Medical Devices
FDA Workshop: Best Practices For Pacemaker and ICD Lead Postmarket Surveillance, October 2-3, 2008
Date, Time and Location
The meeting was held on October 2, 2008, from 8:00 a.m. to 5 p.m., and on October 3, 2008, from 8:00 a.m. to 12:00 p.m. at the following location:
Hilton Washington DC North / Gaithersburg
620 Perry Parkway,
Gaithersburg, Maryland, 20877
(301) 977-8900
Registration
Online registration is now closed.
Final Agenda
Thursday, October 2, 2008
| 7:45 AM | Arrive at Gaithersburg Hilton Registration / Check-in |
| 8:00 AM |
Welcome and Introduction Jonathan Sackner-Bernstein, MD, Associate Center Director for Postmarket Operations, FDA Center for Devices and Radiological Health |
| 8:30 AM |
Topic 1: Clinical Diagnosis of ICD and Pacemaker Lead Problems
|
| 9:15 AM |
Topic 2: Returned Product Analysis
|
| 10:00 AM | Break |
| 10:15 AM |
Topic 3: Adverse Event Reporting
|
| 11:00 AM |
Topic 4: Post-Approval Studies, Registries and Home Monitoring
|
| 11:45 AM | Instructions for Breakout Sessions |
| 12:00 noon | Lunch |
| 1:20 PM |
BREAKOUT SESSIONS Topic #1: Clinical Diagnosis of Pacemaker and ICD Lead Problems Topic #2: Returned Product Analysis Topic #3: Adverse Event Reporting Topic #4: Postapproval Studies, Registries, Home Monitoring |
| 3:00 PM | Break |
| 3:15 PM | Resume Breakout Sessions |
| 5:00 PM | Adjourn from Breakout Sessions |
Friday, October 3, 2008
| 8:10 AM |
Introduction for Day 2 |
| 8:30 AM |
Best Practices Report, Topic 1: Clinical Diagnosis of Lead Problems Q&A Session |
| 9:15 AM |
Best Practices Report, Topic 2: Returned Product Analysis Q&A Session |
| 10:00 AM | Break |
| 10:15 AM |
Best Practices Report, Topic 3: Adverse Event Reporting Q&A Session |
| 11:00 AM |
Best Practices Report, Topic 4: Postmarket Studies, Registries, Home Monitoring Q&A Session |
| 11:45 AM | Wrap-Up |
| 12:00 PM | Adjourn |
Contacts
The workshop organizers may be contacted at:
Megan Moynahan
Network Leader, Cardiac Electrophysiology and Monitoring Network
Office of the Center Director, CDRH
Food and Drug Administration
7520 Standish Place, HFZ-4
Rockville, MD 20855
Phone: (240) 276-8707
Fax: (240) 276-8855
e-mail: megan.moynahan@fda.hhs.gov
Mark Fellman
Scientific Reviewer
Division of Cardiovascular Devices
Office of Device Evaluation, CDRH
Food and Drug Administration
9200 Corporate Boulevard
Rockville, MD 20850
Phone: (240) 276-4060
Fax: (240) 276-4002
e-mail: mark.fellman@fda.hhs.gov
Proceedings
BACKGROUND
The October 2-3, 2008, FDA Workshop, Best Practices for Pacemaker and ICD Lead Postmarket Surveillance brought together stakeholders from FDA’s Center for Devices and Radiological Health, the medical device industry, and the clinical community to hear some of the challenges to postmarket surveillance of these products and to identify best practices. Four topic areas were covered: clinical diagnosis of lead problems, returned product analysis, adverse event reporting, and postmarket studies, registries, and home monitoring. Participants separated into breakout sessions to discuss questions specific to these topic areas. All groups were tasked with answering two Key Questions: What best practices can be adopted? What critical steps require improvement? The results of these discussions were presented to the larger group on October 3, and are summarized here.
