# Lesson 7: Communities, Populations, and Settings *Companion-podcast transcript • Sarah & Kiffer* *Office Hours episode to listen to after working through the lesson* --- **Sarah:** Here's a made-up case. A school vaping prevention program gets good results in a large school board near Toronto, and a rural district in northern British Columbia wants to use it. **Kiffer:** One of its core pieces is a set of role-plays where students practise refusing a vape and get feedback. Albert Bandura would call that a mastery experience. **Sarah:** And the students in the district say the role-plays feel fake and embarrassing. They'd rather hear older students tell their own stories. **Kiffer:** So do you keep the component that was tested, or the one students will actually do? Sarah and I don't land in the same place on that. **Sarah:** Welcome back to Office Hours. I'm Sarah. **Kiffer:** And I'm Kiffer. We're on Lesson seven, Communities, Populations, and Settings, and we'll start with two names that are close enough that people mix them up. **Sarah:** Question one. What separates a community intervention from an intervention in a community, and where do universal, selected and indicated prevention fit? I'll admit the names blurred together for me at first, because both of them happen in a community. **Kiffer:** They do. The difference is who the program is aimed at. A community intervention aims at the community in general, everyone regardless of their own risk, so it's a universal intervention. An intervention in a community happens inside a community and aims at a particular group or setting. If that group is at high risk for one problem, it's selected. If it's people with several risk factors at once, served with a bundle of activities and treatments, it's indicated. **Sarah:** So a law banning smoking on restaurant patios is a community intervention, and a peer program for people who inject drugs in one neighbourhood is an intervention in a community, even when both happen in the same city. **Kiffer:** Right. Here's one for everyone to try, with three made-up programs that all aim at alcohol-related harm in one city. The first puts transit ads in front of every adult about standard drink sizes. The second offers brief counselling to students living in campus residences where heavy drinking is common. The third serves young adults who've had repeated emergency visits for intoxication and who are also living with depression and without stable housing, and it combines housing support, mental health care and addiction referral. Which prevention category does each one fit? Take a few seconds. *(Pause)* **Sarah:** The transit ads are universal, so that's a community intervention. The residence program is selected, because it aims at a group at high risk for one problem. The third is indicated, since it serves people with several risk factors at once and bundles services to match. The second and third are both interventions in a community. **Kiffer:** That's how I'd sort them. The one people tend to get wrong is the residence program. Heavy drinking sounds serious enough to call it indicated, but the program is offered to a whole high-risk group for one problem, with one activity, which makes it selected. In the lesson's terms, indicated needs the combination of risk factors and the bundle of services. **Sarah:** And counting how many people change is only part of choosing between the two kinds of program. With HIV or hepatitis C, each person in a bridge group who stops sharing equipment also protects the people they might otherwise have exposed. **Kiffer:** That's part of the reasoning behind a program like Insite, which opened in Vancouver's Downtown Eastside in 2003 as the first sanctioned supervised injection site in North America. Measured on the scale of the whole metro region, it serves a small group. Measured by the people who inject drugs in that neighbourhood, and their role as a possible bridge for infection, it looks very different. So which kind of program is better depends on the assessment, on whether the group is a bridge, and on baseline risk, which is where the surprises are. **Sarah:** Question two. If people at high risk are the most likely to get sick, how can a small change across a whole community prevent more cases than a program aimed at them? **Kiffer:** This goes back to the epidemiologist Geoffrey Rose. In a 1985 paper called Sick Individuals and Sick Populations, which began as a lecture he gave in Vancouver in 1984, he argued that a large number of people at a small risk may give rise to more cases of disease than the small number who are at high risk. Let's test that with the hardest example in the episode, and every number in it is made up. **Sarah:** Good, because I know that quote and I still find it hard to believe until I see the numbers. Set it up for us. **Kiffer:** A health region has one hundred thousand adults over forty. Five thousand have very high blood pressure, and each has an eight percent chance of a stroke over the next ten years. Twenty thousand have moderately high blood pressure, at four percent. The other seventy-five thousand have blood pressure in the usual range, at 1.6 percent. Before any program, what share of the region's strokes would come from the high-risk group? Sarah, you go first after the pause. Take a few seconds. *(Pause)* **Sarah:** The high-risk group's risk is five times the usual-range group's, so I'd guess most of the strokes come from them. Maybe two thirds? **Kiffer:** Check that before you rely on it. Count the expected strokes in each group. **Sarah:** Okay. Five thousand times eight percent is four hundred. Twenty thousand times four percent is eight hundred. Seventy-five thousand times 1.6 percent is one thousand two hundred. That's two thousand four hundred strokes, and the high-risk group has four hundred of them, which is one in six. I had it the wrong way around. Five out of every six strokes happen in people at moderate or usual risk, because there are so many more of them. **Kiffer:** That's Rose's point, and it's the first trap. Now compare two programs. The first is a clinic program for the five thousand people with very high blood pressure. It reaches three thousand of them and cuts each one's risk by a quarter. The second is universal. Say the region works with food producers and restaurants to lower the salt in the food supply, which lowers everyone's blood pressure a little and cuts each person's stroke risk by five percent. Which program prevents more strokes? **Sarah:** For the clinic program, three thousand times eight percent is two hundred forty expected strokes, and a quarter of that is sixty prevented. For the salt program, five percent of all two thousand four hundred strokes is one hundred twenty. So the universal program prevents twice as many. **Kiffer:** Now look at it from one person's point of view. Someone in the clinic program goes from an eight percent risk to six percent. That's two percentage points, or about one stroke prevented for every fifty people treated. Someone in the usual range, under the salt program, goes from 1.6 percent to 1.52 percent. **Sarah:** That's 0.08 percentage points, so about one stroke prevented for every one thousand two hundred fifty people like them. No one person would ever notice a change that small. **Kiffer:** And that's the second trap. Each person in a universal program gets a benefit too small to feel, so the program looks ineffective from every individual's point of view, even while it prevents the most strokes overall. Rose called this the prevention paradox. It's one reason universal measures so often work through the food supply or policy, the third path from the lesson, where they depend less on each person staying motivated. **Sarah:** And in the lesson's categories, the salt program is universal, a community intervention. The clinic program is an intervention in a community, aimed at people at high risk for one problem. If it bundled blood pressure care with help quitting smoking and diabetes care for people who had all three, it would be indicated. **Kiffer:** Right. And Rose wrote that the high-risk and population approaches are not usually in competition, so a region can run both. The clinic program helps a few people a lot, the salt program helps many people a little, and the assessment decides how to divide the effort between them. **Sarah:** Question three. When a big community trial shows only a small effect, what has it actually told us? After Rose, I'm tempted to say a small effect is what a population strategy should look like. I'm thinking of COMMIT, which ran for four years in eleven matched pairs of communities, including Brantford, Ontario, and raised quitting by about three percentage points among light and moderate smokers. **Kiffer:** The published numbers are worth hearing. Among heavy smokers, eighteen percent quit in the intervention communities compared with 18.7 percent in the comparison communities, and that difference wasn't significant. Among light and moderate smokers, it was 30.6 percent compared with 27.5 percent. That's the three points, and it was significant. **Sarah:** And the Minnesota Heart Health Program had a similar story. **Kiffer:** In some ways it had a harder one. Its 1994 report on risk factors described strong background trends across the study communities, comparison communities included, with health promotion activity rising and risk factors falling. Researchers call that a secular trend. Next to it, the program's overall effects were modest in size and duration and generally within chance levels. **Sarah:** So one reading of a small effect is that the comparison communities were changing anyway, maybe because they were hearing a lot of the same messages. **Kiffer:** That's one reading, and there are at least three others. The program may not have added enough exposure on top of everything else people were hearing, which is what the Minnesota team suggested. The program may have missed the people it most wanted, which is the heavy smoker problem in COMMIT. And the design may have had little power to detect an effect. When you randomize communities, the number of communities limits what you can detect, however many smokers you follow. Eleven pairs of communities took an enormous amount of work, and it's still a small sample. **Sarah:** And three percentage points among the light and moderate smokers of eleven whole communities can still mean thousands of people quitting, which is the small change for many that a community intervention is supposed to buy. **Kiffer:** Which is why I'd read COMMIT as a modest success among light and moderate smokers and a clear miss for its primary target. Its own authors concluded that heavy smokers may need new clinical programs and policy changes, which sounds a lot like running a high-risk strategy alongside the population one. **Sarah:** Question four. Why does the setting narrow the choice of theory? I think students sometimes treat it as a rule to memorize, a classroom gets Social Cognitive Theory and a workplace gets social marketing. **Kiffer:** It follows from the assessment. A setting fixes who is present, why they're there, what the organization is for, and who holds power over a program. Those facts show up in a Precede-Proceed style assessment, and they rule some theories in and others out. The common mistake is choosing a theory because it's familiar and then finding that the setting won't accept it. **Sarah:** Let's do a worksite, then, and it's made up again. **Kiffer:** Picture a regional bus company in the Fraser Valley with about two hundred unionized drivers. Management is worried about blood pressure. Here's what the site visit finds. At last year's screening, many drivers had high readings. Drivers work split shifts, with a two-hour break at a depot that has only a vending machine. New drivers say the senior drivers joke that anyone who goes to the health check is soft. And management and the union have been in contract talks for months without agreeing on much. **Sarah:** The readings belong to the epidemiological assessment. The depot with nothing but a vending machine is the behavioural and environmental assessment, since that environment will shape what drivers eat on every split shift. **Kiffer:** Good. Now let's look at the joke. Which assessment phase does it belong to, and which theories does it point to? Take a few seconds. *(Pause)* **Sarah:** It's the educational and ecological assessment. It's a norm about what a good, tough driver does, and the senior drivers carry it, so they're the influencers. That points to Social Network Theory and diffusion of innovations, with respected senior drivers recruited as change agents who model getting checked. And culture theory or the Theory of Planned Behavior would speak to the norm itself. **Kiffer:** That's how I'd sort it. Now think about what a familiar choice would have produced. A lunchtime Health Belief Model talk on the dangers of high blood pressure assumes drivers don't know the risk or don't feel susceptible. The assessment found three other problems: a norm, a break room, and a stalled relationship between management and the union. **Sarah:** And the contract talks are the administrative assessment. Participation tracks the support of both management and unions, so a program that management launches on its own in the middle of bargaining could easily be read as a bargaining move. **Kiffer:** Which is why I'd run it through the joint health and safety committee. In British Columbia, a workplace with twenty or more workers employed for more than a month needs one, and it brings employer and worker representatives together. Then I'd put quick blood pressure checks in the depot during the split-shift break, because these are adults who attend by choice. **Sarah:** Can I push on the business case for a minute? The cost-saving figures for workplace programs, the ones measured in dollars returned for every dollar spent, make them sound like a safe investment. **Kiffer:** The stronger evidence is more mixed. In 2019, two randomized trials of workplace wellness programs in the United States were published. One covered about thirty-three thousand workers at one hundred sixty stores of a warehouse retail chain, with twenty stores randomly assigned the program. After eighteen months, workers in the program reported healthier behaviours, such as more exercise, and there were no significant differences in clinical measures, health care spending, absenteeism or job performance. **Sarah:** And the second one was at the University of Illinois, wasn't it? **Kiffer:** Yes. It randomized nearly five thousand employees. The program raised screening rates and had no significant effect on medical spending, other health behaviours or productivity. It also found something that helps explain older figures. The employees who chose to take part already had lower medical spending and healthier habits in the year before the program began. **Sarah:** So if you compare participants with nonparticipants, a program can look as if it saves money even when it does nothing, because healthier people signed up. Without a fair comparison group, you can't tell the program's effect from the difference in who chose to join. **Kiffer:** That's the selection problem. My own view is that workplace programs are worth making the case for on health and on an employer's duties to its workers, and I'd treat return-on-investment figures from studies without randomization with real caution. **Sarah:** Okay. That brings us to the topic where Kiffer and I don't fully agree. **Kiffer:** It's about adapting evidence-based programs. Let's set it up with a case, and it's made up. A school vaping prevention program was tested in a large school board near Toronto and showed good results. A rural school district in northern British Columbia wants to use it. **Sarah:** It has five components. There are six weekly sessions where students practise refusing a vape in role-plays and get feedback, a session where students compare what they think their classmates do with what surveys show, videos of teenagers in a subway station and a big shopping mall, a schedule that puts every session first thing on Monday morning, and a newsletter for parents in English. Which of those carry the program's theory, and which could you change freely? Take a few seconds. *(Pause)* **Kiffer:** The role-plays carry Social Cognitive Theory. Practising a skill and succeeding at it is what Albert Bandura called a mastery experience, and he described mastery as the most powerful source of self-efficacy. The session comparing beliefs with survey data carries the norms piece, correcting the belief that everyone vapes. Those two are core. **Sarah:** And the other three map onto the tailoring requirements. The videos are about relevant situations, and a small northern town has no subway. The newsletter is language and materials, so it should be in the languages that families in the district read. And Monday mornings are scheduling, which matters if a lot of students ride a bus in from far out of town. **Kiffer:** So far we agree. Here's where we part ways. **Sarah:** The formative research comes back, and students say the role-plays feel fake and embarrassing. They'd rather hear older students tell their own stories about refusing or quitting, a bit like the role model stories in the Long Beach project, which volunteers from the community handed out. I'd make that change. The community is the expert on itself, and a component that students won't take part in can't do its job. **Kiffer:** I understand the case, and I'd be more cautious. The evidence belongs to the program as it was tested. Hearing how an older student succeeded is still Social Cognitive Theory. Bandura called it vicarious experience. It's a weaker source of self-efficacy than succeeding yourself, though, so dropping the practice removes the part of the program that does the most to build confidence, and you can no longer say you're running the program that was tested. **Sarah:** But the program was tested in a big city in another province. The evidence was already a step removed the moment it moved to a rural district. Keeping role-plays that students resist doesn't make the rural version evidence-based either. **Kiffer:** That's fair, and it's the strongest form of your argument. Felipe Castro and his colleagues described this in 2004 as a tension between fidelity and fit, and I don't think either side wins outright. Where I land is that I'd keep a component's mechanism even when its form changes. Older students could tell their stories and then coach younger ones through a short practice of their own, so the mastery experience survives. **Sarah:** I'd still give the community more say than Kiffer would, but we share a practical rule. Before you adapt anything, name the core components and the theory each one carries. Change the surface freely with the community. If the community gives you a reason to change a core component, replace it with something that works through the same mechanism, and then evaluate the adapted version as a new program. **Kiffer:** Let's pull it together with three things to take away. **Sarah:** First, before you choose between a program for everyone and a program for the people at highest risk, work out where the cases actually come from. Very often most of them come from the large group at moderate risk, and a benefit too small for any one person to notice can still prevent the most cases. **Kiffer:** Second, read a small effect from a community trial with care. Ask whether the comparison communities were changing anyway, whether the program added enough exposure to matter, and whether it reached the group it was aimed at. **Sarah:** And third, let the assessment of the setting choose the theory, and when you adapt a program, keep the mechanisms that carry that theory. At a worksite, the assessment often points to norms, networks and the relationship between management and the union before it points to anyone's knowledge. **Kiffer:** If you'd like more practice, go back to the stroke example from question two and change one number at a time. Shrink the salt program's effect, or have the clinic program cut each person's risk by more than a quarter, and work out the point where the clinic program starts to prevent more strokes than the universal one. **Sarah:** Next time, it's Lesson eight, Communications Campaigns, where we look at when mass media and messaging work, and how a campaign is built and tested. **Kiffer:** Take care, everyone. **Sarah:** See you in Lesson eight.