Are You One in a Million?
Precision medicine meets public health.
Posted Jul 25, 2019
“The patient with the disease is not synonymous with the disease the patient has.”
This statement was made by Charles Scriver, a renowned geneticist who helped establish newborn screening programs in Canada, in 1988.
What he means is that what we know about the various “causes” of disease does not translate one-to-one into being able to predict who will become a “case” of disease. It reflects that the causes of disease can—and do—change over time and vary across populations. This variation affects how common a disease is, who it is most likely to be affected, and how amenable it is to prevention.
The corollary is the following: the factors that cause two different populations (that is, groups of people who live in a defined geographic area at a specified time) to have different health outcomes are often not the same as those factors that cause two different people within a single population to have different health outcomes.
Sometimes I have trouble wrapping my head around this: After all, it seems logical that the factors that make my health different from yours would be the same as the factors that make a group of people like me different from a group of people like you, right? But those factors often are different (usually in degree of importance rather than in an absolute all-or-nothing sense).
Let me illustrate this with the example of smoking and lung cancer from Geoffrey Rose’s seminal 1985 article “Sick individuals and sick populations.”
If everyone in the US smoked two packs a day, two things would be true: (1) The incidence of lung cancer would be much higher than it is, and (2) there will be no apparent association between cigarette use and whether an individual developed lung cancer or not. Instead, we would conclude that lung cancer was caused primarily by genetic liability.
Why? Because in a population where everyone has the same environmental exposure (i.e., tobacco use), genetic liability would be the main source of variation between individuals who developed disease and those who did not. But, if we then compared the US to a different country where only 20% of people were heavy smokers, we would see that smoking was indeed a much more important cause of population differences in risk of lung cancer than genetics.
Scriver provides a second example: rickets. Rickets is a skeletal disease of impaired calcium/phosphate metabolism that primarily affects children. With the identification of vitamin D and its role in supporting bone metabolism, and the subsequent fortification of milk with vitamin D in western countries in the mid-20th century, two things happened: (1) the incidence of rickets declined, and (2) nowadays most cases that occur in these places are due to genetic liability.
That is, rickets went from being a disease of primarily “environmental” causes (i.e., related to diet and sunlight exposure) to one of primarily “genetic” causes in these populations. In contrast, rickets is still a relatively common disease in many parts of the world (i.e., several African and Middle Eastern countries), largely due to dietary issues.
So, what does this mean for precision medicine?
Well, for individuals, it means not all cases of disease are equally preventable, because not all cases have the same set of (potentially modifiable) causes.
For populations, it means that prevention strategies that are effective in one population may not be as effective in another, and vice versa. Populations have features that individuals do not have, such as residential segregation, cultural norms, health disparities, and contagion transmission rates. These population characteristics do not have any individual-level equivalent, akin to the difference between looking at a single pixel on your screen vs. the pattern of pixels that make up the words you are reading right now.
In short, the features that shape population health may have little to do with the causes of individual cases of disease.
The hope of this century’s “precision medicine” initiatives (e.g., the NIH All of Us 1 million person cohort) is to improve the resolution of our understanding of the causes of disease within the US population and to understand which causes are most relevant to which groups in our nation.
[I know I’m hedging here. Precision medicine is usually described by the axiom of getting “the right treatment to the right patient at the right time,” but I’ll consider these initiatives a success if they simply give us “accessible and acceptable treatments with an appropriate balance of risks and benefits to the patient in a timely manner.” And I’m an optimist.]
While the future may be precision medicine, right now there is no way to know which bucket you, as an individual, are in: That is, there is no way to know the relative importance of causes of your personal future health or disease. This means that everyone’s best bet is to support policies and programs that improve the health of our population as a whole.
Even under ideal circumstances, precision medicine won’t eliminate the fundamental nature of uncertainty in healthcare decision-making, and it won’t make up for studies that are poorly designed. But it has the potential to both widen the spotlight of where scientists look for factors that impact our nation’s health, and pay homage to the fact that those causal factors are always tied to a particular time, place, and population.
Scriver CG. Cases are not incidence and vice versa. Genetic Epidemiology 1988;5:481-487.
Rose G. Sick individuals and sick populations. International Journal of Epidemiology 1985;14:32-38.
Creo AL, et al. Nutritional rickets around the world: an update. Paediatrics and International Child Health, 2017; Vol 37 (Issue 2)