Medical Device Risk Classification in Depth

Why classification is the hinge of the whole submission

Ask an experienced regulatory team where most device registrations go wrong, and you will rarely hear "the paperwork". You will hear "the class". A device placed in the wrong risk class at the start carries that error through every step that follows: the wrong dossier is built, the wrong evidence is gathered, the wrong fee is paid, and the first the applicant hears of it is a query from the regulator weeks or months later. By then the cost of the mistake is no longer a few hours of thinking; it is a resubmission and a delayed launch.

Classification deserves the same care you would give a contract clause. It is a defined, rules-based exercise with a right answer that you can reason your way to and defend. This article walks through the framework Kenya uses, the rules that actually decide a class, a fully worked and anonymised example from a recent engagement, many examples across all four classes, the mistakes that cost applicants time, and the repeatable method we use to get the class right the first time.

The regulatory basis for classification in Kenya

In Kenya, the Pharmacy and Poisons Board (PPB) is the national regulatory authority for medical devices. It carries out this mandate through its Health Products and Technologies (HPT) department, which handles the evaluation, registration, and post-market oversight of devices placed on the Kenyan market. The legal foundation sits in the Health Act 2017 and the Pharmacy and Poisons Act (Cap 244), which together establish the PPB's authority over health products and technologies, a category that expressly includes medical devices and in-vitro diagnostics.

Kenya does not classify devices through a bespoke local scheme. It uses the internationally harmonised, risk-rules approach first developed by the Global Harmonization Task Force (GHTF) and now carried forward by the International Medical Device Regulators Forum (IMDRF). Regionally, this approach is adopted through the East African Community (EAC) Medical Devices Regulations 2014, which align member states around a common four-class framework. Many African national regulatory authorities also map their systems to the WHO Global Model Regulatory Framework for Medical Devices, the reference model the World Health Organization publishes to help maturing regulators build consistent, evidence-based controls.

The most important thing to understand is that this is a rules-based, risk-based system, not a lookup list. You do not find your product on a published register and read off its class. Instead, you run the device through a defined set of classification rules that turn on how the device interacts with the body: whether it is invasive, whether it is surgically invasive, how long it remains in contact with the patient, whether it is active, whether it delivers or measures, and whether it acts on vital organs or the central circulatory or nervous systems. Two superficially similar products can land in different classes once these factors are applied.

One parallel track is worth flagging early. In-vitro diagnostics (IVDs) are not classified on the same invasiveness logic as physical devices. They run on their own A to D scheme, graded by the combined public-health risk and individual-patient risk of an incorrect result. A test whose false result endangers the wider population, such as a screen used on the blood supply, sits at the top of that scale regardless of how simple the cartridge looks.

The four risk classes, A to D

Kenya's framework sorts every device into one of four classes. Class A is low risk, Class B is low to moderate, Class C is moderate to high, and Class D is high risk. The class reflects the harm a failure of the device could realistically cause to the patient, the user, or the public.

Class A: low risk

Devices that are non-invasive, simple, and pose minimal risk in normal use. Examples: reusable surgical instruments, hospital beds, non-sterile examination gloves, stethoscopes, wheelchairs, and reusable bowls and trays.

Class B: low to moderate risk

Devices that are briefly invasive or single-use sterile items where a failure has limited but real consequences. Examples: hypodermic needles, suction equipment, sterile single-use gauze, basic infusion sets, hearing aids, and wound dressings for injured skin.

Class C: moderate to high risk

Active devices, longer-term implants, and diagnostics where an error carries serious clinical weight. Examples: surgical lasers and electrosurgery units, orthopaedic bone plates and screws (long-term implants that do not contact the central nervous system or heart), infectious-disease screening IVDs, blood glucose monitoring systems, and lung ventilator circuits.

Class D: high risk

Life-sustaining and life-critical devices, central circulatory or nervous-system implants, and the highest-risk diagnostics. Examples: heart valves and coronary stents, pacemakers, implantable defibrillators, HIV, hepatitis, and blood-typing IVDs, drug-eluting stents, and dialysis machines.

These lists are a starting point, not a verdict. The same generic device can move between classes the moment its characteristics change: a non-sterile item supplied sterile, a passive item given a measuring function, an inert item made active, or an implant whose contact duration shifts from transient to long-term. Those triggers are exactly what the classification rules are built to capture, and they are where we turn next.

The rules that actually decide the class

Under the GHTF, now IMDRF, framework adopted into Kenya through the EAC Medical Devices Regulations 2014, a device is not assigned a class by guesswork or by analogy to "similar" products. The class falls out of a set of formal classification rules, and each rule is built on a small number of core factors. Once you know how a device reads against each factor, the rule set does the rest. The four classes run from Class A, the lowest risk, to Class D, the highest. What follows is the anatomy of those rules, factor by factor.

