By skyforbes 
It’s #Pinktober. This #Pinktober, I’m feeling old: I’m finally both engaged and starting to hit the approximate age window where my maternal grandmother was diagnosed with breast cancer. While I am doing all the generic screenings (mammograms, ultrasounds, MRIs with contrast dye!) on the recommended schedule, I still have not gotten my genetic testing out of the way. It irritates me to no end to wait days while wasting time, lab resources, and money for results that I’m pretty sure are going to be useless, then repeat the cycle, hoping eventually I get where I need to go. In theory, Obamacare/The Affordable Care Act was supposed to to prevent this sort of cycling, this waste of money, this useless testing. In practice, Obamacare did not revolutionize health care at its core, and therefore it always will have problems with serious waste somewhere, as well as the potential for death spirals.
How could it not? It never identified the one singular issue that is the keystone to why healthcare costs are what they are.
My grandmother died of breast cancer metastasizing into her bones in her early-mid 40s (or bone cancer — she died around 1973, knowledge has changed a lot since then), leaving two daughters, my mother and my aunt. By the time I started college, both of them also had breast cancer.
If you haven’t guessed from the curly hair in the photo and the name Shana — yes, I’m of Ashkenazi Jewish Extraction. Ashkenazim are slightly more predisposed to being BRCA1/2 carriers than the general population — enough that just being Ashkenazi is considered an extra risk. 90% of familial Breast/Ovarian cancer among Ashkenazi Jewish families are traceable to 3 BRCA1 or 2 mutations:
- BRCA1: Ex2:c.68_69delAG (aka 185 delAG/187delAG)
- BRCA1: Ex19:c.5266dupC (aka c.5266dupC/5382insC/5385insC)
- BRCA2: Ex11:c.5946delT (aka: c.6174delT)
The remaining 10% fits into the following groups:
- Lynch Syndrome — While Lynch Syndrome primarily affects the colon, many women who have Lynch Syndrome also do get Breast and/or Ovarian Cancer
- Other BRCA1/2 mutations that are NOT Founder mutations
- PALB2 carriers
- The exceedingly rare Cowden Syndrome (PTEN) or Li-Fraumeni Syndrome(CHEk2) sufferer
- The Rare Hereditary Diffuse Gastric Cancer — CDH1 mutations are closely associated with Lobular Breast Cancer
- Nothing. It’s a sporadic, random breast cancer.
- Nothing, it’s a familial breast cancer with an unknown cause.
When my aunt and my mother both ended up with breast cancer in the early 2000s’ (and furthermore, slightly different kinds of breast cancer, with slightly different outcomes) within 3 years of each other, they were co-enrolled in a genetic study with each other alongside my grandfather. My aunt and my grandfather are part of that 90% of Ashkenazic jewish people with familial breast/ovarian cancer syndrome. They carry one of the 2 BRCA1 variants. My mother: has normal BRCA1/2 genes.
My family, and my mother’s extended family is Orthodox. We’re not talking so much right now. Life is complicated, and family dynamics can make it both more and less complicated. One of the complications is my engagement to a great, wonderful, smart person named Shawn who is not Jewish. In theory, this fact shouldn’t matter at all. In practice, that right now we are not close is part of the reason healthcare costs are so high.
When I finally realized my grandmother’s age of diagnosis was creeping up on me, I sat down with a genetic counselor. I told her a slightly more abbreviated version of the story above. However, without a HIPAA disclosed version of all the times (2x) that my mother has been tested experimentally. ideally, I should bring my mother in, and get her to sign off on co-testing clinically with me, thereby skipping the “starter” Ashkenazi founder mutation only panel. Because we’re not really talking, we’d be starting with the ~$250 Ashkenazi founder mutation panel, and if/when that comes back negative, a broad BRCA panel (approximately $1000 from Myriad, who has a stranglehold on data about what is and isn’t a deleterious BRCA1/2 mutation), and if/when that comes back negative, a broader breast cancer gene panel (~$5000 via Myriad)
That is 3 rounds of genetic testing just for familial breast cancer related screening, not including if I also decided to participate in genetic studies. That’s before premarital related genetic testing (and related costs), which is also customary where I come from, due to the high incident of dangerous recessives.
I’d like to skip the round after round of essentially the same test with a different analysis applied to it. I want to move to a much more modern technology — full genome sequencing, and have the sequence file available in my medical file for interpretation as time goes on, research happens, and more becomes known about my particular genome and the genes within it. Veritas Genetics offers full sequencing for $999.
Despite the fact that it is cheaper and more beneficial for my health over the long term, Full Genome Sequencing is not a covered service.
Due to a bunch of personal factors, I happen to be on Medicaid. However, this would be true for nearly any medical plan in the US. Full Genome Sequencing in a lab alongside separate companies to offer analysis of existing genome files is NOT something that any insurance company will pay for, despite the fact that it would save them millions of dollars. A source inside Myriad mentioned to me that the main reason this happens is there is no way to bill an insurance company for interpreting an already genetic data file, and no real will to push insurance companies, doctors, hospitals, and other labs to offer such a code. As a result, despite the technology around sequencing genes growing ever cheaper, prices for testing don’t drift down nearly as fast as one would expect.
We’ve been careening in this direction for my entire life. Healthcare costs keep going up here in the US, and we seem to be getting less for the money. Economists talk about a variety of different factors, everything from we don’t have true mass bargaining power for drugs, healthcare in the US is unusually tied to jobs, patients do not compare prices for MRIs and knee surgeries, and that we under-invest in the public health. All of these facts are true, and are contributors to why generically healthcare costs are so high.
