The boundary between a wellbeing tool and a medical device, why it can be difficult to draw, and a two-step test that weighs both what a product does and what it claims.
Common issues: Some products perform clinical functions but are marketed as wellness; others claim clinical benefit they cannot support. This module gives advisors a two-step test for medical purpose that covers both.
Mental health conditions sit on a spectrum, and the line between subclinical distress and a diagnosable disorder is gradual, contested, and often invisible to the person experiencing it. Someone who today calls themselves stressed, low, or unable to sleep may tomorrow meet full diagnostic criteria.
That is what makes the boundary between a wellbeing tool and a digital mental health technology genuinely difficult to draw, and important to draw correctly.
The distinction is not about pathologising normal life. Wellbeing tools serve a real purpose, and framing a product as wellness rather than treatment can lower the barrier to engagement for people who would otherwise avoid help.
The distinction turns on what the product is actually for: one built to help healthy people through ordinary life sits in different regulatory territory from one addressing established conditions, whatever the label says. Mislabelling is not always deliberate; it can be drift, optimism, or simply not knowing where the line sits.
But the boundary does not move either way, and the two-step test below is how you locate it.
The medical device definition in the UK MDR 2002 and the EU MDR is the only rule in play: a product has medical purpose if it is intended for the diagnosis, prevention, monitoring, prediction, prognosis, treatment, or alleviation of disease, among other purposes. The MHRA's DMHT guidance v1.3 applies this to digital mental health and is explicit that medical purpose does not require diagnostic language.
On classification it states: “DMHTs that provide the probability of mental health conditions or probability of symptoms highly correlated to mental health conditions, where there is a reliance on this to determine clinical management options, may be considered as providing decisive information. As such, they would be considered as providing a ‘direct diagnosis’ as per rule 10 of the EU MDD and should be regulated as a class IIa medical device.”
The full definitions, function tables and worked examples are in the sources below.
The primary source. Section 6 covers medical purpose; Tables 1 and 2 map the device-definition terms and the conditions they cover.
Open MHRA guidance →The legal definition of a medical device and the Annex IX classification rules carried forward from the EU MDD.
Open legislation →Examine what the product does, not what it claims. This step follows the MHRA's own first pass, set out in Figure 3 of the guidance, which asks two questions in sequence; medical purpose attaches only when the answer to both is yes.
A product can pass one and fail the other: a tracker with a clear sports mental performance goal performs a monitoring task but targets no clinical condition, so it has no medical purpose; the same tracker pointed at anxiety levels passes both. (The guidance calls this pair its two-step process for medical purpose; this module's Step 2 below is a different thing, the claim-side analysis of the manufacturer's signals.)
A yes or possibly to both questions means gate one of the MHRA two-gate test is likely met; gate two, sufficient functionality, is covered in Module 2.
Table 1 of the guidance maps the device-definition terms to the tasks advisors actually see.
The task must be aimed at conditions or symptoms at levels considered diagnosable or clinically relevant.
Function is one route to medical purpose. The manufacturer's own claims are another, and the two do not always agree.
The gap between what a product does and what it claims runs in both directions, and both are dangerous.
A product can understate: it performs a clinical function but is marketed as wellness to stay outside the framework. Step 1 catches this, because the function attaches medical purpose whatever the marketing says, and the result is an unregulated medical device on the market.
A product can also overstate: it does little of real clinical substance, or does it poorly and without evidence, but claims a clinical benefit in order to sell. Here the claim itself attaches medical purpose, and the user is left relying on a product for something it cannot safely deliver.
This step assesses the claim side, and the rule is plain: only signals the manufacturer controls can attach medical purpose, because intended purpose is the manufacturer's own statement of what the product is for. That covers the company website, the app store listing, investor and partnership material, and the company's social media.
Voices the company does not control, such as user reviews, journalists, or clinicians recommending the product on their own initiative, do not count unless the company endorses them. The table grades the signals advisors most often encounter.