Invasiveness and route

The first question is whether the device is invasive at all, and if so, how it enters the body. A non-invasive device that only touches intact skin or sits outside the body generally starts low, at Class A, and escalates only when another factor is triggered. An invasive device divides into two routes. A device that enters through a body orifice, the mouth, ear, nose, or other natural opening, is treated more gently than one that is surgically invasive, breaching the body surface through an incision or puncture. Surgical access raises the floor immediately.

Duration of contact

How long the device stays in contact with the body is the second escalator. The rules use three bands: transient, less than 60 minutes; short-term, up to 30 days of continuous use; and long-term, more than 30 days. The longer the contact, the higher the risk reading, because a device that lingers has more opportunity to cause harm. The same physical instrument can land in different classes purely on how long it is intended to remain in place.

Where the device contacts the body

Contact site is the sharpest escalator of all. The two structures the rules guard most heavily are the central nervous system (CNS), the brain and spinal cord, and the central circulatory system (CCS), the heart and the great vessels. A device that contacts, monitors, or diagnoses conditions in the CNS or CCS is pushed sharply upward, typically to Class C or Class D, regardless of how benign it would otherwise look. If your device goes anywhere near these structures, assume the top of the ladder until proven otherwise.

Active versus non-active

An active device depends on a source of energy, electrical, mechanical drive, or software, to perform its medical function, as distinct from energy supplied directly by the body or by gravity. Active therapeutic devices that deliver energy into the patient, and active devices that monitor or control vital physiological parameters, escalate above their passive equivalents. The presence of a powered, energy-delivering function is almost always a step up.

Implantable

An implant is intended to remain in the body after the procedure. Implantable devices escalate to Class C as a baseline, and rise to Class D when they sit in the CNS or central circulatory system, or when they are long-term and life-sustaining. Implantation combines duration and location into a single high-risk reading.

Special triggers

Several features each lift a device by a class on their own: a measuring function; supplied in a sterile state; an integrated medicinal substance, where the rules distinguish an ancillary action from a primary pharmacological one; software intended to drive diagnosis or therapy; and the incorporation of animal, human, or biological derivatives. Any one of these can move a device off the floor it would otherwise occupy.

Accessories

An accessory is classified in its own right, not inherited from the device it supports. Non-active accessories such as instrument trays, bowls, and holders are Class A. This is why a sterilisation tray and the implant it carries can sit several classes apart.

The escalation ladder at a glance

A worked example: classifying a consignment of reusable surgical instruments

A Kenyan importer brought us a consignment of roughly thirty reusable stainless-steel surgical instruments to register with the Pharmacy and Poisons Board. The brief was narrow but real: confirm the risk class of every SKU, then organise the lot for an efficient submission. The two tasks are linked. You cannot decide how to file until you know what you are filing, and the classification work is what surfaces the data problems that would otherwise stall the dossier later.

We started by characterising each instrument against the questions that drive classification. Every item in the consignment was manual, meaning hand-operated and non-powered. None was implantable. Each was used in transient contact, in and out of the surgical field within a single procedure. Critically, none of them contacted the central nervous system or the central circulatory system. That last point matters more than any other, because contact with those systems is the single feature that lifts a hand instrument out of the lowest risk band.

Applying the reusable-surgical-instrument rule, the reasoning is clean. A reusable surgical instrument defaults to Class A unless it makes contact with the central nervous system or central circulatory system, is active (powered), or is intended for permanent implantation. None of those triggers applied to anything in the lot. All thirty instruments classified as Class A.

A tidy table can hide where the actual work was. Two genuine judgement calls came up, and both teach something the rule on its own does not.

Judgement call one: bone instruments

Bone instruments, such as a bone-cutting saw or bone-holding forceps, stay Class A only if they are purely manual. The moment a variant becomes powered, or is designed for permanent implantation, it leaves Class A and re-classifies into a higher band. The instrument name on a packing list does not settle this. We held the bone instruments open until the manufacturer confirmed, in writing, that every unit in this lot was manual-only with no powered or implantable variant included. That confirmation is what let them sit in Class A with confidence.

Judgement call two: hollowares

Hollowares such as kidney dishes, gallipots and instrument trays are not invasive devices at all. They are non-invasive accessories. Under the accessories rule they take Class A, and we listed them explicitly rather than dropping them, so the submission accounts for the full consignment and nothing reads as missing during review.

Grouping SKUs into device families

The PPB permits family-based registration. A single family application can cover multiple models that share the same intended use, technology, manufacturer and risk class. The benefit is direct: instead of filing thirty separate applications, you file a handful, which cuts both the fee load and the review burden, and gives the Board one coherent submission to assess rather than thirty fragments.

We consolidated the roughly thirty instruments into four families.

