Latest / Truth in Motion with Dr. Jocelyn / What “Linked To” Really Means

Transcript

What “Linked To” Really Means · Truth in Motion with Dr. Jocelyn

• 0:02
  SPEAKER_01: Welcome to Truth in Motion with Dr.
• 0:05
  SPEAKER_01: Jocelyn, where insight, clarity, and wisdom help you thrive.
• 0:10
  SPEAKER_00: Welcome back.
• 0:12
  SPEAKER_00: Yesterday, we stepped into epistemology, the study of how we know.
• 0:19
  SPEAKER_00: We stood on a British warship in 1747 where James Lind split 12 sailors into six pairs and asked the question that runs through this whole week.
• 0:32
  SPEAKER_00: Compared to what?
• 0:34
  SPEAKER_00: And we laid out the ladder of evidence, from expert opinion at the bottom to the systematic review at the top.
• 0:43
  SPEAKER_00: Today we climb the first rung in depth.
• 0:46
  SPEAKER_00: Studies that look back in time.
• 0:49
  SPEAKER_00: They're called retrospective studies.
• 0:52
  SPEAKER_00: And the classic kind is the case control study.
• 0:56
  SPEAKER_00: We'll sit in on one of the most famous studies ever done.
• 1:00
  SPEAKER_00: We'll see how eight young women in Boston changed the way a drug was prescribed, and we'll see how a looking back study can point the wrong way.
• 1:11
  SPEAKER_00: And why?
• 1:12
  SPEAKER_00: Why does this matter?
• 1:14
  SPEAKER_00: Because so many of the health warnings we hear come from looking back, linked to, associated with, people who did this were more likely to get that.
• 1:26
  SPEAKER_00: Those phrases are everywhere, on the news, in our feeds, in the waiting room.
• 1:32
  SPEAKER_00: Some of those warnings are the first sign of a real danger.
• 1:36
  SPEAKER_00: Some are false alarms.
• 1:38
  SPEAKER_00: And from the headline alone, they sound exactly the same.
• 1:42
  SPEAKER_00: Many of us have changed what we eat or stopped something we loved because of a headline like that.
• 1:48
  SPEAKER_00: And many of us have ignored a real warning because we've heard so many that came to nothing.
• 1:54
  SPEAKER_00: When we know how to look back and we know how to look at looking back studies, we can tell the difference.
• 2:02
  SPEAKER_00: That's what today is for.
• 2:04
  SPEAKER_00: Let's go to London, just after the Second World War.
• 2:08
  SPEAKER_00: Something strange was happening.
• 2:11
  SPEAKER_00: Between 1922 and 1947, the number of recorded deaths from lung cancer in England and Wales went from 612 a year to 9,287.
• 2:26
  SPEAKER_00: Roughly 15 times as many.
• 2:30
  SPEAKER_00: What was causing it?
• 2:32
  SPEAKER_00: Everyone had a theory.
• 2:34
  SPEAKER_00: Some said doctors were simply getting better at diagnosing it.
• 2:37
  SPEAKER_00: Others blamed the air, car exhaust, the dust from newly tarred roads, gas works, factories, coal fires.
• 2:47
  SPEAKER_00: A young physician named Richard Dahl and a statistician named Austin Bradford Hill set out to find the answer.
• 2:55
  SPEAKER_00: And here's what I love about this story.
• 2:58
  SPEAKER_00: Dahl later said that he and Hill first suspected the new tarmac on the roads or motor car fumes, not cigarettes.
• 3:07
  SPEAKER_00: They didn't go looking for the answer they expected.
• 3:10
  SPEAKER_00: They went looking for what was true.
• 3:13
  SPEAKER_00: Here's what they did.
• 3:15
  SPEAKER_00: They asked 20 London hospitals to tell them whenever a patient was admitted with lung cancer.
• 3:22
  SPEAKER_00: Then four trained interviewers, hospital social workers called almonders, sat down with each patient and asked detailed questions.
• 3:33
  SPEAKER_00: Where they had lived, what work they had done, their exposure to fumes and gas, and how much they smoked and for how long.
• 3:41
  SPEAKER_00: Now here's the key.
• 3:43
  SPEAKER_00: For every lung cancer patient, they also interviewed another patient who did not have cancer.
• 3:50
  SPEAKER_00: Same sex, same five-year age group, same hospital at about the same time.
• 3:57
  SPEAKER_00: That's the control in the case control.
