Medical Device Design Transfer

Medical Device Design Transfer connects an established device design to the specifications, people, equipment, materials, and acceptance activities used in production. Its practical purpose is not merely to send engineering files to a factory. It is to demonstrate that the receiving organization understands what to manufacture, can execute the intended process, and can produce evidence showing whether the resulting devices meet their specified requirements.

The broader framework belongs to Design Controls. This guide focuses on the design-to-production interface: translating design outputs into usable manufacturing instructions, assigning responsibilities, checking readiness, and preserving the connection between the approved configuration and actual execution. Transfer may involve an internal production department, another site, a contract manufacturer, or several organizations performing different operations.

A useful starting question is: could a trained production team build and evaluate the device without relying on undocumented engineering knowledge? If the answer depends on someone remembering an assembly trick, selecting a firmware file from a personal folder, or interpreting an ambiguous drawing, the handoff still contains unresolved work. Those gaps belong in the transfer plan, not in informal shop-floor conversations.

Separate three decisions that schedules often combine. Design evidence addresses the device design. Manufacturing readiness addresses the ability to execute the defined process. Product disposition addresses whether particular units or lots can proceed or be released. Evidence can support more than one decision, but approval of one does not automatically establish the others. Keeping these decisions distinct makes restrictions and unresolved issues easier to understand.

The tangible output is a controlled baseline plus evidence that the intended operation can use it. A transfer report should therefore identify its scope, supporting records, remaining restrictions, decision authority, and the conditions that would require reassessment.

For United States operations, begin with the current regulation rather than a legacy design-control checklist. FDA states that the Quality Management System Regulation became effective on February 2, 2026, incorporating ISO 13485:2016 by reference. Current 21 CFR 820.10(a) requires a manufacturer subject to Part 820 to document a quality management system complying with applicable ISO 13485 requirements and other applicable requirements of Part 820.

The scope in 21 CFR 820.10(c)) is specific: manufacturers of class II, class III, and the class I devices identified in paragraph (c)(1) and table 1 to paragraph (c)(2) must comply with Design and Development, Clause 7.3 and its subclauses in ISO 13485. Do not reduce this to either ‘all devices’ or ‘only class II and III devices.’

• The class I scope includes devices automated with computer software.
• The listed devices are Catheter, Tracheobronchial Suction (868.6810); Glove, Non-powdered Surgeon's (878.4460); Restraint, Protective (880.6760); System, Applicator, Radionuclide, Manual (892.5650); and Source, Radionuclide Teletherapy (892.5740).

Historical 820.30 is not the current design-and-development provision. The current Part 820 text reserves sections 820.20–820.30. Legacy procedures may still contain useful technical content, but their regulatory references need to be understood in the current framework. The incorporation of ISO 13485 should not be confused with a requirement to use the particular transfer forms or gate names proposed here.

FDA's QMSR frequently asked questions explain that investigators may review QMS records created before February 2, 2026. FDA also says a comparative analysis may be useful for demonstrating how earlier documents and records meet QMSR requirements. That is an agency suggestion, not a mandatory transfer-document format.

As a practical first step, record the device classification, relevant operations, receiving sites, and applicability assessment in the transfer plan. This prevents a generic checklist from silently substituting for a product-specific determination.

A transfer coordinator can organize the work without becoming the technical authority for every deliverable. As a practice, assign both a preparing owner and a receiving reviewer for each important output. Engineering should not simply declare that production is ready, and manufacturing should not silently reinterpret design intent to make an assembly easier to build.

The following allocation is an example, not a prescribed organizational structure. Smaller manufacturers may combine roles, while larger programs may divide them among sites and specialist teams. What matters is that the decision rights, evidence responsibilities, and escalation route remain explicit.

Give each handoff a clear acceptance statement. For example, manufacturing accepts an assembly instruction only after confirming that the listed tools exist, component orientation is unambiguous, and the specified inspection can be performed at the intended station. Acceptance of the document is different from approval of the device design or release of units produced with it.

