apply local law and service policy on children's data and parental consent; the age at which a child can consent to data processing differs by jurisdiction (for example under UK GDPR and US COPPA) [uncertain: not verified]. (4) Prefer apps designed and evaluated for the age group, with moderated communities and clear safeguarding routes. (5) Agree with the young person and, where appropriate, the family, what the app is for and what happens to risk disclosures.
  • Example language (illustrative): "This app is meant for people 16 and over, so it isn't the right fit. Here's one made for people your age; let's look at who can see what you put in it."
  • Watch for: Adult apps with adult communities; parental monitoring that undermines safety (e.g. LGBTQ+ young people, abuse at home); social comparison, sleep loss and night-time use. Working if the young person knows who sees their data and the plan respects their competence [synthesis].

4.9 Digital-inclusion check and non-digital alternatives

  • Purpose / when: Before recommending any app, so that apps do not widen inequalities. APA's Access step asks about offline use, operating systems, accessibility, languages, reading level, support needs and intended users, and the April 2025 update added health equity and social factors [source: APA App Evaluation Model; APA hub]. Paywalls are among the limitations the MIND developers name [source: Lagan & Torous 2021].
  • Steps: (1) Ask about device ownership, data allowance, connectivity, shared devices, confidence with apps, sight, hearing, dexterity, literacy and language. (2) Ask about cost tolerance, including free trials that convert to subscriptions. (3) Offer an equal non-digital route (paper diary, printed handouts, phone check-ins). (4) If the client wants an app but lacks confidence, demonstrate it in session and set it up together. (5) Never make access to therapy conditional on app use [synthesis].
  • Example language (illustrative): "Some people love apps and some don't, and either is fine. If paper works better for you, we'll use paper."
  • Watch for: Embarrassment about digital skills or poverty; older clients, people with intellectual disability, refugees or people without a private device being steered away from care. Working if the chosen format is one the client actually uses.

4.10 Agreeing an informed app trial

  • Purpose / when: Once a candidate is chosen, to turn it into a time-limited, reviewable part of the plan [synthesis].
  • Steps: (1) Summarise the evaluation: strengths, gaps accepted, data practices, crisis limits. (2) Agree purpose, which features to use (and which to ignore), frequency and duration (e.g. a 2–4 week trial). (3) Agree what, if anything, the client will show you and how (screen in session, exported report); do not ask for log-ins or unsecured transfer of data. (4) Set a review date. (5) Document the conversation and consent.
  • Example language (illustrative): "Let's try just the breathing and thought-record parts for two weeks, a few minutes most days, and look at how it went on the 14th. You choose what to show me."
  • Watch for: Client agreeing to please you; unclear expectations; feature overload. Working if the client can restate the purpose and review date.

4.11 Using an app for homework and skills practice

  • Purpose / when: To extend practice of skills taught in session (e.g. thought records, behavioural activation scheduling, exposure hierarchies, relaxation, sleep routines) between sessions [synthesis]. For exposure work, keep hierarchy planning, consent and safety in session; use the app only for logging and reminders.
  • Steps: (1) Teach the skill in session first. (2) Show where the matching feature is in the app and do one entry together. (3) Link the task to the client's own goal and situation. (4) Set reminders the client chooses. (5) Begin the next session by reviewing the homework, praising attempts and problem-solving obstacles. (6) Translate app content into the client's language and formulation where the app's framing differs.
  • Example language (illustrative): "Let's do your first entry together now, about Tuesday's argument, so you know how it works when you're on your own." "What got in the way this week? Was it the app or the task?"
  • Watch for: App content that contradicts your approach or is clinically inaccurate (re-check face validity); generic advice replacing the formulation; perfectionism about completing every entry. Working if the client brings examples that show the skill used in daily life, not just entries completed.

4.12 Monitoring and reviewing app data in session

  • Purpose / when: To use mood, sleep, symptom or activity tracking for collaborative case formulation and review. APA's Integration step asks whether users can access, export and securely share data and whether data are presented usefully [source: APA model Step 6].
  • Steps: (1) Agree a small number of measures that matter to the client. (2) Agree how data come to sessions (client shows the screen or brings an export; use your service's secure systems if records are uploaded). (3) Look at the data together, asking the client to interpret first. (4) Link patterns to events, thoughts, behaviour and the formulation. (5) Decide what to file in the record and tell the client. (6) Keep validated outcome measures your service uses separate from app scores; app scores are not validated clinical measures unless shown to be [uncertain: varies by app].
  • Example language (illustrative): "What do you notice when you look at these two weeks?" "Your low days cluster on Sundays; what happens on Sundays?"
  • Watch for: Over-monitoring that increases anxiety or symptom focus; data gaps read as failure; the client feeling surveilled; wearable data driving health anxiety. Working if the data help the client notice patterns and choose actions. If tracking shows deterioration, a manic shift, rising risk or side effects of medication, act on it clinically and pass information to the prescriber or GP with consent [synthesis].