• 4:01
  SPEAKER_00: The cases are the people with the disease.
• 4:03
  SPEAKER_00: The controls are people as much like them as possible without it.
• 4:08
  SPEAKER_00: And then we look back and ask, what was different about them before?
• 4:14
  SPEAKER_00: Why match them so closely?
• 4:17
  SPEAKER_00: Because older people get more cancer than younger people.
• 4:21
  SPEAKER_00: Men in 1950 smoked more than women, and a patient in one part of London may have breathed different air than a patient across town.
• 4:30
  SPEAKER_00: Matching by age, sex, and hospital takes those differences off the table.
• 4:36
  SPEAKER_00: So whatever is left over stands out.
• 4:39
  SPEAKER_00: It's James Lynn's question again, compared to what?
• 4:44
  SPEAKER_00: By 1950, they had interviewed 709 lung cancer patients and 709 matched controls.
• 4:54
  SPEAKER_00: Most of them were men, 649 in each group.
• 4:58
  SPEAKER_00: Here's what they found.
• 5:00
  SPEAKER_00: Of the 649 men with lung cancer, only two had ever smoked.
• 5:06
  unknown: Two.
• 5:07
  SPEAKER_00: Among the 649 men without cancer, 27 had never smoked.
• 5:14
  SPEAKER_00: And the more someone smoked, the stronger the pattern.
• 5:18
  SPEAKER_00: Among the men with lung cancer who smoked, about one in four smoked 25 or more cigarettes a day.
• 5:26
  SPEAKER_00: Among the smokers without cancer, it was about one in eight.
• 5:30
  SPEAKER_00: In September of 1950, they published their results in the British Medical Journal.
• 5:36
  SPEAKER_00: They wrote, and I'll quote them exactly, quote, smoking is a factor and an important factor in the production of carcinoma of the lung, unquote.
• 5:47
  SPEAKER_00: And Richard Dahl, who had been a smoker, stopped smoking.
• 5:51
  SPEAKER_00: There's one more part of this study I want you to hear, because it shows what honest research looks like.
• 5:58
  SPEAKER_00: The interviewers almost always knew which patients had lung cancer.
• 6:03
  SPEAKER_00: So a fair critic could ask, did they lean on the cancer patients a little harder?
• 6:09
  SPEAKER_00: Did they hear more smoking because they expected to?
• 6:14
  SPEAKER_00: Dolan Hill had a way to check.
• 6:16
  SPEAKER_00: Some patients had been sent to them as lung cancer cases, but later turned out not to have lung cancer at all.
• 6:24
  SPEAKER_00: There were 209 of them.
• 6:26
  SPEAKER_00: When they were interviewed, everyone, the interviewers and the patients believed they had lung cancer.
• 6:34
  SPEAKER_00: So if the interviewers' expectations were shaping the answers, those 209 people should have looked like smokers too.
• 6:42
  SPEAKER_00: They didn't.
• 6:43
  SPEAKER_00: Their smoking looked like the other patients without cancer.
• 6:47
  SPEAKER_00: That's what good science does.
• 6:50
  SPEAKER_00: It tests its own bias.
• 6:52
  SPEAKER_00: It asks, how could we be fooling ourselves?
• 6:56
  SPEAKER_00: And then goes and checks.
• 6:58
  SPEAKER_00: So why do we look back?
• 7:00
  SPEAKER_00: So why use a looking back or a retrospective study at all?
• 7:05
  SPEAKER_00: First, it's fast.
• 7:07
  SPEAKER_00: Lung cancer takes decades to develop.
• 7:10
  SPEAKER_00: A study that waited for it to appear would take decades too.
• 7:15
  SPEAKER_00: Looking back, Dolan Hill had their answer in about two years.
• 7:20
  SPEAKER_00: Second, it's the best tool we have for rare diseases.
• 7:24
  SPEAKER_00: If a disease strikes one person in 100,000, we could follow thousands of people for years and never see a single case.
• 7:33
  SPEAKER_00: Looking back, we start with the people who already have it.
• 7:37
  SPEAKER_00: Here's how powerful that can be.
• 7:40
  SPEAKER_00: In the late 1960s, doctors at Massachusetts General Hospital in Boston saw something they had almost never seen, a rare cancer of the vagina.
• 7:51
  SPEAKER_00: A kind that normally appears in older women was showing up in young women in their teens and early 20s.