Forceps family (clamp-style)

Scissors family

Retractors family

The rule for splitting families is just as important as the rule for forming them. Models go together only when they genuinely share intended use, technology, manufacturer and risk class. When models differ in manufacturer, or one carries a higher-risk feature such as a powered or implantable variant, they go into a separate family rather than being forced into a group where they do not belong. Forcing a mismatched item into a family is the kind of shortcut that draws a query and slows the whole application.

Splitting also buys you flexibility on timing. Families that share a manufacturer and have complete paperwork can be filed now. A family with a different manufacturer, or one still waiting on a certificate or a manual-only confirmation, can be filed in a separate, later application window so it does not hold up the families that are ready. The ready instruments reach the market while the incomplete ones catch up.

Worked examples across all four classes

Class A (low risk)

Class A is the resting state for devices that touch the body only on the outside or transiently, do not run on a power source in a way that matters to the patient, do not stay in the body, and take no measurement that drives a clinical decision. If none of the escalators apply, the device stays here.

Class B (low to moderate risk)

One feature, invasiveness, a sterility claim, or the channelling of body fluids, is usually enough to lift a device one step from the baseline. The device is still simple, but a failure now reaches inside the body or relies on a sterility guarantee the importer must stand behind.

Class C (moderate to high risk)

Now the device either delivers energy at levels that can harm, stays in the body long term outside the most critical sites, or produces information on which serious clinical decisions turn. An error here does real damage, although usually with some time or warning before harm becomes irreversible.

Class D (highest risk)

Class D is reserved for devices whose failure tends to be immediately life-threatening, that contact the most critical sites, that sustain life, or that combine a device with a medicine where the medicine does the main work. These attract the deepest scrutiny because there is little margin if anything goes wrong.

Common misclassification mistakes

Most rejected or delayed applications we see do not fail on paperwork. They fail because the product was placed in the wrong class at the very start. The traps below are the ones worth checking your dossier against before you file.

• Assuming every surgical instrument is Class A. A single powered or implantable variant hidden inside a mixed instrument lot quietly carries a much higher class; review each line item, not the family name.
• Ignoring the sterility claim. A generic single-use item that would be Class A jumps to Class B the moment it is supplied sterile; check the label and the marketing claim, not just the material.
• Overlooking a measuring function. A device that displays a quantified clinical value is treated as a measuring device and escalates accordingly; confirm whether any reading drives a decision before you call it simple.
• Forcing mixed-risk models into one family to save on fees. Bundling a higher-risk model under a lower-risk sibling invites a query or a rejection; classify each model on its own merits and group only genuinely equivalent variants.
• Classifying an accessory by its parent device. An accessory is classified in its own right, not inherited from the system it plugs into; assess what the accessory itself does to or for the patient.
• Missing that CNS or central-circulatory contact is the sharpest escalator. Contact with the central nervous system, the heart or the central circulation almost always lands a device in the top class regardless of how simple it looks; map the contact site first.
• Treating an IVD on the device scale. In-vitro diagnostics run on their own four-tier rule keyed to the consequence of a wrong result, not on the general device rules; switch to the IVD scale before you classify.
• Copying a classification from another market. A class assigned in another jurisdiction may not match the EAC and PPB rules, which differ in their cut-offs; verify against the Kenyan and East African Community framework rather than trusting an overseas certificate.
• Writing a vague intended-use statement. Loose wording lets a reviewer assume the broadest, highest-risk reading of what the device does; write a precise, bounded intended-use statement that claims exactly what you can support and no more.

A repeatable method: how we classify, step by step

Classification is not a single judgement call. It is a disciplined sequence that produces the same answer no matter who runs it. Here is the working method we apply to every portfolio, from a handful of models to several hundred.

1. Build a clean master list. Capture every SKU or model with its product code, commercial name and a clear intended-use statement. Resolve duplicates, re-badged variants and obsolete codes before you count anything, so the list reflects real, distinct products.
2. Write an intended-use statement per item. One precise sentence describing what the device is for, how it contacts the body and for how long. This single sentence drives the class more than any other factor, so it is worth getting exactly right.
3. Group into tentative families. Cluster similar models by intended use, technology, manufacturer and risk profile. Keep mixed-risk models apart; if two models in a proposed family would land in different classes, they are not one family.
4. Run the Class A test. A family is Class A only if EVERY criterion holds: non-invasive or transient contact only, non-active, non-implantable, no measuring function affecting diagnosis or therapy, supplied non-sterile or with routine reusable sterilisation only, contact only with intact skin or external surfaces, no integrated medicinal substance, no animal or human derivative, no central nervous system or central-circulatory contact, and consistent risk across the family.
5. Test the Class B conditions. If any Class A criterion fails, check the Class B qualifiers, for example surgically invasive transient use, short-term orifice use, a sterile single-use generic device, an active fluid-channelling device, a reusable instrument sold sterile for transient tissue contact, or a dressing on injured skin.
6. Check the Class C triggers. Long-term surgical invasion outside the central nervous system and heart, active therapeutic energy at harmful levels, active monitoring of vital parameters, a moderate-risk IVD, a medicinal substance with ancillary action, or moderate-risk diagnostic or therapeutic software.
7. Check the Class D triggers. A central nervous system or central-circulatory implant, a long-term implant, an ionising-radiation therapy device, life-support, a high-risk IVD such as HIV, hepatitis or blood-typing assays, a drug-eluting or combination product where the medicine has the primary action, or an active biological derivative.
8. Record the rationale and sign it off. For each family, write a one-line rationale and the rule used, then have it reviewed and signed before submission. This becomes your classification decision record and your defence if a query ever arrives.