Topic 1: Clinical Diagnosis of Lead Problems
To get a firmer understanding of current practices, the group began with a series of questions designed to characterize typical practices of lead diagnosis. Clinicians typically follow several basic procedures to obtain information about the patient and the implanted lead. Simple observational practices to detect constant or intermittent abnormal lead behavior include taking a patient history and exam, conducting a device interrogation, going through archival data and trends, and collecting remote follow-up data. Active investigation practices can be used to detect constant or intermittent abnormal lead behavior. These practices include obtaining measures of pacing / sensing threshold and impedance, having the patient undergo physical maneuvers while taking repeated measures, looking at static and dynamic x-ray imaging, inducing arrhythmias and conducting defibrillation testing, and surgically exploring the site to examine the implant (insulation, set screws, pins). Other practices include developing education and procedure standards for physicians, trainees, physician assistants, nurses, and technicians.
Several types of pacemaker and ICD lead failures can be assessed by clinicians, including dislodgement, perforation, conductor fracture and insulation breach, all of which can potentially be assessed passively through remote monitoring. Through active assessment, problems such as micro-dislodgement, partial rupture or fracture of conductors, intermittent problems, terminal pin and set-screw problems can also be diagnosed. Some problems such as abrasion and terminal pin and set-screw issues can only be definitively evaluated intraoperatively.
Participants discussed what data are needed to make a definite clinical decision that there is a lead malfunction that is not attributable to the patient’s clinical condition. The discussion centered on the fact that the patient’s underlying condition can mask lead malfunction in some cases. Understanding the patient’s history is key, including what types of medications have been taken recently, what procedures they have been exposed to (particularly those that may induce electromagnetic interference to the pacemaker such as MRI scans, cardiac surgery that can damage the lead, or exposure to radiation), and whether they have a local or systemic condition such as acute renal failure or myocardial infarction, which can impact sensing / pacing, and the presence of arrhythmias.
While the clinical diagnosis of lead malfunctions was acknowledged to have some challenges, the group felt that the bigger obstacle was in the documentation of lead problems. There are several impediments that prevent consistent documentation of lead malfunctions. Most leads that do not function according to specifications are abandoned and not removed, and are typically not reported to the manufacturer. For example, the majority of patient follow-up visits occur in the outpatient clinic and device information may not be integrated with their hospital electronic medical record. Patients may be transferred from physician to physician and the previous interrogation may not be sent to the next physician. Many hospital records, procedure reports, and operative notes lack details about the lead itself and, importantly, there are no fixed standards for reporting. When adverse events or device failures are noted, the company representatives submit the reports, sometimes with limited input by the attending clinician. This can be exacerbated if a different manufacturer’s lead is used with the pulse generator. Overall, lead diagnosis findings are not consistently documented.
When explanted leads are returned to the manufacturer, the clinical findings are often incomplete, absent, or not sent back with the device. When leads have lasted close to their expected survival, they are generally not returned to the manufacturer. The best data are often obtained when calls are placed to the company’s technical services department, but this is often not done. Sometimes leads are partially extracted or damaged during extraction and physicians may not know whether these samples should be returned. Likewise, physicians many not report cases of lead failures when they are successful creating a work-around that does not compromise patient therapies (for example, not reporting cases of a dual-chamber device being reprogrammed VVI when there are atrial lead malfunctions).
The group identified specific obstacles and challenges to submitting complete MDR reports. There was felt to be a lack of physician training in this area to maximize the chance of physicians understanding and using best practices (fellowship training, ABIM prep and testing, IBHRE exam prep and testing, HRS training and education materials, HRS/ACC/AHA guidelines). Physicians may have uncertainty or confusion about which abnormal lead behaviors or root causes of failure are actually potentially related to lead design (for example, underreporting dislodgements and perforations). There was concern expressed of the potential for overlapping or multiple reports sent in for the same case, although this was acknowledged to be a rare problem. One example of this would be the scenario where the lead and pulse generator are both returned and the manufacturer submits two separate reports to FDA. This happens despite the fact that the 3500A form can be submitted to FDA for more than one device used in the same case. Another example would be if the clinician submits the report to FDA and the manufacturer also submits a report for the same case. When leads are physically returned to the manufacturer, there may be separate reports sent in for the initial report of the problem and the follow-up findings from the company. When leads are not returned, or if accompanying information is lacking, it is very difficult for the manufacturer to confirm the failure or malfunction described by the physician.
