Headline: “How I used math to conquer my cancer” by Michael Kaplan.
Gist: fellow named Reitzen, 45, discovered he had kidney cancer.
“The doctor came up to my house and had fantastic bedside manner,” says Reitzen, now 57. “He told me that the tumor was larger than the kidney itself, which would necessitate removing my kidney and the lymph nodes around it.”
But Reitzen didn’t want a body part removed — unless data showed it to be absolutely necessary — so he sought a second opinion.
So Reitzen used “data mining”, i.e. he conducted a search of the Internet, to find the kind of surgeon he wanted. He found two. One “thought that it would be better for me not to lose [the kidney] because I had high blood pressure…He had no bedside manner, but I liked his opinion of there being a 70 percent chance that I would not be left with just one kidney.”
The cancer was duly hacked out. “But the oncologist/hematologist who gave him the original diagnosis suggested chemotherapy to avoid a recurrence.”
Using decision-theory math, Reitzen took into account the likelihood of surviving with and without chemo. With the chemo, there was an average life expectancy of 8.1 years; without it, he would be looking at 7 years.
But taking chemo’s side effects into account, his quality of life would be only 70 percent, which he based on information from health-related quality-of-life studies on chemotherapy patients. Without the chemo, he would have no side effects and 100 percent quality of life. After doing the math, quality-adjusted life years came to 5.7 with chemo and 7 without.
“Would you want to exchange a 15 percent life-expectancy increase for a 30 percent drop in quality of life?” he asks.
Quality-adjusted life years? Yet another attempt to quantify the unquantifiable. Wikiwik’s dry statement puts it best: “To be dead is associated with 0 QALYs, and in some circumstances it is possible to accrue negative QALYs to reflect health states deemed ‘worse than dead’.”
Reitzen’s calculation is simple: 8.1 years of “expected” life left under chemo times 70% (0.70) of “worthy living” equals 5.7 QALY. And 7 years “expected” life without chemo times 100% (1) of full life equals 7 QALY. Since 7 is greater than 5.7, the calculation says to pick no chemo.
Which was his choice: no chemo. “Ten years later, Reitzen is cancer-free, with two functioning kidneys, and did not have to endure the misery of chemotherapy. The treatment that he was initially offered has been deemed ineffective for kidney cancer.”
To which we say, God bless him. But do note, and do pause, at the statement The treatment that he was initially offered has been deemed ineffective for kidney cancer. And then recall that the “expected” 8 years he was to live with that now-deemed-ineffective treatment was calculated on the belief (or assumption) that the treatment was effective. All probability is conditional.
While Reitzen routinely makes critical investment decisions with the guidance of math, and found comfort in his cancer-by-the-numbers strategy, doctors don’t think his approach to treatment should be relegated only to statistics-loving professionals.
To which we say capital-A-men. A doctor’s “loss” and “gain” are different, and sometimes far different, than a patient’s. The internal “calculations” the doctor uses to recommend a specific treatment might be best in his mind and for him, but not for the patient. Only the patient can decide what a “worthy” life is.
Now the real problem is only partially the attempt to quantify the unquantifiable. A life’s worth cannot be put to a number, and doing so can only be the crudest of approximations. Collapsing a life’s worth to a single number necessarily strips away vast amounts of information. And that means bad decisions can be made.
The second difficulty is, even is all can be quantified, all should be quantified in its fullness. When it was believed (by at least that one doc) that chemo would work, this doc’s calculation said patients would live an “expected” 8 years. That number is found by multiplying the probability (conditional, as all these probabilities are, on the doc’s belief in the treatment) a patient would live one year by 1, the probability the patient would live 2 years times 2, and so on, the result being a weighted average.
Collapsing these probabilities to one “expected” number again strips away information. The whole swath should be presented to the patient—not necessarily at the finest levels. Do we really need information per year 10, 20 years out?
Well, you get the idea. What is most crucial to grasp is not all this numeric mumbo jumbo, but that not accepting harsh treatment is an solid option. Though it cannot be quantified, a shorter life without the brutal suffering caused by a treatment can be much better than one with it. Especially when it is, as it always should be, recalled that All men are mortal, etc.