The potential, and limits, of science

On shaping a science that supports a new, post-war vision for public health.

Improving the health of populations depends on science, on building a foundation of knowledge that helps guide our actions. I have previously written about how the work of public health rests on the work of population health science. My book, Population Health Science, co-written with Dr Kerry Keyes, and several other commentaries such as this 2017 article in Epidemiology, were efforts to help advance the fundamentals of the field—the science at the heart of all we do. 

At this post-war moment, when we are looking to build new foundations for public health, a focus on science to help us to ground these efforts has never been more important. Therefore, today’s Healthiest Goldish is the first of two essays on the role of science in the work of public health—on its strengths, its limits, and its intersection with the values of those who practice it. 

To say science is at the heart of all we do is to make a statement which is more complicated than it may at first appear. Because science is produced by a clearly defined method, it can be easy to think its conclusions can be reduced to an equation in which X leads to Y and therefore the work of public health is simply the work of acting on X. Such a view of our science might say, for example, that if smoking increases the risk of lung cancer at the population level (an equation in which smoking equals X, and lung cancer equals Y), we should devote our resources to getting rid of X, working to end smoking among populations. Through this process, the thinking may go, scientific conclusions inform the work of public health and shape better outcomes.

But this simple view belies complicated challenges to how science shapes the actions that support a healthier public. First, we are dealing with very complex systems in public health. This complexity means that, while we can say X equals Y, we cannot always know what will happen if we act on X, or even how best to act on X. We are, after all, dealing with people with all their idiosyncrasies, inconsistencies, and unpredictability. Further complicating matters is the broader context of politics, economics, technological advances, and environmental change that is inextricably linked to every aspect of the health of populations. Changing a variable, even in ways that seem self-evidently beneficial for the public’s health, can lead to unintended consequences with unforeseeable effects. 

While I have written on complexity in the context of population health systems in prior work, this is perhaps well illustrated by an example from the field of ecology. Wolves play a key role in shaping ecosystems. They do this by controlling the population of their prey, such as elk, which, in turn, eat plants. The more wolves in an ecosystem, the less prey, the more plants remain. This process—in which predators shape the populations of their prey, which then affects the population at the next level of the food chain—is known as a trophic cascade. 

The trophic cascade model has been attractive for many ecologists and to the public. It is not hard to find stories about wolves being removed or reintroduced to a given ecosystem and how this apparently influenced the environment in one way or another. Yet the reality is more complicated than stories suggesting a clear X equals Y relationship between wolves and their environment. Ecosystems are complex; their changes cannot be reduced to a single variable. While the presence of wolves certainly influences ecosystems, this influence is not the only factor that determines whether a given plant population thrives. Wolves are less a link in a chain and more a node in a web in which organisms are interconnected in dynamic, constantly evolving ways. Yet the simple view of the trophic cascade continues to appeal in many stories we hear about wolves and their environment. As Emma Marris wrote in Nature:

“This story is popular in part because it supports calls to conserve large carnivores as ‘keystone species’ for whole ecosystems. It also offers the promise of a robust rule within ecology, a field in which researchers have yearned for more predictive power.”

We, too, yearn for more predictive power in population health science. We, too, struggle to capture the full complexity of systems in our models. As scientists, we daily face the reality that data are rarely straightforward. Data are often conflicting and confusing, and it can take years to reach a consensus. Along the way, this process is shaped by a range of biases and priors, to say nothing of the intentional muddying of the waters by bad faith actors. To return to the example of smoking, the tobacco industry spent years working to sow doubt about the harms of smoking and market to the public the illusion that the medical profession even endorsed the practice. This context ensured that the process of getting to X (the understanding that smoking increases the risk of lung cancer) was long an uphill climb.  

But science does not need to be influenced by bad faith actors to be confusing, contradictory, and inconclusive. It can be that all on its own. Consider the example of salt, which I have written about. Is salt at the population-level harmful to health? It may be surprising to learn that there are two schools of thought about that question. Each camp lives in its own self-reinforcing bubble, citing papers that reflect the preferred view, with little intellectual cross-pollination between them. This has meant that the variable X (being the health effects of salt at the population level) continues to equal…it depends on whom you ask.

Faced with uncertainty, the most constructive course would seem to be for scientists to acknowledge their limits, be transparent about what they do not know, and work, with humility, towards deeper understanding. Yet, too often, uncertainty can lead to scientists taking unreasonably firm positions, based on their favored interpretation of ambiguous data. This is understandable, and deeply human. Most of us have had times when we have responded to uncertainty by clutching ever tighter our assumptions and biases, trying to maintain a stable worldview in the face of disconcerting change. The problem is when this tendency leads to scientific dogmatism and the doubling down on positions that turn out to be, well, wrong. Or, worse, when we exploit uncertainty to expand our power. This was arguably the case, at times, during COVID-19, when we took strong stands based on data that did not support such certainty, pushing for sweeping policies that touched the lives of millions. Whether these measures were, in the end, the best steps to take from the perspective of supporting health remains an open question. What is clear, however, is that there was real currency during the COVID-19 moment in projecting an aura of certainty that policymakers could use to justify the steps they took. As public health gained power on these dubious grounds, I was reminded of a line from Dostoyevsky’s novel Demons:

“I got entangled in my own data, and my conclusion directly contradicts the original idea from which I start. Starting from unlimited freedom, I conclude with unlimited despotism.”

Without truth, without accountability and a recognition of limits, there is only power. Such a context is inimical to the healthy pursuit of science. Scientific progress relies on a context of freedom—to reason, to inquire, to pursue truth wherever it leads. This process can be messy, inconclusive, and characterized by ambiguity. When our conclusions do not match our initial ideas, it is time for us to reassess, incorporate new data, and try a different approach. Science demands an openness to correction, to being wrong, and to the capacity to course correct without defensiveness or the wearing of blinders that let us continue to think we are right when we are not.

This brings me, as much of my work in population health has, to the insight of Geoffrey Rose, who wrote:

“[Our task] is not to tell people what they should do. That is a matter for societies and their individual members to decide. [Rather, our task is] to analyze the options, so that such important choices can be based on a clearer understanding of the issues.”

This, to me, is right, reflecting both the potential of science and the importance of understanding its limits. Science can be a tremendous force for good, helping us better understand ourselves and the world around us. It has helped drive progress that has created a world which is, in many ways, the healthiest it has ever been, even as much work is still to be done. But science remains inextricably linked to complexity and the fallible humanity of those who practice it. The values, biases, and assumptions of scientists will always be a factor in the work of science, as will our human tendency to occasionally ignore complicating nuance, double down on incorrect ideas, and insist we are right when we are wrong. For science to reach its full potential, it must avoid these blind alleys. I will explore this in next week’s essay, taking a closer look at the values that guide our science and the challenges that can arise when we blur the line between our values and our data.