Ask a medical device injection molder for a first article inspection report, the CMM data sitting behind it, and the resin lot record for a given production run, and you’ll learn more in an afternoon than a stack of certificates and a walk through the plant can tell you.

Certifications and tours describe the building. Whether cavity three on your tool has drifted since the last audit? That answer lives in the paperwork — assuming the paperwork was ever generated.

So if you’re a quality or operations engineer sizing up a molder, what follows covers what each of those three documents should contain, what the QMSR changed about why they matter, and what to put in writing before the PO goes out.

The Certificate Audit Is the Wrong Audit

A certification tells you a quality system exists. It can’t tell you whether the critical dimension on your housing holds a Cpk above 1.33 across three resin lots — only measurement records answer that.

We see a version of this fairly often. An OEM signs with a molder, and eight months into production, in the middle of a complaint investigation, asks for dimensional data. What comes back is a one-page summary stating that inspection passed. Nothing about the certificate was false. There just wasn’t much measurement behind it, and the OEM finds that out at the point where it costs the most.

Run the audit the other way around. Start with the artifacts and treat the certifications as corroboration, not as the argument itself.

QMSR Made ISO 13485 Federal Law

February 2, 2026, is the date that matters. The Quality Management System Regulation amended 21 CFR Part 820 to incorporate ISO 13485:2016 by reference, which turned the standard from a voluntary benchmark into the regulation itself. FDA retired QSIT the same day and moved device inspections onto compliance program 7382.850.

For your supplier file, the operative change sits in purchasing. What used to live in 820.50 now comes from ISO 13485 Clause 7.4: evaluate and select suppliers against criteria proportionate to the risk the purchased part carries into the finished device, then monitor and re-evaluate them on an ongoing basis. A one-time qualification packet doesn’t satisfy that. A certificate in a folder doesn’t either.

Which puts you back on the artifacts. Clause 7.4 wants objective evidence that a supplier can do the work, and an FAI with actuals cavity by cavity, a control plan with real check frequencies, and a lot trace that runs both directions are what objective evidence looks like when an investigator asks to see it.

Process validation is where this gets expensive. Molding is a process whose output you can’t fully verify by inspecting the part — dimensions pass while shear history quietly degrades the polymer — which drops it squarely into Clause 7.5.6. IQ/OQ/PQ. Ask which of the three your molder has run on a comparable tool, and ask to see the OQ study. The process window gets defined there or it doesn’t get defined at all.

First Article Inspection Means Every Dimension, Measured

Everything downstream references the first article inspection. Which is exactly why a summary is the least useful thing a molder can hand you at that stage. A proper FAI starts with a ballooned drawing generated from your Build-To file: every dimension numbered, every one actually measured, actuals written next to the nominal and the tolerance.

Cavity by cavity, not once for the tool. Cavities in the same tool don’t come out identical, so a four-cavity tool should give you four sets of actuals — and the spread across those four tells you as much as the numbers themselves.

Ask which instrument was used for each dimension. If the full layout went across a CMM, the answer comes back right away. If someone spot-checked a few features with calipers and filled in the remainder, the answer gets vague, and that vagueness usually costs money later in the program.

The FAI is also your reference point when a dimension drifts at month nine. Without it, you’re arguing about what nominal meant.

PPAP Discipline Without the Automotive Baggage

PPAP came out of automotive, and some medical device injection molders will tell you it doesn’t apply to device work. The framework travels fine. A PPAP-style package gives you the process flow diagram, the control plan, capability studies on the dimensions your device’s function depends on, and dimensional results tied to a specific production run — most of which your own quality system will ask for eventually anyway.

Most buyers never read the control plan. It’s the best predictor in the package of how the next two years will go. It sets out which features the molder checks, at what frequency, and what the response is when a check fails. Say it calls for a visual check once per shift on a dimension your device depends on — raise that before production starts. The same conversation after a complaint is filed costs a great deal more.

