Pages

Wednesday, January 17, 2024

Biopharmaceuticals And Medical Devices: A General Overview III (Medical Devices)

Drugs Part I

Drugs Part II

(Note:  As with Drugs above, this is a very high level view of a very complex and multi-year process and is only intended as an overview.)

A proposed Medical Device definition is "A contrivance designed and manufactured for use in healthcare, and not solely medicinal or nutritional".  A more "conventional" regulatory definition is here: in short if your item is intended to treat, cure, mitigate, or prevent disease and is intended to impact a human or animal system not using a chemical action internally and/or not activated via metabolic systems, it is a device.

So to answer the question, lots of things are a medical device.  A tongue depressor counts.  So does a scalpel.  So does a hospital bed.  So does an MRI machine.

But, all devices are not created equal and so there are classifications for devices.  In the United States, there are three classes (Class I, Class II, Class II) which are identified based on their risk to patient and user.  Class I is the "least" risky, Class III is the "most' risky.

There are distinct steps for designing a medical device - perhaps unlike Drugs, the process is a lot more standardized.  These steps are found in the US Document CFR 21 Part 820 (Quality System Regulations) and in the ISO Standard 13485 (Medical Devices - Quality Management Systems).

Design

All devices start with identifying User Needs:  what does the end user (patient, medical personnel, even marketing) need the product to do or be?   This is much more of an activity that it may seem, as it can involve getting input from a great many sources.   The User Need will be phrased in the form of a statement - for our purposes today, we will identify a User Need as "The product casing should be blue".

User Needs are then usef to generate Design Inputs, which are "The physical and performance requirements of a device which are used as a basis for device design" (21 CFR 820.3 (f)).  Each Design Input must be individually identified and must be clear, unambiguous, and non-conflicting with any other Design Input. As you can imagine, this will lead to hundreds or even thousands of Design Inputs.  Our example above will be translated into "The product casing must be Pantone Classic Blue Part Number #19-4052".   That may seem very specific - and it is intended to be.  Each input must be able to be objectively verified.

At this point a Design Review Team will be called together.  This will consist of personnel from all concerned departments - R&D, Engineering, Quality, Regulatory, Marketing/Business Development, Manufacturing, Supply Chain - and an Independent Reviewer who is not directly involved in the project but has the educational and scientific basis to assess the project.  All will be reviewed, unclear inputs corrected, additional needs identified.  When all agree on the User Needs and Design Inputs, the project will move to Design Output/Design Verification.

Design Outputs are "the results of a design effort at each design phase and at the end of the total design effort" (21 CFR Part 820.3 (g)).  The outputs will form the basis of the final design documentation - batch records, testing, labeling, etc.).  Each Design Input will have a Design Output which will need to be independently verified.  All of these Design Outputs will be putting into a Design Verification Protocol - again for our example, the Design Output would need to be "Product casing color is verified as Pantone Classic Blue Part Number #19-4052" - likely pulled from the manufacturing documentation for the casing in this case.

Another Design Review will take place to ensure that there are no new User Needs or Design Inputs, that all Design Inputs have Design Outputs, and that the Design Outputs are verifiable.  If approvable, it moves to Design Verification.

Design Verification is (rather straightforwardly) ensuring the Design Input meets the Design Output.  Personnel will move through the Design Verification protocol and verify and/or test every single Design Input.  Each Design Output will be recorded as meeting the Design Input, partially meeting the Design Input, or not meeting the Design Input.  At the end of this process the Design Verification Protocol will be reviewed and those Design Outputs which are found to not fully meeting the Design Inputs will be reviewed for additional clarification, correction, and testing.  In our example, if the manufacturing paperwork could not verify that Pantone #19-4052 was used on the casing, this would need to be resolved by an investigation and perhaps a remanufacture of the part - or, the User Need could be re-examined to see if that shade of blue is really needed, or just any "blue".

Assuming all items have been identified, a Design Verification Report is generated, to be presented to the Design Review Committee for approval. 

As you can imagine, there is a lot of documentation (physical or electronic) generated by this.  One of the most critical items is the Design Trace Matrix, which traces each User Need to Design Input to Design Output to Design Verification Result.  This document allows complete traceability up and down the design process.

A note:  This process takes place for each major component of the system.  For a system that has hardware/instrumentation, software and firmware, and consumables/reagents, every item will have a design history.  