Participants discussed possible incentives to improve physician reporting of lead problems. Some suggestions included having a standardized reporting system, making a simple-to-use reporting form (to be added to the NCDR form currently used for ICDs), conducting education and outreach efforts to providers (by FDA, HRS, and industry) directed to grass-roots level or training centers, considering an HRS or FDA award for compliance in reporting, closing the loop with providers so they know their reports are valued, making improvements to FDA’s MAUDE database, encouraging a sense of belonging to a group such as MedSun, developing consortiums, and considering CMS penalties for not reporting.
Currently several disincentives exist that contribute to physician under-reporting. Physician “ego” was felt to be one contributing factor, with clinicians not wanting to admit that problems occurred with their case. Additionally, physicians may be concerned about malpractice, they may lack knowledge about how to report, they often have little time for additional steps needed to diagnose and report lead problems, they may not be interested or they may not see how reporting problems impacts them directly. When physicians do send reports, they often do not receive feedback from the company on their case in a timely way, and this may discourage future reporting.
This group devised a simple mnemonic to facilitate getting lead diagnosis information more consistently: LEADS, refers to:
- Loss of function: documented whether and how any suspected or confirmed lead defect results in compromise of clinically needed therapy, features or diagnostics
- pacing therapy
- sensing characteristics
- shock therapy
- features or diagnostics
- indicate SUSPECTED or DEFINITE
- Evidence: describe measurements, trends, visual or imaging observations that support diagnosis of lead defect.
- Action taken: including any reprogramming or surgery, including revision, removal, replacement or abandonment.
- Diagnosis: which of the few most common and important lead failure mechanisms (at least potentially relating to lead design) is causing abnormal lead behavior
- dislodgement (often heralded by loss of sensing and/or pacing, abnormal x-ray imaging)
- perforation (often heralded by loss of sensing and loss of pacing, high impedance, abnormal x-ray, abnormal echocardiography)
- conductor fracture, manifest constantly or intermittently (sensed high frequency noise, high impedance)
- insulation breach, manifest constantly or intermittently (low impedance)
- pseudo-lead failure – assure that conditions masquerading as lead defects are not the problem (i.e. set screw, scar)
- indicate SUSPECTED or DEFINITE
- Serial/model number.
This group was comfortable categorizing lead diagnoses into two levels of evidence, “suspected” and “definite”, but agreed that more work would be needed to guide physicians on how to make this distinction. Part of reporting should include a description of how the problem was resolved, for example through programming, replacement, revision, or removal. Lead malfunctions could be further classified as with and without compromised therapy.
When answering Key Question 1: What “Best Practices” can be adopted to improve the clinical diagnosis of lead problems?, this group suggested several measures to improve reporting, including better documentation, collecting complete data, using interrogation data, assessing patient symptoms, and following up on suspected or known lead problems with a device interrogation and chest x-ray. Overall, it was felt that lead diagnosis is not the problem, per se, but rather that reporting practices are not optimal. This group wondered whether future enhancements to programmers could include prompts for lead status, or suggest warnings if lead data are out of range. Patient education was also felt to be a key best practice.
In answering Key Question 2: What critical steps require improvement?, this group identified the need for better patient education, education of all health care providers, standardizing lead assessment steps, perhaps in combination with automated programming prompts, creating a simplified reporting scheme, reporting findings that don’t result in a procedure (e.g., “workarounds”), providing feedback to the reporting clinician, and tackling some of the other obstacles to reporting. Synchronizing with other ongoing efforts to create a quality assurance culture characterized by ingrained system performance reporting was also felt to be important.
