What does this positive mammogram mean?

I’m confused. If they don’t get the screening, how do they know the “test” is positive. Isn’t the mammogram the “test?” Maybe I haven’t had enough coffee yet this morning. Can you clarify?

Clarification:

About 1% of all women have breast cancer. When those who do have cancer get a mammogram, the screening will be positive in 90% of all cases. When those who don’t have cancer get a mammogram, the screening will be positive in 9% of all cases.

10%? My research showed that most positives were not cancer upon further testing.

I had a questionable mammogram once but it was cleared with a second one. Would that first test have been considered “positive” in your study?

Matt Browne, This guy’s language is inaccurate.

Mammograms show masses, lumps, calcifications and abnormal densities. They do not come back either positive or negative. If anything looks suspicious on the films, or today onthe imaging, the next step is to do a biopsy.

Many women have fibrocystic breasts, which show lumpiness on films. So there is a lot of unnecessary biopsying going on, but it does lead to peace of mind.

My breast cancer was discovered when the micro calcifications on the mammo films 17 years ago were biopsied and turned out to be malignant.

@gailcalled the substance of Matt’s question is correct. Although it is true that mammograms read many different things, they do come back positive and negative—a positive mammogram means radiologically detected cancer. Matt’s question is a classic medical Bayesian problem, and is vitally important to understand.

Every test is not perfect, and each test has a sensitivity and specificity. For example, fecal immune testing tests for blood in the stool. It is sensitive—it detects minute traces of blood, but the presence of blood does not mean colon cancer. You run the risk of false positivity Then you go to a higly specific test, such as a colonoscopy. Specific tests tend to be expensive, and higher risk. But if it’s positive, it’s positive. You run the risk, however, of a false negative. A mammogram is a screening test.

To make matters more complicated, and where Bayesian analysis comes in, is that you need to know the incidence of disease in a population. What I like to teach people is let’s say you test a population of cloistered nuns in rural Canada who have never had sex nor done drugs with an HIV test. The chance that a positive test in that group is a true positive test is low. Contrary, if you do the same test to a group of promiscuous young drug abusing gay men in San Francisco, the chance that a positive test is a true positive is much higher.

So getting back to Matt’s question, a mammogram that says cancer will be a true cancer in 9% of the cases, meaning that 91% of women got unnecessary biopsies.

This is the essence of why the USPSTF recommended mammograms to low risk women only after age 50. The incidence of disease is low, so the chance that a mammogram was a true positive was very low.

That Bayesian logic didn’t keep my wife from having mammograms, though.

I’ve heard that the new Halo test has even MORE false positives.

@Rarebear: I stand corrected. Thank you for the clear (and funny) answer. I love concrete examples after the abstracions.

I will be forever grateful for my mammogram, because the cancerous microcalcifications were not palpable. Buy the time of the lumpectomy (very shortly thereafter), I had two nodes involved and was considered one of the lucky ones.

@gailcalled – He wrote a book for the masses, this is how I learned of this. Please interpret “positive” as suspicious and “negative” as not suspicious.

He was also on German television complaining that most educated people don’t understand risks and probabilities. He recommends changing school curricula.

Right. Easier to understand than medical Bayesian problem, however. (My new word of the day.)

edit: “abstractions”

AP Statistics time!

99% * 9% = 8.91%
1% * 90% = 0.9%
0.009 / (0.0891 + 0.009) = 9.1743%

The closest answer is D.

@Judi It is probably true that the new test will increase the rate of false positives. It will, however, reduce the rate of false negatives. A false positive costs money, but a false negative kills people.