Introduction
Speaking before colleagues at Manchester University in 1923 in the aftermath of the recent influenza pandemic, British bacteriologist William Whiteman Carlton Topley addressed ‘a tendency in some quarters to believe that influenza, at least, will leave us in peace for many years to come’. This tendency was ‘based partly on a belief that we have gained some measure of immunity as the result of the last severe visitation, partly on trust in measures of international quarantine’. But Topley, who alongside younger colleague Graham Selby Wilson had been investigating the factors driving the rise, fall, and subsequent resurgence of epidemics among populations of laboratory mice, doubted that ‘any considerable degree of immunity is acquired by an individual’ from influenza. However, when dealing with ‘herds instead of individuals’, he believed ‘the general rise in immunity, even if it be of slight degree, may have some effect in retarding epidemic spread’. Immunity offered limited protection to the population but the ‘case for international quarantine is infinitely worse’ and ‘little or nothing can be hoped for from proceedings of this kind’.Footnote 1 Topley quoted approvingly from the Ministry of Health’s report on the pandemic, which underlined ‘the essential solidarity of all mankind in the matter of epidemic sickness’ meaning that ‘[n]o sanitary cordon, no quarantine, will shield us from this danger’.Footnote 2 For Topley and Wilson, the recent pandemic and their own experiments modelling it with mice captured both the futility of attempting to prevent the spread of such a pathogen and the possibility of reducing its impact by discovering how herd immunity should ‘be distributed among the individuals at risk, so as best to ensure against the epidemic spread of the disease’.Footnote 3 Almost a century later, appeals to infection-acquired herd immunity alongside criticism of attempts to contain the COVID-19 pandemic, would raise considerable scientific debate about the meaning and policy implications of the concept.
This article analyses the history of herd immunity in light of the controversy surrounding it during the pandemic. I argue the debate reflects a tension between observing and cultivating herd immunity that captures the concept’s long-running historical entanglement with technologies for enhancing the immunity of populations. When the COVID-19 pandemic was declared in March 2020, many scientists and government advisors referred to herd immunity as an inevitable and beneficial mechanism by which the pandemic would recede as the population acquired immunity and the virus became endemic. In mid-March 2020, epidemiologist and advisor to the government of the UK, Graham Medley, stated that ‘we’re going to have an epidemic and then it will become endemic and join in with all the other coronaviruses that we all have all the time, but don’t notice’. He linked this predicted trajectory to the government’s plan to ‘manage this acquisition of herd immunity and minimise the exposure of people who are vulnerable’.Footnote 4 Many advocating for such a response were sceptical of the efficacy of measures intended to suppress the new virus, such as school closures, and critical of the social harms these produced. The rise and fall of the pandemic, they argued, would reflect the accumulation of immunity across populations, a process which would proceed irrespective of attempts to stop it.Footnote 5 Other scientists, however, disputed this understanding of herd immunity, arguing that the concept refers to a vaccination strategy whereby transmission of a pathogen is either greatly reduced or eliminated, thereby protecting a small number of remaining susceptible individuals.Footnote 6 As the World Health Organization’s Director General, Tedros Ghebreyesus, stated in October 2020: ‘Herd immunity is a concept used for vaccination, in which a population can be protected from a certain virus if a threshold of vaccination is reached.’Footnote 7 For those who understood the concept in this way, measures to curb transmission were indispensable until a vaccine was developed and deployed at scale.
Soon after Medley and other experts referred to herd immunity in mid-March 2020, governments in the UK and elsewhere dramatically changed course, seemingly in response to backlash equating the strategy with ‘letting the virus rip’ and sacrificing those most at risk.Footnote 8 Countries around the world with the notable exception of Sweden implemented ‘lockdowns’, which included quarantines, restrictions on movement, and the widespread closure of schools and businesses in an attempt to eliminate or greatly reduce transmission of the virus.Footnote 9 In the USA, epidemiologist and future advisor to President Joe Biden, Michael Osterholm, questioned ‘what we hope to accomplish with limited self-quarantines and shelter-in-place directives’ and suggested a more appropriate strategy would be to ‘gradually build up immunity’ by letting ‘those at low risk for serious disease continue to work’ while ‘advising higher-risk individuals to protect themselves through physical distancing and ramping up our health-care capacity as aggressively as possible’. As had others, he predicted that in the absence of a vaccine, the pandemic would ultimately ‘burn itself out as the spread of infection comes to confer a form of herd immunity’.Footnote 10 By the time he wrote these words, however, lockdowns had become a lasting fixture of the policy response around the world.Footnote 11 In the minds of much of the public and many experts, herd immunity in the absence of a vaccine became tantamount to doing nothing at all. How was it possible for scientists to reach opposing conclusions about the definition and policy implications of a fundamental epidemiological concept?
