Earlier this month, a group of leaders from the world’s in vitro fertilization (IVF) community and a group of engineers from IBM convened a virtual meeting to discuss safety; in particular, the need to reduce misidentification of IVF specimens to zero using a solution called ivfOPEN.
To place the importance of effective specimen identification into perspective, consider the names Jessica and Jennifer Smith.
According to Social Security administration records, about two and a quarter percent of female babies born in the United States in the 1990’s were named Jessica or Jennifer. During that same decade, the most common last name was Smith, which accounted for 1% of all babies. That means that there are about 4,500 Jessica or Jennifer Smiths between 20 and 30 years old in the United States.
In the coming decade, if we apply a 15% prevalence of infertility, 675 of these women could need IVF. Assuming a generous 50% success rate per IVF cycle in the coming decade, these women could account for 1350 cycles. Each cycle will produce between three and six different types of tissue that may need storage for the later use (eggs, embryos, semen, trophoblast, testicular biopsy, ovarian biopsy), meaning that the 1350 cycles could produce an aggregate 8,100 samples, all labeled Jessica Smith or Jennifer Smith. Depending on the penmanship of the labeler, it may be difficult to tell one person’s specimens from another’s. More alarming: each of these 8,100 samples generates straws for eggs (assume 12 per egg freezing cycle), one or more sperm vials when sperm is frozen, straws for embryos (assume 4-6 per cycle), and one vial per embryo when biopsied for PGT. Each of these is a potential labeling error, as are the multiple dish and other vessel-labeling steps. In sum, the 675 Jessica and Jennifer Smiths could represent tens to hundreds of thousands of separate instances where a specimen could be mislabeled or transferred to another patient.
We know that patients switch clinics and that their frozen sperm, eggs and embryos travel with them, so we cannot tell for sure how these 8,100 samples will be distributed between the 450 or so IVF clinics in the country. Nor do we know how many Jessica or Jennifer Smith specimens may be shipped to the United States from other countries.
But we can assume that as the number of stored specimens continues to rise, the risk of mistaking one Jessica or Jennifer Smith’s specimen for a different Jessica or Jennifer Smith’s specimen in a given IVF lab will rise too.
There are a lot of common names; there are almost as many Ashley Smiths as Jessica Smiths, and there are more Emilys, Sarahs, and Samanthas together than there are Jessicas and Ashleys. And Smith is far from the only common last name. Census data shows that in 1990, .8% of babies were Johnsons, .7% were Williams, and .6% Jones and .6% Brown. Add Davis and Miller and we are close to 5% of the population. And as IVF cycle volumes increase for infertility, fertility preservation in oncology, genetic disease prevention, LGBTQ fertility management and proactive management of risk for future infertility by using egg freezing and storage, specimen mixups will increase for these names too.
Mistakenly fertilizing eggs with a stranger’s sperm, transferring another patient’s embryo to a woman expecting her own genetically-related embryo, or having to destroy a specimen because of uncertainty of origin should be never events, like amputating the wrong leg or operating on the wrong side of the brain. Sadly, however, a Google search for “IVF mix-up” yields 349,000 results.
And ivfOPEN aims to fix that. How? Some background first.
IVF and computers
The information age arrived a few years too late for IVF.
Mary Louise Brown, the first IVF baby, was born in 1979 into a world of typewriters, clipboards, pencils and erasers. IVF laboratory data over most of the next decade consisted of a few basic numbers (number of eggs, concentration of sperm) that everyone understood, and descriptions that meant something a little different to everyone who encountered them. One embryologist’s “viable day two embryo” was another’s “mildly fragmented asymmetric four-cell embryo” and yet another’s “grade AB-.” But since no one else saw the data, it didn’t really matter how we recorded it.