The product is ready to move to Clinical Validation.

Risk Management

As noted above, the consideration of risk is a primary factor in determining the classification of product.  A risk - in Medical Device terms - is the combination of the probability of the occurrence of harm (injury or damage to people, property, or the environment) and the severity (measure of the consequences of a hazard) of that harm.  The ISO standard 14971 defines the risk management standard and is really the gold standard for all risk management (and largely a universal consensus standard).

A very short definition (for a very long process) is that intended use of the device and its use conditions are identified, along with what safety would look like and potential hazards and harms identified (including not just impact on the patient, but on the user (electrical hazards, chemical hazards) and the environment (spills, disposal of reagents)).  Using a numerical system, the risk are graded and evaluated.  Where risks exceed an acceptable level due to impact or frequency, risk mitigation and/or controls will need to be introduced, which may very well impact the Design Inputs/Design Outputs/Design Verification which in turn may then require another Design Review Meeting.

Another useful tool is the Failure Mode and Effects Analysis (FMEA) or the Failure Mode Effects and Criticality Analysis (FMECA).  Every potential failure mode is identified and evaluated against probability of occurrence, severity of that occurrence, and impact of the occurrence (for the FMECA, criticality of the failure is also evaluated).  Again, a rubric is used to evaluate unacceptable levels of failures.

The point of this entire exercise is to identify all risks and move them to acceptable risk levels (sometimes called "residual risk levels").  Like drugs, all medical devices entail some kind of risk; the goal is to remove as much risk of failure as possible and control the rest. 

Using our example above, the risk would be assessed under "What is the impact if the casing is not Pantone #19-4052"?  Possibly nothing - but perhaps in a clinical lab setting, that shade of blue may be not be able to be discerned for certain vision issues and could conflict with other similar pieces of equipment.

As with the Design Process, there are multiple iterations of the Risk Management Process.  The final documents will be held in a Risk Management File, which should contain every known risk for the product.  When a new risk is uncovered (either in development or in use), it should get added to the Risk Management file and assessed - in some cases, it may entail a  redesign.

(A note for general use:  Risk can only be avoided, mitigated, transferred or shared (to someone/something else), or accepted. True in real life as well as in medical devices.)

Validation

Validation (or as it is often known "Clinical Validation") is the use of the product in the potential use environment.  The difference between verification and validation, if helpful is:

Verification:  Did we make the thing right (e.g., per the Design Inputs)?

Validation:  Did we make the right thing (e.g. per the User Needs)?

A protocol will be developed and clinical sites engaged as well as all items approved by the Design Review Board.  Similar to the clinical process as identified in Drug Overview Part II, there is a whole set of processes involved to insure patient safety and control of the clinical trial. 

At the conclusion of the Clinical Validation, a final report will be generated. The results of the Design Verification will be added to the Design Trace Matrix so there is complete transparency up and down the design chain from User Need to Design Validation.  Any Design Issues or new Risks will need to be addressed, and possibly a second Clinical Validation conducted  Once all issues have been resolved and the report issued, the product will be ready for regulatory submission.

Quality Systems

Underlying all of this, of course, are Quality Systems. Just as with Drugs, there needs to be a system which ensures that all products are documented, manufactured, and released per written procedures and per the approved Design Documents. Just as with drugs, one needs a Quality Manual (corporate statement and discussion of the systems to ensure Quality), Standard Operating Procedures to cover systems like documentation, manufacture, material receiving and testing, product testing, and release.  Materials have to purchased and verified per specification.  Manufacturing facilities need to be maintained for cleanliness and environmental control.  Manufacturing equipment has to be purchased, qualified, and maintained.  Records have to be retained.  The individual Device History Records (batch records for manufacture) need to be created, approved, and executed.  The Device Master File, including the Design History, the Risk Management File, and all documents need to make the product, needs to be controlled and maintained.  Personnel need to be qualified and trained.  Variances like deviations and nonconformities need to be captured and assessed.  Improvements in the form of Corrective Actions and Preventive Actions need to be identified, investigated, and executed.

In short, everything that was done for Drugs is effectively also done for Medical Devices.

Tomorrow, we will discuss Intellectual Property, Regulatory Submissions, and Commercialization.

