How do you audit design controls using the process approach?

A new connection I made on LinkedIn joined the RA Review Group, and they suggested that anything related to the topic of Design and Development would be of interest for a blog topic. Therefore, I thought I would share a secret with everyone reading my blog…

This blog has been moved to the following location and the name has been changed: http://bit.ly/AuditDesign.

This blog website and the blogs within it are gradually being transferred over to my new website: http://www.MedicalDeviceAcademy.com. The titles may change, and there may be minor revisions to the content as the blogs are reviewed and edited. There will be a subscription list created for the new blog site. If you would like to be added to the list for the new blog site, please email me directly at: rob@13485cert.com.

I have left the links to the videos I love.

Entertainment for this week is Diana Krall‘s recording of Bésame Mucho. I have also included another recording by Andrea Bocelli with English and Spanish subtitles for anyone that wondered what the words meant. The song loses a little of it’s appeal in translation, but English is not one of the Romance Languages.

Advertisement

“Triage” for 510(k) – I’m underwhelmed

This week I pulled another song from the movie August Rush.

Thursday, Congress voted 96 to 1 for bill to increase FDA user fees. The rationale is that the FDA needs more funding in order to be strong enough to properly regulate foods, drugs and medical devices. One of the commitments linked with this new funding is to shorten the review of 510(k) submissions. To this end, OIVD has created a new program called “Triage.” The goal of this program is to accelerate the review of certain traditional 510(k) submissions to 30 days instead of 90 days.

In theory this pilot program will help some companies get their 510(k) clearance letter faster, but simultaneously the FDA will be able to concentrate resources on high-risk 510(k) submissions. This entire strategy seems to be the opposite of triage. Triage involves sorting sick patients into three categories:

1)      those who are likely to live, regardless of what care they receive;

2)      those who are likely to die, regardless of what care they receive; and

3)      those for whom immediate care might make a positive difference in outcome.

If we apply the triage analogy to 510(k) submissions, we see three categories:

1)      510(k) submissions that are likely to be approved, regardless of how much time the FDA spends;

2)      510(k) submissions that are likely to be rejected, regardless of how much time the FDA spends; and

3)      510(k) submissions whose approval or rejection is not clear, but the FDA’s earlier involvement in the design and development process would substantially improve the review time.

The FDA’s “triage” program is intended to demonstrate improvement in the time required to approve medical devices by sorting submissions into two groups: group #1 above and group # 2/3 from above. This will make the numbers look good, but the FDA should be spending even less time on the #2 than it spends on the #1 category of submissions. The FDA should also get involved in group #3 submissions much earlier.

The types of submissions that need more FDA reviewer time are devices that are higher in risk and where special controls guidance documents and/or ISO Standards have not already been established for performance and safety testing criteria (i.e. – Category #3 above). In these cases, when a company tries to get some feedback from the FDA the company is asked to request a pre-IDE meeting. The company will not be necessarily performing a clinical trial, but this is the only vehicle the FDA has for justifying the time it spends providing feedback on proposed verification and validation testing plans. The FDA needs to develop something new that is ideally suited for 510(k) products where guidance and Standards do not exist. This would also have the effect of reducing the number of “Not Substantially Equivalent” (NSE) letters the FDA issues.

If a company is developing a device that already has an applicable special controls document or ISO Standard, then the 510(k) pathway should be well defined without the FDA’s help. Unfortunately, there is no easy mechanism for ensuring compliance with these external standards. This type of submission would benefit from software controlled submissions and/or pre-screening of submissions by 3rd party reviewers. The Turbo 510(k) software tool could lend itself to software controlled submissions, but proliferation of the Turbo 510(k) has been limited.

If a company does not submit a 510(k) with all the required elements of a guidance document the submission should not be processed. Implementation of validated software tools for each 3-letter product code would prevent incomplete submissions. At the very least, companies should be required to provide a rationale for any sections of a submission that are not applicable.

One example of a possible software solution is currently used by 3^rd party auditors at BSI. BSI uses a software tool that will not allow the auditor to generate a final report unless all the required elements have been completed. The FDA could use the existing screening checklist and convert this into a similar “SmartForm”. If the submission does not have all the required elements of the checklist, the submission form could not be generated from the software. This forces the task of pre-screening reviews back upon the submitter with the aid of a validated software tool.

The biggest shortfall of the Triage program is the target product types. IVD devices are quite different from other device types. Each IVD has unique chemistry, there are a limited number of Guidance documents for IVDs, and IVD submissions represent only 10-20% of all submissions. Orthopedic, cardiovascular, general/plastic surgery, and radiology devices each represent more than 10% of the submissions and collectively they represent half of the submissions. These types of devices also have both Special Controls Documents and ISO Standards defining the design inputs for design verification. Therefore, these four device types would be a better choice for a pilot program to expedite reviews.

The Ultimate Design Control SOP

Disclaimer: There is no need to create the Ultimate Design Control SOP. We need medical devices that are safer and more effective.

If Adele is worthy of six Grammy Awards, she’s probably worthy of a blog link too. Rumor has it that this is my personal favorite from Adele.

In my previous blog posting, I indicated six things that medical device companies can do to improve design controls. While the last posting focused on better design team leaders (WANTED: Design Team Needs Über-Leader), this posting focuses on writing stronger procedures. I shared some of my thoughts on writing design control procedures just a few weeks ago, but my polls and LinkedIn Group discussions generated great feedback regarding design control procedures.

One of the people that responded to my poll commented that there was no option in the poll for “zero”. Design controls do not typically apply to contract manufacturers. These companies make what other companies design. Therefore, their Quality Manual will indicate that Clause 7.3 of the ISO 13485 Standard is excluded. If this describes your company, sit back and enjoy the music.