In summary, although there are infrequent difficulties with making diagnoses of lead problems, the bigger obstacle is with documenting and reporting those problems. FDA, industry, and HRS should work together to standardize definitions and nomenclature for lead failures. Nomenclature should be developed into a simple stand-alone form and incorporated into the ACC/NCDR form for ICDs. A mnemonic was proposed as a simple way to capture most of the relevant summary information needed to characterize a lead concern. In the future, it would be desirable to have lead performance information incorporated into the pulse generator to allow surveillance by remote follow-up systems, and prompts on programmers to facilitate interrogation, documentation, and trouble-shooting. Ideally, there should be a change in culture among physicians that it is not only acceptable to report lead problems, but a good thing.
Topic 2: Returned Product Analysis
The group began with a general discussion of how returned product analysis (RPA) is used and why it is important that manufacturers, physicians and FDA use RPA for signal identification, trend information, and root cause determination. As part of the Quality System, RPA contributes to product improvements and new product development. Identification of failure modes can lead to new bench test methods. Physicians who return products can get direct feedback on their diagnosis of failed leads and may alter their practices with knowledge of RPA results. Physicians also can view returned product analysis summaries presented in most manufacturers’ product performance reports. FDA uses RPA results for its analysis of pre-market submissions for corrective fixes and for evaluation and classification of recalls. It was widely agreed that returned product analysis was a valuable tool. The group proceeded to follow the returned product process from clinical diagnosis to reporting.
The group discussed where returned leads come from. It was felt that a majority of returned leads come from the EP lab, operating room, or hospital. A high percentage of leads which are implant failures are returned for facility reimbursement. A lower percentage of leads get returned from operating suites at the time of lead revisions or extractions. In both cases, the lead is returned at the request of the physician, but the return is usually executed by the manufacturer’s technical representative present at the time of the procedure. It was noted that technical representatives are trained to return explanted leads, including competitor’s leads. It was felt that few leads are returned from the mortuary due to there being no requirement to extract them, and the added work, expense, and family sensitivity issues surrounding extraction. A few in the group helped to make a rough guess that about 80% of all explants occur at about 50 extraction specialty centers nationwide. The group agreed that a focus for increased rate of returned leads at these centers might provide the most potential improvement in return rate.
The working group next discussed who makes the decision to return leads that are extracted or implant failures. The group agreed that the physician makes the primary decision to return leads. The technical representative is usually tasked by the physician to make the return, and he/she fills out the returned product form which accompanies the returned lead. Though technical representatives are trained to return leads, they are not in a good position to initiate the return and may not have access to clinical diagnostic information that would be helpful to return with the lead. Morticians also make decisions whether to return leads. In most cases, even when the pulse generator is explanted, such as for instances of cremation, the leads are usually left implanted. AdvaMed has recently been advocating for more returned pulse generators with mortician associations with little improvement. It was felt that this was not a good area to focus on for getting back more returned leads. Physician training/education on the importance of returning leads was discussed as a potential way to increase the return rate for removed leads.
The group next discussed why removed leads would not be returned to the manufacturer. From the physician perspective, returning the lead is not a high priority compared to the patient welfare at the time of the procedure. Some physicians may not return leads as their routine procedure or they may not be interested or may not feel that there is any benefit from their effort. In some cases physicians may not return leads thinking they are not of value, such as leads that are extracted but fully functional, or leads that are severely damaged or in multiple fragments. The group agreed that all leads or lead fragments removed from the body are of value for returned product analysis. Technical representatives present in the operating area are trained to accept and return leads but often are busy supporting the physician and not comfortable initiating the return of a lead. Manufacturers insisted that technical representatives are adequately trained to return leads from the field when they are available but acknowledged that the rate of return could be improved. Another set of reasons why leads would not get returned included potential legal issues such as liability, facility policy, device holds, and HIPAA restrictions. Some members of the group suggested that the limitations posed by these legal issues, as perceived by user facilities, may not be real and should be investigated further. Lack of physician feedback on the results of the returned product analysis was also given as potential reason for lack of returns. Manufacturers discussed their procedures to nearly always provide the returned product results to the physician. The level of detail of these reports was discussed. Many physicians were said not to want detailed reports but a smaller percentage were thought to consider a short summary inadequate. It was agreed that lack of feedback to physicians could reduce their incentive to return leads.
