by 
Biostatistician Andrew Vickers, a cancer researcher, uses math and the latest research to help the doctors at the hospital where he works do their best for their patients.
When has a doctor helped you? How much do you trust doctors?
“I love my job: it’s fun and hugely satisfying,” says Biostatistician Andrew Vickers, a cancer researcher at Memorial Sloan Kettering Cancer Center, in New York City. “There have been studies of statisticians,” he adds, “and they have some of the highest levels of job satisfaction of any profession.” Why do people in his line of work like their jobs so much? Vickers explains that he gets to “peer into columns of numbers and extract from them information that helps people live longer, healthier lives.”
When Andrew was in high school, one of his teachers told him that his legs felt stiff. The teacher had been jogging, in order to prepare for a walking trip he planned to take. Andrew, an avid runner, explained how to stretch after jogging. His teacher tried it out and reported “‘That was amazing: I feel so much better.’”
Around the same time, a friend of Andrew’s, a good runner, had a particularly bad experience when he ran a half-marathon. He told Andrew that he had taken glucose tablets before the race. Said Vickers, “‘Toby, you don’t take glucose until half an hour into the race,’” and he cited studies that had taught him how to replenish his blood sugar in a more effective way. Remembering how he helped his teacher and his friend, Vickers points out that “I’ve always been interested in giving people information that will allow them to live healthier lives.”
These days, Vickers researches the best ways of finding and treating cancer. He points out that the cure for cancer “was actually found some time ago”: surgery. But “how do you find it and how do you cut it out?” Vickers and his colleagues study numbers that tell them about what happens to people around the world as they get older and some of them get sick.
Since the late 1970s and early ‘80s, biostatisticians looking at the results of different surgeons’ work have noticed that different doctors tend to get different types of results. 20% of one surgeon’s patients might still be in pain a year after their surgery. 30% or more of another surgeon’s patients might still be in pain.
Part of the problem was that surgeons didn’t tend to keep track of how their patients were doing after they had recovered from surgery. Car dealers, says Vickers, keep track of how many cars they have sold. But surgeons do not tend to have the time or the skills necessary to survey their patients and compare their recoveries with those of the patients of other surgeons. Vickers and his colleagues do this work for the surgeons at his hospital, helping figure out what helps them get the best results for their patients. This makes patients healthier, surgeons and the hospital more successful, and insurance companies happy as well (because it costs them less money when patients recover more quickly).
What kinds of changes do Vickers and his colleagues help surgeons make? It all depends on what the data, the information about how patients do after surgery, indicates. If patients with kidney cancer tend to do better when only part of their kidney is removed, Vickers and his colleagues let surgeons know that it’s better not to take out the whole kidney. And if the ways doctors test for a patient’s risk of getting cancer turn out not to work as well as they might, they suggest better ways to test for the disease.
For instance, Vickers and other scientists did a very large study of men in Malmo, a small town in Sweden. They were able to examine the blood of many of these men, who had their blood drawn in the town’s one hospital over many years. Would the tests being done for prostate cancer predict who would and wouldn’t get the most dangerous kinds of that cancer?
“We had all this great data” about who did and didn’t get different forms of prostate cancer, and “we could go back and pull all of those original blood samples and thaw them out. A chemist looked at the blood to try to detect levels of this protein, which is called PSA,” prostate-specific antigen, a protein used to test for a person’s chances of having or getting prostate cancer.
Vickers used sampling to determine which blood to test for PSA. He and the other researchers chose the blood of men who had developed prostate cancer and matched it with the blood of men of similar ages who visited the hospital at similar times, creating a control group. “We would get the blood for both of them and compare the levels of PSA in the blood samples of these men that had been kept in that freezer in Sweden for all those years.”
Using those samples, they created “a spreadsheet thousands of rows long with numbers” telling them whether each man got cancer, how bad it was, and what his PSA levels had been. And they found a correlation: men with higher PSA levels tended to get the more dangerous forms of the cancer. This led Vickers and his colleagues to suggest changes to the ways doctors test for PSA as men get older. “You should start by giving a single test to men about the age of 45 or 50 and if their PSA is above average, then you focus in on those men and ask them to come back every year for screening. The others you don’t have to screen quite so intensely.”
What math should future biostatisticians be sure to learn? Vickers suggests that it’s not the calculations you should focus on, because computers and calculators will help you do those. “But what computations should you be doing? Which calculations should you be doing?” And when doing all of that math gives you a bunch of numbers, “what do those numbers mean? You have to understand the concepts of math, rather than just how to do the calculations.” And then, like Andrew Vickers, you may be able to help improve people’s health, perhaps even saving their lives.
