When I use a word . . . . Medical counterfactualsBMJ 2022; 377 doi: https://doi.org/10.1136/bmj.o1267 (Published 20 May 2022) Cite this as: BMJ 2022;377:o1267
- Jeffrey K Aronson
- Centre for Evidence Based Medicine, Nuffield Department of Primary Care Health Sciences, University of Oxford, Oxford
- Twitter @JKAronson
A counterfactual statement or narrative expresses an assessment of what might have happened if certain unfulfilled conditions had actually occurred: “If such-and-such a thing had happened [typically different in a particular way from what actually happened], such-and-such another thing might have occurred as a consequence [instead of what actually occurred].” Since the protasis, or premise, of a statement of this kind is false, as implied by the use of the conditional subjunctive mood, the truth or falsehood of the apodosis, the proposed outcome, whether a single statement or a prolonged narrative, is indeterminable.
The idea of a counterfactual statement was originally a philosophical concept called a contrary-to-fact condition or supposition, derived from the algebra of logic,1 which was introduced by George Boole in The Mathematical Analysis of Logic (1847).2 The logic led to the conclusion that a false proposition implies any proposition, true or false. A contrary-to-fact conditional statement was first described as counterfactual in 1946,3 although the idea of a counterfact was much older, implying evidence against a particular view.4
The use of counterfactual ideas is a popular device in fiction, and there are many examples.
In Les Miserables (1862), for instance, Victor Hugo speculated on what would have happened if Napoleon had won the Battle of Waterloo. And even earlier, in Napoleon and the Conquest of the World (1836), Louis Geoffroy speculated on what would have happened if Napoleon's invasion of Russia in 1811 had succeeded.
Nineteenth century American history has often figured in counterfactual novels. In Bring the Jubilee by Ward Moore (1953) the conceit is that the `Confederates win the civil war. In Colson Whitehead’s Pulitzer Prize winning novel The Underground Railway (2016), the history of American slavery is highlighted in the context of an imaginary escape route for runaways.
The Second World War has also been a rich source of counterfactual speculation. In The Man in the High Castle (1962), Philip K Dick imagined a world in which the allies lost the war, Giuseppe Zangara having succeeded in his attempt to assassinate FDR in 1933. Other novels that depict imagined outcomes in which Germany won the war include Hitler Has Won by Frederic Mullally (1975), Fatherland by Robert Harris (1992), Dominion by C J Sansom (2012), and a posthumously published short story by Kingsley Amis, 1941/A. Len Deighton’s 1978 novel SS-GB imagines wartime Britain under Nazi occupation. In The Plot Against America by Philip Roth (2004), the anti-Semitic Charles Lindbergh defeats Roosevelt in the 1940 presidential election and then makes a pact with the Nazis.
Hitler’s fate has also inspired counterfactual fiction. In his 2009 film Inglourious Basterds [sic], Quentin Tarantino imagined the assassination of Hitler in 1944. Conversely, that Hitler might have survived the war has been imagined by George Steiner in The Portage to San Cristobal of AH (1981) and by Beryl Bainbridge in Young Adolf (1987). Other novels depict what would have happened had Hitler not become the politician that he did. In The Iron Dream, Norman Spinrad describes how Hitler emigrates to America in 1919 and becomes a science fiction writer; the novel includes a narrative, Lord of the Swastika, supposedly written by this fictional Hitler just before his death in 1953. And in an extreme example, Making History (1996), Stephen Fry describes imagined outcomes in Germany in the 1930s and 1940s had Hitler never been born, the counterfactual being achieved by a form of time travel and a pharmacological intervention.
An important element in all of this is the assumption that events that actually happened were replaced by alternative events. Otherwise all fiction could be regarded as counterfactual.
Some regard counterfactual speculation as more than just an entertaining game, considering it also a mode of historical analysis.5 It might, for example, promote reflection on past actions, suggest possible causes of outcomes, and lead to different actions in the future, or lead to the repetition of previously successful strategies, avoiding other possibilities.
However, it is a commonplace that those who say that they must learn the lessons from their recent experience rarely show subsequent signs of having done so. Of course, on the next occasion the conditions are not quite the same as they were before and the necessary actions may alter. Furthermore, counterfactual analysis may attribute to individuals characteristics that would have allowed them to have behaved differently, when they were not actually possessed of those characteristics.