After discussing why leads do not get returned, the group considered how returned product might be promoted. A concern was that any promotion needed to be clearly aimed at returning leads that are explanted or not implanted, but not in any way promote unnecessary extraction of leads. Options were discussed about how organizations could promote returned product by the Heart Rhythm Society (HRS) to their physician members, by the FDA through public health notifications, and by industry to their technical representatives and physician users. Processes to provide returned product analysis reports to physicians were discussed, including the potential to unify the analysis reports, which was not embraced by industry due to the differences in leads and analysis methods. A question was raised as to whether FDA could provide feedback to reporters for MDR submissions that involve returned product. Some group members considered the current rate of lead returns acceptable for signal detection and that the quantity of leads returned was not as important as the quality of return information coming back with the lead. Others in the group emphasized that higher return rates would yield higher probability of signal detection and potentially better trend information for multiple failures.
The topic next shifted to what information should be sent back with a returned lead. It was stated by many in the group that inadequate or even no information coming back with a returned lead was common. It was agreed that the availability of complete clinical diagnosis information together with the returned product could be critical to making a failure determination, especially if the lead was damaged during explant. Suggested information of value included: clinical history, physical observations, interrogation information, operative report, images, and data from remote monitoring prior to explant. The group discussed why this type of information typically did not come back with returned leads. It was stated that the information was difficult for the technical representative to collect as the procedure was in process, and that this type of information could be requested by the RPA lab retrospectively. The extra work involved in getting the information at the time of the procedure or after the fact is a big contributor to why supporting data is often not collected at all. An associated factor could be that many physicians do not recognize the importance and value of this information.
The group briefly talked about their perception or knowledge of rates for returned leads. It was estimated that about 40 to 50% of registered high-voltage (defibrillation) leads known to be explanted are returned, compared with an overall return rate of about 10 to 15% of out-of-service high-voltage leads (including deaths). The group felt that return rates increase during advisories and that return rates are currently up. In general, leads have a higher return rate soon after implantation. The group also felt that low voltage leads (non-defibrillation leads) are returned less frequently. Reasons for this included that pace-only systems may be implanted by non-EP specialists and that right atrial leads may not be replaced when they fail.
The industry representatives did not believe it appropriate to develop a standard set of inspections, measurements, and images to obtain for all returned leads because the nature of the clinical complaint drives the testing that is performed on the lead. Each company has their own set of tests and procedures; there are some similarities but they are not identical. Manufacturers said the depth of investigation depends on the condition of the lead and the suspected or reported problem. They expanded that all anomalies are noted regardless of reason for return as part of the RPA and quality system process. Further, all returned leads were retained in storage for potential further analysis of trends or failure modes.
The industry representatives also did not believe that a standard returned product analysis report was appropriate. RPA reports vary by manufacturer and are used as part of their internal quality system. Most companies stated that their RPA report has a summary which includes event description, summary of testing, and conclusion. In most cases the critical RPA report content was considered to be the answer to the question: Was the reason for return confirmed? It was mentioned that this summary often includes terms like ‘no anomalies found’, ‘met specifications’, or ‘inconclusive’, which could be hard to interpret if the lead was damaged or incomplete and if no clinical information was obtained. It was discussed that there is some work underway with AdvaMed to standardize nomenclature, definitions, and reporting categories for returned product reports.
