Thursday, October 23, 2008

Follow-up to McCain and melanoma

Kyle makes a couple of arguments below in the comments of my last post, and my reply is long enough and separate enough from my original post that I'm posting it separately:

Kyle,
I don't disagree with your political conclusion of what to do at the end of the day. The point is that any president might die. The fallacies of the Altman argument are that:

1) the medical chart does not contain information on what is most likely to kill a president; and
2) John McCain has some unknown risk of recurrent melanoma, which further information might allow us to calculate slightly differently than what we know to be his generic risk without further information. But John McCain either will or will not die in office if elected. And, with or without melanoma, he has a reasonable probability of dying in office because US presidents generically do have a much higher probability of dying in a given four year span than many other people, entirely because of the risks associated with the office rather than the officeholder.

So whether the possibility is x% chance of melanoma combined with y% chance of cardiac disease combined with z% chance of lung cancer, the generic risk to the officeholder already meets a test of likelihood. That is, statistically, taking any president at any age, a generic voter should assume--regardless of what is in the medical chart--that the president is at least as likely to die before the end of the term as the voter herself, or at least, the voter's children.

So, the reason this is relevant to a medical blog and a medical argument is that melanoma in particular should not change the intervention--i.e., your vote. Whether or not he has a given chance of recurrent melanoma, Sarah Palin is not a qualified vice president, and she has a high generic chance of becoming president if elected as vice president.

Incidentally, though the two candidates' generic risks of mortality can be influenced by their age, this is an easily discernible and intuitively obvious risk which does not require Lawrence Altman digging through colonoscopy reports.

Because the baseline presidential risk of death is high, the issue of Sarah Palin's competence is more or less exactly the same as the issue of Joe Biden's competence. If Obama had picked Palin, it would be just as bad a pick, for the exact same reason.

Now, take the low-probability but possible chance of Lawrence Altman finding something in the chart that a group of Mayo Clinic doctors deliberately concealed or misrepresented when they vouched for McCain's health. When going through McCain's chart, the greatest likelihood is that whatever Altman would find would be a "false-positive"--that is, it would raise concerns for voters but would not actually change the outcome of the next four years.

That would be unfair to McCain, and more importantly to democracy in general. It would represent a medicalization of democracy--a completely out-of-proportion ability of a few doctors and medical screening tests to influence democratic elections.

Also, at least as importantly, it would be unfair to anyone with "concerning" things in their medical charts who wanted to be in a powerful position. There is no magic about a president; to the workers of a company, the CEO and CFO might well have greater power to determine their quality of life in the next four years. So, should boards of directors have the right to examine every bit of the medical record of any executive? And how far down the management trail are you willing to go with that logic? And what does that mean for people with family history of genetic diseases, for people with past medical issues, for people with high epidemiological risk?

Aside from this, I want to make sure I am clear about the statistical and medical argument I'm making, so let me take a less-loaded and very common medical parallel. Let's say someone comes into the emergency department with shortness of breath and a fast heart rate. There are various possibilities. One is a blood clot in the lung, known as a pulmonary embolism (PE).

To diagnose PE, I can get a CT scan with contrast, which can damage the kidneys. I can also do a blood test, the result of which will increase or decrease my estimate of the chance that the person has a PE. If the blood test increases my estimate, I'll get a CT scan. If it decreases my estimate, I won't get a CT scan.

So, if I think there is a clear alternative explanation and no reason to suspect PE, should I get the blood test? No, because it will not change my decisionmaking. I am more likely to get a false positive than a true positive, and therefore, the CT scan is more likely to be dangerous than helpful.

What if it's someone has a very high risk of pulmonary embolism--a known clot in the veins of the leg, and cancer which predisposes to PE? Will I do the blood test? No, because I don't care if the test is negative. I will still not be satisfied until I do the CT scan.

So it turns out that the only time to do the blood test is if you're not sure whether or not a PE is likely--the risk is indeterminate, or "medium-risk." The argument Altman is making relies on the idea that a medical chart is the equivalent of this blood test. That is, you will cast your vote--the potentially wise or unwise and highly consequential decision, the equivalent of CT scan with contrast--based on the information in the medical chart. This is a common issue in medicine: will a given test change your medical choices? If not, why are you doing the test?

Given that the Altman chart review is the equivalent of the "medium-risk" blood test, where we are trying to convince ourselves that a candidate will very likely die in office or will very likely not die in office. But this is not the right choice from the point of view of the "change your choice" test. That's because a president is not medium risk for death in the next four years. At least historically, a president is high risk. So, I certainly will not fail to care about the qualifications of a vice president. But I will not use the medical chart to decide how much I care. I already care a lot, because I know no matter what the chart says, I care. Any reasonable person should vote with the assumption that a president has a high chance of dying in office.

If examining the chart had no impact, then it wouldn't matter. And I'm the last person to defend McCain.

But I'm not defending McCain or a decision to vote for him. I'm defending people with a history of melanoma, and a lot of other people too. For anyone with higher probabilities of disease (whether through genetics, behavior, or past medical history), the risk of making the argument Altman is making is actually quite high. So, thinking through the issue of whether you do a test, and whether the Altman chart review is a test we should be doing:

the test does not change my ultimate decision
the test has a high chance of falsely influencing my thinking
the test has a high chance of other bad effects.

And therefore, the test should not be conducted. And Lawrence Altman doesn't need to look through every page of John McCain's medical chart.

[edited for clarity later in the day]

1 comment:

Vasha said...

Thanks for this explanation. It may come in handy if I get sucked into political discussions.