Time and chance: the stochastic nature of disease causation

Abstract

On Dec, 31, 1664, Samuel Pepys, taking stock of the past year, confided to his diary: “I bless God, I never have been in so good plight as to my health … but I am at a great loss to know whether it be my Hare's foot, or taking every morning a pill of Turpentine, or my having left off the wearing of a gowne”.1 We would be wrong to think that our current preoccupation with health is new. People have always been intensely concerned with what makes them sick and what keeps them well. What has changed since the 17th century is our expectations, not only of effective treatment, but also of a satisfactory explanation as to why we have become ill. For Pepys, illness and disease had to be accepted as part of the human predicament. Their causes were a matter of conjecture, and avoiding them required magic and good fortune. Nowadays, luck and magic have been replaced by risk factors. Health promotion messages lead people to believe that by adopting the right lifestyle they will stay well. Naturally, if they become ill, they demand a reason. If such a reason is not forthcoming, they imagine that this is because medical knowledge is still incomplete, and that further research will soon reveal the answer. The view that better knowledge of causation will allow us to understand why some people develop a disease whereas others do not, is widely shared by doctors and medical scientists. Think how often the introduction to a grant application contains the phrase, “the causes of X are poorly understood”, by which the authors imply that there are important, but as yet unidentified, causes waiting to be discovered and that, provided the committee has the vision to award funding, better prediction of disease occurrence and more effective prevention could be possible. Though we smile at Pepys' hare's foot amulet and turpentine pill, his doubts about their effectiveness might contain more truth than the aspirations of the grant applicants. Obviously, a patient with a disease must have been exposed to a combination of causes sufficient to induce that disease, but it does not necessarily follow that these causes are measurable or even identifiable. An analogy can be drawn with the throw of a die. The fact that a six is rolled does not mean that the die was heavily exposed to any risk factors for that outcome. Instead, this result is due to the operation of a complex set of circumstances (dimensions, shape, and weight of the die; the exact position of its centre of gravity; forces acting on it and its height above the ground when thrown; viscosity of the air; contour and elastic properties of the surface on which it landed, and so on), none of which could be shown to have a consistent effect on its own on whether the die rolls six—except, of course, in the trivial circumstance of dropping it very gently from minimum height with the six face up. In this essay we review the nature of disease causation, and argue that the role of stochastic processes is underappreciated. Advances in understanding of pathogenic mechanisms or genetic susceptibility might not enable us to explain why one person gets a disease and another does not.