After talking about returned product reports, the group discussed the process for providing the results back to the submitter. Manufacturers stated the normal practice is to provide a short summary letter approximately 4 to 6 weeks later, either directly to the physician or via the technical representative. Some companies provide detailed information by request and others included both a summary and details in one report. Manufacturers reported that many physicians specifically asked for summaries and not full reports. It was suggested that obtaining an electronic RPA report with a summary and detail by email would be desirable to some physicians. Companies could poll physicians to see what level of detail and delivery format they prefer for RPA reports.
A final discussion area concerned how returned product results are presented in product performance reports. A summary chart was projected from Dr. Sharma’s morning presentation which originated from an AdvaMed subcommittee for leads terminology, definitions, and reporting. Common reporting formats were considered beneficial by the group. Further, the group discussed the desire to identify and link when RPA data is associated with Medical Device Reports (MDRs) and Product Performance Reports (PPRs). Some suggested RPA acknowledgement letters that would indicate that the information had been used within an MDR or PPR report.
Through the duration of the working session for the Returned Product topic, best practices and key areas for improvement were collected. These ideas were discussed at the end of the session and a set of best practices was prepared and presented on the second day of the workshop.
- Promote collection and return of more complete data with returned leads. The group generally agreed that information coming back with leads was limited or missing in many cases and that collecting it weeks later as part of the returned product analysis was not working well. The group recommended developing an ideal set of information to collect for returned products including clinical history, physical observations, interrogation information, operative report, and images. The group recommended identifying a process to make this information available to the technical representative at the time of return so that the data can be included when the product is returned. The group recommended that physicians and technical representatives should receive training/education on the new process that is developed. Common lead return kits should be considered to standardize the process.
- Promote increased return rate for extracted leads. The group recommended that physicians be educated on the importance of returning leads as part of their routine practice with emphasis that return of intact working leads and lead fragments are just as important as return of a complete defective lead. The group recommended that industry should examine and improve their process to provide feedback to physicians for all returned leads, possibly adding electronic/email reports with both a summary and full details. The group recommended continued education and emphasis to technical representatives on the importance of returned product analysis of leads from unsuccessful implant attempts and for leads that are extracted, including intact, damaged, or lead fragments. The group recommended that a focused effort to promote returned product should be implemented at specialized extraction centers.
- Promote lead returns with extraction device manufacturers. The group noted that extraction device manufacturers usually have representatives on-site at the time of lead removals. The group recommended that these firms be brought into the discussion concerning the promotion of returned leads. The group recommended education and training for technical representatives from these companies as well as the provision of common lead return kits and forms.
Topic 3: Adverse Event Reporting
The group began by characterizing the typical steps involved in reporting adverse events for pacemaker and ICD leads. It was noted that companies have a decision point at which they need to determine whether an event is reportable to FDA. That decision is made based on the information known at the initial investigation. The initial report is sent to the FDA within the mandatory 30-day window or the next scheduled Electronic Data Import (EDI) submission date. This may lead to over-reporting of events when subsequent investigation may determine that the complaint is not required to be reported. Companies may request additional information from the physician or hospital, either to satisfy their own questions about the case, or in response to FDA queries. It was felt that there is a kind of “sweet spot” for collecting this type of follow-up information; it needs to occur soon after the report is filed, so that it is still relevant and readily acquired. The follow-up information is obtained and sent to the FDA in what are called “supplemental reports”. The information in the supplemental reports sometimes indicates that the event did not actually meet reportability requirements. In general, the supplemental reports contain more complete information about the case: the initial report typically has the “symptoms”, while the supplemental report has the “diagnosis” of the problem. FDA currently experiences significant delays in processing and data entry for supplemental, but not initial, reports.
Cardiac rhythm device companies also avail themselves of other reporting techniques for implantable leads and generators. Alternative Summary Reporting (ASR) refers to the practice of submitting certain kinds of reports in a summary fashion, rather than as a collection of individual cases. The only kinds of reports allowable under this method are for the patient problem codes for pocket erosion or infection. The summaries include limited individual event information. Companies must request and receive permission from FDA’s Division of Surveillance Systems (DSS) to utilize ASR for these well-defined adverse events.
