George Rebane
[This introduction to modern medical decision making is intended as the first of a series of such essays on RR. Subsequent pieces will cover simplifying nomograms (to avoid the arithmetic), personal utility (always subjective), and planning sequential diagnostics and treatments. The aim here is to prepare the intelligent healthcare customer to communicate effectively with the healthcare industry and make prudent decisions for his own benefit. A related background tutorial by the author for this piece is 'Hello Bayes Theorem!']
The healthcare industry woefully lags in many areas, but helping patients make correct decisions about their individual healthcare is where healthcare providers, led by attending physicians, are pathetically behind the times. Let me explain.
Almost all of us have been in a clinic or doctor’s office and encountered a quandary on how to decide on what diagnostic or treatment procedure to choose next, or how to meaningfully interpret the results of a test that just came back. The usual response of the layman in this situation is to give up trying to understand the gobbledygook from the healthcare professional, and just put everything into their hands. Arguably for most people this is the wrong choice. The medical profession has a terrible record of mistakes that result in tens of thousands of deaths annually (mortality rate), and even a much higher rate of morbidity (making things worse).
However, it doesn’t have to be that way if the patient has an elementary understanding of how uncertainties are measured and can be dealt with when they inevitably encounter their next critical decision in their own or a loved one’s healthcare. Numeracy on the part of the patient is the only real requirement.
The Meaning of a Diagnostic Test
To firm up our understanding of this topic, let’s consider a couple of scenarios. A patient could have a dreaded disease, say DD4, and has undergone a diagnostic test for that disease. The attending physician reports that the results came back ‘positive’. What to do now? The first question should be ‘then what are the chances that I really do have DD4?’ Unfortunately, that question is seldom asked, and when asked, the answer at best is an ambiguous prevarication such ‘well, the chances are pretty good that you’re suffering from DD4.’ Now what are you supposed to do with ‘pretty good’.
On another occasion a patient’s blood test comes back during a routine physical, and the physician notices an anomaly that may harbinger DD9. He points this out to the patient, says not to worry, ‘it’s probably nothing, just a normal aberration’, and tells the patient that they’ll repeat the test six months from now just to make sure. We all know what the patient does with that kind of a report – he goes away and worries for six months, making all kinds of plans for the contingency that he really does have DD9.
In both scenarios a more complete report would have served the patient better in deciding what to do next or how to respond.
To begin, the patient can do his part with a little homework on the reliability of tests in general. No test is 100% reliable in the sense that if it comes back ‘positive’ or ‘negative’, one is then assured that they do or don’t have some dreaded disease. This means that the chances, or probability, of suffering from DD is somewhere between zero and one (or 100%). One reason for this is that most tests come back with a number – e.g. your PSA test for prostate cancer returns 5.7 – instead of just ‘positive’ or ‘negative’. It is where the physician or the clinical literature puts the threshold on such a number that determines their pronouncement ‘positive’ (if the PSA is at or above, say, 4.0) or ‘negative’ (if it is below 4.0).
But let’s not forget that just getting a positive/negative report on a test result doesn’t answer the burning question on the patient’s mind – ‘given the test result, what’s the chance that I have the DD?’ To answer that we have to encounter some more realworld details about test results. The easiest way is to look at the following graphic from Wikipedia (Figure 1) that summarizes the clinical data for a specific test given to diagnose a specific DD.
[To continue reading, please download a complete copy of this article Download Making Medical Decisions.]
However, it doesn’t have to be that way if the patient has an elementary understanding of how uncertainties are measured and can be dealt with when they inevitably encounter their next critical decision in their own or a loved one’s healthcare. Numeracy on the part of the patient is the only real requirement.
The Meaning of a Diagnostic Test
To firm up our understanding of this topic, let’s consider a couple of scenarios. A patient could have a dreaded disease, say DD4, and has undergone a diagnostic test for that disease. The attending physician reports that the results came back ‘positive’. What to do now? The first question should be ‘then what are the chances that I really do have DD4?’ Unfortunately, that question is seldom asked, and when asked, the answer at best is an ambiguous prevarication such ‘well, the chances are pretty good that you’re suffering from DD4.’ Now what are you supposed to do with ‘pretty good’.