4.13 Reviewing, switching or stopping

  • Purpose / when: At the planned review, after an update or data incident, when engagement falls, or at the end of therapy. APA notes apps change often [source: APA hub].
  • Steps: (1) Ask what helped, what did not and how often the app was used. (2) Decide jointly: continue, adjust features, switch, or stop. (3) On stopping: help the client export anything they want to keep, use the app's deletion or opt-out route, cancel subscriptions, and remove permissions [source: APA model Step 3 on opt-out and deletion; synthesis]. (4) At the end of therapy, plan how the client will use the app (or not) for relapse prevention, with a human route back to help.
  • Example language (illustrative): "It sounds as if the reminders helped but the community part didn't. Shall we keep one and switch the other off?" "Before you delete it, is there anything you want to keep?"
  • Watch for: Abandoned accounts still holding data; the app becoming the client's only support after discharge. Working if decisions are deliberate and the client knows how to seek help again.

4.14 Talking about general-purpose AI and chatbot apps

  • Purpose / when: When a client uses AI chatbots or AI 'companions' for support. APA's Privacy step now asks whether an app uses AI, how, and how bias is monitored and mitigated [source: APA App Evaluation Model; April 2025 update]. Mozilla's 2023 review rated one AI companion app among the worst for privacy [source: Mozilla 2023]. Full guidance on AI chatbots is in the AI entries of this family; regulatory concerns such as the FDA advisory committee's November 2025 discussion and state laws (e.g. Illinois, 2025) [uncertain: not verified here].
  • Steps: (1) Ask without judgment what the client uses it for and how it feels. (2) Explore benefits and limits: availability at night versus accuracy, privacy and absence of a human response to risk. (3) Check what personal information has been shared. (4) Include it in the safety plan: the chatbot is not a crisis service. (5) Bring useful 'conversations' into therapy as material. (6) Never present a chatbot as a replacement for therapy.
  • Example language (illustrative): "Lots of people talk things through with AI. What does it give you that's helpful? And when it gets really bad, what would you do that involves a real person?"
  • Watch for: Withdrawal from human relationships, dependence, reinforcement of delusional or hopeless beliefs, unsafe advice, sharing identifying information. Working if the client uses it knowingly and keeps human supports.

4.15 Meditation and mindfulness apps

  • Purpose / when: Clients often use meditation apps for stress, sleep or anxiety. Evaluate the app (4.3, 4.5, 4.7) as for any app; for the practices themselves, their teaching and their risks, use Domain XIII (contemplative).
  • Steps: (1) Apply the evaluation, privacy and safety checks. (2) Ask how the practice affects the client, including any distress. (3) Refer to the Domain XIII guide for practice-related guidance.
  • Example language (illustrative): "Before we build this app into your plan, let's check its privacy settings and talk about how the practices actually feel for you."
  • Watch for: Distress, dissociation or trauma-related reactions during practice (see Domain XIII); subscription costs. Working if the client reports the practice as usable and not distressing.

4.16 The clinician's supervision and governance role

  • Purpose / when: Throughout; to keep app use within professional standards [synthesis].
  • Steps: (1) Keep a short, dated evaluation note for apps you recommend often, and re-check them periodically. (2) Follow your service's approved list or digital library and its DTAC position where one exists [source: NHS England DTAC page]. (3) Do not ask clients to send app data through insecure channels; use service systems. (4) Be clear in your records and with the client that you do not provide real-time monitoring. (5) Declare any conflict of interest (e.g. affiliate links, company payments). (6) Bring uncertain cases (high risk, minors, novel AI tools) to clinical supervision. (7) Report safety incidents involving apps through your service's channels and, where the product is a medical device, to the relevant regulator [uncertain: reporting routes not verified].
  • Example language (illustrative), to a colleague: "I've used this app with a few clients; here's my evaluation note from March and what's changed since."
  • Watch for: Recommending apps from habit or marketing; untracked changes in ownership or privacy policy. Working if your recommendations are dated, documented and reviewed.