Yet they also don’t explain why median costs for a given service are high, and why that median is mediocre compared to what it could be. For example, a lipid panel, a fairly commodity lab test, will have outlier high and low cost lab providers. Yet still the median costs will be overtly higher than one would think compared to the cost of running the lab panel and producing all the material needed to do a lipid panel. It should be for the median price of a lipid panel, we’d get more than just a lipid panel. And yet instead, we have ever more providers who price themselves at approximately the median for the exact same lipid panel, with no real improvements. And that median price seems to illogically be edging up bit by bit every year. So much for the free market.
A historical peculiarity drives the mediocrity problem. The very first software that became electronic health records was healthcare administrative software packages for large hospitals that ran on mainframes. Of the subset of healthcare administrative software a hospital could buy, by far and away the bestselling packages were around billing.
The second generation of this software was extraodinary. Not only could it automate billing, it could automate normalization within a bill and across all bills for a given procedure or medication for a given insurer for the maximum price, particularly the US government. For example: If before aspirin cost between one cent and one dollar depending on the person, day, procedure, dose, insurer, doctor, etc., the second generation of software normalized the price so it always costed the maximum cost that insurer would pay out per dose of aspirin. This drove up revenue for hospitals that it caused the rise of the HMO and the development of Hillarycare. (Yes, That Hillarycare, from the early 1990s.)
Ironically, my mother was the project lead for that very feature at McDonnell Douglas Health Systems, before they were bought by American Express, in late 1989, right before the birth of my brother. She left the company right after his birth, and right as that feature became mainstream.
By the 90s, this sort of software ran on home PCs , allowing your average doctor to run complex billing off a computer. It became normal to hire consultants to help doctors figure out how to maximally bill insurers. These consultants have yet to disappear.
When computerized patient health records were introduced, they were designed to be compatible with billing software. The data structure and databases were at their core the same, if not linked together by the software vendor. By linking patient to revenue in the data structure, and by keeping a fee-for-service based system, the data structure made it very easy for even small practices of one doctor to start looking at how they should code procedures slightly differently in order to maximize revenue. At the same time, that same data structure made it very difficult for most doctors to look at their procedures and see which causes similar patients to have better outcomes. It never occurred to anyone to have a database design, and therefore software design, about a given patient’s health, or group of patients health, especially over long periods of time, say a decade, and making the link to costs/revenue secondary.
Even if modern EHR software is designed to put recording patient data first, fundamentally, the structure underneath does not fully allow “recording and therefore understanding patient data” to be fully true to this day. Every standard for medical data interoperability is still designed around the concept that patient data needs to integrate with billing and administration. The HL7 standards, a standard of data encoding for medical software, have lots of precise interoperability definitions around administrative tasks, and far less structure about defining the status of the patient, especially as patient data gets ever more complex.
Furthermore, another remnant of computerization: Nearly everything, from diagnosis to treatment, is defined by ICD-10 codes inside software, despite the fact that patients, doctors, and nurses don’t talk about diseases in terms of ICD codes (though if your provider starts talking to you about your W55.22XA status, I want to talk to you). If you look at photos of medical bills on the internet, you’ll often see an ICD code alongside (or instead) of the procedure or diagnosis, despite the fact that the person receiving the bill has no idea what the ICD code refers too. And that is before classes for doctors about how to hack the ICD-10 for maximum revenue.
We clearly caused our doctors and hospitals to climb to the top of the wrong mountain of optimizing in healthcare. Why didn’t we stop them earlier? Or better, why didn’t we get them to go up the right mountain, making us healthy over the long term?
All insurance companies start their process of what to cover and how to pay for it based on two primary sources:
- How Congress, via the Centers for Medicare and Medicaid Service, determines the value of a given service or type of doctor
- The rating for a procedure given by The US Preventative Service Task Force.
This is why heart surgeons get paid more than therapists that specialize in CBT. The work of a heart surgeon easy to prove as valuable compared to a therapist, especially if we look at shorter term parameters for a patient, such as 6 months.
If we were to ask as a country if sending someone to therapy at 30 for depression makes it less likely that they will need heart surgery at 60, no one could tell you. This is despite the American Heart Association and American Psychiatric Association explicitly saying for almost a decade that depression and heart disease are often comorbid for unknown reasons, and if someone is checked in for a heart attack, you should also screen them for depression. As a country, there is no organization that tracks such a statistic, and the NIH and NSF doesn’t have the money to conduct longitudinal studies at scale for tracking health interventions for heart disease 30 years later, let alone for thousands of other conditions that people suffer.
Therein lies the problem. Today we are sick from diseases of civilization — heart disease, cancer, diabetes, obesity, what have you. They are diseases with multiple underlying causes interacting with each other. If you block the keystone cause(s), you potentially stop the disease, but we don’t track that data en mass longitudinally.
Essentially, we don’t define “healthy” and we don’t pay our doctors and hospitals around a definition of “healthy”. If we pushed them to make us healthy, doctors would make decisions based on if a medication, a procedure, a recommendation would help us 30 years from now.
It is unlikely that once this is published, suddenly every doctor I see will care about what my health will be like closer to when I’m supposed to retire, and make decisions with me based on that time scale.
It is unlikely that Myriad and Vertias Genetics will tell me that they’ve worked with my insurance company, and I’ve saved the US taxpayer about $4k.
And I’m definitely not going to wake up the day afterwards to find out that it’s become a national priority to measure healthy for the long term, and then to code all databases and reimburse all doctors based on how healthy they make their patients over their lifetimes.
I can say I tried to warn people. Costs will always rise unless we start drastically reevaluating what our healthcare system is at the source, and reassess what we track about it administratively. If you can never pay a lab, a doctor, a hospital, a nurse, a therapist, a social worker, whomever, to make a decision to ensure someone’s health for the next 30 years, don’t be surprised when the system pays more for unnecessary, mediocre care.