In the early years of IVF, we collected sperm and eggs and made embryos, which were all either used or discarded within a few weeks. Some embryos turned into term pregnancies. Over time these pregnancies accumulated identities: names and numbers and pictures. Embryos that arrested in the lab or failed to develop after transfer left a legacy of a few lines of obscure, arbitrary text, scribbled into the patient’s chart.
Prior to 1979, we already knew how to freeze, store and thaw sperm. Over the next two-and-a-half decades, we figured out how to do the same for embryos, and later for eggs. Then we learned how to query the eggs and embryos, to test for genetic information. These new skills coincided with gradual, steady improvement in the rate of successful pregnancy for each embryo transferred, an efficiency that resulted in transferring fewer embryos per cycle, which in turn resulted in more embryos that could be saved for future use.
This new proficiency created a new challenge, however, that of developing a common language to precisely and consistently describe, track, inventory and quality control what became an enormous worldwide inventory of embryos, eggs and sperm, a collection that is growing at increasing rates, and has gradually overwhelmed the rudimentary systems that have been developed piecemeal from one isolated IVF center to the next.
We have yet to conquer that challenge.
Accustomed to tracking outcomes of entire cycles, both as a legacy of the early, pre-freezing IVF era and to comply with outcome reporting requirements of the United States Centers for Disease Control (CDC), IVF clinics’ systems for data collection, quality control, and inventory management on the individual specimen level varied from clinic to clinic, or changed with changes in laboratory leadership.
We have always been good at answering “pregnant: yes or no?”
We’re not great at “egg or embryo: good, bad, or what? And why?”
So while our science has progressed, permitting us to perform intracellular and intra-embryonic surgery, to infer ever greater amounts of information from the eggs and embryos that we store, and make possible one-embryo-at-a-time efficiency, our data collection remains in an information wasteful past.
This relative neglect of specimen level data analysis was exacerbated by the nature of the data itself, which remains descriptive and subjective, far from ideal for standardization, rigorous analysis and rational decision making.
In an ideal world, our ability to retrieve and store data about the eggs and embryos that we keep frozen in laboratories would have been developed at a similar pace as the bioengineering that created these new sources of data. Unfortunately, we have not yet transitioned from the analog/descriptive/pattern recognition/I know it when I see it language of the 20th century.
And this haphazard approach to keeping track of sperm, eggs and embryos is causing problems at the most basic level: keeping track of what’s where and what belongs to whom.
Enter ivfOPEN: the concept
The combination of a huge and rapidly growing number of frozen and stored specimens spread among thousands of IVF labs and storage facilities worldwide, labeled inconsistently, lacking a best practices method for inventory management or protocol for specimen and label transfer from one site to the next points to one very simple and obvious solution: an industry-wide system to generate a unique identifier for each specimen at its point of origin, generating a universally queryable source of true identity that follows the specimen anywhere in the world, with no possibility of identifier duplication.
Shifting IVF record-keeping from the cycle level to the specimen level should be the first, and arguably the most important, step in eliminating specimen identification mistakes and the “never errors” that follow those mistakes.
But how do we implement this?
Before thinking about a solution, what’s most notable about the need for a robust identifier and tracking system, a “single source of truth” regarding specimen identity, is that this concept has been in existence in one form or another, throughout industry, public health and government, for decades. The first social security numbers were assigned in 1936. VIN numbers for automobiles were introduced in 1954. ISBN numbers for books started in 1970. The first UPC grocery scan was performed in 1974 on a pack of gum in Troy, Ohio. How can we justify not implementing a single source of truth system for IVF?
That said, how do we implement this?
Enter IBM, and blockchain.
Unique specimen tagging and tracking is not new; it has been successfully implemented in much larger industries; food is an excellent example. Bacterial contamination in the produce aisle of a New Jersey supermarket can be quickly traced to a lettuce field in California and every transport vehicle in between, using a system designed by IBM, part of an overall safety and surveillance infrastructure that identifies and records voluminous data about our food supply, but that limits access to sensitive data through an unimpeachable permissions architecture. IBM also recently piloted a similar system with the FDA using blockchain to track prescription medication in the US. (Perform a Google search for IBM and blockchain for a much more thorough discussion of blockchain and how it works to organize and protect the integrity and confidentiality of massive datasets.)