The ambiguity surrounding herd immunity was evident well before the COVID-19 pandemic. Often, the term simply denotes the degree of immunity in a population, irrespective of its quantity or whether it is acquired by infection or vaccination.Footnote 12 At other times, it is taken to mean the protective effect on remaining susceptible individuals of a growing proportion of immune people around them. As epidemiologist Paul Fine recognised over thirty years ago, however, even this understanding ‘lends itself to different interpretations’ which may be quantitative and ‘reflected in reductions in frequency of disease’ or qualitative, suggesting ‘total resistance’ and ‘implying a threshold number or percentage of immunes above which an infection cannot persist’.Footnote 13 Others argue that this conflates herd immunity with a ‘herd effect’ which increases in proportion to the development of herd immunity but is not synonymous with it.Footnote 14 To further complicate the issue, herd immunity is sometimes used interchangeably with ‘herd immunity threshold’, a distinct concept that traditionally denotes the ‘proportion of immunes in a population, above which the incidence of the infection decreases’.Footnote 15 While that threshold often describes transient, infection-acquired immunity driving fluctuations in the prevalence of endemic pathogens, in recent decades, many have invoked it as the point at which there is sufficient immunity such that ‘transmission of the agent is largely prevented’.Footnote 16 Clearly, much confusion and slippage surround herd immunity and contributed to its contested meaning during the COVID-19 pandemic. While history cannot resolve contemporary scientific disputes, it can help explain how these disputes developed over time.
Recent historical scholarship has deepened our knowledge of herd immunity. Historian David Jones and anthropologist Stefan Helmreich argue that its origins reside in the work of American veterinary scientists in the 1910s, who wrote of it in relation to contagious abortion, a disease of cattle. They note the concept’s adoption by British epidemiologists in the interwar era, beginning with Topley and Wilson.Footnote 17 Subsequently, I pursued its development into the postwar era, highlighting its growing association over time with vaccination and suggested that recent debates reflect the inherent tension of relating the individualistic notion of immunity to populations.Footnote 18 Contesting the veterinary origins of herd immunity, historian Warwick Anderson has proposed that the term’s adoption by Topley and Wilson reveals the influence of British social theorist Wilfred Trotter and his theory of the ‘herd instinct’ which ‘made the herd a supple and powerful metaphor for the social dynamics and connectedness of human populations’.Footnote 19 These histories have sketched the development of the concept and raised ongoing uncertainty about its origins. There is, however, no compelling explanation as to why the concept became so contested in recent years.
In contrast to the suggestions offered by other scholars, I argue that recent debates reflect a tension between observing and cultivating herd immunity that has been a long-running feature of the concept. Jones and Helmreich propose that herd immunity was contested during the COVID-19 pandemic because of its implied associations with the sacrifice of livestock, which ‘could have contributed to the objections in March to policies that would have asked many people to be sickened or killed by SARS-CoV-2 in pursuit of herd immunity’.Footnote 20 Meanwhile, Anderson suggests that one group of scientists who recommended leveraging infection-acquired herd immunity to protect those most at risk did not consider ‘the possibly limited duration of any postinfection immunity’.Footnote 21 Whatever the merits of ‘herd immunity strategies’ against COVID-19, these explanations for the disagreements surrounding them fail to elucidate the terms of the debate. That debate was not about whether the vulnerable should be sacrificed to attain herd immunity. Nor did proponents of herd immunity argue that immunity would be permanent – they repeatedly stated that it would not be. Rather, the debate concerned how effective disease control measures would be at protecting those most at risk and whether or not herd immunity was relevant to infections providing transient immunity. Epidemiologists disagreed on both questions, and it is the nature of that disagreement which requires an explanation. The long-running tension between observing and cultivating herd immunity, I argue, explains how the definition and implications of the term could become so contested.
