Hands-on or Hands-off? Outsourcing MDR/510(k) Work

Medical professional undertaking med tech compliance paperwork

Most MedTech founders treat outsourcing their cybersecurity evidence as a single switch: build it all in-house, or hand it all over. It is neither. Treating a spectrum as a switch is what causes the trouble, and that trouble usually shows up late: a stalled submission, duplicated work, or a gap nobody owned until a reviewer found it. 

This article is for founders and regulatory leads at Software as a Medical Device (SaMD) and connected device companies deciding how much of their security evidence, verification and validation (V&V), and documentation to build in-house, and how much to bring in outside support for. That evidence sits behind an EU Medical Device Regulation (MDR) submission or a US Food and Drug Administration (FDA) 510(k), the FDA’s premarket clearance route, so getting the split right matters well beyond the security team. 

Key Takeaways

  • It’s a spectrum, not a switch. Support can sit anywhere from light-touch advisory input to a fully managed evidence package. Most teams land somewhere in the middle. 
  • Outsourcing delivery doesn’t outsource accountability. As the manufacturer, you remain legally responsible for what’s submitted, whoever built the evidence behind it. 
  • The right split is about capacity, not preference. The question isn’t how hands-on you’d like to sound, it’s where your team’s time is genuinely best spent right now. 
  • Clear boundaries matter more than the split itself. Whichever tier you choose, both sides need to agree in writing exactly who owns what. 
  • The split should move as you grow. A sensible arrangement for a pre-revenue team on its first submission is often the wrong one two years and a funding round later. 

Why this matters now

The evidence expected in a MedTech cybersecurity submission keeps growing. A Software Bill of Materials (SBOM), a post-market vulnerability monitoring plan, and a secure update and patching process are now baseline expectations for FDA cyber devices under Section 524B of the US Federal Food, Drug, and Cosmetic Act, reflected in the FDA’s premarket cybersecurity guidance (updated 3 February 2026). On the EU side, MDR’s General Safety and Performance Requirements (GSPR) 17.2 and 17.4 call for the same kind of ongoing, evidenced security work folded into your ISO 14971 risk file. None of this is a box you tick once before submission. It’s sustained work across the whole product lifecycle. Our companion piece on choosing between full conformity and reference use of IEC 81001-5-1 covers the standards side of this in more depth. 

Small SaMD teams are usually clinically led, understandably. That means the security-specific slice of this work, the part with nothing to do with whether the device works clinically and everything to do with whether it can be trusted on a network, often has no natural owner inside the business. That’s the actual gap this decision is about. 

The outsourcing spectrum, not a binary

Once you accept it’s a spectrum, the useful question is which point on it fits your team right now. Three broad tiers cover most of what we see. 

Tier What we do What you keep Best fit when 
Advisory Review specs, answer standards questions, spot-check work Technical file, risk file and V&V, written and run by your team Security-literate team that mainly needs a second opinion 
Blended Lead on the parts you lack capacity for, usually V&V and evidence write-up Design decisions, clinical risk ownership, final sign-off Capable team under deadline pressure, no spare capacity for compliance work 
Fully managed Own the V&V and evidence package for cybersecurity end to end Product ownership, clinical safety file, final sign-off Pre-revenue or lean team, wholly focused on the clinical problem 

Who owns which artefact when you outsource

The tiers above describe how much support you bring in. What matters day to day is more specific: for each artefact and each decision, exactly one side is accountable. The table below takes the core cybersecurity artefacts a submission relies on and shows how ownership typically shifts across the three tiers. Read it as a starting point for your own boundary document, not a fixed rule. 

Notice the bottom two rows. Clinical and product risk ownership, and final sign-off, stay with you at every tier. That is the practical shape of the accountability point: a partner can build the evidence, run the testing, and draft the plans, but the manufacturer still owns the decisions and the submission. This is the same shared-responsibility thinking that applies to your live system, where some controls are yours, some sit with the hospital, and a few are genuinely shared. Undocumented assumptions become disputes. 

Artefact or decision Advisory Blended Fully managed 
Security architecture and system boundaries You draft, we review We co-draft, you approve We own, you approve 
Threat model and risk register You draft, we spot-check We lead, you validate the assumptions We own, you approve 
SBOM and vulnerability monitoring You own, we advise on tooling We build the process, you run it We own and run it 
Verification and validation (V&V) evidence You write and run, we sanity-check We write and run, you review We own it end to end 
Secure update and patch policy You own We draft, you commit to the timelines We own, you resource the fixes 
Vulnerability disclosure and incident response You own We draft, you operate We own, you are the escalation point 
Clinical and product risk ownership (ISO 14971) You own You own, never outsourced You own, never outsourced 
Final sign-off and Declaration of Conformity You own You own, never outsourced You own, never outsourced 

What determines where you sit on the outsourcing scale?

  • Internal security expertise. Have your engineers built this kind of evidence before, or would this be their first time? 
  • Timeline pressure and submission stage. A hard deadline with no slack argues for more support, not less. 
  • First submission or mature programme. First-time submissions carry more unknowns. Teams on their second or third product often need less hands-on help. 
  • Cost versus opportunity cost. Weigh the day rate of support against what your own engineers aren’t doing while they build compliance evidence instead. 
  • How settled your architecture is. Early-stage products with a design still in flux usually need more hands-on support than a mature, stable one. 

A pattern we see often: a SaMD company approaching its first FDA 510(k) submission had a software team fully committed to clinical validation, against a fixed deadline. Rather than diverting engineers onto V&V evidence they’d never built before, they asked us to take that work on directly, while keeping design decisions and clinical risk ownership firmly in-house. The point wasn’t that they couldn’t have done it themselves. It was that the fastest route to a stronger submission was letting each side do what it already does well. 

