[BostonCard]

Bad Blood also explains that capillary blood from a fingertip isn't suitable for accurately doing many tests. The composition is notably different from blood captured from a vein. Sorry I don't recall enough details to do justice to the reasons why.

The former doesn't necessarily flow from the latter.  The composition of capillary blood can be different from venous blood and still be useful for diagnosing disease.  The important question is whether the result of the test, no matter how you get it, is reliably within a predictable range in healthy patients, and reliably outside of that range in disease.  That range could be different depending on what the analyte measured is.  The key is not so much showing concordance between capillary blood and venous blood (though that makes life easy) but rather to show concordance between a particular test no matter where the blood is obtained, and health or disease.

There is precedent for using capillary blood; glucose monitoring for diabetics.  It turns out that capillary blood glucose is higher than venous blood glucose taken simultaneously.  However, capillary blood glucose is still useful because the difference is fairly low, and the difference between the two is small enough compared to important clinical differences.  You might want to use venous blood from a fasting patient to diagnose diabetes, but for a diabetic managing their insulin, portable glucometers using capillary blood have proven that they are good enough.

In terms of Theranos, if they had actually managed to develop their system (a brobdingnagian if, of course), there probably would have been many blood tests for which the system would have generated accurate enough measurements to be useful, and some things for which the system would have been too inaccurate to use.  That is not unlike venous blood tests currently under development.  We can measure thousands of analytes in a tube of blood; however not all of them will be useful.  Venous blood just has a about a century head start, and so clinicians have become comfortable with it.

BC

[BostonCard]

Bad Blood also explains that capillary blood from a fingertip isn't suitable for accurately doing many tests. The composition is notably different from blood captured from a vein. Sorry I don't recall enough details to do justice to the reasons why.

The former doesn't necessarily flow from the latter.  The composition of capillary blood can be different from venous blood and still be useful for diagnosing disease.  The important question is whether the result of the test, no matter how you get it, is reliably within a predictable range in healthy patients, and reliably outside of that range in disease.  That range could be different depending on what the analyte measured is.  The key is not so much showing concordance between capillary blood and venous blood (though that makes life easy) but rather to show concordance between a particular test no matter where the blood is obtained, and health or disease.

There is precedent for using capillary blood; glucose monitoring for diabetics.  It turns out that capillary blood glucose is higher than venous blood glucose taken simultaneously.  However, capillary blood glucose is still useful because the difference is fairly low, and the difference between the two is small enough compared to important clinical differences.  You might want to use venous blood from a fasting patient to diagnose diabetes, but for a diabetic managing their insulin, portable glucometers using capillary blood have proven that they are good enough.

In terms of Theranos, if they had actually managed to develop their system (a brobdingnagian if, of course), there probably would have been many blood tests for which the system would have generated accurate enough measurements to be useful, and some things for which the system would have been too inaccurate to use.  That is not unlike venous blood tests currently under development.  We can measure thousands of analytes in a tube of blood; however not all of them will be useful.  Venous blood just has a about a century head start, and so clinicians have become comfortable with it.

BC