In the first section, I discuss the work of Topley and Wilson with long-running mouse epidemics, focusing on the growing importance they placed on herd immunity as the core mechanism undergirding the wave-like pattern of the epidemics they observed. In the second, I turn to efforts by others to relate the concept to human diseases, primarily to outbreaks of diphtheria. I suggest that various immunological technologies – most notably, the Schick test for immunity to diphtheria – provide insight into the underlying conceptual shift such technologies produced. Permitting the quantification and enhancement of a population’s immunity, these tools pushed researchers to think of herd immunity as a quality that in addition to being observed could also be actively cultivated. This connection between technologies like the Schick test and subtle shifts in the scope of herd immunity points to developments in the postwar era, when vaccination increased and some came to refer to the concept interchangeably with disease elimination. In the third section, I turn to the work of epidemiologists Roy Anderson and Robert May who, beginning in the 1970s, began to model infectious diseases through the predator-prey framework of ecology. Their work included important contributions to discussions of herd immunity and disease elimination but in contrast to some others in the period, Anderson and May were careful to distinguish between these concepts. In addition to the enormous impact of their work on vaccination programmes and the mathematical modelling of infectious diseases, Anderson and May directly influenced many of the scientists who would come to contest herd immunity during the COVID-19 pandemic, a subject to which I return in the conclusion.
Endemic epidemics: conceiving herd immunity
Just over a century ago, in the aftermath of the influenza pandemic of 1918–19, the Evening Standard reported on experiments being undertaken on a ‘mouse colony’ at Charing Cross Hospital, London. With annual funding from the Medical Research Council of £10,000, these experiments were expected to uncover ‘the laws of epidemics […] as they emerge in a population of mice under controlled conditions’ with the ultimate aim of producing ‘knowledge effective for the control of epidemics’.Footnote 22 Initiated in 1918 by British bacteriologist William Whiteman Carlton Topley (1886–1944), the experiments ran continuously over the next two decades, taking the lives of up to 200,000 mice and reshaping knowledge about the mechanisms of epidemic disease.Footnote 23 Quoting Topley’s speech at Manchester University in 1923, The Manchester Guardian reported that the mouse epidemics suggested it was ‘extremely improbable’ that influenza ‘would relieve us of its presence in the future’. The only reasonable scientific course to pursue was to discover ‘the factors which determined the change from the unwelcome but relatively harmless acquaintance of inter-epidemic periods to the destructive fury of times of epidemic prevalence’.Footnote 24 Discussing his experiments before the Royal College of Physicians in 1919, Topley stressed that their purpose was to ‘explain the constant presence of a specific cause of disease’, its ‘periodic reappearance’ in epidemics and ‘the characteristic form of each such wave of disease in its rise, crest, and subsidence, leading to another disease free period’.Footnote 25 As historian Olga Amsterdamska noted, while at first these experiments seemed to confirm the primary role of variations in the virulence of pathogens in the waxing and waning of epidemics, by 1923 Topley and Wilson were directing their attention to the growing immunological resistance of the host population.Footnote 26
Topley’s ill-fated mice followed him from Charing Cross to Manchester University, where he was appointed chair of bacteriology in 1923, and on to the London School of Hygiene and Tropical Medicine (LSHTM), where he became the first chair of bacteriology and immunology in 1927.Footnote 27 While at Charing Cross, Topley enlisted the support of younger bacteriologist Graham Selby Wilson (1895–1987), who in 1919, having returned from military service in India while ‘convalescing from malaria and dysentery’, was ‘[m]otivated subconsciously by Topley’s dominating personality’ and decided to ‘hitch my waggon [sic] to a star’.Footnote 28 Wilson followed Topley and his mice to Manchester and LSHTM. From their extensive mouse work, the pair laid the foundations for the concept of herd immunity, fundamentally reshaping the understanding of epidemics as what their colleague, Major Greenwood, would refer to as the study of ‘crowd diseases’.Footnote 29 Just as Greenwood had suggested, the epidemiologist was concerned ‘not with individuals, but with aggregates’, so too Topley and Wilson sought to extend the conceptual parameters of immunity beyond the individual.Footnote 30
Though over the years, Topley and Wilson conducted different experiments with a variety of mouse pathogens, the basic setup and purpose of the experiments remained consistent. Cages with varying proportions of immune and susceptible mice would be exposed to a pathogen and deaths would be tracked over the coming days, weeks, and months. Invariably, the result would be the familiar ‘wave-like’ pattern of human epidemics, deaths tending to rise and fall in a symmetrical pattern. By creating populations with different proportions of immune and susceptible mice or, alternatively, by introducing different numbers of mice into each cage at variable intervals, the pair found that they could alter the timing and patterns of these waves. Instructions for the lab assistants uncovered in Wilson’s archive reiterate the importance they placed on the standardisation of their experiments.