Supplier and contractor interfaces need the same clarity. Specify who approves substitutions, investigates failures, owns process changes, retains execution records, and communicates deviations. A purchase order alone may not express these operational boundaries. A practical quality agreement can organize them, provided its content matches the actual work and established responsibilities.

Record disagreements as decisions awaiting resolution, not as vague action items. ‘Confirm fixture suitability’ is weaker than ‘manufacturing engineering will evaluate fixture F2 against drawing revision C before pilot authorization.’ The latter identifies an owner, an object, an evidence expectation, and a decision boundary.

A production specification package should tell the receiving team what is approved and how that approval becomes an executable operation. As a practice, build the package around relationships rather than a folder count. A drawing, bill of materials, software image, work instruction, inspection method, and label can each be approved individually while still describing incompatible configurations.

Start with a characteristic-to-control map. Identify the design characteristic, the production document that preserves it, the activity that controls or evaluates it, and the resulting evidence. The table illustrates useful connections; it is not a universal list of required records.

A bill of materials deserves particular attention. Engineering and manufacturing structures may differ because production needs intermediate assemblies, consumables, or packaging levels. Document those differences and reconcile quantities, units of measure, component identities, and permitted alternatives. Do not assume that matching finished-device part numbers proves that the two structures are equivalent.

Instructions should distinguish a target setting from an acceptance limit. A machine setpoint, an allowed process range, and a product specification answer different questions. For example, recording that a fastening tool reached its programmed setting does not necessarily establish every relevant property of the resulting joint. Define the supporting evidence appropriate to the characteristic and process.

Use document control to establish revisions, approval status, effective dates, and removal of superseded instructions from use. Before a pilot build, perform a desk-to-station walkthrough: select one device configuration and follow every referenced document through the actual work area. Missing references and contradictory units become visible before they affect production.

Staged transfer allows a team to learn without pretending that early builds establish routine production readiness. In this practice framework, every gate authorizes a defined next activity. The authorization names the configuration, site, equipment, personnel conditions, and intended use of the resulting units. It also states what remains prohibited, such as distribution or use of an unassessed alternative component.

Agree on entry criteria, evidence expectations, and stop conditions before execution. Otherwise, a pilot that uncovers missing instructions can be retrospectively described as successful simply because some units eventually passed inspection. Learning is valuable, but it should not be confused with satisfying a readiness criterion.

Evidence reuse can be sensible when its relevance is demonstrated. A receiving site may be able to use an existing method study, but should evaluate differences in equipment, fixtures, materials, operators, or environment. Record why the evidence remains applicable and what additional work addresses the differences. Reuse should be a reasoned decision, not an assumption based on a familiar document title.

For each gate, write a short decision statement supported by record identifiers. Distinguish a pilot-only authorization from unrestricted production readiness, and distinguish both from permission to release product. This makes staged progress visible without weakening the meaning of the final handoff.

Supplier readiness is not just whether an organization appears on an approved list. As a practical transfer question, ask whether approval covers the actual material, component revision, manufacturing site, process, and intended use involved. A supplier may be suitable for prototype machining without having demonstrated the controls or capacity needed for the intended production operation.

Define purchase specifications that preserve the characteristics on which the design depends. Where relevant, these may include material grade, finish, dimensions, cleanliness, packaging, storage conditions, or software revision. Make alternatives explicit. A commercially equivalent part description is not, by itself, evidence that a substitution preserves the relevant device characteristics.

Connect supplier qualification to incoming acceptance. Identify what the manufacturer evaluates directly, what evidence it obtains from the supplier, and how discrepancies are handled. A certificate should be associated with the supplied lot and specification. Review whether its statements and results actually address the planned acceptance decision rather than treating the presence of a certificate as the decision itself.

