Erik Swain03.31.09
No medical device company wants to feel the wrath of the U.S. Food and Drug Administration (FDA), so a number of them have implemented software systems that help them comply with quality regulations.
But that by itself is not good enough anymore.
As always, the FDA wants firms to comply with quality regulations, handle complaints properly and take the appropriate corrective actions. But if recent warning letters are any indication, the agency also wants device makers to take proactive steps to prevent quality problems from reaching the field in the first place. Device manufacturers and their software outsourcing partners anticipate these sorts of expectations will grow even greater under the Obama administration.
"There is a consensus among medical device manufacturers that with the new administration, FDA will apply more regulatory oversight to the product development process. And software is a key aspect of that process," said John Carey, vice president and general manager of the medical division of Foliage Software Systems Inc., a software product consulting and development firm based in Burlington, Mass. "Software will get a lot of focus. Companies will look to improve their own internal software development processes, and will expect suppliers to do the same. I expect that there will be a more rigorous approach to software safety and hazard analysis along with a focus on maintaining a proper design history file."
Many of the latest software systems can help device companies become more proactive about quality issues. In order to prevent quality issues from recurring, information about failures and problems must be fed back into the redesign process. Software can help with that. In order to prevent quality issues from ever happening in the first place, potential design hazards must be analyzed, and problematic manufacturing trends must be spotted before the product makes it to the user. Software can help with that, too.
"A lot of quality solutions come from the old mentality that quality equals compliance. The fact is, that's not true anymore," said Karim Lokas, vice president of product strategy and marketing for Charlotte, N.C.-based Camstar Inc., a supplier of software that advances product quality. "You can meet the letter of the regulations but not produce a top-quality product. Software that is geared to compliance does not cut it anymore. It has got to be geared to producing a safe and effective product. You can't get great quality by focusing on compliance, but you can satisfy compliance requirements as a byproduct of focusing on quality."
Chris Parsons, Camstar's senior director of marketing, said a recent research survey of device manufacturers that Camstar sponsored found that the top driver regarding implementing quality management software was to achieve "more consistent processes and higher product quality." Yet, many companies are in a reactive state of using quality management software mainly to "document regulatory compliance."
The device industry isn't quite there yet, said Nikki Willett, vice president of marketing and regulatory affairs for Pilgrim Software, a Tampa, Fla.-based supplier.
"We're seeing a greater expansion of quality issues into engineering and design," she said. "More effort is being put into design transfer issues. This is good news. It means the medical device community is seeing that quality is not just an end result. It's starting to expand its view.
Willett said the device community has done a better job than the drug industry in tracking a lot of these issues. It has better procedures for corrective and preventive action (CAPA) and complaint handling, and has done a very good job of being reactive and responsive. But neither has done as good a job in becoming more proactive, Willett noted. "It's not just about risk management. You have to make quality datamore visible to management so that they can do predictive modeling, which better controls risk throughout the process."
So how can software help device manu-facturers adopt a more proactive vision of quality? There are several ways, including enabling firms to identify and measure the metrics that are most critical to the quality of a product, allowing them to merge and analyze data across different departments, and making it easier for them to feed back quality data into design and manufacturing.
"If you have several departments who are not on a closed-loop system, you will be inefficient at bringing data together, which is very important," explained Tamar June, vice president of strategic marketing for Morgan Hill, Calif-based AssurX Inc., a provider of quality management software. "It is very important in today's manufacturing environment that you be on top of everything. Management must be able to review and evaluate data at a glance, then take action quickly."
Software must be usable, adaptable and flexible, said Lokas. "By adaptable, it means that a system needs to have configurable best practices built in, not provide a tabula rasa or rigid business processes," he said. "You must determine issues early, analyze root causes, and conduct proper investigations. But products and processes vary, and the software must be adaptable to that. You also have to make sure your software is designed for a high-quality environment and can support effective computer systems validation. And it has to be flexible in regards to change control. How can you have all this? It comes back to the architecture. It is done through configuration, not code."
Software also must be able to offer a device manufacturer a window into its supply chain, so it can identify any supplier-specific adverse events quickly, said Tim Lozier, marketing manager for Farmingdale, N.Y.-based ETQ Inc., a software provider.