• 7:59
  SPEAKER_00: Seven of them within a few years, and then an eighth.
• 8:03
  SPEAKER_00: A gynecologist named Arthur Herbst and his colleagues wanted to know why.
• 8:09
  SPEAKER_00: So for each of the eight young women, they found four women born in the same hospital within five days of her, 32 controls.
• 8:18
  SPEAKER_00: Then they looked back at the mother's pregnancies.
• 8:22
  SPEAKER_00: Seven of the eight mothers had taken a drug while pregnant, a synthetic estrogen called DES, or diethylstilbestrol.
• 8:34
  SPEAKER_00: Doctors had prescribed it for years to prevent miscarriage.
• 8:38
  SPEAKER_00: Among the 32 control mothers, none had taken it.
• 8:43
  SPEAKER_00: They published it in 1971.
• 8:46
  SPEAKER_00: That same year, the FDA told doctors to stop prescribing it to pregnant women.
• 8:52
  SPEAKER_00: Eight women, eight women and a careful look back changed medicine.
• 8:57
  SPEAKER_00: A study that followed people forward could never have found it in time.
• 9:02
  SPEAKER_00: The cancer was too rare and it took nearly 20 years to appear.
• 9:07
  SPEAKER_00: What a retrospective study can't do, now for the other side.
• 9:12
  SPEAKER_00: Because a looking back study has real weak spots and we need to see them clearly.
• 9:18
  SPEAKER_00: The first weak spot is memory.
• 9:21
  SPEAKER_00: Most looking back studies depend on what people remember, and memory bends.
• 9:26
  SPEAKER_00: Dahl and Hill knew this.
• 9:28
  SPEAKER_00: When they asked 50 of their patients the same smoking questions six months or more later, many of the answers had shifted.
• 9:36
  SPEAKER_00: And think about the mothers in a study like the DES study.
• 9:40
  SPEAKER_00: A mother whose child is seriously ill searches her memory far harder than a mother whose child is well.
• 9:48
  SPEAKER_00: So the two groups may not remember with the same care.
• 9:52
  SPEAKER_00: That's called recall bias, and it's built into looking back studies.
• 9:58
  SPEAKER_00: The second weak spot is the comparison group.
• 10:02
  SPEAKER_00: A looking back study is only as good as the people it compares against.
• 10:07
  SPEAKER_00: Choose the wrong comparison and we can find a link that isn't there.
• 10:11
  SPEAKER_00: Here's a famous example.
• 10:13
  SPEAKER_00: In 1981, a team of researchers at Harvard published a looking back study in the New England Journal of Medicine.
• 10:21
  SPEAKER_00: It found that people with pancreatic cancer were more likely to be heavy coffee drinkers.
• 10:28
  SPEAKER_00: It made headlines.
• 10:30
  SPEAKER_00: But look at who the comparison group was.
• 10:33
  SPEAKER_00: The controls were other patients of the same doctors, and many of those doctors were stomach and digestion specialists.
• 10:40
  SPEAKER_00: Their other patients often had digestive problems, and people with digestive problems are often told to cut back on coffee.
• 10:48
  SPEAKER_00: On top of that, the researchers left out patients with diseases linked to smoking or alcohol, and smoking and drinking tend to travel with coffee.
• 10:58
  SPEAKER_00: So the comparison group drank less coffee than ordinary people, and that made coffee look guilty.
• 11:05
  SPEAKER_00: Later studies didn't confirm it, and in 1986, the same research team published new data from their own study and wrote that it did not support the link.
• 11:17
  SPEAKER_00: That's the honest end of the story.
• 11:19
  SPEAKER_00: They checked and they said so.
• 11:22
  SPEAKER_00: The third weak spot is the biggest one.
• 11:25
  SPEAKER_00: A looking back study shows a pattern.
• 11:28
  SPEAKER_00: It can't on its own prove a cause.
• 11:31
  SPEAKER_00: People who share one habit often share others.
• 11:34
  SPEAKER_00: People who smoked in 1950 may also have worked in different jobs or lived in different places.
• 11:41
  SPEAKER_00: Dalin Hill asked about many of those things to rule them out, but we can only rule out what we think to ask about.
• 11:49
  SPEAKER_00: So linked to means exactly that.
• 11:53
  SPEAKER_00: Linked.
• 11:54
  SPEAKER_00: It's a question worth asking.
• 11:56
  SPEAKER_00: It's not yet an answer.