The mouse enclosures were ten inches in diameter with perforated lids and could be ‘fixed in series, so as to form together a cage of any desired size’.Footnote 31 Each day, assistants would move surviving mice into clean cages, collect and catalogue dead animals, and sterilise filthy cages for the following day’s work. The instructions stressed that ‘the object in view is always to avoid the passage of infection from cage to cage, by dust, by the hands, by dirty instruments or feeding vessels, by contaminated food, by flies, or in any other way’. Gloves were worn throughout the work and tools sterilised immediately after use. Flies presented an ‘especially dangerous’ problem, and assistants were encouraged to keep ‘a careful look-out for flies every time you enter the room, and when you see one kill it immediately’. Each morning, assistants collected the dead mice and placed them ‘on a tin tray with a label bearing the number of the cage’. ‘On the corresponding card are entered the date on which the dead mice are found, and the number dead’, before the mice were promptly ‘brought to Professor Topley’s laboratory before 10.30 a.m., 10 a.m. on Saturdays’.Footnote 32 The animals would then be autopsied, daily deaths in each cage tallied, and charts produced visualising the epidemics.
In their first published reference to ‘herd immunity’ in 1923, Topley and Wilson discussed an experiment with 150 mice that had run for 120 days. After immunising half the mice against B. enteritidis, they put groups of thirty into five separate cages, the first containing thirty ‘normal’ mice (i.e. not immunised), the last, thirty immunised mice, and different proportions of each in the remaining cages. Taking ten mice ‘which had recently passed through a considerable epidemic of enteric infection’, they confirmed the presence of B. enteritidis and then added two infectious mice to each of the five cages. Closely observing them over four months, they tallied and autopsied the dead mice, identifying those that had died following infection with the pathogen. The results showed that
a considerable epidemic was produced among the 30 normal mice leading to a mortality of 96.7 per cent. within 120 days, and a specific mortality of 70 per cent. Among the immunised mice there was no evidence of spread of specific infection. Although a total mortality of 60 per cent. occurred during 120 days, no death could be attributed to enteric infection. Among the population, half of which were normal and half immunised, an epidemic was produced of equal severity with that which occurred among the normal population. The total mortality was 100 per cent. and the specific mortality 53.3 per cent. It should, however, be noted that, while the specific mortality among the 15 normal mice was 66.7 percent. that among the immunised mice was only 40 per cent.Footnote 33
One obvious finding was that ‘spread of infection occurs with difficulty among a population, each individual of which has been actively immunised’. More significant, however, was that in a population of both susceptible and immune individuals, ‘the relative immunity of the latter does not save them from infection and death, when epidemic spread occurs’. This unusual finding pointed to the connection between the immune status of individual mice and the proportion and degree of immunity in those surrounding them. ‘It appears that a degree of immunity, which may save individual hosts when living among equally resistant companions,’ Topley and Wilson reasoned, ‘is rendered of no avail when they are surrounded by highly susceptible individuals’.Footnote 34
This finding was of crucial importance for understanding epidemics among humans, as it framed the immunity of an individual as relative to and interconnected with the immunity of others in their proximity. Furthermore, such experiments emphasised that the distribution of immunity in a population was at least as important as the quantity, prompting them to ask a question which has remained central to discussions of herd immunity to this day:
Assuming a given total quantity of resistance against a specific bacterial parasite to be available among a considerable population, in what way should that resistance be distributed among the individuals at risk, so as best to ensure against the epidemic spread of the disease, of which the parasite is the causal agent?Footnote 35