IVF procedures include dozens of procedures and interventions, most of which involve some degree of identity check. By replacing handwritten labels or sharpie-generated petri dish identification systems that may rely on some combination of patient name, date of procedure, or birth date; inputs that will almost certainly cause duplication errors as the sheer number of specimens increases and they move from one lab or storage facility to another ― by replacing these with a computer generated and electronically tracked (think bar codes or RFID tags) identification system we can rationally engineer the elimination of specimen disposition mistakes.
This digitalization of IVF specimen identity, applied throughout the IVF ecosystem, is the vision of ivfOPEN. Implementation will involve creation of a point-of-specimen creation system for unique identifier assignment, and application/verification of that identifier at each step in specimen contact, as well as management of the data generated and control of access to those data through a strict permission structure.
Integral to the success of ivfOPEN is maintaining its narrow scope. IVF is a series of interventions, performed by a consortium of professionals, service providers, medication, reagent, equipment and device producers and managers. Ideally, ivfOPEN will improve the efficiency of each step of this operations chain, improving the safety and certainty of all of them without competing with any of them.
Naturally, a cooperative plan like this raises questions.
Not every clinic writes patient names on sticky notes. Some have automated systems already. Where does ivfOPEN fit there?
ivfOPEN would integrate with the existing system to generate the identifiers themselves, insuring that no two specimens anywhere in the world had the same identifying label. The ivfOPEN system should be value-added to any existing labeling or witnessing system.
Is this one more IVF cycle “add-on” that will increase costs to patients?
Quite the opposite. The plan for now, pending legal and engineering advice, is to make the ivfOPEN a non-profit. After an initial build-out period paid for by industry-leading companies, the system will support itself with tiny fees (likely $ 1 or less) each time the system is used for an identification verification (the equivalent of using an iPhone to scan a bar code.)
Why would companies invest in this technology if they don’t own it?
The short answer is that these businesses profit from the expansion of the IVF industry. I have modeled the US IVF underlying demand for IVF to be well over one million cycles per year; currently we do less than 300,000. Recently, expansion in employer-based IVF insurance coverage for IVF and egg preservation, increased use of IVF to prevent genetic disease, preserve fertility for oncology patients and broaden family building in the LGBTQ community together presage an acceleration in IVF demand. To the extent that this increased demand can be met efficiently, and — more importantly — safely, avoiding specimen loss and accidents from specimen identification errors, using innovation, then that innovation will be supported by industry incumbents.
How about the doctors and embryologists? What’s in it for them?
Three benefits immediately come to mind: 1) risk mitigation. Whether or not the safety benefit stemming from precision specimen identification translate into actuarial tables and insurance savings, they will reduce the risk of rare catastrophic events that can ruin a clinic’s reputation and result in lawsuits 2) the ivfOPEN process will replace a legacy specimen identification system that was not designed to eliminate duplicate identifiers from specimens transferred in from other facilities and 3) the system will be engineered for rapid and seamless workflow (recall the iPhone – bar code comparison.)
What about privacy issues?
As we have learned from IBM, the combination of “scrubbing” of the identities underlying the identifiers and establishing a strict, non-tamperable permissions architecture so that the only queries that are possible are those that preserve the integrity of the data.
So how do we make this happen?
The most important aspects: simplicity and humility.
ivfOPEN should solve one problem: design the most efficient method for assigning unique identifiers to specimens obtained or created during IVF and serve the entire IVF community: patients, doctors, scientists, staff and businesses, in a value-added fashion that does not interfere with or compete with their goals; to facilitate a future of safe and accessible IVF.
disclosure: The author is co-chair of ivfOPEN.