Resident/ninja level diagnostic test interpretation

On Wednesday we discussed a case of a man with cirrhosis who developed acute kidney injury. It was a diagnostic mystery that is now thought to be ATN. In figuring out what was and wasn’t happening, we got deep about the utility and limitations of common diagnostic tests.

Framework for interpreting diagnostic tests

About 6.5 million times this year, you’ve heard a chief ask “what’s your framework for thinking about x” where x may be dyspnea, altered mental status, or March Madness. Here’s a framework interpreting diagnostic tests with lots of examples.

1) what is my pretest probability?

  • As this awesome low-down on likelihood ratio explains…
    • when you have a very low pretest probability, unless the test is extremely specific, a positive result is more likely a false positive than a true positive.
      • This is why rheum poo-poos positive ANAs in patients without symptoms. ANA is a non-specific test for lupus.
    •  when you have a very high pre-test probability, unless the test is extremely sensitive, a negative test may not adequately rule out disease.
      • For example, in a patient with active cancer, a swollen, red leg, tachycardia and hypoxemia, a negative D dimer does not adequately rule out a pulmonary embolism
2) What assay does my lab use and what are the test characteristics (sensitivity/specificity)?
  • Every assay has its own test characteristics – sensitivity, specificity etc, rules about how it should be drawn, etc.  You can find them in your hospital’s lab manual
  • Example: influenza testing
    • Check out this post from december for a refresher on the different influenza test options:
    • At Moffitt, the ED uses the 90% specific, 50% sensitive, super fast influenza antigen test. If someone has a good story for flu and a negative, 50% sensitive test, you should consider treating empirically until their (much more sensitive) respiratory virus panel returns.
    • At ZSFG, the ED uses the very sensitive and specific influenza A/B PCR. So you should have a much higher threshold to treat for flu despite a negative test here @ZSFG.
3) Are there any patient or operator factors that make the test less reliable?
  • Patient factors
    • who was the reference population? 
      • Many basic lab tests like CBCs are validated in healthy, white, male volunteers. Women and non-white people may have a slightly higher than “average” white count. 
        • This entity used to be called “benign ethnic neutropenia.” That’s a really awkward name
        • Don’t forget that sensitivity and specificity are dependent on population prevalence.
    • Is the patient taking medication known to interfere with the assay?
      • The methamphetamine component of the Utox at ZSFG is only 90% specific because many common medications cause a false positive. These range from things you would expect, like OTC cold medications, to things you wouldn’t expect, like trazedone. By contrast, the cocaine assay is >99% sensitive.
      • As a primary care doc, this most commonly comes up with the renin/aldosterone ratio assay when evaluating for hyperaldosteronism. _Many_ common antihypertensives can cause false positives, which is a real bummer when you are sending it to work up resistant hypertension. Either consult the lab manual to figure out what meds to switch them to or ask your friendly neighborhood endocrinologist.
    • Are there comorbidities or acute processes that would interfere with an assay
      • This grab bag includes things like cold agglutinins leading to a lower measured hemoglobin and pseudohyponatremia in the setting of hyperglycemia.
  • Test factors – how persnickety is the lab test itself?
    • Is it hard to draw correctly? 
      • ammonia level – leaving the tourniquet on for too long can cause a false positive
      •  cort stim – very time sensitive protocol required, specially trained nurses do it best
      • a renin/aldo level (has to be drawn in the morning, affected by patient positioning
    • Does the assay require special laboratory expertise?
      • thin and thick smears for malaria are _much_ more reliable in places that see a lot of malaria. The ZSFG lab is pretty darn good, better than most labs in the US, but pales in comparison to a lab in Nairobi.
    • how often does the lab run the test and can they expedite?
The ZSFG lab manual! And the brains of some senior clinicians.

One thought on “Resident/ninja level diagnostic test interpretation”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s