For an outsourced process, transfer the boundaries as well as the instructions. Identify the incoming condition, permitted configuration, process specification, expected output, required records, and escalation route. Discuss how the supplier will communicate equipment changes, site moves, material substitutions, unexpected process results, or subcontracting. These arrangements are practice recommendations to tailor to the relationship and applicable requirements.

A useful supplier exercise is to trace one representative delivery backward and forward. Starting with a received component lot, retrieve its specification, associated supplier evidence, incoming decision, storage status, and the builds that consumed it. Then simulate a supplier notification affecting that lot. Determine whether the team can identify affected work in progress and finished units without reconstructing the chain from personal emails.

Close the loop before transfer approval. Unresolved supplier questions need an owner and a disposition: obtain evidence, restrict use, change the source, or postpone the affected activity. Schedule pressure is not a substitute for understanding the dependency, especially when a single outsourced step can affect several downstream product configurations.

Readiness evidence becomes stronger when the intended people use the intended instructions with the intended equipment. As a practice, avoid evaluating these elements in isolation. An approved instruction is not necessarily usable; an available machine is not necessarily configured correctly; a completed training acknowledgment does not necessarily demonstrate competence at a consequential task.

Plan training around activities and revisions. Use a training record to connect the individual, task, instruction version, completion date, and assessment appropriate to the work. For a demanding assembly or inspection, observed execution can reveal misunderstandings that a reading acknowledgment would miss. Include supervisors, inspectors, maintenance personnel, temporary staff, and other affected roles where relevant.

For equipment, check identification, setup, tooling condition, maintenance arrangements, and applicable calibration status. For measurement, check that the method and instrument are suitable for the characteristic and decision. A limit expressed more precisely than the measurement can meaningfully distinguish is a warning sign. Link relevant measurement system analysis to the actual method rather than merely filing a generic study.

Assess process-validation needs under the applicable quality system. Practical process validation planning should consider what subsequent inspection can establish, what it cannot establish, and the consequences of process variation. A successful pilot or equipment acceptance test should not automatically be relabeled as validation evidence without assessing its purpose, conditions, and adequacy.

Finally, define acceptance before testing: method, limits, sampling rationale where used, data recording, failure handling, and disposition authority. Preserve initial failures and authorized retests. Repeating a test until a passing value appears is not a defensible substitute for understanding the original result.

A transfer baseline should be recoverable after production begins. As a practice, define a baseline identifier that connects the approved product configuration with its production documents. Capture the baseline actually used for each build. A link that always opens the newest instruction is insufficient for explaining an earlier operation performed under a different revision.

The relationship diagram below is an example evidence structure, not a prescribed database schema. Identifiers can be maintained in controlled paper records or electronic systems. The important feature is that each connection answers a concrete question about what was authorized, what happened, and how the result was evaluated.

Apply change control to the relationships, not just the drawing. A changed component can affect purchasing, incoming inspection, assembly, labeling, software compatibility, or existing evidence. Define the first affected work order or other effective boundary, and decide what happens to inventory and work already in progress.

When evaluating medical-device manufacturing software, ask vendors to demonstrate historical revision retrieval, lot relationships, training applicability, exception handling, and disposition boundaries. These are evaluation questions, not assumptions about any product. Test the proposed configuration with representative records and failure cases before relying on it for the intended workflow.

Packaging and labeling belong inside the transfer boundary when they are part of the operation. Under 21 CFR 820.45, manufacturers must document and maintain procedures describing activities to ensure labeling and packaging integrity, inspection, storage, and operations under the conditions specified in the regulation. Section 820.45(c) also requires operations that prevent mixups, inspection before use, and documented labeling inspection results.

As a practical handoff, connect the device configuration to the correct artwork, packaging materials, instructions, and inspection criteria. Walk through issuance, workstation setup, changeover, rejected labels, and unused materials. Verify that a label cannot be considered correct merely because its printing is legible; its identity and content also need to match the intended product configuration.