One proposal of the working group members was to consider extending the 30-day window in order to obtain more complete reports to begin with. This would only affect companies not involved with established agreements for bi-monthly or quarterly EDI schedules. Other proposals were to encourage future use of alternate electronic submission processes such as eMDR. The group identified a critical step needing improvement: supplemental reports should be connected better to the initial report. This is a recommendation to FDA to examine the process by which this occurs and to institute improvements in the process. Encouraging industry’s use of eMDR would also help because the supplemental reports go directly into MAUDE and don’t require FDA to process them individually. The group agreed that eMDR has potential, but requires improvement and debugging. New problem codes are anticipated soon and there was hope that the new codes, along with standardized terminology, would make reporting practices better.
The group tackled the question, “How is the decision made that an adverse event is reportable to FDA?” Industry’s complaint handling departments are responsible for this step. The group discussed whether the terms “suspected (failure)” and “confirmed (failure)” would be useful. There was some concern expressed over what constitutes confirmation of a failure. One possible best practice would be to incorporate those terms into the standard codes used for adverse events.
Participants discussed implementation of the new coding manual for device problem codes and patient codes. Industry members asked FDA how it uses the MDR reports. It was explained that FDA uses MDRs to track or trend product performance similar to the way industry follows their Corrective and Preventive Action (CAPA) process. FDA uses MDRs to connect the dots with other activities of firms, such as device modifications. Lately, FDA has been requesting information about MDRs in some premarket applications. When MDR information comes in unsolicited in a premarket application, FDA conducts its own analysis to confirm the associated trend reported by the firm. FDA expressed that event problem codes provided in reports are frequently non-informative. Different codes are used for similar events across industry. FDA believes it is reasonable to expect a specific patient problem code to be provided in reports for death or serious injury, and a specific device problem code to be provided in reports for malfunction. The group proposed that FDA and industry work together to develop code usage guidelines, or some other educational piece, to help industry understand what codes are expected for the typical failures seen in these products.
The group discussed what criteria are used to determine that a device failure was related to the reported event. Often, causality is inferred from the patient narrative, which may provide details obtained from the clinician. In terms of how a CAPA is opened, most firms use a risk-based method that is part of an ongoing quality systems and statistical trending. Trends in adverse events are tracked by the manufacturer through their CAPA process. The group offered that one best practice would be to include MDR summaries in premarket applications for device modifications that were motivated by field issues. Another best practice that would help both industry and FDA to connect the dots would be to include premarket submission numbers in response to request-for-additional-information letters coming from FDA’s Office of Surveillance and Biometrics.
When available, information from returned products is used to complete the report. The group identified several obstacles to receiving returned products including the fact that hospitals won’t return a device without patient consent or for legal reasons. Companies’ technical representatives are primarily responsible for recovering devices, as this is part of their training. While some pulse generators are now being returned from funeral homes, this is not the case for leads. There is a perception that Health Insurance Portability and Accountability Act (HIPAA) concerns prevent the return of devices to manufacturers. A best practice would be for manufacturers to create a standard letter to send to user facilities that are concerned about releasing devices due to HIPAA. An analogous letter could be considered in an effort to reach out to funeral homes.
With respect to remote monitoring, this group felt that baseline values for certain adverse events could be obtained using data from this source. The group encouraged industry to continue exploring ways to use the data from remote monitoring systems to confirm performance problems.
Other recommendations from the group included having FDA learn about or adopt the existing terminology that manufacturers use for adverse event reporting, and encouraging industry to continue publishing Product Performance Reports on leads.
Topic 4: Postmarket Studies, Registries, and Home Monitoring
The group began by characterizing the purpose of postmarket studies and how the data are typically analyzed. Most postmarket studies are used to inform premarket decisions, to support the design of new leads. The data are analyzed every six months and published in Product Performance Reports. Data coming to the FDA are also typically provided every six months but it is important to note that this is not because of pre-specified interim analyses. Manufacturers perform a nightly integration of adverse event reports they receive.