"You need to take postmarket data and integrate it back into changes in the product. But does that include the necessary data? The design team has to know about the adverse events and complaints in the field and the areas that need improvement in the supply chain," he said. "A quality management software system is perfect for this. It is workflow based and can show who needs to approve something, who needs to execute it and how to get it distributed to the relevant departments. And if one product line has to undergo a change, you can go in to see how it relates to other product lines, so you can be proactive about it. You need to be able to drill down to the low levels and then roll it up to the corporate level. We also have a system whereby a quality manager can set it up to trigger quality alerts when certain conditions are met, from a certain number of CAPAs."
Such a scheme won't work if CAPAs are not properly prioritized, though.
"You have to make sure that not only is your postmarket data properly handled, but properly prioritized," said Lozier.
"For a lot of companies, CAPA is sometimes prioritized by what's overdue. Instead, you must be able to determine which CAPAs you must get to because they could lead to bottlenecks or even recalls if not addressed. Focus on that with the most weight. The right software system can automate the whole process for you, ensuring visibility and traceability."
"You should have CAPA, audits, change control, document management, nonconformances and other things all in one closed-loop system to prevent things from slipping through cracks," said June.
"Set up key performance indicators and show when they are not being met. That way, you can see where issues arise across the organization. If top management sees that something is not being taken care of, it can take action immediately. That is crucial in the device environment. Failure to do so can impact profitability, prompt lawsuits and bring about other consequences. You can simplify the task by making it automated with software."
In fact, the task is almost impossible without software, said Willett.
"If you don't have an automated system, by the time you retrieve everyone's data, format it correctly and do the analysis, you could be several weeks late," she said. "And by that point, who knows where that lot might have gone to. If you have real-time data that can help you raise and analyze issues immediately, that can help prevent nonconformances from getting into the distribution pipeline. And you will be able to fix the problem immediately."
Feeding back information from CAPA in-vestigations into the design, engineering and manufacturing processes is very important and can help firms prevent errors and quality problems from recurring. But a more effective system involves finding a way to prevent quality errors from occurring in the first place, or at least making sure the affected products never get out into the field.
This requires an intense regimen of statistical analysis and statistical process control, as well as the ability to detect and correct problems in real time. For those things to happen, the right software system is required.
"When you look at the manufacturing process from a statistical process control perspective, you have to monitor the manufacturing process in real time," said Michael Lyle, CEO of Chantilly, Va.-based InfinityQS, a quality management software provider. "For example, to provide statistical information about a component lot and how it affects output lots, you can use that to point to factors within manufacturing that are in need of improvement or adjustment or control."
Lyle noted that the software package must allow change controls and enable the client to make real-time changes and spot trends that affect the end product.
"When events occur, all appropriate people must be notified. Using statistical analysis, you can look at trends associated with quality," he said. " You can see what is likely to happen if no changes are made and trigger statistical changes prior to a product being produced that does not meet conformances or internal specifications. It can alert you so you can address it before it becomes a failure."
Device companies should make sure their software systems have a tracking database to enable this kind of information to come to the fore, said Marshall Wenrich, president of Realtime Group, a software design and validation test service based in Plano, Texas.
"For company operations and development/maintenance process, a common, centralized accessible problem tracking database should be used to capture all observations about the device," he said. "In order to avoid 'clutter' in the database, the categories of entries should include 'enhancement request,' 'problem,' and 'observation.' True problems should be dispositioned by a multi-disciplined change control board, with any implemented change being tracked to closure with supporting test evidence."
"Software charting techniques to show the level of concern versus the probability, much like a system FMEA (failure modes and effects analysis) shows that a hazard has been mitigated to a lower risk value, are a high-level indicator of which areas you need to focus on," he added.
A similar approach should be employed when dealing with software testing, said Carey of Foliage, which also supplies services dedicated to software verification and automated testing.
"It is always best to get quality engineering folks responsible for developing test cases involved as early as possible in the process," he said. "For example, Foliage always gets the quality engineering group involved in the review of software requirements. Requirements can be full of ambiguity and thus not testable or implementable in a straightforward fashion. But if quality is involved early on in the development process, they can help to identify weaknesses in the requirements even before the development phase begins. From a cost perspective, it is much easier to catch and fix defects up front so dedicated approaches to automated testing and simulation can make a big difference in detecting quality issues before they happen."
It takes a lot of coordination to implement, however, and for it to succeed, a device company must have different departments that are willing to share information with each other.
"Whenever Foliage delivers software architecture strategy services, we interact with a broad array of product stakeholders: clinical, senior management, engineering, manufacturing, regulatory, QA [quality assurance], and so forth. We interview all of them to find out what is most important to them about the product under development," said Carey.