Sterile devices add interfaces that should be planned explicitly. FDA's 2024 sterility submission guidance is nonbinding guidance for 510(k) review, with a scope limited to devices labeled sterile that undergo industrial terminal sterilization based on microbial inactivation. It should not be treated as a universal transfer procedure for every sterile or aseptically manufactured product.

FDA's ISO 11607-2 recognition entry describes a scope covering development and validation of packaging processes for terminally sterilized devices, including forming, sealing, and assembly. Its ASTM F1980-21 recognition entry identifies machine-process validation, sterilization effects, and distribution events as outside that aging guide's scope. Aging evidence therefore should not be used as a catch-all answer to packaging-process readiness.

In practice, map the relationships among package configuration, sealing equipment and settings, sterilization configuration, transport conditions, and shelf-life evidence. Identify which configuration each study supports and assess later changes against that scope. A revised pouch, loading arrangement, or sealing station deserves an explicit impact assessment rather than an assumption that earlier evidence automatically applies. Keep the product-specific acceptance and release decision separate from completion of the packaging workstream.

Consider a fictional nonsterile handheld electronic instrument transferred to a receiving assembly line. The pilot baseline identifies hardware revision C, firmware 2.4.0, instruction revision D, and fixture F2. All units are designated for engineering evaluation, not distribution. The example uses three runs for illustration; neither that run count nor the numerical criterion is presented as a regulatory requirement.

At G1, the team authorizes 120 units across three runs after checking the build package, material status, training, and equipment prerequisites. Before execution, it sets an internal readiness criterion of at least 97% first-pass yield, together with closure of issues that could undermine repeatable manufacture. The yield criterion supplements, rather than replaces, product acceptance and technical assessment.

First-pass yield is 114 ÷ 120 × 100 = 95.00%. Final conformance yield after authorized rework is 118 ÷ 120 × 100 = 98.33%, rounded to two decimals. Scrap is 2 ÷ 120 × 100 = 1.67%. Reconciliation is complete: 114 first-pass units + 4 successfully reworked units + 2 scrapped units = 120 units started.

The six initial failures concern latch-fit acceptance. Investigation identifies a fixture-location problem; four units are recovered through an authorized rework route and pass the specified reevaluation, while two are scrapped. The team records the original failures, the investigation, rework instructions, repeat results, and unit dispositions. Passing after rework does not erase the initial process signal.

At G2, readiness remains on hold: 95.00% is below the predefined 97% first-pass target, and the fixture issue needs controlled correction and supporting follow-up evidence. The team assesses effects on instructions, training, and earlier results before planning further builds. The 118 conforming units remain engineering units. Their acceptance results neither authorize distribution nor demonstrate statistical sufficiency for routine production.

Use the following practice checklist as a working decision tool, not a retrospective document collection exercise. For every row, record an owner, evidence reference, status, and closure date. Permit ‘not applicable’ only with a reason tied to the device and transfer scope. A green dashboard is useful only if its underlying decisions can be explained.

Do not average away a critical gap. Ninety-nine completed checklist items cannot compensate for an unresolved issue that prevents a reliable acceptance decision. Distinguish readiness blockers from follow-up improvements, and document the basis for that distinction. Administrative cleanup should not disguise missing technical evidence, and a restricted authorization should say exactly what activity it allows.

After the handoff, monitor a focused set of indicators: first-pass yield, recurring defects, rework, instruction deviations, supplier discrepancies, and configuration errors. Define who reviews them and what triggers escalation. A change in supplier site, tool, software, material, or acceptance method may call for reassessment of affected transfer evidence; it does not necessarily require repeating every activity.

Close transfer with a concise decision record that identifies the accepted scope, evidence package, remaining restrictions, and ongoing owner. Then verify retrieval using one actual build. The practical test is whether another qualified person can reconstruct its configuration, execution, exceptions, and disposition without relying on the original project team's memory.