The output of the method is not a hunch but a record. For every family we capture a fixed set of fields, so the reasoning is auditable months later and consistent across the whole portfolio.

From class to dossier: what the decision unlocks

Once the class is agreed, it sets the shape of everything that follows. Classification is not paperwork for its own sake; it is the switch that determines how much you build and what you must prove.

The class fixes the dossier depth. Submissions follow a common technical document structure organised into Modules, and a higher class means more Modules populated in more detail, from administrative information through to clinical and performance evidence. A Class A device carries a light dossier; a Class C or D device carries a substantial one.

The class also fixes the evidence set you must assemble. Across the classes this typically includes an ISO 13485 quality-management certificate, a manufacturer authorisation letter, a free sale certificate or certificate to a foreign government, a declaration of conformity, instructions for use, and labelling artwork. Higher classes pull in deeper safety and performance data on top of this core set.

It then drives the application structure on the PPB PRIMS portal. Families are submitted together, so a clean grouping at the classification stage means fewer, tidier applications rather than a scatter of single-model filings. The class also lands the submission in a particular fee band, and it determines whether a foreign manufacturer needs to appoint a Local Technical Representative to hold the registration in Kenya and act as the point of accountability with the Board.

This is why the class has to be right first time. Set it too high and you over-build the dossier, paying for studies, documents and a fee band the device never required. Set it too low and you under-build it, inviting queries, resubmissions and weeks of avoidable delay. Both errors cost money; one of them also costs time you cannot get back.

Getting the class right at the outset is the single cheapest piece of insurance in the whole submission. A few hours of disciplined classification, recorded and signed off, protects every cost and timeline downstream.

Frequently asked questions

Is medical device classification in Kenya done per SKU or per device family?

You classify every SKU, but you can register by family. The PPB permits a family application to cover several models that share the same intended use, technology, manufacturer and risk class. You still confirm each model's class; you simply file the equivalent ones together to save fees and review time.

Are reusable surgical instruments always Class A?

Usually, but not automatically. A reusable surgical instrument is Class A unless it contacts the central nervous system or central circulatory system, is active or energy-delivering, or is designed to be implanted. A powered variant, a sterile-supplied claim, or a measuring function can lift it out of Class A, so each line item is checked rather than assumed.

Who makes the final classification decision?

The Pharmacy and Poisons Board makes the final determination after reviewing the full dossier. The applicant's job is to arrive with a correct, documented and defensible rationale. A clear classification record is the best protection against a reclassification query.

How is an in-vitro diagnostic classified differently?

IVDs run on their own A to D scale based on the public-health and individual-patient risk of a wrong result, not on invasiveness or contact duration. A blood-supply screening assay sits at the top of that scale even though the cartridge is simple to handle.

Does a classification from another country carry over to Kenya?

Treat an overseas class as a useful reference, not a decision. Kenya applies the EAC Medical Devices Regulations 2014 and the PPB rules, whose cut-offs may differ from another market. Always re-run the device against the Kenyan framework before you file.

Related reading and tools

• Medical Device Classification Kenya: practical guide with real examples
• Free medical device classifier tool
• Dossier and eCTD preparation
• Local Technical Representative for foreign manufacturers

References

• East African Community, Medical Devices Regulations, 2014.
• Global Harmonization Task Force, Principles of Medical Devices Classification, GHTF/SG1/N77:2012 (carried forward by the International Medical Device Regulators Forum, IMDRF).
• World Health Organization, Global Model Regulatory Framework for Medical Devices including in-vitro diagnostic medical devices, 2017.
• Pharmacy and Poisons Board (Kenya), Health Products and Technologies, Medical Devices evaluation and registration procedure, web.pharmacyboardkenya.org.
• Republic of Kenya, Health Act, 2017; Pharmacy and Poisons Act, Cap 244.

This article is general regulatory guidance based on a real, anonymised engagement and does not identify any client. It is not a regulatory determination; the final classification of any device rests with the Pharmacy and Poisons Board. Written by the team at The Pharma Partner (Pharma Verixa Limited), a pharmacist-led PPB compliance and Local Technical Representative practice in Nairobi, Kenya.