On another occasion a patient’s blood test comes back during a routine physical, and the physician notices an anomaly that may harbinger DD9. He points this out to the patient, says not to worry, ‘it’s probably nothing, just a normal aberration’, and tells the patient that they’ll repeat the test six months from now just to make sure. We all know what the patient does with that kind of a report – he goes away and worries for six months, making all kinds of plans for the contingency that he really does have DD9.
In both scenarios a more complete report would have served the patient better in deciding what to do next or how to respond.
To begin, the patient can do his part with a little homework on the reliability of tests in general. No test is 100% reliable in the sense that if it comes back ‘positive’ or ‘negative’, one is then assured that they do or don’t have some dreaded disease. This means that the chances, or probability, of suffering from DD is somewhere between zero and one (or 100%). One reason for this is that most tests come back with a number – e.g. your PSA test for prostate cancer returns 5.7 – instead of just ‘positive’ or ‘negative’. It is where the physician or the clinical literature puts the threshold on such a number that determines their pronouncement ‘positive’ (if the PSA is at or above, say, 4.0) or ‘negative’ (if it is below 4.0).
But let’s not forget that just getting a positive/negative report on a test result doesn’t answer the burning question on the patient’s mind – ‘given the test result, what’s the chance that I have the DD?’ To answer that we have to encounter some more realworld details about test results. The easiest way is to look at the following graphic from Wikipedia (Figure 1) that summarizes the clinical data for a specific test given to diagnose a specific DD.
[To continue reading, please download a complete copy of this article Download Making Medical Decisions.]
George, well done. I am looking forward to future installments to this series. It is easy to imagine an interactive software program that could deploy such info in a personal, logical and easy to understand format. Imagine if there was a diagnosis 'report card' for each Dr. and history of diagnosis/treatments/outcomes. We have yet to harness such technology in many fields... healthcare being one of the most important. Furthermore, imagine having the capabilities to review treatments more in depth before deciding (for example, this treatment will make the next x months living hell yet could extend your life by x months/years). Imagine the possibilities.
Let's hope that it is not just the Health care czar that gets to implement such technology (within her agenda).
Posted by: THEMIKEYMCD | 20 April 2012 at 03:56 PM
MikeyMcD 356pm - you nailed it. All of this can be computed ahead of time before making critical decisions so as to see the 'tree' of possibilities each weighed with expected costs of dollars, discomfort, time delays,..., and, of course, ultimate success. More to come.
Posted by: George Rebane | 20 April 2012 at 04:54 PM
So repeat the test at different vendors, and see what is happening on at least three attempts.
Posted by: Douglas Keachie | 20 April 2012 at 10:09 PM
DouglasK 1009pm - Brilliant! and so revealing.
Posted by: George Rebane | 21 April 2012 at 08:40 AM
George, I love your post and your approach. Perhaps in general medicine there is a bigger market for it, but my experience with pediatricians and obstetricians is that you can count on downright kitten-birthing if you dare to bring up the cost (emotional, physical or economic) of any test or procedure. Physicians use their status and mysterious voodoo knowledge to brow-beat parents into accepting whatever is "standard of care," which I strongly suspect is more about what pharmaceutical companies are selling than anything resembling math. I don't see how that happened unless patients have been willing participants. A black market has already sprung up for child birth, but most of those practitioners are just selling a different variety of voodoo from their medical counterparts. How are you going to sell math to people who are looking for magic?
Posted by: Ben Mavy | 23 April 2012 at 06:34 PM
BenM 634pm - That's a valid concern, and one that has been in the forefront of keeping the patient as an ignorant buyer. My own experience with my mother's decision making on having a valve replacement was Exhibit A; I had to get pretty rough with her cardiologist to squeeze the clinical data out of him so I could present it clearly to my mom. When she saw the numbers (and she was by no means mathematically astute, but she understood odds), it was easy for her to make a decision and she appreciated the clarity of the alternatives.
Going into a reasonable buyer's market in healthcare will be difficult and take a lot energy/ingenuity. The priesthood will want to maintain status quo; no peeking at the man behind the curtain. Good comment.
Posted by: George Rebane | 23 April 2012 at 08:14 PM