Monday, May 16, 2016

Can you have too many tests?






(Hint: Yes.)

Screening and diagnostic tests are valuable tools in helping doctors to prevent illnesses, find medical conditions at early treatable stages, and identify or exclude medical conditions as causes of patient symptoms.  By definition, screening tests (such as annual cholesterol testing: can we keep you from having a heart attack) are done to prevent problems or to identify them before they have actually started to cause problems.  Diagnostic tests (such as heart muscle enzyme tests: are you having a heart attack) are done due to symptoms or other problems.

Doctors learn as early as medical school that just the right amount of testing is the best.  Too little can lead to delays in diagnosis and treatment, too many can confuse the matter by raising "red herrings".

Theranos is one of the most heavily financed start-ups in history.  The company proposes break-through technology which would allow most blood tests to be performed on no more blood than a relatively painless prick of the finger (rather similar to what diabetics do to check their blood sugar).  Their eventual goal seems to be to set up free-standing labs in retail locations such as Walgreen's that would allow people to pay for any lab test as often as they want without requiring a doctor's order.  Presently, the company is beset by serious questions regarding data they seem to have generated using standard lab testing equipment and a number of other major issues. (The Wall Street Journal has quite a bit of coverage on this, but it is pay-walled.)

I don't have any financial stake in any labs or X-Ray facilities and am generally in favor of transparency and the exercise of free will.  Frankly, I imagine I would be seeing patients who are concerned about abnormal tests as often as I already see patients for abnormal symptoms if Theranos were already up and running.  In other words, I would not find their existence threatening.

My issue with Theranos' proposal is that selecting and interpreting tests is complex.  (As in if it wasn't so complex, pathology wouldn't me a medical specialty and medicine wouldn't be a profession.) 

  • If you are concerned about a specific condition, you have to know what tests will help to identify and/or exclude it.
  • You have to know when a test result that is outside the reference range is significant and when it is benign.
  • You have to know when results that are technically within the reference range are significant.
  • You have to be prepared to act on expected and on unexpected results.
Katherine Hobson at FiveThirtyEight does a great job at addressing the greater concern; we don't necessarily need more tests.

No comments: