# Lesson 5: Sexual and Reproductive Health

*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 five, Sexual and Reproductive Health.

**Sarah:** It's best heard once you've finished the lesson. We'll bring some perspective on why these histories still shape public health practice, offer some critique, work through the questions students most often find tricky in this material, and try a few extra worked examples, including at least one that's harder than anything on the lesson page.

**Kiffer:** Along the way there are a few places where we'll ask you to work something out before we give the answer. You'll hear a few seconds of quiet when that happens, and you're welcome to pause the audio if you'd like more time to think it through.

**Sarah:** Here's the plan. Question one asks why maternal mortality figures are so hard to read when deaths are rare. Question two asks what it means to say Black women in the United States die at three times the rate of white women, and whether that gap is growing. Question three takes on contraceptive failure rates, and why a figure for one year can mislead. Question four asks whether a study like Tuskegee could happen today. And question five is where Kiffer and I disagree, on whether the push for faster drug approval that came out of HIV activism has had costs.

**Kiffer:** Question one, then. The maternal mortality ratio counts maternal deaths per one hundred thousand live births. The lesson points out that a country with three hundred and fifty thousand births a year and a ratio of ten would see only about thirty-five maternal deaths a year. A single province or health region sees far fewer, and that makes the numbers jumpy.

**Sarah:** I'd like to try one of these myself, because the units always catch me.

**Kiffer:** Here's a made-up health region with twenty-five thousand live births every year. Last year it recorded three maternal deaths. This year it recorded five. What's this year's maternal mortality ratio, and how much has it changed since last year? Take a few seconds.

*(Pause)*

**Sarah:** Okay. Five deaths divided by twenty-five thousand births is 0.0002. So the ratio is two per one hundred thousand live births.

**Kiffer:** Before you go on, compare that with the Canadian figures in the lesson. Does two per one hundred thousand sound like a plausible number for a Canadian region?

**Sarah:** No. The lesson puts Canada at roughly eight to ten. Two would be far below the national figure, even though this region's deaths went up. So I've slipped somewhere. Right, 0.0002 is per birth. To get per one hundred thousand births I multiply by one hundred thousand, and that gives twenty. I think I moved the decimal the wrong number of places.

**Kiffer:** That's it. Now work out last year's ratio the same way, so we can compare the two years.

**Sarah:** Three divided by twenty-five thousand, times one hundred thousand, is twelve. So the ratio went from twelve to twenty. Twenty divided by twelve is about 1.67, which is a rise of about two thirds. I can already picture the headline.

**Kiffer:** And the headline would describe two extra deaths. Each of those deaths is a family's loss and deserves a full review. As evidence of a trend, though, a change from three to five can easily happen by chance alone. At ten per one hundred thousand, this region would expect about two and a half deaths a year, so three one year and five the next is well within the ordinary bounce of small numbers.

**Sarah:** So what do surveillance teams do about it?

**Kiffer:** Three things. They pool several years, so five years in this region would give one hundred and twenty-five thousand births and a steadier ratio. They review each death in detail, which is what maternal mortality review committees do. And they track severe maternal morbidity, the near misses, which the lesson notes are roughly fifty to one hundred times more common than deaths. Near misses give enough events to see patterns while there's still time to act.

**Sarah:** One more thing about the denominator. It's live births, which means a woman who dies early in a pregnancy that never reaches a live birth is in the numerator, and her pregnancy isn't in the denominator. The ratio is convenient, and it's worth remembering that its denominator counts births.

**Kiffer:** Right. Some reports, including the American ones we're about to use, call it a rate anyway, so check the denominator whatever the label says.

**Sarah:** Question two. Black women in the United States are often said to have about three times the maternal mortality of white women, and that's the figure in the lesson. I'd like to check what that "three times" actually rests on.

**Kiffer:** Let's use the official figures from the National Center for Health Statistics. In 2021, the maternal mortality rate was 69.9 per one hundred thousand live births for non-Hispanic Black women and 26.6 for non-Hispanic white women. In 2023, the figures were 50.3 and 14.5. Here's the question. Between 2021 and 2023, did the gap between Black and white women get bigger or smaller? Take a few seconds.

*(Pause)*

**Sarah:** My instinct is that it got smaller, because both numbers fell. Let me check. In 2021, 69.9 divided by 26.6 is about 2.6. In 2023, 50.3 divided by 14.5 is about 3.5. So as a ratio, the gap grew.

**Kiffer:** Now try it a second way, by subtracting the white rate from the Black rate in each year.

**Sarah:** In 2021, 69.9 minus 26.6 is 43.3 per one hundred thousand. In 2023, 50.3 minus 14.5 is 35.8. So as a difference, the gap shrank. Both answers come from the same four numbers.

**Kiffer:** And they're both correct. The rate for white women fell by nearly half over those two years, while the rate for Black women fell less, so the ratio rose even as the absolute difference narrowed. The report itself found that the change for Black women between 2022 and 2023 was small enough to be chance. Anyone who wants to say the disparity is getting better or worse can pick the measure that suits them.