Historian Olga Amsterdamska has highlighted that Topley and Wilson conceived epidemics in collective terms, ‘as a herd phenomenon that could not be reduced to individual cases of disease’, a perspective at odds with American researchers conducting similar experiments in this period.Footnote 36 As Topley and Greenwood complained in 1925, ‘the great majority of investigators have been so preoccupied with the individual that they have neglected the herd’.Footnote 37
Treating an epidemic as a collective process had important implications. The analogy of herds as ‘immune’ fed back into Topley and Wilson’s conception of individual immunity. In the first edition of their textbook, The Principles of Bacteriology and Immunity (1929), they distinguished ‘six categories of individuals among any infected herd’. The first four of these were, to varying extents, infectious to others, and included the ‘typical case’, the ‘atypical case’, the ‘latent infection’, and the ‘healthy carrier’. The remaining two classes – the ‘uninfected resistant’ and the ‘uninfected susceptible’ – were neither infected nor infectious. Understanding immunity not as a binary but as a spectrum of resistance, they noted that many of those in the categories of latent infection and healthy carrier ‘are relatively resistant’ and the only one that was entirely susceptible was the ‘uninfected susceptible’.Footnote 38 These immunological states were in constant flux as immunity was lost and replenished, meaning that the measurement of the average immunity of individuals could not provide ‘a true measure of the immunity of the herd’. A herd comprised of some highly resistant individuals and some ‘highly susceptible, would differ from that of a herd in which each individual possessed a resistance of intermediate grade’.Footnote 39
Another implication of framing epidemics as collective, dynamic processes leaving in their wake a variable spectrum of immunological resistance was that Topley and Wilson situated the individual’s immunity as intertwined with its surrounding community. One did not have to be entirely resistant to infection to be considered immune, and the immunity of a herd rarely implied the complete absence of infection and disease.Footnote 40 This is why Topley’s scepticism of the individual protection offered by prior infection with influenza did not diminish his belief that ‘the general rise in immunity, even if it be of slight degree, may have some effect in retarding epidemic spread’.Footnote 41 This understanding of herd immunity included changes in the environment, and Topley and Wilson distinguished between ‘herd immunity which is mainly dependent on environmental factors’ and that ‘which is mainly dependent on reactions in which the host’s tissues play a dominant part’. ‘All natural herd conditions, and all administrative modifications of these conditions, which tend to render more difficult the spread of a parasite from host to host,’ they explained, ‘must be regarded as factors making for herd immunity against the disease’.Footnote 42 While other researchers continued to emphasise the role of variations in a pathogen’s virulence as a critical driving factor in epidemics, Topley and Wilson insisted that this hypothesis ‘received little support as the result of direct experiment’ and they instead emphasised ‘the importance of variations in host resistance, and particularly in the ratio between immunes and susceptibles in a herd exposed to infection’.Footnote 43
A closely related outcome of treating epidemics as collective processes was an embrace of ideas of balance and equilibrium. From this perspective, herd immunity constituted an adjustment of the host population to the presence of an infective agent. As historian Andrew Mendelsohn has demonstrated, in the immediate aftermath of the 1918–19 pandemic, concepts of equilibrium ‘moved from margin to mainstream’.Footnote 44 By 1936, when Topley, Greenwood, Bradford Hill, and Joyce Wilson published a report drawing insights from almost two decades of mouse experiments, they wrote: ‘After the first wave of disease and death that always follows the aggregation of an infected herd, the epidemic settles into a state of unstable equilibrium.’Footnote 45 Intermittent waves of infection would continue, replenishing the immunity lost before once again receding, awaiting a loss of resistance. According to The Times, Topley and Wilson’s mice ‘seemed to establish an equilibrium with disease’.Footnote 46 In the textbook culminating from almost two decades of this research, they summarised the relevance of these findings to human populations:
