October 15, 2015
On September 24, the Food and Drug Administration (FDA) convened a public advisory committee meeting to discuss safety concerns about the Essure — a permanent birth control device that is implanted in the Fallopian tubes in order to create a tissue barrier and prevent pregnancy. The meeting of the Obstetrics and Gynecology Devices Panel was prompted by complaints submitted to the manufacturer (Bayer) and the FDA about unusual side effects—including patient deaths—and other health concerns that were not reflected in the labeling.
The ad hoc nature of such meetings, as a reaction to over 5,000 complaints, is a product of the U.S.’s current device surveillance system which itself is ad hoc in nature. Once a device is approved for market, the FDA relies heavily on these voluntarily submitted complaints in order to flag any potential adverse effects not identified during initial testing — a system that, the agency acknowledges, may be “incomplete, inaccurate, untimely, unverified, or biased.” In recent years, the FDA has renewed its efforts to actively strengthen or revamp this system.
At the September 24 Essure meeting, medical device experts reviewed the complaints alongside scientific data from two post-market studies conducted by the manufacturer. Conspicuously absent from the discussion at this meeting, however, were the volumes of real-world objective data that should theoretically be collected—systematically, quickly, and cheaply—from health systems and health plans like ours.
Unfortunately, the main reason we at Geisinger Health System in Pennsylvania and other similar systems cannot quickly and routinely access that data today is that most clinical information systems (e.g., claims databases, billing systems, and electronic health records) reliably track surgical procedures but not the specific brand and model of the implantable device a patient receives. Essure, however, is a unique case which provides us with a preview of what a truly robust device tracking system would allow through the use of device-specific information integrated into these databases.
Essure As A Unique Test Case For Unique Device Identification
Electronic health records, billing, and claims records are not currently focused on tracking individual devices; instead, they depend upon standardized procedure codes (e.g., Current Procedural Terminology [CPT] or the Healthcare Common Procedure Coding System [HCPCS]) to track a patient’s health care encounters. For most procedures, knowing what surgery a patient had doesn’t help you know what particular device they received. The procedure code for a hip replacement surgery could mean the patient received any one of dozens of brands and models of prosthesis currently on the market.
Essure is a serendipitous exception to that typical scenario. Since its approval by FDA in 2002, there has been only one competing device on the market for the same indications, and thus only one other product associated with the same procedure codes as Essure (e.g., CPT 58565). That competing device, the Adiana system manufactured by Hologic Inc., was approved in 2009 and withdrawn in May 2012, meaning that for the majority of its lifespan, the Essure has been the only medical implant on the market using a particular subset of procedure codes.
With unique procedure codes, it is possible to track the patients who received Essure (and their medical care post-surgery) in a way that is currently prohibitively difficult for the vast majority of other medical implants. Because claims and medical records lack specific device information, they are typically not a very useful resource for device researchers or policymakers. But in this rare instance of Essure, it may have been possible to utilize those data sources to analyze this product’s safety profile.
What Claims And Electronic Records Can Tell Us
At Geisinger Health System, we have easy access, via a centralized data warehouse, to data from both the managed care company (Geisinger Health Plan) and a provider/facility network (Geisinger Clinic) that have used electronic health records since 1996 via a centralized data warehouse. We examined these records to answer a few key questions about the history of Essure use among our patients and members.
Searching the health plan’s records from January 2006 until June 2015, we found 67 health plan members who unambiguously must have received the Essure because of their procedure code. (This excludes 20 patients who had the procedure during 2009-2012 when the Adiana device was competing with Essure, although those patients may have received the Essure, too.) When searching the provider/facility network, there were an additional 118 patients who had an Essure procedure at a Geisinger hospital during that timeframe using a different type of insurance, bringing the total number of Essure patients to 185.
So what do we know about these 185 patients?
- All 185 are still alive.
- None of the 185 women have had the device subsequently explanted.
- We did not obtain contact information (e.g., phone, address) for these 185 patients for the purpose of this article, but we could have if needed, and in central Pennsylvania where the out-migration rate is traditionally very low, that contact information is very likely (though not guaranteed) to be current.
- Taking an ever-so-slightly deeper dive into the data, we can see that between their initial surgery and today, there were a total of 38 women (20 percent) with some evidence of unwanted events: 20 (11 percent) were newly diagnosed with abdominal or pelvic pain, 16 (9 percent) with heavier menstrual bleeding, 1 with device complications such as migration or breakage, 2 with postoperative infection, 2 with uterine perforation, and 2 became pregnant after receiving the device.
We realize this broad analysis is overly simplistic, and we do not imply causation between the new diagnoses and the Essure device itself, particularly the pain and bleeding which can be more common symptoms of other conditions.
Note, however, that we completed all of the above analysis—from raw data extract to final counts—in a matter of hours, not days or weeks, highlighting the minimal time and effort needed to answer basic questions about this device’s utilization history. The answers to those basic questions could be a first step toward determining whether more rigorous investigation is justified. We believe that our ability to quantify these “real-world” outcomes so easily shows what is possible when standard codes in claims, billing, and electronic health data can link individual patients to a specific device.