**Sarah:** So which one should we report?

**Kiffer:** I'd report both. The ratio tells you how unequal the risks are, and the difference tells you how many extra deaths that inequality produces. It's also worth noticing that the lesson's "about three times" sits between the two years. The 2021 figures give about 2.6, and 2023 gives about 3.5.

**Sarah:** There's another point in the lesson that I think students rush past, which is the education finding.

**Kiffer:** Yes. A Centers for Disease Control and Prevention analysis of deaths from 2007 to 2016 found that among women with a college degree or more, Black women had 5.2 times the pregnancy-related mortality of white women. Black college graduates also had 1.6 times the mortality of white women who hadn't finished high school. Across all education levels in that analysis, the ratio was 3.2. If education explained the gap, the ratio should shrink among the most educated women, and in these data it grew larger. That's a large part of why researchers look to racism within health care and the wider society as an explanation.

**Sarah:** Question three moves to contraception. The lesson gives one-year failure rates of about seven to nine percent for the Pill under typical use, and about 0.1 to 0.5 percent for long-acting reversible methods like IUDs and implants. Those are already very different. Kiffer wants to push it further.

**Kiffer:** This is the harder example. Picture one thousand people who start the Pill and keep using it for five years. Say the typical-use failure rate is seven percent a year, the low end of the range in the lesson. The numbers are hypothetical, and we'll assume the rate stays the same each year. About how many of the thousand would have at least one unintended pregnancy over five years? Take a few seconds.

*(Pause)*

**Sarah:** The quick answer is seven percent times five years, which is thirty-five percent, so three hundred and fifty people.

**Kiffer:** That's the trap. Once someone has had a pregnancy, they leave the group still at risk for a first failure. Can you take it year by year?

**Sarah:** Sure. Year one, seven percent of one thousand is seventy, so nine hundred and thirty remain. Year two, seven percent of nine hundred and thirty is about sixty-five, which leaves about eight hundred and sixty-five. Year three, about sixty-one, leaving about eight hundred and four. Year four, about fifty-six, leaving about seven hundred and forty-eight. Year five, about fifty-two, leaving about six hundred and ninety-six.

**Kiffer:** Good. So how many people in total have had at least one pregnancy by the end of year five?

**Sarah:** Seventy plus sixty-five plus sixty-one plus fifty-six plus fifty-two is three hundred and four. So about thirty percent over five years. The shortcut is 0.93 multiplied by itself five times, which is about 0.70, and one minus that is about 0.30.

**Kiffer:** Now do the same calculation for a hormonal IUD with a failure rate of 0.2 percent a year.

**Sarah:** 0.998 multiplied by itself five times is about 0.990, so about one percent, or ten people in a thousand. Thirty percent compared with one percent. That's a much starker contrast than seven compared with 0.2.

**Kiffer:** It is. The one-year figure tells you about the first year. A person deciding on a method is often thinking about the next five or ten years, and the gap compounds over that time.

**Sarah:** Can I push on the assumption, though? We held the rate at seven percent every year.

**Kiffer:** Please do, because that assumption is the second idea built into this example, and it matters.

**Sarah:** The lesson's Methods Spotlight says the Pearl Index assumes a constant failure rate, which often doesn't hold. In real use, the people who find a daily pill hardest to take tend to have their failures early, or switch methods. The people who keep going are more practised, and on average somewhat less fertile. So the yearly failure rate usually falls with time, and thirty percent probably overstates the five-year risk for people who stay on the Pill.

**Kiffer:** That's right, and it's the reason longer studies tend to produce lower Pearl Index values. Life-table methods follow the cumulative risk directly, month by month, so they don't need that assumption. The practical lesson is that the compounding is real, and a constant yearly rate gives you an upper guide. For an IUD or implant, typical use and perfect use are nearly the same, so the estimate barely moves either way.

**Sarah:** Which also explains the CHOICE Project result. If cost and counselling are what keep people from the most effective methods, removing those barriers changes outcomes for years.

**Kiffer:** Question four. The lesson's reflection asks whether a study like Tuskegee could happen today. I'd like to answer it by applying the three Belmont principles to a made-up proposal.

**Sarah:** Before you do, it's worth recalling that the Belmont Report, from April 1979, names Tuskegee directly. In its section on justice, it describes the study as having used disadvantaged, rural Black men to study the untreated course of a disease.

**Kiffer:** Here's the hypothetical proposal. A research team wants to test a text-message reminder system for people finishing syphilis treatment, at a drop-in clinic for people experiencing homelessness. First, they'll recruit only at this clinic because it's convenient, though the system is meant to be sold to private clinics serving wealthier patients. Second, the nurse who provides each person's care will do the recruiting, using a seven-page consent form. Third, to make the comparison cleaner, the control group won't receive the clinic's usual follow-up phone call. Which Belmont principle does each problem fall under? Take a few seconds.

