[lvs]

> blood constituents are not evenly distributed at small volumes

You're saying they're running into statistical stochasticity? Are you claiming they were sampling femtoliter volumes? That's the how small you'd have to get for your claim to be true.

[Fede_V]

Actually, yeah. The variance in the amount of quantities present in the blood draws by Theranos is too high to get useful measurements (without very many replicates, defeating its purpose) according to published research: http://ajcp.oxfordjournals.org/content/144/6/885.abstract

I was also utterly astonished when I read this. Even if the experimental error of the instrument was zero, you cannot reduce the biological variance short of drawing more blood or doing multiple measurements. However, this is such a no brainer that the idea that Theranos skipped this step seemed too crazy to contemplate - apparently not though.

[hooloovoo_zoo]

Actually, that paper states that the standard deviation from finger pinpricks is about 5% of the mean value of, for example, platelets. The normal range has a standard deviation of about 25-50% of the mean https://www.nlm.nih.gov/medlineplus/ency/article/003647.htm. So, if anything, this study supports the conclusion that fingerprick tests have clinical utility.

[lvs]

Well you also seem to have skipped what that paper is taking about. For cell counts, yes, the stochasticity claim would be true for microliter scale volumes. For molecular constituents of blood, the claim cannot be true until you drop to the femtoliter range. I don't think the company was ever approved for cell counts, but I could be wrong.

[adevine]

The point, though, is that things like proteins ARE bound up in individual cells, so there are a whole host of tests that will vary based on number of constituent cells. Hemoglobin is one of the examples in the paper.

When you say "I don't think the company was ever approved for cell counts" - well, the company wasn't ever really approved for anything besides a herpes test, that's the issue.

[Fede_V]

Are you trying to model metabolites as perfectly diffusing molecules with D^2 proportional to an idealized Stokes radius?

[lvs]

No. I multiplied 1 nM by Avogadro's number. Do you have any reason to believe that free IgG is not uniformly distributed at steady-state in circulation? ... Because that's what they're testing for.

[alsocasey]

Many (most) of the things we're interested in measuring in blood are bound to some sort large macro-molecular assembly... be it cells, lipid vesicles, large protein agglomerates, etc.

[lvs]

In the clinic the most common testing is metabolic paneling (levels of various salts, urea, glucose, creatine, etc.), serology (for circulating antibodies), or cell counts (which they weren't doing, as far as I know).

[ChrisLomont]

You're wrong; a single red blood cell is several femtoliters. To get a good statistical count of much more rare (yet important) features one can sample much, much more than this and still get anomalies from having too small a sample.

Here's [1] a cool graphic showing the tip of the iceberg.

[1] https://en.wikipedia.org/wiki/File:Reference_ranges_for_bloo...

[lvs]

No, I'm not wrong. You're talking about testing for cell counts, which is not -- as I understand it -- what this company was approved to do. For molecular and protein components of blood on the nanomolar range, you would not encounter this problem until you were sampling in the femtoliter regime. Just use Avogadro's number, and you'll get it.

[Fede_V]

A lot of those molecules are bound up in cells, they are not freely diffusing in blood. Even those diffusing in blood are not uniformly distributed, blood is highly dishomogeneous.

[lvs]

Routine metabolic paneling and serology does test for molecules that are uniformly distributed in blood. I can't say it any plainer than that. Of course many possible testing targets are inside cells, but these tests were not claiming to look at any intracellular targets, at least as far as I know.

[learc83]

They weren't really clear about what tests they could perform with their technology, but they did say they planned on offering the "full range" of blood tests.