Participants discussed whether there is a role for remote monitoring of devices for postmarket surveillance, and it was generally agreed there was. However, the manner in which remote monitoring could inform postmarket surveillance depends on the postmarket questions being asked and expectations for the study. It is important to understand how monitoring information relates to clinical performance, and the data are not a substitute for physician intervention. These systems provide information on device performance. Remote monitoring allows for screening out patients who are doing well. Alerts can be passed on to physicians so that they may pursue additional follow-up as necessary. Currently, the primary role of industry is to follow up on transmission errors, not to triage the reports themselves, which are monitored by clinicians. The remote monitoring alerts are sent to physicians via an Internet portal. In order for remote monitoring to provide data to analyze and predict lead performance, the identifying source of data (lead type, serial number, implantation site, etc.) must be reliable. There is currently no automatic method available to link this data to the home monitoring measurement information because manufacturers have no processes to verify the validity of manual entries.
The group discussed the obstacles that prevent more widespread use of remote monitoring. To sift through the information to find cases of device failure, one needs to develop tools and algorithms with acceptable sensitivity and specificity to assess performance. How do we define lead abnormalities? Each manufacturer’s devices capture different data and none of them currently capture information about which lead is being used in conjunction with the pulse generator. There are numerous failure modes and alerts, all of which prohibit standardization. There is a need for standardizing reporting. Moreover, right now these systems do not have information linking an alert to respective clinical events. Information gleaned from remote monitoring needs to be integrated into FDA decision making.
Members of this group discussed how registries, postmarket studies, or remote monitoring systems could be used as early-warning sentries for emerging postmarket safety issues. It was generally felt that each surveillance system has a role, but that none of them are specifically designed to address early-warning issues. Typically, companies’ CAPA systems address early warning issues. It was felt that there was potential to use remote monitoring in this way by accumulating enough data to identify emerging issues.
Several best practices were identified to improve the utility of postmarket registries and studies. First, the group identified that the last few postmarket studies have not been ideal for FDA or for industry. Future studies can have differing nuances depending on what questions need to be answered. The studies should be designed to address questions that all stakeholders want answered. Postmarket plans submitted to FDA should have sufficient detail. It was agreed that study designs could be designed more “generically” to capture endpoints of common interest across leads. If other questions need to be addressed, the study could be expanded or modified. The group recommended developing guidance for postmarket study and registry design, including specifying common endpoints and common design elements. It was also suggested that criteria be established to know when postmarket studies are required.
There were several identified critical steps that were felt to require improvement. Companies were encouraged to have early and frequent communication with FDA about potential postmarket questions. Home monitoring could be considered a complementary tool to answer some postmarket questions. The group felt there could be improvements in the operational timeline for devising and carrying out a postmarket study. There is a need for improvement to the costs and time associated with obtaining IRB approval for these studies. There has been previous consideration for a centralized IRB, but many IRBs do not want to give sole authority to a centralized IRB. In the past, Dr Daniel Schultz, Director, CDRH, has provided a letter addressed to IRB Chairs providing general information about why postmarket studies are conducted. One recommendation was to revise that letter to be more specific with respect to authorization of a central IRB. The group asked, What steps can be taken on behalf of postmarket staff and sponsors to alleviate concerns about selection bias? While generalized recommendations were not identified, it was felt that this issue could be addressed on a case-by-case basis. The possibility of using home monitoring data for non-consented patients was also raised.
In terms of next steps, this group felt it was important to define expectations around postmarket studies. For example, what types of leads would require a postmarket study? What kinds of modifications to leads would require a postmarket study? What type of postmarket plan should be in place at the time of PMA/Supplement submission? Are there core minimum dataset requirements? Would there be different sample sizes and follow-up time as a function of risk? What are the clinically relevant point estimates and confidence intervals for primary endpoints? Participants felt it would be important to put out these recommendations in an informal guidance, perhaps in conjunction with an FDA advisory panel meeting to allow discussion of the issues.