Useful Mind Tools: The Concept of Type 1 and Type 2 Error

 

Medicine is filled with the usage of the term “alpha”, especially in statistics and journal articles.  Medical literature and literature across multiple disciplines often uses alpha to reference the quantifiable rate of a type 1 error.  A type 1 error is considered, in a word, tampering.  That is, saying that there is something wrong or different, and actively addressing the perceived issue when in fact there is no such issue, is a type 1 error.  Again, if we were to describe a type 1 or alpha error in one word that word would be “tampering”; there is simply nothing wrong with the situation and adjusting it would be improper–but we commit a type 1 error when we doso anyway.

 

There may be some confusion over the term alpha because there other usages of the term you may see online.  Alpha is also used to describe the rate of return on a mutual fund in excess of the quantifiable risk inherent in that mutual fund’s distribution of underlying assets.  Meaning the alpha can indicate the bang for your buck you get above and beyond the risk inherent in a certain investment.  There’s an interesting investment blog called “Seeking Alpha”.

In this entry we discuss alpha from the more common standpoint of tampering and its utility in making decisions.  This is readily applied to how we develop new business models and how we re-work old ones.  The importance of considering type 1 and type 2 error in medicine and in decision making can’t be overstated.

 

A type 1 error, again placing a chest tube when in fact no chest tube is necessary, frequently has less harm inherent in it than a type 2 error which is under-controlling or under-recognizing a situation and not treating the very real issue.  Telling a patient that they are fine only to have an issue later is clearly under-controlling or a type 2 error.  Also, type 2 errors are often heavily focused on by the legal system. The legal system tends to really frown on type 2 errors.  For that reason fluid, acute care and trauma situations are often very demanding in terms of decision-making. We often do not have enough data, time, or other information to know every direction a scenario can take.  Despite the conditioning of M&M and the retrospectoscope (which often give the illusion of perfectly available information, etc.) we should take a moment to recognize the realities of decision-making as it progresses forward through time with the uncertainties the come from real-world sources.  Of course, there is our years of education that encompasses multiple different scenarios and which can help minimize the unknowns.  Nowadays it is more and more challenging to educate residents and fellows in the same manner in which many of us were educated. That means, it is hard to make sure they see every direction a scenario can go.  That is, in part, because they have less hours spent in hospitals to take in all possible scenarios as residency classically teaches those points by a brute force methodology.

 

Often, in trauma and acute care surgery, type 1 errors are less devastating than type 2 errors in my opinion.  You and I likely agree that this is sort of a philosophic point.  Maybe you feel, for example, that “first do no harm” means our prime directive indicates we should not have intervened with a central line if a patient is later found to not require a central line.  In my opinion, that type of thinking fails to recognize decision-making moves forward in time with relevant uncertainties.  The retrospectoscope should be discarded.  However, others may feel that the risks inherent in failing to intervene constitute doing harm.  It’s always interested me how people feel differently about the modern sense of “first do no harm”.  Given the choice, and broadly speaking, I’d rather commit a type 1 error in my specialty than a type 2 error.  (Of course, I’d rather avoid errors all together.)

 

I am interested in finding new and different ways to educate residents and fellows given the constraints on their time now, and thinking of type 1 and type 2 errors helps do that.  Rather than lamenting the current state of affairs with medical education, I do feel this is an evolve or die scenario wherein we need to focus on new and better ways to ensure excellent patient care in the future via innovative techniques to educate our residents. I think one of these is explicitly teaching about type 1 and type 2 errors.  Thinking of issues framed in terms of type 1 and type 2 errors is just one part of a larger framework that encompasses decision-making in uncertainty.

 

Decision-making in uncertain situations can often carry a very negative connotation to physicians and surgeons.  In short, we may have been educated with the philosophy of being ‘right, wrong, but never in doubt’. However, that focus on certainty etc. may lead us to immediately react, in my opinion, negatively to the term uncertainty.

 

So let’s clarify:  in this context “uncertainty” does not mean personal uncertainty on the part of the physician or surgeon.  “Uncertainty” here means that we all recognize a scenario has multiple different ways it can progress. For example placing a chest tube when we think there is a pneumothorax can have several consequences. We can have been incorrect despite testing (pCXR) and with our exam that there was in fact pneumothorax.  This is unusual but is possible.  We may have a patient in PEA, or have a trauma patient “code” in front of us.  We may feel placing chest tubes in those scenarios has little downside risk.  However, we can have other issues with chest tube placement, including hemorrhage, diaphragm injury, abdominal organ injury and some of the other catastrophic and rare downsides to chest tube placement that we have all seen in our practice.  So, what are we to do? Analysis paralysis is one issue that probably each of us has seen arise.  That is a situation where all the multiple ways in which a scenario can unfold leads to us not making a decision.  In my opinion, not making a decision is as important of a problem as making an improper or poor quality decision.

 

In short, thinking of issues in terms of type 1 and type 2 error rate can help us frame and deal with the fluid situations that arise in acute care surgery, investing, and business model innovation.  Further, we can use tools to quantify scenarios and make the best quality decisions we can despite probabilistic influences.  This type of advanced decision-making is not typically taught in medical schools of reinforced in residency.  However, seeing it, and using it, can be performed on a daily basis given the multiple decision tools that exist.  Further, we can always utilize the framework ahead of time to formulate high quality decisions based on different scenarios when they arise.

 

The concepts of type 1 and type 2 errors are useful mental tools to frame just what to do in trauma, acute care, investment, and other important high-stake decisions in our lives.  I invite you to read more about type 1 and type 2 errors at your leisure and have found this personally to be a very useful tool for my clinical, investment and general decision-making toolbox.

0 comments