# Lesson 6: Human Development and the Life Course

*Companion-podcast transcript • Sarah & Kiffer*  
*Office Hours episode to listen to after working through the lesson*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. This episode goes with Lesson six, Human Development and the Life Course.

**Sarah:** It's meant for after you've worked through the lesson. We'll add some perspective on how life-course ideas get used, some critique of the studies and scores behind them, a run at the questions students tend to find thorny here, and a few extra worked examples, including one that's harder than anything on the lesson page.

**Kiffer:** There are a few places where we'll ask you to work something out before we give the answer. When that happens, you'll hear a few seconds of quiet. Pause the audio if you'd like more time to think it through.

**Sarah:** We've got four questions. First, how do you tell the three life-course models apart when a real finding could fit more than one? Second, why is the Dutch Hunger Winter such strong evidence, and what can still mislead us in a natural experiment? Third, what does an ACE score tell you about one person, and should clinics be screening for it? And fourth, what changes when we judge aging by function, and what should we make of the Blue Zones?

**Kiffer:** Question one, then. Sarah, the three models are critical or sensitive period, accumulation, and pathway. Why do students find it so hard to sort real findings into them?

**Sarah:** Because the definitions sound clear until you meet an actual example. Almost any early exposure seems to add up over time, and almost any chain of events seems to start in some important window. So everything starts to look like all three.

**Kiffer:** That's fair, and the lesson itself says most research combines them. What helps me is to turn each model into a question you can ask of the evidence. For a sensitive period, does the timing of the exposure matter beyond the amount? For accumulation, does the total dose predict the outcome best, whatever the timing? And for a pathway, does the early exposure work mainly by changing what happens later, so that its link to the outcome shrinks once you account for the later steps?

**Sarah:** Can we start with a real case of the first one?

**Kiffer:** Folic acid is a good Canadian one. The neural tube closes very early in pregnancy, often before a person knows they're pregnant, so folate has to be adequate at that moment. Canada made folic acid fortification of flour mandatory in 1998, and a study of seven provinces published in 2007 found that neural tube defects fell from 1.58 to 0.86 per thousand births, a drop of forty-six percent. That's a critical period at the population level. The policy works because it reaches people before the window opens.

**Sarah:** Okay, let me set up some made-up scenarios, all about hearing, so the outcome stays the same and only the model changes. Scenario A, in a hypothetical study, adult hearing loss is best predicted by the total number of years someone has worked in loud jobs, and it doesn't matter whether those years came at age twenty or age fifty.

**Kiffer:** That one's accumulation. Total dose wins, and timing doesn't add anything.

**Sarah:** Scenario B. Children with a hearing problem between ages one and three show lasting delays in language, while children with the same hearing problem at age ten show little lasting effect on language. Scenario C. Young people who leave school early are more likely to end up in noisy, low-wage jobs, and those jobs are where their hearing damage happens. Listeners, which model fits B, and which fits C? Take a few seconds.

*(Pause)*

**Sarah:** B is a sensitive period. The same problem has a bigger effect when it falls in the window when language is being built. And C is accumulation, because the damage keeps adding up over years in those jobs.

**Kiffer:** B is right. For C, try the third question I mentioned. If you took two people with the same number of years in loud work, one who left school early and one who didn't, would leaving school early still predict their hearing loss?

**Sarah:** Probably not. In this scenario the school leaving matters because of where it sends people. Once you hold the noisy work constant, there isn't much left for it to explain. So the school leaving sets the path, and that makes C a pathway. The noise inside the jobs is accumulating, but the early exposure we were asked about is the start of a chain.

**Kiffer:** Exactly. And that's why the models aren't mutually exclusive. C has a pathway that leads into an accumulation. The classification depends on which exposure you're asking about, so name the exposure first and then ask the three questions.

**Sarah:** Question two. The Dutch Hunger Winter. Kiffer, why does it carry so much weight compared with Barker's original maps?

**Kiffer:** Barker's first evidence compared areas. Places with high infant mortality in the early 1920s later had high heart disease mortality among people born there. That's a strong clue, but areas differ in many ways at once, including poverty, housing and later diet. The Dutch famine gives you something closer to an experiment. A well-fed population was hit by severe hunger with a clear start and a clear end, and the records were good enough to know who was in the womb at the time, and in which trimester. Cohorts conceived just before or just after serve as comparisons.

**Sarah:** And because nobody chose to be exposed, people's own habits can't explain who was exposed.

**Kiffer:** That's the strength. But a natural experiment can still be distorted by who ends up in each group, and the famine gives a good example. Work by Stein and Susser in the 1970s found that the famine markedly reduced the number of conceptions that led to births, and that the fall was steeper among manual workers than among non-manual workers. That changed the social class mix of the babies born after famine conceptions.

