New study says single finger-prick tests aren’t necessarily accurate | Hacker News Reader

New study says single finger-prick tests aren’t necessarily accurate

ajcp.oxfordjournals.org

19 pointspavornyoh19 comments

The article actual says that anything less than 4 drops is inaccurate for measuring haemoglobin.

If you can’t get an accurate result measuring the most common protein in blood then the concept is not going to go far unless you can get a few ex-generals and war criminals on your board of directors.

That's one weird thing about the theranos thing: Won't a good lab-on-a-chip device to replace medical labs be profitable enough ? risky enough ?

Why add an extra huge risk in the form of finger pricks ? Why did Holmes and investors even began to consider it ?

What you're referring to has existed since the 90s[0].

0: https://www.abbottpointofcare.com/products-services/istat-ha...

But the iStat focuses solely on clinical chemistry assays. Theranos was trying to add microfluidic immunoassays, which is something even Abbott has been trying to do since the first microparticle ELISA technology was invented.

Sure, but most blood tests today aren't done via lab-on-chip, which should be cheaper.

I don't get your last sentence. What does taking 5 drops of blood have to do with ex-generals or war criminals?

Theranos reference

thanks

Considering how much blood they have on their hands, it's ironic how little they know about the substance.

This is no surprise whatsoever to anyone (e.g., diabetics) who uses glucose test strips. Small sample + cheap small battery-powered measuring device = high variation.

Is it good science? The study had a mere 11 "donors"! I know Rice is a small school, but there are plenty of people around the university to volunteer. Its within throwing distance of Houston's Medical Center!

That struck me too. Not only did they have a laughably small sample size, they didn't compare the results they got against gold-standard venous blood tests.

Part of could be that it is easier to fuck up a finger prick blood draw and cause inaccurate results. Improper cleaning [1], not getting the blood out in the right way, or even certain patient conditions. Less likely to mess up (in a way that causes test inaccuracy) when you drain a whole vial out of their arm.

[1] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1238762/pdf/brjg...

Good. Maybe they should consider getting blood from somewhere within the 99% of the body's surface area that doesn't have as many nerves as the tips of the fingers.

I thought this has been known for quite a while. I keep seeing articles and comments by people who have worked in this field for years saying finger-prick blood is unreliable. I'm unclear why another study is needed.

Is it just me, or is "not necessarily accurate" just weasel words for "not accurate"?

Either something is accurate, or it is not.

You can't argue that one failure to achieve accurate results mean that it's inherently impossible to achieve accurate results.

In this case, it's conceivable that there could be a tip geometry, tip composition, cleaning technique, normalizing strategy, prick location/depth, etc (or combination of the above) that leads to good data. Of course, that's not a guarantee that it is possible, and it still puts the burden of proof on Theranos (for example) to prove that they have actually overcome this problem if they want to run tests that way.

It is mighty suggestive though.

I don't think I'd be worried about this if I were running a company like Theranos. There's probably still a huge market for that kind of multitest product even if the draw was venous.

And you can always draw multiple small samples, as recommended with actual pinprick glucose meters.

Or you might be able to take smaller vials.

Generally speaking, phrases like "not necessarily accurate" are an attempt to describe (in words) the shape or qualities of a much more complex relationship.

For example, a scale is often an accurate measure of mass, but not necessarily, because it assumes 1G of force. It can easily give you the wrong results in an elevator or on another planet.

Reply on news.ycombinator.com