Another popular vote was “one”. If you only have one procedure for design controls, this meets the requirements. It might even be quite effective.

When I followed up to poll respondents asking how many pages their procedures were, a few people suggested “one page”. These people are subscribing to the concept of using flow charts instead of text to define the design control process. In fact, I use the following diagram to describe the design process all the time: The Waterfall Diagram!

From the US FDA Website.

I first saw this in the first AAMI course I took on Design Controls. This is on the FDA website somewhere too. To make this diagram effective as a procedure, we might need to include some references, such as: work instructions, forms, the US FDA guidance document for Design Controls, and Clause 7.3 of the ISO Standard.

The bulk of the remaining respondents indicated that their company has eight or more procedures related to design controls. If each of these procedures is short and specific to a single step in the Waterfall Diagram, this type of documentation structure works well. Unfortunately, many of these procedures are a bit longer.

If your company designs software, active implantable devices, or a variety of device types—it may be necessary to have more than one procedure just to address these more complex design challenges. If your company has eight lengthy procedures to design Class 1 devices that are all in the same device family, then the design process could lose some fat.

In a perfect world everyone on the design team would be well-trained and experienced. Unfortunately, we all have to learn somehow. Therefore, to improve the effectiveness of the team we create design procedures for the team to follow. As an auditor and consultant I have reviewed 100+ design control processes. One observation is that longer procedures are not followed consistently. Therefore, keep it short. Another observed I have made is that well-design forms help teams with compliance.

Therefore, if you want to rewrite your design control SOP try the following steps:

1. Use a flow chart or diagram to illustrate the overall process
2. Keep work instructions and procedures short
3. Spend more time revising and updating forms instead of procedures
4. Train the entire team on design controls and risk management
5. Monitor and measure team effectiveness and implement correct actions when needed

The following is a link to the guidance document on design controls from the US FDA website.

Refer to my LinkedIn polls and discussions for more ideas about design control procedures:

1. Medical Devices Group
2. Elsmar Cove Quality Forum Members Group

In addition to the comments I made in this blog, please refer back to my earlier blog on how to write a procedure.

What is the Design Input?

Micky, this is for you.

I have been directly involved in dozens of design projects throughout my career, and during the past three years I have audited 50+ Design Dossiers for CE Marking of Medical Devices. Throughout most of these design projects, I have noticed one common thread—a misunderstanding of design inputs.

ISO 13485 identifies the requirements for Design Inputs. These requirements are:

1. Functional (7.3.2a)
2. Performance (7.3.2a)
3. Safety (7.3.2a)
4. Statutory / Regulatory (7.3.2b)
5. Previous and Similar Designs (7.3.2c)
6. Essential Requirements (7.3.2d)
7. Outputs of Risk Management (7.3.2e)
8. Customer Requirements (7.2.1)
9. Organizational Requirements (7.2.1)

The most common error seems to be the failure to include the outputs of risk management. For those of you that have used design FMEA’s—that’s what the right-hand columns are for. When you identify suggested actions to mitigate risks with the current design, these actions should be translated into inputs for the “new and improved” model.

The second most common error seems to be failure to consider regulatory requirements. There are actually two ways this mistake is frequently made: 1) Canadian MDR’s were not considered as design inputs for a device intended for Canadian medical device licensing, and 2) an applicable ISO Standard was not considered (i.e. – “State of the Art” is Essential Requirement 2 of the Medical Device Directive or MDD).

The third most common error, and the one that drives me crazy, is confusion of design outputs and design inputs. For example: an outer diameter of 2.3 +/- 0.05 mm is not a design input for a 7 French arterial catheter. This is a design output. The user need might be that the catheter must be small enough to fit inside the femoral artery and allow interventional radiologists to navigate to a specific location to administer therapy. Validation that the new design can do this is relatively straight forward to evaluate in a pre-clinical animal model or a clinical study. The question is, “What is the design input?”
Design inputs are supposed to be objective criteria for verification that the design outputs are adequate. One example of a design input is that the catheter outer diameter must be no larger than a previous design that is an 8 French catheter. Another possible design input is that the catheter outer diameter must be less than a competitor product. In both examples, a simple measurement of the OD is all that is required to complete the verification. This also gives a design team much more freedom to develop novel products than a narrow specification of 23 +/- 0.05 mm allows for.

If you are developing a Class II medical device for a 510(k) submission to the FDA, special controls guidance documents will include design inputs. If you are developing a Class IIa, Class IIb or Class III medical device for CE marking, there is probably an ISO Standard that lists functional, performance and safety requirements for the device. Regulatory guidance documents and ISO Standards usually reference test methods and indicate acceptance criteria. When you have a test method and acceptance criteria defined, it is easier to write a verification protocol. Therefore, design teams should always strive to document design inputs that reference a test method and acceptance criteria. If this is not done, verification protocols are much more difficult to write.

In my earlier example, the outer diameter of 2.3 +/- 0.05 mm is a specification. Unfortunately, many companies would document this as an input and use the final drawing as the output. By making this mistake, “verification” is simply to measure the outer diameter to verify that it matches the drawing. This adds no value and if the specifications are incorrect the design team will not know about it.

A true verification would include a protocol that identifies the “worst-case scenario” and verifies that this still meets the design input requirements. Therefore, if the drawing indicates a dimensional tolerance of 2.3 +/- 0.05, “worst-case” is 2.35 mm. The verification process is to measure either a previous version of the product or a competitor’s catheter. The smallest previous version or competitor catheter tested must be larger than the upper limit of the design output for outer diameter of the new catheter.