Medtech’s Recall Epidemic Still Rages?

FDA medical device recalls hit an all-time high again last year. Now we really know why.

Peter Lomas/Pixabay.com



  
Back in 2016, MD+DI published an article titled “What’s Behind Medech’s Recall Epidemic?” by Joshua R. Dix, Suraj Ramachandran, and Darin S. Oppenheimer. It reported that FDA recalls hit an all-time high in 2014. It happened again in 2017 (Figure 1). Last year now holds the record. The 2016 paper blamed risk management. A valid claim, but that was only a symptom. In Figure 1, the “%Recalls” trendline (red) shows that growth of recalls correlates 95% with growth in new device introductions (blue lines). The solid blue line has two spikes averaged out. The dashed line shows the raw values. The result is the same: More devices, more recalls. New manufacturers are no better at risk management, and older companies haven’t improved in 14 years.

Figure 1: Recalls by year and class

 



Figure 2 highlights another symptom: In more than one third of FDA site inspections, either voluntary or official actions are required. The black trend line betrays the percentage of adverse inspections, which has been stuck between 30% and 40% for almost a decade. There is no real improvement here either.

Figure 2: Percent action trends

Citation information like that shown in Figure 3 describes the situation but does not point to a solution. The assumption is that if these observations are corrected, all would be well. That is wrong. Acting on these observations individually can be overwhelming. Chasing broken processes and root causes takes you in the wrong direction. It only represents effort when results are needed. Analysis needs to look forward. What opportunities are open? Why must new medical device companies construct support systems from scratch? And why must they build them while developing their product?



Figure 3: Device observations

The 2016 MD+DI article blamed risk management for recalls: the culture, the shallow understanding industry-wide, the inability to tie risk management into quality management systems. But poor risk management is a symptom, not the problem. Insufficient documentation and broken processes are also symptoms and not the problem.

In the paper, the authors quote a Francisco Poliodoro Harvard Business Review article: “safety-related behaviors fade over time and other motivating forces come to the fore, gradually launching the seeds of the next error.” They conclude their article by accusing medical device manufacturers of having an “inability to disengage from the reactive cycle” and calling them “the main impediments to the reduction of adverse events.”

This solitary blame is unmerited. No one in the industry is guilt free.

Which experts, you should ask, recommend making detection ranking part of risk prioritization? Who is saying to use both fault tree analysis (FTA) and failure mode and effects analysis (FMEA) to make devices safe and reliable?

What currently available solutions apply to all stages of medical device manufacturing? Which ones integrate with all the other non-QMS related work that goes on inside manufacturers? How many integrate planned work with reactive work like problems, corrective and preventive actions, non-conformances, requests, risks, and failure modes on the basis of relative priority? Name the solutions that help companies build a foundation for growth and a means to reach maturity and excellence.

Manufacturers are forced in many cases to fit their procedures to hard-coded solutions, and not the other way around. Is all CAPA handling centralized? Or, distributed? It is no wonder that many medical device manufacturers still use paper or Microsoft products. At least it lets them produce their devices, and document compliance, just the way they have been doing or want to do.

The problem is complacency. Figure 1 infers there is no epidemic of recalls. Figure 2 infers there is no crisis of bad inspections. The medical device industry has been in the same situation for 10 years or more.

That means all in the industry need to gauge their sense of security and self-satisfaction regarding compliance and the depth and precision given to risk management.

The solution is innovation. Management consultant Peter Drucker famously said, “Results are obtained by exploiting opportunities, not solving problems.”

What opportunities are being missed or ignored? Think more in terms of absolutes. Set targets to zero. Zero recalls. Zero citations. Take concrete actions now that can have a big impact. Then establish feedback and learning loops where faults are eliminated and performance is continually improved incrementally and in great leaps in a program of process improvement.

Risk Management Innovation

Medical device manufacturers probably have all the product knowledge and risk management expertise that they need in house. It has just never been brought together, so its value can be maximized. Why has this message been lost?

Synergy happens when elements combine to produce a total effect much greater than their simple sum.