*(Pause)*

**Sarah:** The first one is justice. The burden of research falls on a vulnerable group chosen for convenience, and the benefits are aimed at people who are better off. That's the same pattern the Belmont Report describes in Tuskegee.

**Kiffer:** Agreed. What about the second problem, the nurse doing the recruiting with a seven-page form?

**Sarah:** I'd put the long consent form under respect for persons, because people have to be able to understand what they're agreeing to. I was going to put the nurse under beneficence, though now I think it's respect for persons too. When the person who controls your care asks you to join, saying no may not feel safe. That's a problem with voluntariness.

**Kiffer:** Yes. Comprehension and voluntariness are both parts of informed consent. What about the third problem, the missing phone call for the control group?

**Sarah:** Beneficence. Dropping the usual phone call adds risk to the control group in order to make the result look cleaner. It's a small echo of withholding care for the sake of the data.

**Kiffer:** That's how I'd classify them. A review board would very likely catch all three. The harder point in the lesson is that the Tuskegee study was published in medical journals for decades, and hundreds of physicians knew about it. Rules only work when the people around a study are willing to speak up, so the review system is one safeguard among several.

**Sarah:** And the lesson's point about trust follows from that. Alsan and Wanamaker found measurable drops in Black men's use of health care after the study was disclosed.

**Kiffer:** Canada has its own version of this history. In 2013, the historian Ian Mosby described federal nutrition studies that began in the late 1940s at six residential schools. At one school in Port Alberni, children received less than half the recommended daily milk for two years while researchers collected baseline data, and dental services were withdrawn at the schools so they wouldn't interfere with the results. The children had no say in any of it.

**Sarah:** And that's the justice problem again, a captive and marginalized group carrying the burden of research. It's part of the reason TCPS 2, the policy that governs federally funded research in Canada, has a whole chapter on research involving First Nations, Inuit and Métis peoples.

**Kiffer:** Question five, ACT UP and the speed of drug approval. On October 11, 1988, about fifteen hundred activists protested at the Food and Drug Administration's headquarters in Rockville, Maryland, demanding faster research and approval of AIDS drugs. In 1992, the agency launched its accelerated approval pathway. It lets a drug for a serious illness reach the market on the basis of a surrogate measure that's reasonably likely to predict real benefit, with a confirmatory trial to follow.

**Sarah:** And this is where I'll make the case for speed. In December 1995, the first protease inhibitor, saquinavir, got accelerated approval in ninety-seven days, based on CD4 cell counts. Protease inhibitors became part of the three-drug combinations that transformed HIV treatment the following year. People who were dying had every reason to accept more uncertainty in exchange for earlier access, and they were right.

**Kiffer:** I agree about HIV in the 1990s. My concern is what happened to the pathway later. In June 2021, the agency gave accelerated approval to aducanumab, sold as Aduhelm, for Alzheimer's disease, on the basis of reduced amyloid plaque in the brain. Its own advisory committee had voted overwhelmingly against approval. In January 2024, the company discontinued the drug and ended its confirmatory trial. The drug was on the market for more than two years, and its benefit to patients was never confirmed.

**Sarah:** One drug doesn't discredit a pathway that has approved more than three hundred. And the two situations are very far apart. People with HIV in 1995 had few options and were dying within a few years.

**Kiffer:** That's a fair point. My view is that speed carries an evidence cost, and someone has to keep paying attention to it. In HIV, the combination therapies built on those early approvals cut deaths sharply within a few years, as the lesson describes. For aducanumab, the link between less plaque and real benefit to patients was never confirmed. In December 2022, Congress passed a law that lets the agency require confirmatory trials to be under way before approval, and that formalizes how it withdraws approvals.

**Sarah:** I'm still more sympathetic to speed than you are, especially when people have no other options. Where we agree is that early approval should come with a confirmatory trial that's actually under way and actually finished.

**Kiffer:** And with a clear public statement of what's known and what isn't. To me, that seems a fair way to keep both the early access and the evidence.

**Sarah:** Let's pull it together. First, when deaths are rare, a single year's maternal mortality ratio can swing widely by chance. Check the counts behind the ratio, pool years, and look at near misses.

**Kiffer:** Second, a disparity can grow as a ratio and shrink as a difference at the same time. Report both, and be wary of anyone who reports only the one that suits their argument.

**Sarah:** And third, a one-year contraceptive failure rate understates what a person experiences over years of use, because risk compounds. The constant-rate assumption is an approximation, and life-table methods avoid it.

**Kiffer:** If you'd like more practice, rework the five-year Pill example with a typical-use failure rate of nine percent, the top of the lesson's range, and then think about why the true figure is probably lower.

**Sarah:** Next time, it's Lesson six, Human Development and the Life Course, where we follow health from before birth through older age.

**Kiffer:** Take care, everyone.

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