| Signal | What it indicates, and why |
|---|---|
| Validated clinical instrument (PHQ-9, GAD-7) where every score permits continued use | The product is designed for clinically unwell users. A validated instrument measures clinical severity, and not restricting access on the basis of that severity means the product is intended for people across the clinical range. The MHRA uses this scenario as its own example of inferred medical purpose. |
| Explicit treatment or prevention claim in company content | The company has stated its intended purpose directly. Treatment and prevention are named in the legal definition of a medical device. This is not a signal to weigh; it is a declaration. |
| Deployment in a clinical pathway for patient care: NHS formulary, NICE pathway recommendation, or a commissioned patient-facing clinical service | The company has contracted to serve patients in a clinical context. Note 1 of the guidance lists use or recommendation as part of an NHS service among the aspects that infer clinical intent: commissioned care for named patients with a defined clinical presentation is its clearest case, general staff-wellbeing procurement its weakest. NICE recommendation is the clearest standalone marker. |
| The company celebrates a clinical deployment, for example in an investor deck or on LinkedIn | Publishing clinical use converts third-party activity into the company's own claim of intent. A company cannot celebrate a clinical deployment and later argue the clinical use was incidental. |
| Clinical condition language (depression, anxiety, PTSD) in marketing or in non-validated instruments | Naming clinical conditions identifies a clinical population as the intended user. Weight depends on how consistently and prominently the language appears. Occasional use in a blog post carries less weight than language embedded in product onboarding. |
| Founders or clinical advisors describe the product as a digital therapeutic; outcome claims in clinical metrics (PHQ-9 reduction, relapse rates) | Outcome claims in clinical metrics are a statement of clinical function. A single conference remark carries less weight than language repeated across marketing and investor materials. |
| Trans-diagnostic symptom targeting (worry, rumination, sleep) with no named diagnosis | Not determinative on its own. The question is whether people at clinically relevant severity are implicitly included or actively excluded by design, eligibility criteria, or screening at onboarding. |
| Partnerships with charities serving clinical populations | The deciding factor is whether the partnership puts the product in front of patients as users, or is general brand sponsorship. Distribution context determines the weight. |
| Generic wellbeing language, a founder with a clinical background, or app store category placement | None of these establishes a clinical function. They add context to stronger signals but cannot carry the analysis alone. |
| User reviews and journalist descriptions | Not manufacturer-controlled, so not part of the statement of intended purpose. Irrelevant unless the company later endorses or incorporates the description. |
| Off-label clinician or NHS use the company knows about but has not endorsed | Does not attach medical purpose by itself, but it is not inert either. Note 1 lists use or recommendation as part of an NHS service among the aspects that infer clinical intent, and Example 6 of the guidance expects a manufacturer that becomes aware of clinician recommendation to either regulate the product or act to stop that use. Awareness creates a duty to choose; endorsement converts the use into the company's own claim. |
Strong indicators in either step are enough. A product that performs a diagnostic or decisive-information function has medical purpose through function alone, whatever the marketing says.
A product with weak function but strong manufacturer signals has it through the signals, because the manufacturer has claimed the medical function. Both steps must be run and read together.
The most common real situation is drift: a product launched as wellness that accumulates medical purpose over time, as marketing softens, a partnership is announced, or a clinical deployment is celebrated. The moment the totality of signals crosses the line, the product is a medical device on the market without a pathway.
Denied intent does not change the regulatory position. If a product performs a function the MHRA treats as direct diagnosis or decisive clinical information, and it meets the sufficient-functionality gate, it is Software as a Medical Device.
Placing it on the UK market without registration, a clinical safety case, and the required conformity assessment is placing a non-compliant medical device on the market.
That is a criminal offence under the UK MDR 2002, enforced through the Consumer Protection Act 1987. The MHRA can prosecute, and the penalty on conviction is up to six months' imprisonment, an unlimited fine, or both.
The system relies on the manufacturer's stated intended purpose and runs on a presumption of honesty: that the manufacturer does its due diligence, including taking clinical and regulatory advice, to understand medical purpose and classify accordingly. That is not the same as the MHRA being passive.
It does not accept whatever is asserted and let a wrong classification stand because the manufacturer says it did not realise. Many unregulated products sit on the market today; that is an enforcement-capacity problem, not permission, and it offers no individual protection.
The MHRA can engage at any point, prompted by a complaint, an incident, a procurement query, or its own market surveillance, and the manufacturer then has to explain, cleanly and on the record, why the product was not treated as a medical device.
Form your own view first. Reveal the reference answer to compare reasoning.
A meditation app posts 20 times in a quarter. 18 are wellness-framed (breathwork, focus, sleep tips). 2 say the app helps 'reduce anxiety symptoms'. The posts are written by the company's marketing team. Does medical purpose attach?
A mindfulness app is informally recommended by 3 GPs in one borough to patients with mild depression. The company is unaware. Then the company learns about it and tweets: 'Honoured to be supporting NHS patients.' At which point does medical purpose attach?
A product includes the GAD-7. Scores above 15 trigger a hard block: 'Consider speaking to your GP', and the app's therapeutic features become inaccessible until the score drops below 15. Does medical purpose attach via the GAD-7?
Educational resource. Not formal regulatory or legal advice.
