Part 1: artificial intelligence in IVF, private equity, the unexpected legacy of Covid-19 on IVF practice and the minefield of “add-on” pricing
Regarding AI in IVF, the good news for the IVF world is that the people developing artificial intelligence for IVF are outstanding: visionary, knowledgable about data, science, patient care and the how IVF cycles work. There are two problems though. The first is that, having met with over a dozen start-ups and early stage companies applying AI to IVF, I note that they are ALL visionary, knowledgable about data, science, patient care and the how IVF cycles work, and that none have yet sufficiently differentiated themselves versus their peers. This is great for the field, but difficult for each business.
Don’t get me wrong: I think IVF needs AI. The engineering part of IVF is starved of measurements and metrics, is awash in confounders and, despite steady improvement over the past decades, is inconsistent in its data collection — all factors that make process optimization difficult. AI, with an agnostic approach to data collection, measurement, and correlation, will over time make connections that that our eyes and brains and biases overlook. Unfortunately this mass analysis needs a tremendous number of observations, and “big data” in IVF isn’t, well, all that big. And that’s the second problem for AI in IVF.
We do, at most, 3 million cycles a year worldwide, and the cycles themselves vary quite a bit both in terms of the patients treated and in the protocols themselves. The algorithms that form the core of AI become more specific over time, as different patterns of uncategorized data points emerge and previous patterns thought to be predictive are discarded or de-emphasized.
We will undoubtedly reach a point where AI is an integral part of both short-term quality control and long-term process improvement, the holy grail of definable and reproducible best practices. In the meantime: more data is better, and all proposed solutions are early and immature. The silver lining is that a true A team is working on it.
Still no signs of Evil Private Equity: while we are unlikely to see any PE firms on the ASRM exhibit floor in Baltimore, private equity continues to invest in in vitro fertilization worldwide, attracted to its decentralized marketplace, high gross margins, enormous number of underserved and unserved patients, management systems that were optimal for stability but unsuited for growth and legacy owners who, in many cases, are looking for diversification of their assets and exits.
Depending on your vantage point and previous impressions of PE, this expanded footprint is either consistent with a natural and organic evolution of the IVF industry, given its size, need for scale and financial metrics, and can be good or bad or indifferent for patients and IVF professionals depending on the specific where and particular who; or it is a worrisome trend that will no doubt result in the type of “greed is good,” Gordon Gekko management style that layers unsustainable debt on the cash flows of the IVF programs, slashes (the word “slashes” is always used when discussing private equity) costs and cuts jobs with impunity, for the sake of short-term profit-and-loss statement improvements, and de-emphasizes quality of care and the patient experience.
For now, any movie about PE in IVF is likely to be rather dull: the transactions that I have been involved in (disc: not as an investor, usually as a participant in diligence calls) have emphasized quality of patient service and IVF cycle outcomes as the most valuable attributes in the investment marketplace, and that preservation of standards of excellence is the best way to participate in this business. Moreover, the PE-catalyzed consolidation that the industry has experienced for a decade has resulted in an overall improvement in the availability of optimally managed IVF cycles, and has contributed to the overall rise in pregnancy rates in the United States and elsewhere.
And, uncontroversial or not, the PE in IVF story will continue; you can make the argument (I have) that it will be grow more important. Consolidation of the industry is still in its early stages, the industry remains attractive to large pools of capital for the same reasons, the number of untreated patients remains one of the largest in healthcare, and IVF one of the most attractive business of healthcare opportunities in the economy. There may be a plot twist, however: Covid’s effect on IVF, and how it may affect how practices are run moving forward.