"These engagements ensure that the software architecture will align with the product and business drivers. It also produces a consensus document, or feedback loop, that allows all of the various stakeholders to see the alignment with business drivers and understand how their specific role plays into that alignment. That feedback loop can often help product groups make proper tradeoffs during development and avoid decisions that disrupt the alignment."
Many of the same principles also apply to software that is part of a medical device or used to automate a medical procedure, said Robert Ruppenthal, president of RTEmd, a Rochester, N.Y.-based firm that provides software to automate diagnostics systems, among other things.
"Preventing quality issues from happening in the first place is part of a good design process. There's an emphasis on a hazard analysis process, especially in the design cycle," he said. "You can also use software tools to identify defects and dispose of them. If we find something in the process that impacts the requirements of the product, we take it back to the beginning and fix the flaws in the manufacturing and design processes and retest it. The emphasis is on catching defects as early as possible. Our approach is that FDA and the public assumes that this is a safe and effective product. It is dependent on us during development and testing to make sure that the process produces the safety and effectiveness that the consuming public and FDA expect."
Another consideration device companies have to make when buying software is its ability to generate and submit electronic medical device reports.
"We are also working with FDA on the Medwatch Plus program, formerly eMDR [electronic medical device reporting]. We have a gateway set up. Now the information can get to the proper hands in six minutes, not fiveto seven business days," noted Lozier of EtQ.
As part of its postmarket transformation initiative, the FDA has made it known that it would like MDRs to eventually be filed electronically.
"We are waiting for MDR guidance documents to be published. They don't know when that's going to happen. Device companies know this is going to happen. It's a question of when, not if," said June of Assurx. "FDA has invested a significant amount of money into its IT infrastructure. So you have to track and trend potential issues before they happen and while they happen. FDA uses an electronic submission gateway (ESG), so our new eMDR product builds in ESG functionality. You don't have to buy a separate application and slap it on. That's three to four months of validation and training that can be eliminated."
But most importantly, device companies must take steps to minimize MDRs and recalls from being implemented.
"We are working on showing how tighter management of the quality system can get quality right the first time," said Lozier. "That can mitigate risks and recalls. And in this environment, a recall can be a potential catastrophe."
But that by itself is not good enough anymore.
As always, the FDA wants firms to comply with quality regulations, handle complaints properly and take the appropriate corrective actions. But if recent warning letters are any indication, the agency also wants device makers to take proactive steps to prevent quality problems from reaching the field in the first place. Device manufacturers and their software outsourcing partners anticipate these sorts of expectations will grow even greater under the Obama administration.
Web-based supplier quality management delivers instant collaboration and visibility across the supplier network for greatly improved product quality.With an Internet connection and a few mouse clicks, the entire supplier network is cost effectively included in all the quality management initiatives. Image courtesy of AssurX. |
"There is a consensus among medical device manufacturers that with the new administration, FDA will apply more regulatory oversight to the product development process. And software is a key aspect of that process," said John Carey, vice president and general manager of the medical division of Foliage Software Systems Inc., a software product consulting and development firm based in Burlington, Mass. "Software will get a lot of focus. Companies will look to improve their own internal software development processes, and will expect suppliers to do the same. I expect that there will be a more rigorous approach to software safety and hazard analysis along with a focus on maintaining a proper design history file."
Many of the latest software systems can help device companies become more proactive about quality issues. In order to prevent quality issues from recurring, information about failures and problems must be fed back into the redesign process. Software can help with that. In order to prevent quality issues from ever happening in the first place, potential design hazards must be analyzed, and problematic manufacturing trends must be spotted before the product makes it to the user. Software can help with that, too.
"A lot of quality solutions come from the old mentality that quality equals compliance. The fact is, that's not true anymore," said Karim Lokas, vice president of product strategy and marketing for Charlotte, N.C.-based Camstar Inc., a supplier of software that advances product quality. "You can meet the letter of the regulations but not produce a top-quality product. Software that is geared to compliance does not cut it anymore. It has got to be geared to producing a safe and effective product. You can't get great quality by focusing on compliance, but you can satisfy compliance requirements as a byproduct of focusing on quality."
Chris Parsons, Camstar's senior director of marketing, said a recent research survey of device manufacturers that Camstar sponsored found that the top driver regarding implementing quality management software was to achieve "more consistent processes and higher product quality." Yet, many companies are in a reactive state of using quality management software mainly to "document regulatory compliance."
The device industry isn't quite there yet, said Nikki Willett, vice president of marketing and regulatory affairs for Pilgrim Software, a Tampa, Fla.-based supplier.