Another problem with counterfactual analysis is the butterfly effect,6 by which small changes in outcomes in a complex system lead to large overall changes, militating against accurate predictions. For example, the presumption that the attempted assassination of President Roosevelt in 1933, if successful, would have led to victory of the Axis powers in the Second World War implies that America would not have entered the war, even after Pearl Harbor; in fact America might have entered the war sooner. History would undoubtedly have been changed, but we cannot know how and to what extent.
There are many counterfactual texts presented not as fictions but as academic analyses. For example, J C Squire's 1931 collection of essays, If It Had Happened Otherwise, included 11 essays, one, by Winston Churchill, titled “If Lee had not won the Battle of Gettysburg,” and another, by Ronald Knox, titled “If the General Strike had succeeded,” postulating Britain under communist rule.
Most counterfactuals consider an event that happened and then posit what would have happened if it had been otherwise; for example, the allies won the war, but what would have happened if they hadn’t? Some of the essays in a modern sequence of books edited by Brack and Dale, all with similar titles, take a different approach—consideration of an event that never happened, contrasted with an event that also never happened. For example, what if Lyndon Johnson had been shot down in 1942? Or what if Britain had joined the Euro in 1976?
This sequence includes Prime Minister Portillo and Other Things that Never Happened (2003), President Gore and Other Things that Never Happened (2007), Prime Minister Boris and Other Things That Never Happened (2011), Prime Minister Corbyn and Other Things That Never Happened (2016), and Prime Minister Priti and Other Things That Never Happened (2021).
In some cases this approach becomes more predictive history than counterfactual. For example, the title essay in the 2011 entry in the series, “What if Boris Johnson becomes Prime Minister in 2016?”7 contemplates something that did not happen then but has since happened. And can we presume that “Prime Minister Corbyn” and “Prime Minister Priti” will not happen? The former seems highly unlikely and, although the latter may not be likely, who knows?
A search for the term “counterfactual/s” in titles of papers listed in PubMed yields just over 500 hits, nearly 50% of which are publications in journals devoted to various aspects of psychiatry and psychology, including behavioural science and cognition. Here I discuss some more general aspects of the roles of counterfactuals in medicine.
Counterfactuals underpin our approach to testing scientific hypotheses. If a given hypothesis is true, we would expect some preventive or therapeutic intervention to produce a predictable outcome; if it does we should test the hypothesis again using a different intervention, attempting to falsify it. Alternatively, if the intervention produces a different outcome to that predicted, we would have to either reject the hypothesis or modify it and test the modified version in further experiments. This implies that we should take a subjunctive approach to hypothesis testing, by assuming that the hypothesis is false (“if it were true …; but we assume that it isn’t”), which is the statistical approach (the null hypothesis). In that case the outcome will be indeterminable in advance. If we assume from the start that the hypothesis is true, we may be seduced into making the cardinal error of trying to prove it rather than testing it, i.e. trying to disprove it.
Counterfactuals are also relevant to the design and analysis of randomised controlled trials, since the average outcome in those given the intervention being studied (the factual outcome) is compared with either a placebo or a comparator that is known to be effective (the counterfactual to the intervention being studied).8 Techniques have been described that use the counterfactual principle in, for example, analysing data from clinical trials in which surrogate markers are used as endpoints9 and in accounting for treatment failures in clinical trials.10
When a preventive or therapeutic intervention is contemplated, counterfactuals also underpin the Precautionary Principle.11 If, in contemplating the use of an effective intervention in a particular patient, I were concerned about the risk of an adverse outcome, the benefit:harm balance in that patient being unfavourable, the Precautionary Principle would dictate that I should not use the intervention.
It is all too easy to fall in love with one’s own hypothesis when testing scientific hypotheses, conducting clinical trials, and contemplating possible therapeutic interventions. This has been much in evidence in the evangelistic ways in which certain preventive and therapeutic interventions have been touted during the recent pandemic. There has been too little consideration of the counterfactual. Too much subjectivity; too little subjunctivity.
Competing interests: None declared.
Provenance and peer review: Not commissioned; not peer reviewed