Watch what the Covid catch-up reveals about the IVF workforce: Was it really just over a year ago that much of the IVF industry essentially shut down for months as we tried to figure out the best way to balance keeping patients and staff safe, triaging resources to those acutely ill with Covid-19, and caring for patients whose probability of successful IVF treatment decreased over time? At the time, predictions for 2020 cycle volumes were scaled back significantly, and it was unclear how individual IVF programs would deal with the resulting backlog of patients that accumulated pre-summer of 2020. Now, a year later, the backlog is a distant memory. While we will not have the final cycles statistics for 2020 until sometime next year, public data from IVF supply and service companies like Vitrolife, Hamilton Thorne (disc: I am a board member and shareholder) and Cooper Surgical indicate that volumes rebounded quickly, as programs accelerated their operations to accommodate their patients quickly.
With United States IVF cycle volumes at around the 300,000 mark and the number of programs slowly decreasing (450 at my last count), the question as to whether this need for a temporary expansion of operations strained the capacity of the clinics, and — if so — what specific input was the constraint? If I had to guess, it would be that the number of oocyte retrievals and embryo transfers put pressure on the reproductive endocrinologists (RE’s) while the clinics ran at increased capacity; I base this on several complementary observations, quotes and anecdotes: hiring plans for large practices (if successful, a handful of the larger practices in the country could hire a very outsized percentage of all of the graduating fellows this year) coupled with conversations with the fellows themselves (who describe what appears to be a sellers’ market for RE’s fresh out of training, although they — a humble and patient-centric group, would never describe it that way.) Scattered quotes from Griffin Jones’ interviews on his podcast Inside Reproductive Health elicited what would have been a stunning fact a few years ago, that programs are increasingly relying, or planning to rely on, non-RE’s for some of the tasks that RE’s themselves did exclusively, in some cases patient intakes, in other cases specific procedures.
Whether this is a first step in the re-alignment of labor within reproductive medicine, and whether this realignment includes parts of the IVF cycle taking place offsite, whether in ob/gyn’s offices, shared monitoring spaces, mini-laboratories that perform oocyte retrievals and vitrification only, remains to be seen, but without an expansion in fellowship spots the labor market for double boarded fertility specialists will be increasingly constrained, and the other pressures for expansion (increased insurance coverage, expansion of proactive egg freezing etc) will dictate some re-engineering of who performs which procedures.
“Add-ons:” The debate about IVF add-on’s, additional procedures of variably defined efficacy, backed up by lots of or little bits of or essentially no convincing or unconvincing data, that are sold to patients in the hopes of improving the probability of a good cycle outcome, is not as loud in the US as it is in Europe, but it always comes up in founder-investor meetings. The debate usually includes:
a) Add-on opponent: IVF is expensive enough as it is
b) Add-on proponent: we need to address non-embryo development factors, particularly implantation and our (fill in the additional test) does that
c) Opponent: the claims seem exaggerated and are based on small studies with unrepresentative inclusion criteria
d) Proponent: the same could have been said for IVF in the 1980’s, ICSI in the 1990’s and have been said about PGD/PGT since it was invented.
e) Opponent: nonsense. You were able to measure IVF outcomes in tubal factor patients from the start; ICSI outcomes were isolated using fertilization rates, and PGD/PGT had convincing data in well-defined subpopulations from the start (translocation patients, >3 cycle failures using cell-stage biopsy and analysis in >35 yo women.) Most of procedures being added on now supposedly affect pregnancy rate, and because of cycle to cycle variability and confounders the studies would have to be much much larger than those being done in order to isolate the effect of the extra intervention (and the value of the extra cost).
Having engaged in these discussion since the first Clinton administration, I look at it more as a marketing issue rather than a healthcare issue, one that reflects the decentralization of IVF care from an everything under one roof performed by one organization model to a general contractor with subcontractor model. Back in the 1990’s I was fortunate (“blessed” is a more accurate description) to have Dr. Jacques Cohen’s “Bell Labs of Embryology” as part of our IVF program at Saint Barnabas in New Jersey, and Jacques and his team had an ever-replenished pipeline of ideas that were vetted as part of our practice. Some were minor variations of technique that the embryology team tested informally and gradually adopted or rejected; others were more major departures from standard operating procedure requiring informed consent, institutional review board review — these innovations were ultimately presented and published and sometimes adopted by others. Back then the economic benefits came from better pregnancy rates which attracted more patients — unless the procedures, once adopted, cost a lot to perform, the way ICSI and PGD/PGT required expensive micro manipulators attached to operating microscopes and an advanced level of training for the embryologists performing the procedures.
