Comment by danso | Hacker News Reader

Comment by danso | Hacker News Reader

Question from an outsider (of medicine and of medical research): Why is this study new? I mean I can understand why it is news (for the mass media)...but isn't this something that would've have been tested long ago to a certain degree? At least to a degree in which, today, doctors are content that a venous draw is a statistically useful amount of blood to derive health results from? Wouldn't that have to based on some study that purported to find the minimum volume of blood needed to reliably represent someone's health?

Not only does it seem like a very fundamental question to have already asked...it doesn't even seem like a very difficult study to do. It's not substantially longitudinal over time -- for every subject, you take several pinpricks, and run the tests. Or logistically difficult to manage.

So I get why it's news, in terms of Theranos and what not...but this has to have been something that was studied many times over many decades. Or is the NYT misinterpreting/signifying the significance, i.e. the Rice scientists found a previously undetectable kind of difference, but which is, yes, technically shows that blood drops are different?

Microfluidics as a field is relatively new (last 10 years or so).

This study is at the "microtiter" scale, which is the scale that companies such as theranos are trying to take advantage of.

Microfluidics is largely an industrial field rather than academic.

>Not only does it seem like a very fundamental question to have already asked...it doesn't even seem like a very difficult study to do.

Yeah, well, that's the difference between Silicon Valley and academia.

SV doesn't want to hear about results that invalidate their business model.

> SV doesn't want to hear about results that invalidate their business model.

I promise you that academics not wanting to hear results that invalidate their models is the rule, not the exception.

otoh, academics love hearing results that invalidate other people's models :)

Hi, pathology resident here. Even with large volume venous blood draws, you can change the values quite substantially. one of my professors likes to tell the story how just by pumping his fist as is usually directed by The phlebotomist, he was able to change his potassium level from 4 to 5 to 6 to 6.5. This is the difference between normal and starting to worry about heart problems. so it is far from shocking at least to the experienced, that this is a problem for microfluidics.

This essentially equivalent to a sampling error problem. A large venous sample (10 mL) is enough to get a pretty good average. Microliters of blood from any given location in the body are likely to be different from microliters somewhere else in the body. I seriously doubt anyone in a clinical lab would be surprised by these results.

As someone who's lived with health professionals my whole life ... if you're smart and have a background in science or engineering, you're only about two or three questions away from stumping health-care providers.

The level of understanding required to build a thing is very different from the level of understanding required to patch a somewhat broken system.

That's a good analogy, but it bears noting that the reason medical professionals only have patch-level knowledge is not for any ignorance or lack of inquisitiveness, but for sheer breadth and lack of source code. There's a reason that where we used to have doctors we now have so many sub specialties; There's just that much to learn and discover, because we don't have the blueprints.

Yep:

- What are the units on blood pressure? 120 of what?

- Would it make a difference if I held my arm up while the blood pressure was taken?

Seriously? Even I knew the first one and unless I'm quite mistaken the second one can be felt by just raising your arm -- or, better, turning your head upside down.

This is missing the forest for the trees a little bit. The actual unit in blood pressure is relatively unimportant; it's so conventional, that no-one produces general medical tools that measure in other units. What is far more important is understanding what the different levels mean in the context of the presenting patient. Is 180 bad? When is it bad? Is it ever good? Is 50 bad? When is it bad? Is it ever good? Is 100 bad &c? If it is bad, what can you do to make it good? Do you do the same thing if it is bad 200 or bad 20? What do you not do when it is bad 200 or bad 20? Which conditions complicate with which medications affect your intended treatment for a bad reading?

Meanwhile, over here in the dev world, our passionate mantra is "we shouldn't be judged poorly for 'everyone knows' shit we can just google"...

Fair points, so let me explain my reasoning there:

First of all, in the case from my past, the machine labeled it -- "mmHg". They could have read it off, had they expected things to have units that matter.

Second, it's part of the conceptual understanding of what blood pressure means -- that e.g. it's not some normalized percentage, that it indicates how much the fluid inside is pressing against the outside, that it's relative to atmospheric pressure, that we live in and expect a pressurized environment.

Third -- I mean, you read it out every day, wouldn't you ever wonder what it means? Imagine a Java dev that doesn't know what System is, only that it's the beginning of the stdout print statements they use.

Someone who merely knows that "180 bad, 120 good" has a disconnected understanding, like the expert who can literally do nothing more than plug numbers into an equation but not know what the numbers mean or whether you're measuring them correctly to be compatible with it. It's not enough.

Its only recently that people have started trying to create diagnostics from blood droplets. Back in the day the technology wasn't sensitive enough.

Ah that makes sense. I was thinking it from the wrong way -- i.e. the way Calvin's dad explains how they discover the weight limit for bridges -- no need to test from small to minimum-amount-needed when there's a certain, reasonable volume of blood that pretty much works all the time (and I'm assuming that's been tested to some degree). Though I'm still surprised blood-drop-variance wasn't just something studied frequently out of scientific curiosity and because it seems relatively easy.

http://www.gocomics.com/calvinandhobbes/1986/11/26

The volume of venous blood for everyday blood test is indeed well known, and has been established decades ago. Take a look at your test tubes then next time you have your blood drawn. The required volumes are always clearly indicated.

This study is about novel tests based on minute volumes of blood. These are not routine in medical practice.

It's indeed not new at least for some markers it was known to be unreliable for a (relatively) long time: http://www.ncbi.nlm.nih.gov/pubmed/9365861

Reply on news.ycombinator.com