Lies, Damn Lies, & Statistical Methods

This post is very long, and boring through significant sections. It is essential that you read the introduction and the I Tell You What big finish. The discussion section is optional, but hopefully informative.

Introduction: Due to some past events, Wifey and I found ourselves at the Yale New Haven Hospital. Right, as if there is a Yale Albany to confuse people. This is a world class institution, beyond a doubt. There those world class medical technicians were, performing tests on Wifey to evaluate if everything was OK with the pregnancy. The tech paused, and really lingered on one shot. Wifey & I exchanged a glance, because you're supposed to breeze through these things, right? The image was similar to the one below, and she was zooming in on the neck area until the resolution looked something like one of those fakes of a Sasquatch roaming the forests of the Pacific Northwest.

We were informed that the nuchal thickness was either 2.8 or 2.9 mm, depending on which side of the fuzzy gray pixelated line the tech chose for her measurement. The doctor was called in to review the measurements, and this questionable data was coupled with a quad screen to inform us that the pregnancy was "positive" for Down Syndrome. Some days you eat the bear, other days, the bear eats you.

Discussion: This doctor (hereafter: Dr. Who) was in for a surprise, as most swamp Yankees or NYC socialites referred to him are not (A) educated in statistical sampling and test methods, or (B) of generally dour and dorky disposition. I proceeded to fire off questions about his criteria for test, for evaluation, and how they compared to other accredited institutions. It seems that:

• Some relationship was developed between this measurement and the likelihood of a 11-13 week-along baby to have Down Syndrome (DS)
• Above this "normal" measurement the likelihood of DS is increased, although not guaranteed, and continues to increase as the measurement increases
• "Normal" is often well over 3mm in terms of a "positive" DS result
• Dr. Who personally adjusted the "positive" criteria back one standard deviation to the mean

Confused? Don't be. A sample spread (many are shifted, or flatter, or steeper, but this is a good example) of any population characteristic may be illustrated as below: I do not recall how many standard deviations this doc pulled towards the mean, but the result is this: more "positive" results. I'm roughing the numbers, but if the normal incidence of DS for a lady in her 20's (best case scenario) is 1/1500 (0.06%), you're almost 5 standard deviations from the mean. Given a normal false positive rate of this sort of screening is 5% (see Table 1 here) combined with ratcheting the criteria to, say, 4 standard deviations, you increased the confidence interval by about 0.06%. This means that on top of the existing 5% false positive rate inherent to the testing methods, Dr. Who has DOUBLED the number of ultrasounds reporting "positive" for DS.

Normally, such a move is great for the patient. By invoking greater scrutiny on a borderline or within-sight-of-the-borderline patients, you are more likely to catch conditions and allow treatment to proceed.

Did I mention there is no cure for DS? It gets better.

We were sat down with a "genetic counselor" to discuss Dr. Who's flux capacitor results alongside a quad screen, which is a blood test of both parents. Let me interject that quad screens, per Wikipedia, have a 7.5% false positive rate. Anyway, Wifey & I Tell You What have pretty good health, and a reasonable genetic slate. No Olympians in our clan, but no complaints. Long story short, something like a 1 in 50 chance of our baby having DS. Our options were presented as follows:

• Abortion
• Amniocentesis
• Wait for further testing in 8-10 weeks and pray for good results

I took option 1 off the table immediately, and amniocentesis, which is essentially sticking a needle in to draw off amniotic fluid for direct DNA testing, was never a realistic option for us since abortion was not in the cards. We informed her that we would wait for Round 2 of testing before deciding to keep our baby anyway. The lady was... lukewarm to my approach, but she had no authority to make any decisions. We put our faith in God to give us the strength to handle whatever happened. Now, through no credit of our own, we have our second beautiful, and quite healthy to this point, daughter from that pregnancy.

I Tell You What!: One critical statistic, where I can site the Washington Post, Wikipedia, or Time Magazine, is that over 90% of pregnancies with a DS diagnosis result in abortion. Why is this critical? For doctors to allow a high false positive rate is often benign, because early treatment is a desirable thing and more invasive subsequent testing can correct false positives. With DS, that is not the case. There is no treatment that can help the baby, other than to prepare the parents for the surely difficult task of raising a DS child. Additionally, if the mother is willing to consider abortion as a "cure", then what are the odds she's going to wait for further results, if that means staying pregnant for 2 more months (no cakewalk) or have a needle stuck in a very uncomfortable place? Judging by the 90%+ number, not very likely at all.

But why are false positives really allowed in the first place? Insurance companies (the guys writing the checks for all this testing) are mavens for cost efficiency. Some live, some die, keep cost down, because that's the way it was meant to be. BUT, extra false positives mean MORE testing, MORE doctor visits, MORE screenings. Insurance companies hate that sort of stuff. Unless... they are fully aware of the likelihood of a positive DS screening result. Based on my knowledge of miscarriages, a D&C is an outpatient procedure. It involves an OB, let's say it costs $1,000 to $1,500 (guessing, folks). It is also the coup de main of an abortion. Now, a full term pregnancy, following the decision point of this DS testing, will involve a handful of doctor's appointments, perhaps including a few extra ultrasounds because of the initial "positive" result. This is assuming no additional complications, mind you. Then there is the delivery, which may involve a few trips to the hospital triage room for monitoring before being sent home. When the baby is actually delivered, you're looking at 2 nights in the hospital, a battery of tests on the newborn and THEN 18 years of dependent care, largely covered by... the insurance companies. And I have to assume that if the child really does have DS those medical costs, especially the portions covered by insurance, skyrocket.

Folks, I haven't even touched on the wellspring of material provided by the argument that people with Down Syndrome are, well, people. I may have gotten kicked out of the living room by my mother when I was a kid for making fun of Corky Thatcher, but I never would have denied him his humanity.

Time to wrap this up, it's late. Special thanks for inspiration to Patricia E. Bauer of the Washington Post of Friday, November 16, 2007; Page A33. What have we learned?

• The bodies that God has designed are not yet fully understood by modern medicine
• Doctors are making assumptions, just like you are when you head to your typical gas station because the price there is normally lower or buy size 11 shoes because they normally fit well
• There are always pressures in our societies coming from directions or levels that we often do not expect or realize that provide motives for the people we interact with. That motive can be expected to involve money.
• Sarah Joy's recent arrival was a blessing, and worth the wait