In any of those common endemic-epidemic diseases from which many or most of our people suffer at one time or another during their lives, but which occur in epidemic form only at more or less widely-spaced intervals, we should regard this ever-varying state of the immunological constitution of the herd as the main factor determining the intervals at which the epidemic waves occur.Footnote 47
These insights had important implications for the effectiveness of isolation and quarantine, which had ‘played a major part in public health administration’. ‘A very cursory consideration of the various types of distribution of a bacterial parasite within an infected herd’, they argued, ‘will raise serious doubts as to the probable efficacy of such measures.’ ‘Once the barrier is passed, the subsequent course of events will depend upon the conditions obtaining within the community into which the infection has been introduced.’Footnote 48
For Topley and Wilson, the accrual and loss of herd immunity explained the rise, fall, and subsequent resurgence of an epidemic like that of 1918–19. Other factors, such as variations in the virulence of pathogens or quarantine measures, were at most secondary in a process that was driven by the uneven accumulation of varying degrees of immunological resistance across a population. While at times the pair explored the role of immunisation in their mouse epidemics, this was an attempt to establish baselines of immunity in different enclosures rather than an effort to discover immunisation rates necessary to control infectious diseases among humans. Soon after they began observing herd immunity among their mice and relating it to epidemics in human populations, however, others began drawing on the term as part of efforts to control epidemics by cultivating herd immunity.
Cultivating immunity: diphtheria and its immunological technologies
In 1923, the same year that Topley and Wilson first referred to herd immunity, British pathologist Sheldon Francis Dudley (1884–1956) analysed recent epidemics of diphtheria and scarlet fever among adolescent boys at the Royal Naval Medical School, Greenwich. Dudley was Professor of Pathology at the school from 1921 until 1933, during which time he witnessed recurrent epidemics of diphtheria, scarlet fever, and influenza and employed the Schick test for immunity to diphtheria to quantify the shifting immunological profile of the population.Footnote 49 His first year at Greenwich brought him face-to-face with an outbreak of diphtheria which ‘reached epidemic proportions’ between September and December 1921.Footnote 50 As the outbreak fizzled out over the first half of 1922, Dudley undertook mass testing of the boys using the Schick test. The test was a tool for assessing susceptibility to the disease that had been developed by Hungarian American paediatrician Béla Schick in 1913 to exclude already immune subjects from unnecessary and potentially harmful treatment with diphtheria serum.Footnote 51 A subject’s right forearm would be injected with a small quantity of diphtheria toxin while the left was injected with an equal quantity of heat-inactivated toxin as a control. If the individual already possessed a degree of immunity from previous exposure their right arm would display little or no redness and they would be deemed negative. Those who displayed redness only on the right arm were deemed positive (susceptible).Footnote 52 By this method, the immunity of populations could be quantified and remaining susceptibles immunised with diphtheria serum or, later, vaccination. The Schick test allowed researchers like Dudley to estimate the quantity of immune individuals required to stop or prevent an outbreak, an ability which encouraged efforts to cultivate herd immunity.
