As a lab industry professional, I've been watching Theranos for a couple or three years. I was skeptical then, but am even more so when I see the hematology correlations on their website:
First: they're limited (It's also not clear what platform was used for the CBCs):
Eos (%, r=0.99)
Hct (r=1.00)
Hgb (r=1.00)
Lymphs (%, r=1.00)
Neuts (%, r=1.00)
Plts (r=0.99)
RBC (r=0.99)
WBC (r=0.99).
Where are MCV, MCH, MCHC, and Basos?
Second: Too good to be true...
The R-values are all either 0.99 or 1.00. This is almost too good to be true for a comparison on Technology X (the Theranos device) vs Technology Y (the predicate). If I'd seen some 0.97s (even one!), I would have been less skeptical. If these data from diluted blood vs same specimen whole blood on their mainstream instruments - that might be more plausible.
So now I look forward to hearing just how they got such good data...but I'm not holding my breath!