Maarten J Wensink, MD, PhD
Institute of Public Health, University of Southern Denmark
Tolstoy in May, 1908, photographed at Yasnaya Polyana by Sergey Prokudin-Gorsky
It is a testimony to the genius of Lev Nikolayevich Tolstoy that a fine selection of concepts later arduously acquired over the course of decades can be found in the first of Tolstoy’s great novels, War and Peace (1). Although Tolstoy was not primarily interested in the origin of disease, in a way he was like all of us: he observed and tried to understand. The number of his concepts that find application in modern epidemiology is as astounding as their diversity. Tolstoy being Tolstoy, he exposed those concepts meticulously, and today’s public health professionals may like to take notice, if only as a matter of curiosity.
In War and Peace(1 [IV,4,8]), Tolstoy describes how the Russian army looked prouder than ever after having been through battle, without supplies, chasing the chaotically retreating Napoleonic army through the Russian winter. The reason for this improvement in the state of the Russian army was not that battle and winter helped the individual soldier, but that “[a]ll the physically or morally weak had long since been left behind and only the flower of the army—physically and mentally—remained”:Tolstoy was aware of selection within a heterogeneous population. In epidemiology and demography too, the frail tend to die first, leaving only the stronger individuals. Mortality of supercentenarians is constant at 50% per year, while individual supercentenarians age observably, because the frailer individuals tend to die first, leaving only the more robust(2). Tolstoy would have understood and marveled at this thinking.
Modern epidemiology distinguishes two types of bias: selection bias and information bias (3).Selection bias includes all forms of bias where the study sample is not representative of the reference population. Information bias refers to those biases where the quality of measurement of exposure or outcome differs between the comparison groups. Tolstoy did not separate these types of bias, nor did he use the word, but he was certainly aware of the problem. He describes how those orders given by (military) commanders that had the required effect were remembered, while other orders, which had seemed equally reasonable and justified at the time, but that did not fulfill their intended function, were forgotten (1 [epilogue 2, 6]). Consequently, the effect of orders on (the success of) affairs was overestimated. Tolstoy ridiculed Napoleon, writing that many of his orders were not even carried out, and that many of those that were carried out remained entirely without effect. Thus, Tolstoy gave a very clear description of survivorship bias, "survivorship" defined as an order being successful. Tolstoy was fully aware of the concept of bias and of the way it distorts our world view.
Another great example of Tolstoy’s incisivenessis is his treatment of correlation without causation. He goes to great lengths to explain this phenomenon (1 [III, 3 , 1]): “Whenever I look at my watch and its hands point to ten, I hear the bells of the neighboring church; but because the bells begin to ring when the hands of the clock reach ten, I have no right to assume that the movement of the bells is caused by the position of the hands of the watch.” He continues to give a number of similar examples, including the emergence of (heroic) historical figures at historical scenes, which, so argues Tolstoy, does not mean that those events are actually being driven by those historical figures. So he touches on something that has long been at the core of epidemiological inquiry: what does a correlation mean, and under what circumstances does correlation mean causation? And clearly he flirted with the concept of confounding: the underlying real causes of both historical scenes and the emergence of Napoleons, so that a relationship may seem to be there, but is not.
Indeed, Tolstoy was obsessed with the process of causation, especially with regard to history. According to Berlin (4), he envisioned some kind of “calculus” of history, in which infinitely many infinitely small causes would be added up to explain the totality of history. Yet these causes are too great in number for men to identify them all. Historians selectively pick those causes that suit their theories, which are compromised as a result. Events took place not because Napoleon so wanted (as historians would say), but because hundreds of thousands of soldiers so felt (1). This model features prominently in Berlin’s canonical essay on Tolstoy’s view of history, The Hedgehog and the Fox (4). But for Tolstoy, all those causes interact, too. According to Tolstoy, an apple falls from the tree not only because of gravity, not only because the apple is moved by the wind, not only because of changes in the physics of the apple’s stem, and not only because a boy standing under the tree wishes to eat the apple [sic], but because of the coincidence of all of the above(1 [III, 1 , 1]). So with history: component causes need to coincide to give some specified effect, and without any one of these component causes, the effect would not occur. And so with epidemiology: Tolstoy’s thinking here is exactly the causal pie model (5), where component causes interact to form sufficient causes. As for the multitude of causes, one can only wonder how Tolstoy would have reacted had someone explained to him the later concept of Principal Component Analysis: would he have changed his view?