Create synergy from product knowledge and risk expertise by integrating team members with risk experts. Bring them all together to share cross-component failure modes, mitigations, and other discoveries. Find analogs of risks and mitigations in other risks and components. Find risks and mitigations in component interfaces and interactions.

Create a continuum of meetings. Design meetings that focus in depth, and completely, on specific or high-ranking risks. Invite risk expert stakeholders and selected team members that own those risks. Have periodic risk management meetings with all hands to share risks and mitigations. Increase the depth of risk understanding by making detection ranking part of discussion and risk prioritization. Synergy will increase the breadth of risk understanding.

Use FTA and FMEA to make your medical devices safe and reliable. Fine tune your FTA and FMEA processes. Find deep failure modes and inventive mitigations.

Plan a campaign to find new methods, like a way to combine FTA and FMEA. And make your own, built not just from the great body of risk management knowledge but also from academic work on the subject. Use this new capability to prevent the fade of safety-related behaviors. Use FTA results in the form of quantifiable probabilities required by benefit-risk assessment. Use benefit-risk assessment to anchor both development and risk management to safety and reliability goals.

If you have not already, create a learning process for your risks, faults, and failure modes. Manufacturers that have been in business a while should have a proprietary risk vault of past risks. Every project should have a post-mortem to mine its history of project risks, product faults, and lessons learned. This creates a precious corporate asset for future risk management that will gradually improve the safety, reliability, and quality of your products.

Compliance Innovation


The compliance model in Figure 4 shows that to enforce rigor and discipline in outcomes, this same rigor and discipline needs expression in procedures, work instructions, and other proprietary assets. Procedure steps should represent detailed goals to reach, assumptions to check, constraints to observe, commitments to promise, dependencies to monitor, decisions to make, information to gather or disseminate, questions to ask and answer, and, of course, tasks to perform.

Figure 4: Compliance model

A second component needed is the prevention and correction of errors. Errors of omission can be prevented with detailed procedures, steps, and work instructions that are ready for allocation, prioritization, optimization, motivation, and tracking to completion for performance guarantees. This requires that they provide the information needed at the points of execution and the means of accountability.

Errors of commission can be prevented and corrected by timely reviews. If reminders and notifications of changes can be sent to the right people, then reviews can be done when the number of changes is smaller and memories are fresher. Skipped review opportunities force review work to accumulate and memories to fade. Work may become overwhelming, and matters and memories a distant resource.

Adopt a Higher Standard

Consider figures 1 and 2 again. Manufacturers need to break out of their conundrum. They should find support for FTA, FMEA, and other methods and practices, and use them now. Also, ask more of providers and experts and not fall victim to low expectations. The medical device industry should adopt higher standards for risk management and compliance. Implement only the solutions that help you build a foundation for growth and a means to reach maturity and excellence.

And stop using Microsoft Office and paper.

Richard L. Bollinger

Richard L. Bollinger is CEO of Menlo Park Associates, a management consulting firm based in Ann Arbor, MI. He is active in process improvement and speaks at various seminars and conferences of the Project Management Institute and the Dr. W. Edwards Deming Institute. Contact him at rick@menloparkassociates.com.

500 characters remaining
Richard L. Bollinger is CEO of Menlo Park Associates, a management consulting firm based in Ann Arbor, MI. Contact him at rick@menloparkassociates.com and visit www.MenloParkAssociates.com to see more about this problem.
A problem with comparing year to year recalls is that we don't know what the number should be in terms of actual recallable events. This might be influenced by the variety of things manufacturers call field actions other than recalls. The FDA then may or may not pick them up and label them a recall.
The FDA answered a similar question from me thus: "When you compare 2017 recall data to 2016, you may see some medical device corrections or removals in 2017 that are “not yet classified”. There will be no “not yet classified” actions in years prior to 2017. If you sort the data by class 1, 2, and 3, the searches for all years will result in data that follows the same rules." I do indeed select by class 1, 2, and 3. Thanks for your comment, William. Others may have had the same query.