0:00:00.147,0:00:02.408
- [Amy] What are the costs[br]and benefits of mammograms?

0:00:02.408,0:00:04.265
How do you weigh, for example,

0:00:04.265,0:00:06.948
the potential savings[br]and healthcare cost

0:00:06.948,0:00:08.857
against the potential lives saved,

0:00:08.857,0:00:11.061
against the increase[br]in psychic costs --

0:00:11.061,0:00:13.566
these are all very,[br]very difficult issues.

0:00:13.566,0:00:15.920
♪ [music] ♪

0:00:23.249,0:00:25.729
My collaborators and I[br]have recently been looking

0:00:25.729,0:00:28.256
into the highly controversial area

0:00:28.256,0:00:30.940
of recommendations[br]for whether and when

0:00:30.940,0:00:33.803
to screen for breast cancer[br]using mammograms.

0:00:33.803,0:00:35.757
There's concerns[br]that not enough women

0:00:35.757,0:00:37.524
are getting screened[br]for mammograms.

0:00:37.524,0:00:40.542
- [Tamar] This is a hugely[br]relevant issue for many women.

0:00:40.542,0:00:43.117
One in eight women[br]get a breast cancer diagnosis

0:00:43.117,0:00:44.298
in their lifetimes,

0:00:44.298,0:00:46.888
so there's this recommendation[br]to get annual testings at age 40.

0:00:46.888,0:00:49.258
- There's also concerns that[br]many women are getting screened

0:00:49.258,0:00:50.862
and having false positives...

0:00:50.862,0:00:53.146
- [Abigail] ...which is when[br]you detect a tumor,

0:00:53.146,0:00:54.380
and you treat it,

0:00:54.380,0:00:57.812
but if you had left it alone,[br]it wouldn't have been a problem.

0:00:57.812,0:00:59.782
Overdiagnosis is a problem

0:00:59.782,0:01:02.537
because you're incurring[br]a lot of costs

0:01:02.537,0:01:05.310
that you really[br]shouldn't have had to.

0:01:05.310,0:01:07.379
It causes a lot[br]of anxiety to people

0:01:07.379,0:01:09.048
if they're diagnosed with cancer,

0:01:09.048,0:01:13.433
so we really want[br]to reduce overdiagnosis.

0:01:13.433,0:01:17.962
- So we're asking how does a person[br]who responds to a recommendation --

0:01:17.962,0:01:19.783
so, in this case, for mammograms --

0:01:19.783,0:01:22.352
differ from people[br]that don't get screened

0:01:22.352,0:01:24.758
or from the average person[br]in the population.

0:01:28.881,0:01:32.286
- We got data on people[br]who were screened,

0:01:32.286,0:01:34.868
so we could see the rates[br]of screening by age.

0:01:34.868,0:01:36.031
- [Abigail] Before age 40,

0:01:36.031,0:01:39.039
about 10% of people[br]were getting screened.

0:01:39.039,0:01:43.348
After age 40, that jumped up[br]to about 35% of people.

0:01:43.348,0:01:45.441
- The share of people[br]getting mammograms

0:01:45.441,0:01:47.360
went up drastically at 40,

0:01:47.360,0:01:49.932
but the share of people[br]who got mammograms

0:01:49.932,0:01:52.989
who tested positive[br]for cancer was going down.

0:01:53.655,0:01:55.403
But that just tells us

0:01:55.403,0:01:59.112
how the people who respond[br]to the recommendation at 41

0:01:59.112,0:02:01.351
differ from the people[br]who get mammograms

0:02:01.351,0:02:03.888
without a recommendation at 39.

0:02:03.888,0:02:06.614
It doesn't tell us[br]what we really wanted to know,

0:02:06.614,0:02:10.664
which is, how do the women[br]who are getting mammograms at 41 --

0:02:10.664,0:02:11.865
when it's recommended --

0:02:11.865,0:02:14.755
differ from the women[br]who aren't getting mammograms?

0:02:15.392,0:02:17.476
That's a very hard problem[br]to answer

0:02:17.476,0:02:21.099
because what you need is data[br]on the underlying cancer

0:02:21.099,0:02:24.303
of people who, by definition,[br]aren't being screened.