Why Can’t We Get The Same Level Of Detail On Other Devices?
In the digital age, many patients likely assume their medical implants (particularly the life-saving ones) are meticulously logged by their health care providers for easy future reference. The reality is much more complex.
To illustrate this complexity, we examined two other commonly-used implants in our health system: hip replacements and knee replacements. Over the past eight years (since 2007), 6,535 members of our health plan received a primary hip or primary knee replacement, 4,191 of which were implanted at non-Geisinger hospitals. An additional 4,544 patients had one of these procedures done at a Geisinger hospital, paid for by a non-Geisinger insurance plan, for a total affected population of 11,079 patients. This is a substantial number of “our” members or patients who could potentially be affected should a safety issue arise with one of these implants. Unlike with the Essure, however, for the 4,191 insurance members who had joint surgery elsewhere, we have absolutely no ability to identify within our existing records which model (or even manufacturer) of implant they received.
For the remaining patients who did have surgery at a Geisinger hospital, discrete implant information theoretically should have been captured in the operating room suite software, but the reality falls short. We were unable to find discrete documentation (i.e., not counting scanned images or text notes) of the manufacturer and model for the devices used in 995 (14 percent) of these surgeries. We attribute this sizable shortfall to the many challenges encountered in the operating room, including non-standardized labeling, sometimes-confusing packaging with multiple identification numbers, manual keyboard entry of information, and the pressure on operating room staff to develop workarounds when necessary to expedite the surgery. In any case, the consequences are that that for roughly half of “our” affected patients with joint replacements, we cannot easily determine what implants they have.
What Can Be Done?
The FDA has taken preliminary steps towards a strengthened medical device surveillance system with the phased introduction of Unique Device Identification (UDI) numbers for all medical products. The UDI will replace the current practice whereby manufacturers choose their own model or serial numbers (which could be duplicative across different manufacturers): a 2013 FDA final rule required labels to bear a UDI that conforms to a standard format and uniquely distinguishes that device from any other on the market.
The implementation of this rule will be phased in over seven years; September 24, 2015 (coincidentally the same day as the Essure meeting) was the date after which labels of all implantable, life-supporting, and life-sustaining devices must bear a UDI. Lower risk devices will follow suit in the coming years. The disconnect, however, from the perspective of health care providers and payors, is that the mere presence of a UDI number on a label may not mean that it actually enters the patient’s medical record or insurance claims.
Adding the UDI number to product packaging is not enough to improve our knowledge about devices, unless:
- Hospitals and health systems are committed to capturing and tracking that information;
- Health care software vendors work with those hospitals to help store and retrieve the information; and,
- Data standards organizations, such as Health Level Seven (HL7) and the Accredited Standards Committee (ASC) X12, help to develop tools for communicating that information between systems.
Integrating UDI all the way along the information pipeline including claims would allow us—and other researchers that utilize health plan data—to evaluate devices much in the same way that we utilize claims information for other purposes, such as examining prescription drug use and subsequent outcomes . Many organizations working to incorporate UDI into claims—including Aetna, the Society of Thoracic Surgeons, the American College of Cardiology, the National Association of ACOs, the Pew Charitable Trusts, and many others—argue that including UDI numbers in claims will provide better real-world data to evaluate long-terms outcomes associated with specific products.
Some have suggested that claims data are imperfect because patients may switch health plans or lack insurance, but as our demonstration highlights, claims data are already used for analysis and can provide earlier warning signs of product failures. Upgrading claims to accommodate UDI requires updates to the standard form used by all hospitals and health plans, a form that is updated by the Centers for Medicare & Medicaid Services (CMS), based on standards proposed by ASC X12. The form is updated infrequently with the next update expected in 2019 or 2020. Missing that window would delay the inclusion of this information—and analyses similar to our Essure demonstration—for at least a decade.
The main outcome of the September 24 FDA meeting—where testimony was heard from the Agency, the manufacturer, and a parade of witnesses and where a formal panel vote was not taken—was the recommendation that further study was needed. Even as the FDA considers the safety of a device that has allegedly harmed thousands of women, we should also consider how to identify problems with products more proactively.
Health care data such as claims and electronic medical records can provide that capability in a fraction of the time required to conduct a traditional surveillance study, but only efficiently and robustly once UDI data are added. Due to its 100 percent market share and an anomaly in its medical coding, Essure gives us a glimpse of how simple it could be if all device types could be discretely and uniquely linked to the experience of the patients who use them.
Response to “How Safety Concerns About Essure Reveal A Path To Better Device Tracking”
- Laura Henze Russell says:
- This is an excellent article. I testified at the FDA Essure Panel hearing. I did not have Essure, but have experience with another device illness due to biocompatibility issues, as have friends and associates. Even with Univeral Device Indicators, FDA regulations are still in the dark ages when it comes to patient safety. Patients need written information on materials composition and health risks, the right of written informed consent, pre-screening for biocompatbility to ensure it is “right for you” before a device is installed, and recourse and responsibility to fix things when they go wrong, instead of federal preemption that blocks responsibility to replace a device that is “wrong for you” through no fault of your own.