Image courtesy of AssurX. |
"We're seeing a greater expansion of quality issues into engineering and design," she said. "More effort is being put into design transfer issues. This is good news. It means the medical device community is seeing that quality is not just an end result. It's starting to expand its view.
Willett said the device community has done a better job than the drug industry in tracking a lot of these issues. It has better procedures for corrective and preventive action (CAPA) and complaint handling, and has done a very good job of being reactive and responsive. But neither has done as good a job in becoming more proactive, Willett noted. "It's not just about risk management. You have to make quality datamore visible to management so that they can do predictive modeling, which better controls risk throughout the process."
Proactivity
So how can software help device manu-facturers adopt a more proactive vision of quality? There are several ways, including enabling firms to identify and measure the metrics that are most critical to the quality of a product, allowing them to merge and analyze data across different departments, and making it easier for them to feed back quality data into design and manufacturing.
"If you have several departments who are not on a closed-loop system, you will be inefficient at bringing data together, which is very important," explained Tamar June, vice president of strategic marketing for Morgan Hill, Calif-based AssurX Inc., a provider of quality management software. "It is very important in today's manufacturing environment that you be on top of everything. Management must be able to review and evaluate data at a glance, then take action quickly."
Software must be usable, adaptable and flexible, said Lokas. "By adaptable, it means that a system needs to have configurable best practices built in, not provide a tabula rasa or rigid business processes," he said. "You must determine issues early, analyze root causes, and conduct proper investigations. But products and processes vary, and the software must be adaptable to that. You also have to make sure your software is designed for a high-quality environment and can support effective computer systems validation. And it has to be flexible in regards to change control. How can you have all this? It comes back to the architecture. It is done through configuration, not code."
Software also must be able to offer a device manufacturer a window into its supply chain, so it can identify any supplier-specific adverse events quickly, said Tim Lozier, marketing manager for Farmingdale, N.Y.-based ETQ Inc., a software provider.
"You need to take postmarket data and integrate it back into changes in the product. But does that include the necessary data? The design team has to know about the adverse events and complaints in the field and the areas that need improvement in the supply chain," he said. "A quality management software system is perfect for this. It is workflow based and can show who needs to approve something, who needs to execute it and how to get it distributed to the relevant departments. And if one product line has to undergo a change, you can go in to see how it relates to other product lines, so you can be proactive about it. You need to be able to drill down to the low levels and then roll it up to the corporate level. We also have a system whereby a quality manager can set it up to trigger quality alerts when certain conditions are met, from a certain number of CAPAs."
Such a scheme won't work if CAPAs are not properly prioritized, though.
"You have to make sure that not only is your postmarket data properly handled, but properly prioritized," said Lozier.
"For a lot of companies, CAPA is sometimes prioritized by what's overdue. Instead, you must be able to determine which CAPAs you must get to because they could lead to bottlenecks or even recalls if not addressed. Focus on that with the most weight. The right software system can automate the whole process for you, ensuring visibility and traceability."
"You should have CAPA, audits, change control, document management, nonconformances and other things all in one closed-loop system to prevent things from slipping through cracks," said June.
"Set up key performance indicators and show when they are not being met. That way, you can see where issues arise across the organization. If top management sees that something is not being taken care of, it can take action immediately. That is crucial in the device environment. Failure to do so can impact profitability, prompt lawsuits and bring about other consequences. You can simplify the task by making it automated with software."
In fact, the task is almost impossible without software, said Willett.
"If you don't have an automated system, by the time you retrieve everyone's data, format it correctly and do the analysis, you could be several weeks late," she said. "And by that point, who knows where that lot might have gone to. If you have real-time data that can help you raise and analyze issues immediately, that can help prevent nonconformances from getting into the distribution pipeline. And you will be able to fix the problem immediately."
Quality in Real Time
Feeding back information from CAPA in-vestigations into the design, engineering and manufacturing processes is very important and can help firms prevent errors and quality problems from recurring. But a more effective system involves finding a way to prevent quality errors from occurring in the first place, or at least making sure the affected products never get out into the field.
This requires an intense regimen of statistical analysis and statistical process control, as well as the ability to detect and correct problems in real time. For those things to happen, the right software system is required.
"When you look at the manufacturing process from a statistical process control perspective, you have to monitor the manufacturing process in real time," said Michael Lyle, CEO of Chantilly, Va.-based InfinityQS, a quality management software provider. "For example, to provide statistical information about a component lot and how it affects output lots, you can use that to point to factors within manufacturing that are in need of improvement or adjustment or control."