4.17 Integrating with prescribers and other care

  • Purpose / when: When an app touches medication (reminders, symptom tracking shared with a prescriber) or is a prescription digital therapeutic [synthesis].
  • Steps: (1) Do not advise on medication changes based on app data; route questions about medication, side effects or adherence to the prescriber. (2) With consent, share relevant patterns (sleep collapse, mood elevation, side-effect reports) with the prescriber or GP. (3) If a prescriber has prescribed a digital therapeutic, support engagement and ask how it fits with therapy goals, and pass on problems to the prescriber.
  • Example language (illustrative): "Your chart shows you've slept much less this fortnight. That's worth telling your psychiatrist; shall we write down what to say, or would you like me to contact them?"
  • Watch for: Clients changing medication on the basis of app feedback; conflicting advice from app and clinicians. Working if information reaches the right professional with the client's agreement.

5. Handling problems

The evaluation sources do not address most in-therapy difficulties; responses below are clinical synthesis unless a source is named [synthesis].

  • Non-adherence and falling engagement. Retention is a known limitation the MIND developers name ('consistency') [source: Lagan & Torous 2021]. Response: treat drop-off as information, not failure; ask what got in the way (task, app design, mood, time, cost); reduce to one feature; move reminders to a better time; switch to paper. Illustrative: "Most people stop using apps after a while. What made it harder this week?"
  • Stuck or 'app-only' clients. A client who uses an app heavily without change: review whether the app targets the right problem and whether the level of care is enough; bring the work back into sessions; consider referral for more intensive treatment. Never accept an app as the whole plan when needed care is missing.
  • Resistance or distrust of technology. Respect it; offer non-digital equivalents; explore specific worries (privacy, surveillance, previous breaches) and answer them with the privacy review (4.5). Illustrative: "Your concern about data is reasonable; let's do this on paper."
  • Overwhelm or harmful effects. Notifications, streaks, symptom tracking or community content can increase anxiety, rumination, health anxiety, body checking or shame. Response: switch off streaks and notifications, reduce tracking frequency, pause or stop the app, process the reaction in session. With eating-disorder risk avoid calorie, weight or body tracking features.
  • Risk disclosed only through the app, or after the fact. If a client says they entered suicidal thoughts in an app, do a full risk assessment using your service's standard procedure, update the safety plan and restate that the app is not monitored by you. Review the app's crisis response (4.7) and whether to continue it.
  • Clinically inaccurate or conflicting content. Discuss it openly, explain where your approach differs, disable that content or switch apps; re-run the Clinical Foundation step [source: APA model Step 4 on face validity].
  • Privacy incident or news of data misuse. Tell affected clients what is known from reliable reports, help them change passwords, request deletion, review what was shared, and consider stopping use; document the conversation. Recent enforcement examples are in 4.5 [source: FTC 2023; The Markup 2024].
  • Ruptures linked to apps. A client may feel judged by data review, monitored, or that you prefer the app to talking. Name it, apologise for the part you played, return choice to the client over what to share, and adjust. Illustrative: "I may have spent too long on the chart and not enough on how you've been. Let's put it aside."
  • Shared devices, coercive control or family access. If a partner or parent can read entries, prioritise safety: use no app or one with a separate lock and no notifications, or paper kept safely; follow safeguarding procedures.
  • Young people and age mismatch. If a young person uses an app above its age limit or an adult community, discuss it without shaming, offer age-appropriate alternatives, and involve carers in line with competence and safeguarding rules [source: APA model Step 2 on intended users; synthesis].
  • Cost and paywalls. Check for free tiers, NHS-provided or library access, or non-digital equivalents; never make care depend on paying for an app [source: Lagan & Torous 2021 on paywalls; synthesis].
  • Discontinued or changed products. Apps are withdrawn, sold or change policies (e.g. Woebot Health's consumer app closure) [uncertain: not verified]. Response: help export data, find a replacement through MIND and the APA steps, and re-check privacy after any ownership change [source: APA hub on apps changing often].
  • Pressure from third parties. Employers, insurers or services may offer an app instead of therapy. Clarify the client's needs, advocate for the appropriate level of care, and treat an app as an adjunct at most.

Sources

← Previous · IX-094AVATAR therapy for voices (Leff, Craig)Next · IX-096 →Rule-based therapy chatbots (Woebot, Wysa)