• 11:58
  SPEAKER_00: So here's what that gives us.
• 12:00
  SPEAKER_00: When we see linked to, associated with, or people who did this had more of that, we can ask three questions.
• 12:09
  SPEAKER_00: First, how did they find out what people did?
• 12:14
  SPEAKER_00: From memory, from records?
• 12:16
  SPEAKER_00: Memory is useful, but it bends.
• 12:20
  SPEAKER_00: Second, who were they compared with?
• 12:23
  SPEAKER_00: Were the two groups truly alike except for the one thing being studied?
• 12:28
  SPEAKER_00: Or was the comparison group already different in some way?
• 12:32
  SPEAKER_00: Third, is this a pattern or a proof?
• 12:36
  SPEAKER_00: A looking back study can sound the alarm.
• 12:39
  SPEAKER_00: It can tell us where to look, but the alarm still needs to be checked.
• 12:44
  SPEAKER_00: Let's try these questions on a made-up headline.
• 12:48
  SPEAKER_00: Quote, night owls linked to higher risk of heart disease, unquote.
• 12:54
  SPEAKER_00: How did they find out who was a night owl?
• 12:57
  SPEAKER_00: Probably by asking, and people may not remember their sleep the way it really was.
• 13:02
  SPEAKER_00: Who are they compared with?
• 13:04
  SPEAKER_00: People who go to bed early may also work different jobs, eat at different times, and drink less late at night.
• 13:12
  SPEAKER_00: Is this a pattern or a proof?
• 13:15
  SPEAKER_00: A pattern.
• 13:16
  SPEAKER_00: A good question to keep asking, not yet a reason to panic.
• 13:21
  SPEAKER_00: And these questions work both ways.
• 13:23
  SPEAKER_00: They keep us from panicking over every new linked to, and they keep us from ignoring the real ones.
• 13:30
  SPEAKER_00: Smoking was real.
• 13:32
  SPEAKER_00: DES was real.
• 13:34
  SPEAKER_00: The difference was in how carefully someone looked and whether they checked.
• 13:40
  SPEAKER_00: This week we keep questioning, one layer deeper.
• 13:44
  SPEAKER_00: Here's today's practice.
• 13:46
  SPEAKER_00: Watch for one linked to this week.
• 13:49
  SPEAKER_00: A headline, a post, something you hear on the radio.
• 13:53
  SPEAKER_00: When you find it, ask the three questions.
• 13:56
  SPEAKER_00: How did they find out what people did?
• 13:59
  SPEAKER_00: Who were they compared with?
• 14:02
  SPEAKER_00: Is this a pattern or a proof?
• 14:05
  SPEAKER_00: You may not find every answer, and that's alright.
• 14:08
  SPEAKER_00: Just noticing that it's a looking back claim changes how you hold it.
• 14:14
  SPEAKER_00: You can take it seriously and still wait to see.
• 14:18
  SPEAKER_00: Today, we climbed the first rung in depth, looking back.
• 14:23
  SPEAKER_00: We sat in on Dalin Hill's study in 20 London hospitals, where 709 patients with lung cancer were compared with 709 patients without it, and smoking stood out.
• 14:36
  SPEAKER_00: We saw them test their own bias.
• 14:39
  SPEAKER_00: We saw eight young women in Boston and a careful look back that changed how a drug was used.
• 14:46
  SPEAKER_00: And we saw how a poorly chosen comparison group made coffee look guilty of something it didn't do.
• 14:54
  SPEAKER_00: And we saw why it matters.
• 14:56
  SPEAKER_00: So many of the warnings we hear come from looking back.
• 15:00
  SPEAKER_00: When we know how to question them, we can take the real ones to heart and let the false ones go.
• 15:07
  SPEAKER_00: Tomorrow, we turn around and follow forward.
• 15:10
  SPEAKER_00: In October of 1951, Dalin Hill wrote to every doctor registered in Britain and asked about their smoking.
• 15:19
  SPEAKER_00: More than 41,000 wrote back and they followed those doctors for the next 50 years.
• 15:25
  SPEAKER_00: Join me tomorrow to hear the rest of the story and learn more about our next rung on the ladder prospective or looking forward studies.
• 15:36
  SPEAKER_00: I'll see you there.
• 15:40
  SPEAKER_01: Keep the conversation going.
• 15:42
  SPEAKER_01: Find your next steps in notes from a fellow traveler, my blog.