The independent news organization of Duke University

Betting against the odds

Last Wednesday night, Cameroon Indoor rocked and rolled as the Blue Devils rallied from a 14-point halftime deficit to defeat hated North Carolina by six points. Afterward, a bonfire raged as the student body celebrated.

At first glance, the win over UNC—and with such a dominant second-half performance!—would seem to prove Duke’s superiority over the Tar Heels.

But according to college basketball statistician Ken Pomeroy, Duke’s close win may have suggested the opposite. Last season, 1,049 conference games were re-matches of games previously played that season. The home team won the second matchup less than half the time. And if the first game was decided by single digits (like last week’s big win), the winner of the first game won the rematch only 33 percent of the time.

The point of this statistical odyssey was not to suggest that Duke won’t beat UNC March 5 (they will, obviously), but to demonstrate the importance of keeping track of results and using real data to guide conclusions. Imagine a situation in which an individual had to use the results of a head-to-head matchup to choose the better of two teams (or better yet, just imagine the NCAA Tournament seeding process). If the winner of the head-to-head matchup was playing in its own arena, it may not be fair to conclude that the losing team is truly inferior.

In the grand scheme of things, the NCAA Tournament is a fairly trivial example. (Back when I was a sports columnist, I never could’ve imagined writing that sentence.) But in my real life, where I try to pretend I don’t care too much about college basketball, the importance of using real data to guide conclusions is extremely important. We call it evidence-based medicine.

I recently finished a rotation in medical oncology clinic. I saw patients with lung cancer, liver cancer, colon cancer, breast cancer, bone cancer, leukemias and lymphomas. In general, when a patient is sent to see a medical oncologist, the prognosis is likely grim—medical oncologists are the doctors that administer chemotherapy to those patients whose cancer cannot be cured by surgery, in many cases because the cancer has already spread throughout the patients’ bodies.

The chances of long-term survival with metastatic cancer are small. Metastatic colon cancer, for example, carries a 5 percent five-year survival rate; for lung cancer, only one percent of patients survive five years, and only 19 percent even survive for a single year.

And yet, every day, the attending physician and I would walk into patients’ rooms, and, armed with pictures from CT scans showing a decrease in the size of our patients’ multiple tumors, we would pronounce that the patient’s cancer was “responding” to our chemotherapy. And every day, I would wonder who we thought we were fooling. We knew the prognoses were bad. The patients, universally, knew their prognoses were bad. And still, we’d say, “The chemo is working, your cancer is responding.”

But after the win over UNC, it all made sense. We can use the data to set expectations, to know that, statistically, Duke only has a 33 percent chance of winning at the Dean Dome or that a metastatic lung cancer patient only has a one percent chance of living for five years.

But, wonderfully, statistics will never tell the whole story. Only one in three teams that win a close home game win the re-match, but maybe Duke will be the one. Only one in 100 patients will truly overcome lung cancer, but maybe this patient will be the one.

Just like I didn’t spend the whole second half of the UNC game fretting over whether a close win would doom Duke in the re-match, my oncology patients shouldn’t spend the rest of their lives thinking about their diseases’ dismal prognoses.

Like any basketball fan, they should worry, but only when things are going badly, and they should cheer their heads off whenever they get the chance.

Alex Fanaroff is a fourth-year medical student. His column runs every Wednesday.


Share and discuss “Betting against the odds” on social media.