This model changed when the individual procedures moved out-of-house; the PGD subcontractor (disc: Reprogenetics eg, of which I was a co-founder in 2000) and provided blastomere or trophoblast analysis to the general contractor IVF program — for a fee, which was in turn passed along to the patient. In the past 20 years, many different tests and services have been offered on this basis; a model where instead of utilizing whatever in-house tools the program had at hand to give the specific patient the best possible chance of conceiving, all for the same bundled fee, the general contractor IVF program performed their in-house basic IVF cycle, added in extra tests and added the subcontractor’s test fee to the total.
Now no one wants their IVF program to be like a car showroom, where your Camry can come with the luxury package or the sporting package or what have you, or — worse — you have to choose whether to upgrade your steering without know a rack-and-pinion from an anti-lock brake. You just want a safe, comfortable and reliable car. In a perfect world the expected benefits of extra input X, whether it is a satellite radio or an endometrial scratch, should be evident and obvious to whomever is making the purchasing decision. In the imperfect world of IVF decision making, where the knowledge difference between the patient and the doctor or clinic is enormous, the discretionary aspect of asking the patient whether or not to do the extra procedure or the added technique seems counter-productive; asking that same patient to pay extra, sometimes many hundreds of dollars extra, is somewhere between aggressive and exploitative. If there is evidence that a procedure improves outcomes and that that improvement can be quantified, then the value proposition can be readily identified and priced in: a 10% higher probability for an IVF cycle with a $10,000 base price suggests a reasonable range for pricing; you can make the argument (and I have made these arguments, back in the day) that every applicable cycle should include the technique and everyone should pay something less that 10% of $10,000 for it — resulting in an aggregate increase in the number of good outcomes and avoiding failing to help someone for whom the procedure had a demonstrable benefit, even though the added benefit did not guarantee a positive outcome. In the absence of reasonable arithmetic certainty of the marginal benefit, the costs of the additional procedure should be allocated the third party “add-on” company, the IVF clinic and the patient, with the overwhelming burden of the risk of uncertainty allocated away from the patient — she is least capable of assessing the potential benefits, and the only one of the three parties to the decision who won’t benefit in any way if the procedure never proves its worth. The add-on company benefits from whatever revenue it accrues and from the data the the outcomes generate, as does the clinic, which may use the data from one patient’s unsuccessful cycle to refine its practices regarding using the extra procedure in the future.
Looked at from a different angle, the “add-on” controversy is part of IVF’s “retail medicine” model, ie an individual patient purchases a cycle and pays the bill, either with no negotiating or with discretion over adding or omitting incremental costs with no ability to assess the value proposition, if any. This is where our industry needs to be careful. There is a big difference between: 1) Do you want to pay an additional two thousand dollars for a polygenic risk assessment of your embryos? And 2) How about another $1000 for filling your tires with nitrogen or advanced grade rustproofing of the undercarriage. Information about rust-proofing is readily available and the stakes of a wrong decision are relatively trivial. For new IVF techniques the information asymmetry is much greater, as are the stakes, and patients, understandingly wary of avoiding a lifetime of regret, are extremely susceptible to “let’s do everything possible” type arguments.
In the near-term, this will play out clinic by clinic, cycle by cycle. Longer term, however, as the retail model morphs into more of a large purchaser — large provider model facilitated by greater insurance coverage and clinic consolidation plus better data collection, some version of a more efficient pricing structure will emerge, hopefully rewarding clinically relevant innovation with compelling (and aggressively collected and presented) data.