What to check before you bring anyone in

Choosing a tier is the easy half. The harder half is agreeing the boundary before any delivery work starts, because most outsourcing problems are boundary problems, not competence problems. A short checklist saves a lot of rework. 

Ask a prospective partner to show, not tell. Good external support can point to evidence they have produced that survived a real review, name the standards they work to (IEC 81001-5-1, ISO 14971, AAMI TIR57), and describe how they hand work back to your team rather than locking it in. Be wary of anyone who implies they can take your regulatory accountability with them. They cannot. 

Write the boundary down before work starts. For each artefact and decision in the table above, record who drafts it, who reviews it, who approves it, and where the evidence lives. Capture the assumptions explicitly, for example that the manufacturer owns clinical risk and that V&V evidence is delivered in the manufacturer’s own template. Assumptions left in people’s heads are the ones that fail a review. 

Agree the handover up front, even if it is months away. Decide now how evidence, tooling and monitoring responsibilities transfer back to you as your team grows, so the shift from more support to less is planned rather than improvised. The teams that outsource well are not the ones that hand over the most. They are the ones whose boundary document a reviewer could read and understand in five minutes. 

Where this goes wrong

Assuming a fully managed engagement removes your own regulatory accountability. It doesn’t. As the manufacturer, you remain responsible for what’s submitted regardless of who wrote it. 

Choosing a tier based on how independent it sounds, rather than on where your team’s time is genuinely best spent. 

Leaving the boundary undocumented, so both sides quietly assume the other’s covering something. This is the single most avoidable failure, and the fix is simply agreeing it in writing before work starts. 

Picking one tier at the start and never revisiting it as the team, product, and funding position mature. 

How the split should move as you grow

The right tier is a snapshot, not a permanent setting. The arrangement that gets a pre-revenue team through its first submission is often the wrong one two years and a funding round later. The useful habit is to revisit the split at each of the moments below, rather than defaulting to whatever you agreed at the start. 

Early, pre-revenue, first submission. The clinical problem is the whole business and every engineer is on it. There is usually no in-house security owner and no time to grow one before the deadline. A blended or fully managed tier is often the fastest route to a stronger submission here, provided the boundary is written down. Keep clinical risk, design decisions and sign-off in-house from day one, because those are the parts that cannot be delegated whatever else is outsourced. 

Post-funding, scaling the team. Once you can hire, the calculation changes. Bringing the recurring work in-house, such as SBOM generation, vulnerability triage and routine V&V, starts to pay off, while you keep external support for the specialist or peak-load pieces. This is the point to move deliberately from fully managed towards blended, and to plan a proper handover of the evidence and the tooling rather than an abrupt cut-off. 

Mature, second or third product. A team on its third submission carries fewer unknowns and usually needs advisory input more than delivery: a second opinion on architecture, a check on a threat model, an independent test before submission. Support becomes narrower and more targeted. The evidence work is now yours, and outside help exists to stress-test it. 

Whichever direction you are moving, treat the handover as its own piece of work. The most common way this goes wrong is not incompetence on either side. It is an external-developer boundary that everyone assumed and nobody wrote down, so a control or an artefact quietly falls between the two teams during the transition. Update the shared-responsibility document every time the split changes, and the move stays clean. 

Stage Sensible starting tier What to bring in-house first 
Pre-revenue, first submission Blended or fully managed Clinical risk ownership, design decisions, final sign-off (keep from day one) 
Post-funding, scaling Blended Recurring work: SBOM, vulnerability triage, routine V&V 
Mature, multi-product Advisory Full evidence ownership, keep external input for review and independent testing 

Free resources

System Boundaries: Who Controls What: a shared-responsibility table and example boundary diagram, useful for documenting whichever tier you choose.

Essential Guide to SaMD and MedTech Device Security: a broader primer if you’re still scoping what the evidence needs to cover.

Book a review 

In 30 minutes, we’ll: 

  1. Map where your current capacity genuinely sits against the three tiers above 
  1. Identify which artefacts you already have and which are still missing 
  1. Agree clear system boundaries in writing before any delivery work starts 

Get in touch

FAQs

Does outsourcing V&V work remove our regulatory accountability? 

No. As the manufacturer, you remain legally responsible for what’s submitted to FDA or under EU MDR regardless of who built the underlying evidence. A support partner can’t sign your Declaration of Conformity for you. 

How do we know how hands-on we need to be? 

It depends mainly on your internal security expertise, your timeline pressure, and whether this is your first submission or a mature programme. There’s no single right answer, only the right one for your team right now. 

Can we start hands-off and bring more in-house later? 

Yes, and it’s a common, sensible pattern. Many teams start with more support under deadline pressure on a first submission, then shift work in-house as their own team grows and gains experience. 

What should we keep in-house even in a fully managed arrangement? 

Product and clinical risk ownership, final sign-off authority, and ownership of your own quality management system. None of that can be meaningfully delegated, whatever else is outsourced. 

Does a more hands-off engagement cost more? 

Usually, per day. The comparison that actually matters is against the opportunity cost of your own engineers’ time spent on compliance evidence instead of the clinical or product work only they can do. 

How do we avoid outsourcing badly? 

Agree the system boundary in writing before work starts: exactly who owns which artefact, which decision, and which sign-off. Most outsourcing problems are boundary problems, not competence problems. 

Related reading

  • Adaptix case study: a worked example of a blended engagement, Cyber Alchemy supporting a connected medical device manufacturer through its compliance evidence.
  • DTAC: The Three Phases of Readiness: a look at Digital Technology Assessment Criteria (DTAC) readiness, the NHS procurement assessment for digital health products, useful alongside this if NHS procurement is also part of your route to market.

Similar Posts