9 comments:

  1. Nylon128:59 AM

    Article in this morning's local newspaper about a startup at my old university to develop drugs to treat several types of cancer. A professor and grad students are involved, this week's postings are rather timely in understanding what was presented in that article. I see they have a website all setup. Thanks for the work you put into these posts TB.

    ReplyDelete
    Replies
    1. That is great Nylon12! This is exactly what I what I was hoping for (sort of, indirectly). This whole thing is not a huge mystery - complex and complicated, yes, but not a mystery. The moer people understand, perhaps the more they can both appreciate the amount of work that goes into the pill or insulin pump the hold and think "All that money for this? How hard can it be?"

      Delete
  2. Interesting article on medical devices. I am reminded of the impact of the validation/revalidation cycle and technology change on such devices. Many years ago a product (I cannot recall exactly what) was brought to market that incorporated a Toshiba laptop in the design. The inevitable happened, the particular model of laptop was discontinued and the supplier could not afford the costs of revalidation with a new laptop, so the product, despite being very useful, was discontinued.

    ReplyDelete
    Replies
    1. Will, it is - and devices are far more prone to the need.

      Of interest, the US FDA come out late last year with the final guidance on Cybersecurity. The speed of the announcement surprised a lot of people. The retrospective work that will need to be done by companies for firmware, software, and interfaces is daunting. I do not wonder that some companies may discontinue products as you have mentioned instead of putting in the time and money to revalidate and update if, in some cases, they are even able to.

      Delete
  3. Oh my goodness. I admire the knowledge you're writing from, TB, but I got hopelessly lost in all the manufacturing jargon. My brain got tickled, though at the mention of choosing a color, and I have questions that you may not have the answer to. Are colors limited, and discontinued, and can you tell when an item was designed and manufactured based on the color? I ask because over the years, I've noticed that things made of plastic and are common to a time period are often the same shade of green, for example. The items don't even have to be related to each other for this to be the case. I might have a toothbrush exactly the same shade of green as a cutting board. The truth is, when my green cutting board was new I suddenly noticed all kinds of things the exact same color. It makes me wonder if plastic pellets (or maybe liquid plastic) are dyed a certain exact shade (of green, in this case), and manufacturers of all kinds or products order that color because that's what's available. In a year or two, does that color go out popularity? Do colors get recycled? How often are colors made to order? In your example of Pantone #19-4052", is that color chosen amongst what is available, or does a company order that specific color? Do you have any idea how that part of the manufacturing process work. It may seem silly to wonder about this, but every time I see several disparate things the same color, and they show up in the same time frame (like the same year), I'm curious how that happens.

    ReplyDelete
    Replies
    1. What I meant by "does a company order that specific color" was really... does a company (ever? often?) custom-order colors to be made up into whatever material is used to make a particular medical device? Clear as mud?

      Delete
    2. Becki - Interesting question about which I know little, but I can guess on - based on a very early job in a print shop.

      Colors can in fact be very specific, almost to the point of uniqueness. Think about something like the "red" in a can of Coke, especially Coke cans when we were young. They were completely recognizable by the logo and color, right? And if you ever went to a place where they had a Coke knock-off and similar but not the same can, it looked wrong, did it not?

      On the whole I am sure that any color can be resurrected if needed, but in point of fact color changes like any fad. The fact that we live in a world where we can make colors that far exceed the ability to generate them out of natural material gives us even more possibilities.

      The Pantone reference is to the Pantone color pallet, which can identify colors based on a series of measurements (Wikipedia link is here https://en.wikipedia.org/wiki/Pantone if you want to follow up).

      Specific colors schemes are part of the user need and design input process, although custom colors may be a bit of stretch. That said, it is worth noting that in Clinical Labs, Color Blindness is something tested for and there may be specific tests or indicator lights that cannot be in certain colors to ensure all personnel can read them.

      Delete
    3. Thank you, TB. That last paragraph is a new thought to me.

      Delete
    4. Becki, it was new to me as well until one of my former employers started a Clinical Testing Lab. Turns out it was a requirement and we had to develop an SOP, buy a commercially available test, and test on an annual basis. One of my good friends in the lab is color blind, so it actually had an impact in that there were certain tests he could not run.

      Delete

Comments are welcome (and necessary, for good conversation). If you could take the time to be kind and not practice profanity, it would be appreciated. Thanks for posting!