This second model of the machinations of causation, interaction, is curiously absent from The Hedgehog and the Fox, possibly because Tolstoy himself did not apply this concept to history as fully as the first. Tolstoy does not allow historical figures to play a role in interaction with the multitude of other causes, instead presenting them as being entirely driven by history (e.g. “A king is history's slave.” (1 [III, 1 , 1]), a view previously criticized (discussed in (4)). Perhaps Tolstoy did not consider interaction between his two models of historical causation because it escaped his notice that he had two models to juxtapose; perhaps, and this seems the more likely reason, not because he disliked historical figures, whom he mostly thought of as arrogant and self-important, too much to acknowledge their potential role.
Frailty selection, bias, correlation without causation, confounding, component causes; this collection of Tolstoy’s insights is not exhaustive. Tolstoy also warned of post hoc explanations (1 [III, 2 , 1]), and makes the case for the heterogeneity of disease (1 [III, 1 , 16]). The famous first sentence of Anna Karenina has been picked up in the form of the observation that there are many ways of being sick, but few ways of being healthy(6).
Tolstoy’s work thus abounds in insights that the modern public health physician should devour with gusto. The fact that Tolstoy was weary of medical doctors should be no barrier for public health professionals to emerge themselves in Tolstoy’s work. Indeed, no longer being guilty of many of Tolstoy’s accusations is one of the major achievements of medicine, and modern medicine being at odds with Tolstoy would indicate room for improvement in the former rather than for the latter. The accusations that Tolstoy hurls at medicine are as rib-tickling as they are accurate. These range from slights-in-passing (“But despite the fact that the doctors treated him, bled him, and gave him medicines to drink, he recovered.” (1 [IV,4 ,12]) to the wondrously ridiculous scene at the very beginning of Part II of Anna Karenina (7), his other great novel: the awkward examination of a young girl by a putatively brilliant doctor(male, of course), the failure to understand that the teenager suffers from a broken heart, the authoritative misbehavior of “the famous doctor” towards his colleagues. Tolstoy’s unequivocal condescension vis-à-vis medical doctors, for reasons he explained very well, while identifying the vulnerability of patients (“He, as a man who had seen life and was neither stupid nor sick” [distrusted the doctor, MW] (7 [part II, chapter 1]), offers rich warnings that have not lost a single bit of their urgency, although medicine no doubt does much better today than it did in Tolstoy’s time.
Had he been concerned with the causes and spreading of diseases, Tolstoy would have been a most excellent epidemiologist. His insights being hidden in a 1000+ page novel is all the better, since the acquisition of epidemiological insights should be far from the only reason to read this monumental piece of work.
I am aware of two ways of partitioning War and Peace (reference (1)). Some versions have 4 parts, of 3, 5, 3, and 4 chapters, respectively, bringing the total to 15 chapters, followed by 2 epilogues, all consisting of a number of paragraphs. Other versions, for instance War and Peace cited here, are partitioned in 15 books and 2 epilogues, all consisting of chapters. I give both references, (1[IV, 3 , 30]) indicating part IV, chapter 3 for the one system, book 14 for the other, paragraph/chapter 30.
- Tolstoy, L. War and Peace. First published 1869, translation by L. and A. Maude, Gutenberg.org;2009.
- Maier, H, Gampe, J, Jeune, B, et al. (eds). Supercentenarians.Springer-Verlag Berlin Heidelberg; 2010.
- Rothman, KJ, Greenland, S.Modern Epidemiology 2nd ed. Philadelphia, US: Lippincott, Williams & Wilkins; 1998.
- Berlin, I. The Hedgehog and the Fox. 2ndedn, first published 1953. Princeton University Press; 2013.
- Rothman, KJ. Causes. American Journal of Epidemiology 1986;104:587–592.
- Cohen, A. Complex systems dynamics in aging: new evidence, continuing questions. Biogerontology2015 DOI 10.1007/s10522-015-9584-x.
- Tolstoy, L. Anna Karenina. First published 1877, translation by R. Pevear and L. Volokhonsky, Penguin Classics; 2006.
I would like to express my profound gratitude to Bernard Jeune for a round of invigorating literary discussions.
Maarten Wensink is a medical doctor who takes an interest in biodemography, such as the evolution of aging. Currently at the Max Planck Odense Center, he is interested in epidemiology, biodemographic models of cancer, cancer etiology, aging, and the relationship between aging and cancer. Maarten is also an aficionado of (Russian) literature.