Lyle noted that the software package must allow change controls and enable the client to make real-time changes and spot trends that affect the end product.
"When events occur, all appropriate people must be notified. Using statistical analysis, you can look at trends associated with quality," he said. " You can see what is likely to happen if no changes are made and trigger statistical changes prior to a product being produced that does not meet conformances or internal specifications. It can alert you so you can address it before it becomes a failure."
Device companies should make sure their software systems have a tracking database to enable this kind of information to come to the fore, said Marshall Wenrich, president of Realtime Group, a software design and validation test service based in Plano, Texas.
"For company operations and development/maintenance process, a common, centralized accessible problem tracking database should be used to capture all observations about the device," he said. "In order to avoid 'clutter' in the database, the categories of entries should include 'enhancement request,' 'problem,' and 'observation.' True problems should be dispositioned by a multi-disciplined change control board, with any implemented change being tracked to closure with supporting test evidence."
"Software charting techniques to show the level of concern versus the probability, much like a system FMEA (failure modes and effects analysis) shows that a hazard has been mitigated to a lower risk value, are a high-level indicator of which areas you need to focus on," he added.
Non-System Software
A similar approach should be employed when dealing with software testing, said Carey of Foliage, which also supplies services dedicated to software verification and automated testing.
"It is always best to get quality engineering folks responsible for developing test cases involved as early as possible in the process," he said. "For example, Foliage always gets the quality engineering group involved in the review of software requirements. Requirements can be full of ambiguity and thus not testable or implementable in a straightforward fashion. But if quality is involved early on in the development process, they can help to identify weaknesses in the requirements even before the development phase begins. From a cost perspective, it is much easier to catch and fix defects up front so dedicated approaches to automated testing and simulation can make a big difference in detecting quality issues before they happen."
It takes a lot of coordination to implement, however, and for it to succeed, a device company must have different departments that are willing to share information with each other.
"Whenever Foliage delivers software architecture strategy services, we interact with a broad array of product stakeholders: clinical, senior management, engineering, manufacturing, regulatory, QA [quality assurance], and so forth. We interview all of them to find out what is most important to them about the product under development," said Carey.
"These engagements ensure that the software architecture will align with the product and business drivers. It also produces a consensus document, or feedback loop, that allows all of the various stakeholders to see the alignment with business drivers and understand how their specific role plays into that alignment. That feedback loop can often help product groups make proper tradeoffs during development and avoid decisions that disrupt the alignment."
Many of the same principles also apply to software that is part of a medical device or used to automate a medical procedure, said Robert Ruppenthal, president of RTEmd, a Rochester, N.Y.-based firm that provides software to automate diagnostics systems, among other things.
"Preventing quality issues from happening in the first place is part of a good design process. There's an emphasis on a hazard analysis process, especially in the design cycle," he said. "You can also use software tools to identify defects and dispose of them. If we find something in the process that impacts the requirements of the product, we take it back to the beginning and fix the flaws in the manufacturing and design processes and retest it. The emphasis is on catching defects as early as possible. Our approach is that FDA and the public assumes that this is a safe and effective product. It is dependent on us during development and testing to make sure that the process produces the safety and effectiveness that the consuming public and FDA expect."
Reporting
Another consideration device companies have to make when buying software is its ability to generate and submit electronic medical device reports.
"We are also working with FDA on the Medwatch Plus program, formerly eMDR [electronic medical device reporting]. We have a gateway set up. Now the information can get to the proper hands in six minutes, not fiveto seven business days," noted Lozier of EtQ.
As part of its postmarket transformation initiative, the FDA has made it known that it would like MDRs to eventually be filed electronically.
"We are waiting for MDR guidance documents to be published. They don't know when that's going to happen. Device companies know this is going to happen. It's a question of when, not if," said June of Assurx. "FDA has invested a significant amount of money into its IT infrastructure. So you have to track and trend potential issues before they happen and while they happen. FDA uses an electronic submission gateway (ESG), so our new eMDR product builds in ESG functionality. You don't have to buy a separate application and slap it on. That's three to four months of validation and training that can be eliminated."
But most importantly, device companies must take steps to minimize MDRs and recalls from being implemented.
"We are working on showing how tighter management of the quality system can get quality right the first time," said Lozier. "That can mitigate risks and recalls. And in this environment, a recall can be a potential catastrophe."