CMM Data: The Difference Between Checked & Verified

“We inspected it” and “here’s the measurement record” are different claims. CMM output gives you actuals — this dimension, this cavity, this date, this value. When a question comes up in device work, and it will, the record either exists or it doesn’t.

Ask how measurement data is stored and for how long. Some shops keep CMM records for the life of the program; others keep the last 90 days. A complaint investigation two years out doesn’t care what the retention policy says. Get the terms in writing with everything else.

Resin Lot Traceability Has to Run Both Directions

The forward direction is straightforward enough: a certificate of analysis for each incoming resin lot, with that lot number recorded against every production run it goes into. Any shop with a functioning quality system can pull this up when asked.

The reverse direction is harder to demonstrate. Take a part out of a box that shipped last March, send it back to your molder, and see how long it takes to get the resin lot, the press it ran on, and the date of the run. A few hours is a reasonable expectation.

That exercise is a small version of the recall scenario. There, the lot record stops being a filing question: the answer decides whether you contain a defined population of parts or go to a full field action.

Ask about regrind while you’re at it. Where regrind is permitted on your part, it needs to be under lot control too — an unlogged regrind stream breaks the trace chain no matter how thorough the rest of the records are.

Why Validation Is the Leverage Point in a Supplier Switch

Mold transfers are where device OEMs get burned. A transfer changes everything except the steel — new press, new water lines, new technician, new dryer — which is a revalidation trigger, and often there’s no baseline to revalidate against, because the outgoing molder never produced a real FAI in the first place.

Transfers where the documentation shows up alongside the tool generally go well. The ones that start with a shipping crate and a print revision nobody can vouch for are the ones that cause trouble. We take in transferred molds often enough to have seen plenty of both.

So ask for a full FAI after the tool arrives. Then capability studies on the critical dimensions, run before anything goes back into production. And where the outgoing molder left a baseline behind, compare against it. That’s the work that gives you something to stand on during a transfer. The receiving shop’s facility credentials, on their own, tend not to help you much here.

What to Ask for in Writing With Medical Device Injection Molding

You don’t have to take any of this on faith, and you don’t need a long questionnaire either. Four documents, requested in writing before the PO goes out, will tell you most of what you need to know about a molder:

  • A first article inspection report from some comparable part, anonymized, with the actuals listed cavity by cavity
  • Cpk values on the critical dimensions, rather than capability data reported as pass/fail summaries
  • Resin lot traceability demonstrated in both directions, plus a stated regrind policy
  • Retention terms for CMM data — which records are kept, and for how long

A shop that does medical device injection molding work will produce these without friction, usually inside a few days. Hesitation on any single item is information. Hesitation on two or more is your answer.

Where H&H Fits and Where We Don’t

A word about what we do and don’t do. H&H Molds holds ISO 9001:2015. Not 13485 — and under QMSR, that’s the first question a device OEM asks, so here is the straight answer on it.

The supplier-control obligation is yours, not ours. Clause 7.4 asks you to evaluate suppliers against risk-proportionate criteria and keep monitoring them; it does not require your molder to hold a 13485 certificate, and FDA doesn’t issue one either. It asks for objective evidence that the supplier can do the work. That’s a records question, and records are the part of this we can put in front of you.

So: first article inspection backed by CMM measurement, PPAP-style documentation packages, resin lot control on every program, and retention terms stated in writing before the PO. If your 13485 system needs a supplier that produces that evidence on demand, we’re a fit. If your procurement policy hard-requires a 13485-certified molder, we’re not, and we’d rather say so now than at qualification.

The other honest limit: no rated cleanroom. If your component has to be molded in one, we aren’t the right shop for that specific need — and you’ll hear that from us on the first call instead of in the fourth month.

What’s left is the part most programs actually get stuck on. Measured baselines, capability data, a trace chain that runs in both directions. None of that is a special arrangement here; it’s how our shop normally operates.

If you’re evaluating molders right now, include the four-item request list in your email and see what comes back. This video from RapidDirect highlights some of the benefits of choosing injection molding over other processes for medical device injection molding.