**Sarah:** So let's put made-up numbers on that. Suppose in a normal month, a city has a thousand births, six hundred to manual-worker families and four hundred to non-manual families. During the famine, births to manual families fall to two hundred and fifty, and births to non-manual families fall to two hundred and fifty. Here's the question. If adults from manual-worker families have higher heart disease risk anyway, which way does that shift push a simple comparison of the famine cohort with the comparison cohort? Take a few seconds.

*(Pause)*

**Kiffer:** The share of manual-worker births drops from six hundred out of a thousand, which is sixty percent, to two hundred and fifty out of five hundred, which is fifty percent. So the famine cohort has relatively more children from better-off families.

**Sarah:** And those children would have had lower heart disease risk anyway. So the famine cohort looks healthier than it should, and a raw comparison would understate the effect of the famine.

**Kiffer:** Right. The bias pushes toward finding nothing, which is part of why the findings that did appear are taken seriously. Researchers handle it by comparing within social class or by using siblings born at other times as comparisons. And there's a second selection problem, mortality selection. The frailest exposed babies may have died before they could be measured as adults. If they would have had the highest risk, the survivors understate the harm again.

**Sarah:** So both selections hide effects here. But I'd guess that isn't always the direction.

**Kiffer:** No, and that's the habit to build. For any natural experiment, ask who was born, who survived and who was traced, and then ask which way each of those pushes the result. Sometimes selection exaggerates an effect, and you only know by thinking it through.

**Sarah:** Question three. ACE scores. The 1998 study reported that adults with four or more adverse childhood experiences had about four and a half times the odds of having had two weeks or more of depressed mood in the past year. People hear that and think a high score means you're headed for depression.

**Kiffer:** Let's test that with a made-up population, which will be our harder example, because it combines a strong association with how common the exposure is. Imagine ten thousand adults. Six hundred of them, six percent, have four or more ACEs. That's close to the original study, where 6.2 percent of respondents reported four or more. The rest, nine thousand four hundred, have fewer than four. Now suppose, hypothetically, that eight percent of the low-ACE group and thirty-two percent of the high-ACE group develop depression.

**Sarah:** So the risk ratio is thirty-two divided by eight, which is four. That's the strong association.

**Kiffer:** Right. Now the cases. Eight percent of nine thousand four hundred is seven hundred and fifty-two. Thirty-two percent of six hundred is one hundred and ninety-two. So there are nine hundred and forty-four people with depression altogether. Here's the question for everyone. Among the people with depression, what share had four or more ACEs? Take a few seconds.

*(Pause)*

**Sarah:** Okay. One hundred and ninety-two with depression in the high-ACE group, out of six hundred. That's thirty-two percent.

**Kiffer:** Check the denominator. What group did you just divide by?

**Sarah:** Oh. Six hundred is everyone with a high score. So thirty-two percent is the risk within the high-ACE group, which we already knew. The question was about people with depression, so the denominator has to be all nine hundred and forty-four cases. One hundred and ninety-two divided by nine hundred and forty-four is about 0.20. So about twenty percent of the people with depression had four or more ACEs, and about eighty percent had fewer than four.

**Kiffer:** That's the point the lesson makes in words. Most people with high scores don't develop the outcome, since sixty-eight percent of the high-ACE group here never do. And most people with the outcome didn't have high scores, because the low-ACE group is so much larger. Now the last step. If the association were fully causal, and we could prevent every high-ACE childhood, how many cases would we avoid?

**Sarah:** The high-ACE group would have the low-ACE risk, eight percent, in place of thirty-two. That's twenty-four percentage points of extra risk, times six hundred people, which is one hundred and forty-four cases. One hundred and forty-four out of nine hundred and forty-four is about fifteen percent.

**Kiffer:** So a fourfold risk, and yet about fifteen percent of cases, and only if the link is causal. One more trap. The published figures are odds ratios. With these same numbers, the odds ratio comes out to about 5.4, while the risk ratio is four. When an outcome is common, odds ratios look bigger than risk ratios, so be careful turning four times the odds into four times as likely.

**Sarah:** Which brings us to the disagreement. I'm more open than Kiffer is to asking about ACEs in primary care. California's ACEs Aware initiative, launched in 2020, pays clinicians in Medi-Cal, the state's public insurance program for people with low incomes, to screen for ACEs once they've completed a certified training. My view is that asking opens a door. Many people have never been asked, and a clinician who knows a patient's history can make care feel safer.

**Kiffer:** I agree that asking can help, and I don't object to clinicians knowing a history. Where I hesitate is the score. We just saw that most high scorers won't develop the outcomes we screen for, and the score treats a divorce and repeated abuse as one point each. In my view, a number in a chart can turn into a label that follows a patient around, especially when there's no service to refer to.

**Sarah:** But the same can be said of blood pressure. Most people with high readings won't have a stroke next year, and we still measure it, because the measurement prompts something useful.

