Cervical Cancer Just Got Much Deadlier -- Because Scientists Fixed a Math Error (arstechnica.com)

An anonymous reader quotes a report from Ars Technica: Cervical cancer is 77 percent more deadly for black women and 44 percent more deadly for white women than previously thought, researchers report today in the journal Cancer. But the lethal boosts aren't from more women actually dying than before -- they're from scientists correcting their own calculation error. In the past, their estimates didn't account for women who had undergone hysterectomies -- which almost always removes the cervix, and with it the risk of getting cervical cancer. We don't include men in our calculation because they are not at risk for cervical cancer and by the same measure, we shouldn't include women who don't have a cervix," Anne F. Rositch, the study's lead author and an epidemiologist at Johns Hopkins told The New York Times. For the study, the researchers looked at national cervical cancer mortality data collected between 2000 to 2012. They also looked into national survey data on the prevalence of hysterectomies. Then, they used those figures to adjust the number of women at risk of dying of cervical cancer. The researchers found that black women have a mortality rate of 10.1 per 100,000. For white women, the rate is 4.7 per 100,000. Past estimates had those rates at 5.7 and 3.2, respectively. The new death rate for black women in the US is on par with that of developing countries. Though the new study wasn't designed to address racial disparities, experts speculate that the large difference reflects unequal access to preventative medicine and quality healthcare.

Comments

Why wasn't this caught in peer review?

[nick13245]

Seems like an obvious error from a statistical analysis standpoint. Makes me wonder how much critical medical research has obvious errors like this.

Both numbers are correct, I would say. Older more

[raymorris]

It seems to me both are true and useful. I would even go so far as to say the original number is more useful.

1 in X women die from cervical cancer. (old number)

Of women who did not have a hysterectomy (prior to cancer), 1 in Y die from cervical cancer. (new number)

Both are true. How might mortality rates be used? One important use is comparisons for policy making decisions:
10 in X women die from heart disease, 1 in X die from cervical cancer. Therefore, we should invest more prevention efforts toward heart disease.

Or:

X% of women die from alcoholism, Y% from cervical cancer. Therefore, we should spend the most money researching cures for ______ ?

For these policy, questions, we want to know how many people are affected. Period. It's not a useful comparison to say "of people who drink, X die from alcoholism, while of people who have a uterus, Y die from cervical cancer". Those numbers don't give us any useful comparison with which to make decisions. The useful numbers for decision making are "how many people could be helped by addressing this issue?"

That's not a math error.

[2fuf]

It's a classification error.