Sqad:
Even if the results of an online blood test are, in terms of the figures alone, 100% accurate, might there not be a possible problem with the way that they're interpreted by an independent lab?
For example, my 'Which?' link above shows that a PSA-testing service is classifying some results as 'normal' that would probably at least be classed as 'requiring investigation' by the NHS.
Further, as far as I can see, each test only looks at a specific result, rather than at any trend across the results of successive tests. For example, I've got incurable prostate cancer and therefore get tested for my PSA level periodically. I suspect that, when the consultant sees that my PSA is rising, he'll be looking to see whether it's just creeping up very slowly or rising dramatically. No individual test could show that.