**Kiffer:** That's a strong comparison. The difference I'd point to is that high blood pressure has a treatment with trial evidence that it lowers stroke risk. For an ACE score, the response is less settled, and the lesson notes that downstream help for adults shows smaller and less consistent effects than prevention does.

**Sarah:** Then the response matters as much as the question. Screening with nothing behind it would bother me too.

**Kiffer:** So here's where I land. I'd support asking about childhood adversity when the clinic is trained in trauma-informed care, treats the answer as a flag for attention, and has something to offer, such as counselling, social supports or a referral. And I'd keep most of the public investment upstream, in home visiting, child benefits and parental leave.

**Sarah:** I'd screen in more settings than Kiffer would, but we agree on the practical rule. Ask only when there's a trained response in place, and read the score as a population risk marker that says little about what will happen to the person in front of you.

**Kiffer:** Question four. Aging, and how much the definition of aging well changes who counts as doing well.

**Sarah:** Let's classify three made-up older adults. Rowe and Kahn's successful aging has three parts, low risk of disease and disability, high mental and physical function, and active engagement with life. The WHO defines healthy aging as developing and maintaining the functional ability that enables wellbeing in older age. Person one is seventy-eight. She has type 2 diabetes and arthritis, walks slowly to a community garden every day, volunteers there, and lives on her own.

**Kiffer:** Under Rowe and Kahn, she misses the first part, because she has two chronic diseases. Under the WHO framework, she's aging well, because she can do the things she values. That gap is the critique the lesson describes, that successful aging labels a lot of ordinary aging as failure.

**Sarah:** Person two is seventy. He has no diagnosed disease and tests well on fitness and memory, but he rarely leaves home and has lost touch with most of his friends.

**Kiffer:** Both frameworks would flag him. He misses active engagement for Rowe and Kahn, and the WHO lists building and keeping relationships among the abilities that make up functional ability. That's a reminder that the absence of disease isn't the whole picture under either one.

**Sarah:** Person three is eighty-two and uses a wheelchair after a stroke. Her building has an elevator and there's accessible transit to the choir she sings in. Here's the question. She moves to an apartment with no elevator, three floors up, and her body doesn't change at all. Under the WHO framework, does her healthy aging change? Take a few seconds.

*(Pause)*

**Kiffer:** It does. In the WHO framework, functional ability comes from a person's own capacities together with their environment. Her capacities are the same, but in the new building she can't get to choir, so her functional ability drops. Under Rowe and Kahn, she was already outside successful aging after the stroke, and the framework has no place for the building itself.

**Sarah:** And that's why the policy direction in the lesson points toward age-friendly environments. Changing the building changes her aging, in the WHO sense.

**Kiffer:** It matters for Canada. Statistics Canada reports that eighteen and a half percent of Canadians were sixty-five or older in 2021, and projects 23.1 percent by 2043 under its medium-growth scenario. Whether that is a crisis depends heavily on environments like the one person three lives in, as well as on long-term care and home care.

**Sarah:** And the CLSA is how we'll learn some of this. It follows 51,338 Canadians who were forty-five to eighty-five at recruitment, recontacted every three years for at least twenty years.

**Kiffer:** Which brings us to the Blue Zones, where the same caution about evidence applies. Saul Newman's 2019 preprint reported that when places began formally registering births, records of people over a hundred and ten fell by sixty-nine to eighty-two percent.

**Sarah:** Let me try a made-up version. A hypothetical region records forty people aged one hundred or older per ten thousand residents, compared with four per ten thousand nationally. Suppose a careful check of birth records found that three quarters of those ages were wrong. Then the region has forty times one quarter, which is ten per ten thousand.

**Kiffer:** Still higher than four, so something may be going on. What's the next question?

**Sarah:** Whether the leftover difference comes from the lifestyle in the brochure. For that, I'd want regions with the same diet and habits that don't have extra centenarians, and regions with different habits that do. If lifestyle doesn't line up with longevity across many places, the story is cherry-picked.

**Kiffer:** That's the right order. First check the count, then check the explanation. Let's pull it together with three things to take away.

**Sarah:** First, to classify a life-course finding, name the exposure and ask three questions. Does timing matter beyond dose, does total dose win, or does the early exposure work through what comes later?

**Kiffer:** Second, natural experiments are strong, but who gets born, who survives and who gets traced can push results either way. Work out the direction of each selection before you trust a comparison.

**Sarah:** And third, a large relative risk can sit alongside a small share of cases, so an ACE score or any risk marker describes populations, and functional ability in older age depends on environments as well as bodies.

**Kiffer:** If you'd like more practice, rework the ACE example with a different share of high scorers, say twenty percent, and see how the share of cases with high scores changes.

**Sarah:** Next time, it's Lesson seven, Genetics, Genomics, and Health, where we look at what genes can and can't tell us about risk.

**Kiffer:** Take care, everyone.

**Sarah:** See you in Lesson seven.
