CompanyGraph
A medical device supplies a clinical action only when its design, materials, sterility or cleanliness, calibration, packaging, users, maintenance, and evidence remain connected from manufacture to the patient.
A device enables a clinical action
A syringe enables delivery of a measured dose, an implant supports tissue, an infusion pump controls flow, and a diagnostic instrument turns a sample into information. The patient needs that function safely and at the required time, not a unit count in a warehouse. A device becomes clinically useful only when the product, accessories, software, instructions, trained people, and care setting work together.
The journey begins with a clinical need, then passes through design, materials, tooling, assembly, software, testing, packaging, sterilization where required, release, distribution, hospital receipt, use, maintenance, complaint handling, recall, and disposal. Each stage creates evidence and closes some options.
Risk classification changes the route
Regulatory classification reflects the risk and intended use of a device. In the United States, FDA describes Class I, II, and III routes with different controls and evidence; classification is tied to the device's intended use, not merely its material or price. FDA classification guidance shows why a familiar object can follow a different route when its clinical claim changes.
Classification does not itself make the device safe. It determines which design controls, performance tests, submissions, manufacturing controls, and post-market duties apply. A cleared or approved device can still be misused, poorly maintained, or incompatible with the surrounding system.
This is why a hospital's stock count is not a clinical-capacity count. Ten pumps without the approved tubing, drug library, charging station, preventive service, or trained staff do not provide ten usable infusions. A procurement record can show that the hardware arrived; only installation, calibration, maintenance, and a procedure-specific check establish the local function.
Materials become a qualified surface and interface
Material choice affects strength, fatigue, corrosion, wear, tissue response, electrical insulation, and cleanability. A polymer, metal, ceramic, adhesive, coating, battery, sensor, or software component has to be joined into a device whose interfaces remain within tolerance. Substituting a resin, supplier, adhesive, or sterilization method can change performance and may require validation or regulatory review.
The finished object also depends on manufacturing history. Injection-moulding parameters, bond cure, machining, calibration, software version, and inspection results are not physically visible in a product photograph. A certificate can connect a lot to a defined test; it cannot substitute for a function test in the patient's actual environment.
Sterilization can be part of manufacture
Sterilization is not a decorative final step. Steam, ethylene oxide, radiation, or other methods must achieve the required microbial reduction without damaging the device, packaging, label, or residue limits. FDA's overview of medical-device sterilization describes the product-specific validation problem.
Reusable devices follow a different route: cleaning, inspection, packaging, sterilization, storage, and reprocessing happen repeatedly. A sterile barrier can be intact while the device is wrong for the procedure; a correctly selected device can be unsafe after a failed cleaning cycle. The record and the physical condition must be kept together.
Packaging preserves sterility until use
A sterile barrier protects a defined condition during transport and storage. Seals, pouches, trays, desiccants, labels, and expiry dates all matter. A package can pass release inspection and later be crushed, punctured, wetted, or opened. A lot number supports a recall, but it does not prove the package remained intact after leaving controlled custody.
Hospitals add their own conditions: receiving inspection, stock rotation, storage temperature, battery charging, software updates, calibration, cleaning, and staff training. An installed device is therefore a system in a location, not simply a box delivered from a distributor.
The hospital receives an installed system
An infusion pump requires the right tubing, drug library, power, alarm settings, maintenance, and user training. An imaging system needs software, shielding, cooling, calibration, and specialist interpretation. An implant needs surgical technique, compatible instruments, patient anatomy, and follow-up. The same model can deliver different clinical results in different settings.
Money and reimbursement shape reachable maintenance. FDA's servicing report notes that many adverse-event reports do not contain detailed servicing histories, while hospitals still need contracts, technicians, parts, and downtime to keep equipment usable. The FDA report describes the information gap; the physical consequence is that a hospital may know preventive calibration, spare batteries, or a software update would reduce risk yet lack the service arrangement or budget authority to perform it. A device may remain physically present while becoming clinically unavailable because the accessory, trained staff, or approved software is missing.
Records establish different boundaries
A design dossier states intended performance. A batch record documents production and test results. A sterilization cycle record observes a defined load and process. A shipping log records custody. A maintenance record reports work performed. A complaint describes an event from one user or patient. These are not interchangeable observations.
A device label communicates intended use, warnings, lot, and expiry. It does not establish present calibration or a patient's outcome. FDA's recall database can identify affected products and actions, but a recall notice is not a repair in every hospital. The cause may be a design, supplier, process, software, installation, or use condition that remains to be reached.
A recall travels through an installed base
When a failure signal appears, correction requires more than notifying a manufacturer. The signal must reach distributors, hospitals, clinicians, and patients; affected units must be identified; alternatives or repairs must be available; and use may need to stop before replacement arrives. FDA's recall explanation distinguishes correction in place from removal and describes inspection, repair, adjustment, relabeling, destruction, and patient monitoring, but the physical work depends on local inventory, records, staff, and money.
Detection may occur years after manufacture. If serial numbers were not recorded at installation, a hospital may need to inspect every unit. If the failure is intermittent, a bench test may pass while the device fails under clinical load. A reported problem becomes useful only when it reaches a cause that can still be changed.
After use, device and data diverge
Explantation, cleaning, refurbishment, reuse, recycling, and disposal preserve different things. A reusable instrument can retain geometry and function; a single-use device may be contaminated; an implant contains patient history that must not travel with an unsecured device. Batteries, electronics, sharps, polymers, metals, and biological residues require different routes.
Disposal records can establish that a handler accepted a defined waste stream. They do not prove that every component was destroyed, recycled, or rendered safe without the required chain of custody. Clinical responsibility also includes the patient's information and the knowledge gained from a failure.
Inside CompanyGraph
Inside CompanyGraph, map designers, material suppliers, contract manufacturers, sterilizers, distributors, hospitals, service firms, regulators, and disposal handlers. The graph can identify handoffs and installed-base relationships; it cannot itself observe sterility, calibration, patient context, or the authority and resources available at the bedside.