0:02:24.303,0:02:27.660
That's where the biologists[br]and the clinicians came in --

0:02:27.660,0:02:28.911
they developed models

0:02:28.911,0:02:32.065
of the underlying incidence[br]of breast cancer

0:02:32.065,0:02:35.022
in, say, a random 25-year-old[br]in the population

0:02:35.022,0:02:37.641
as well as, most importantly,[br]for our purposes,

0:02:37.641,0:02:40.408
how it progresses[br]in the absence of treatment.

0:02:40.849,0:02:42.935
- This is, I think, a wonderful case

0:02:42.935,0:02:46.090
where the medical community[br]has so much to add here,

0:02:46.090,0:02:48.825
and then the economics[br]community can take that

0:02:48.825,0:02:50.411
and build onto that.

0:02:50.411,0:02:54.070
- This model gave us[br]the underlying level of cancer,

0:02:54.070,0:02:58.559
and, using that, we could back out[br]what the cancer level was

0:02:58.559,0:02:59.927
for people who never got screened.

0:02:59.927,0:03:02.093
It's like supposing[br]you have a roomful of people.

0:03:02.093,0:03:05.270
There's ten people, and you know[br]that half of them have cancer,

0:03:05.270,0:03:06.438
but you don't know who.

0:03:06.438,0:03:08.494
So you pick five,[br]and you screen them.

0:03:08.494,0:03:10.412
and only one of them has cancer --

0:03:10.412,0:03:12.549
then you know that,[br]of the other five,

0:03:12.549,0:03:16.238
four of them have cancer,[br]even though you didn't screen them.

0:03:20.411,0:03:23.751
So we find that the people[br]who follow the recommendation

0:03:23.751,0:03:27.507
actually are healthier than[br]the people who don't follow it --

0:03:27.507,0:03:29.293
they're less likely to have cancer,

0:03:29.293,0:03:31.661
and if they're diagnosed[br]with cancer,

0:03:31.661,0:03:35.387
they're more likely to have[br]an earlier stage cancer

0:03:35.387,0:03:37.861
or a smaller cancer[br]that's less dangerous.

0:03:38.456,0:03:41.363
We wanted to study what types[br]of women get mammograms

0:03:41.363,0:03:44.668
to see whether they're the types[br]of women who would benefit most.

0:03:44.668,0:03:46.921
Right now, it seems like[br]the recommendations

0:03:46.921,0:03:49.727
are targeting people[br]who are most healthy.

0:03:49.727,0:03:52.513
Maybe they engage in other[br]preventive health behaviors,

0:03:52.513,0:03:54.300
and maybe they're doing really well,

0:03:54.300,0:03:57.822
but we're not reaching the people[br]who have a higher burden of cancer.

0:03:57.822,0:04:00.109
So we'd like to look into ways

0:04:00.109,0:04:02.630
that we could target[br]those people better.

0:04:02.630,0:04:04.516
- So, in the end, to be clear,

0:04:04.516,0:04:07.588
we don't resolve[br]or even attempt to resolve

0:04:07.588,0:04:10.960
the question that, in some sense,[br]motivates the whole literature,

0:04:10.960,0:04:15.919
which is, should we recommend[br]screening at 40, 45, 50, 35?

0:04:15.919,0:04:19.799
Instead, all we do is add[br]another piece to the puzzle.

0:04:19.799,0:04:22.928
- So our paper brings[br]an additional dimension

0:04:22.928,0:04:24.515
that should be considered

0:04:24.515,0:04:26.017
when sort of designing[br]these policies.

0:04:26.017,0:04:27.478
- You have to worry about

0:04:27.478,0:04:29.290
who the people are[br]that you're reaching,

0:04:29.290,0:04:32.467
and if they're the people[br]who are more or less at risk

0:04:32.467,0:04:33.637
for having cancer

0:04:33.637,0:04:36.037
than a randomly chosen person[br]in the population.

0:04:37.269,0:04:39.375
- [Narrator] Want to see more[br]Economists in the Wild?

0:04:39.375,0:04:41.159
Check out our playlist.

0:04:41.159,0:04:42.262
Are you a teacher?

0:04:42.262,0:04:44.907
Here's some related material[br]for your classroom.

0:04:44.907,0:04:46.895
♪ [music] ♪