Between January and April 1922 Dudley tested the entire population of approximately 1000 adolescent boys at the school in Greenwich, finding that older residents who had passed through recent epidemics had consistently higher rates of immunity than younger, more recently admitted students. Among the older residents ‘only one in seven was susceptible to the disease’, while among the younger ‘more than three new boys out of every seven [had] positive Schick reactions’.Footnote 53 Dudley believed this provided a ‘striking demonstration of how immune a population of this sort can become in a diphtherial environment’.Footnote 54 Generalising from these figures, he calculated
that 45 out of every 100 boys who enter the school are Schick-positive reactors, but on being exposed for 10 months to a diphtherial environment, such as existed in this school, 35 of these 45 positive reactors will become Schick-negative reactors, and in most cases without getting clinical signs of diphtheria, i.e. 90 per cent. of positive reactors become negative reactors in 10 months.Footnote 55
Dudley agreed with Topley and Wilson that a freely spreading pathogen left in its wake an unpredictable distribution of immunity and that this distribution conditioned its future spread. ‘One community may contain a majority of highly immune individuals with an admixture of highly susceptible ones’ while another ‘may consist of individuals who all possess an intermediate degree of resistance’. Which scenario would provide ‘a more suitable soil for the evolution of an epidemic’ depended on the pathogen in question and the environment in which it circulated. By 1926, he was referencing Topley and Wilson’s discussions of herd immunity, noting that it was more than the sum of the resistances of individuals because it ‘depends also on how these resistances are distributed among the individuals who make up the herd’.Footnote 56
Throughout the 1920s and 1930s, communities throughout Britain were tested for immunity to diphtheria.Footnote 57 Following mass testing, susceptible individuals would receive a dose of toxin-antitoxin serum and would have their immunity retested weeks later. That diphtheria was surrounded by a range of immunological technologies for testing and inducing immunity explains why it was the exemplary disease studied through the prism of herd immunity in the interwar era. The Schick test enabled insights from the mouse work of Topley and Wilson to be directly related to disease control efforts in human communities. While Topley and Wilson did immunise some of their mice, their observations were limited to counting and autopsying dead mice. The Schick test, however, permitted one to monitor the quantity of herd immunity before, during, and after an outbreak and thereby obtain more granular insights into an unfolding epidemic. Philosopher of science Jonathan Simon argues that the development of diphtheria serum altered the definition of the disease by linking the common symptoms of diphtheria with a specific pathogen, transforming diphtheria from ‘a collection of symptoms’ into ‘the result of the action of an invading microorganism’. In addition to shifting definitions of disease, diphtheria technologies also transformed the relevance of herd immunity.Footnote 58
Dudley soon understood ‘natural’ diphtheria immunisation as an ongoing process involving the exposure of new susceptible individuals to clinical and subclinical doses of infection as well as ongoing stimulation of the immune responses of individuals who already had a degree of immunity. Already in 1923, he suggested that ‘during an epidemic of diphtheria there is a coincident “epidemic of immunization” to diphtheria’. This rapid spread ‘cuts in two directions’ as it produced a ‘greater concentration of infection’ and ‘a proportionately more rapid dissemination of subinfective doses, which will immunize, without symptoms, some number proportional to that of those in which symptoms appear’.Footnote 59 He proposed that carriers of diphtheria boosted the population’s resistance, preventing ‘large disease outbreaks by supplying sub-infective doses of infectious material to the more susceptible members of a population’. ‘In this way’, he reasoned, ‘the general resistance to epidemic disease is increased, and the herd is vaccinated by its carriers’.Footnote 60
At the same time, the Schick test pushed him to quantify and manipulate a population’s immunity. Following mass testing at Greenwich, he referred to the ratio of immune to susceptible individuals in a community as its ‘immunity index’ and hoped that ‘it may become possible to estimate the herd immunity of populations together with the prevalence and distribution of potential parasites’. Contrasting Schick results of residents at the school with ‘day boys’ who did not sleep there, Dudley estimated an ‘index of herd immunity’ as the ratio of immunes to susceptibles. As ‘techniques improve’, he hoped, ‘it may become possible to estimate the herd immunity of a population together with the prevalence and distribution of parasites’.Footnote 61 In 1927, physician Graham Forbes, speaking about his own research on diphtheria in London, suggested that overcrowded conditions of the housing in which many children lived prevented larger outbreaks ‘by the degree of herd immunity maintained by repeated exposure to small doses of infection’. After a certain point, however, ‘the more crowded the rooms, the greater the risk of close contact with massive infection capable of overcoming acquired partial immunity’.Footnote 62 In 1933, drawing on statistics from the USA, he attempted to estimate the immunity level required to reduce transmission, finding this ‘could not be expected until at least a third of the 0-5 population had been protected’, even if half of a school had been immunised.Footnote 63
My suggestion that the Schick test captures a shift from observing to cultivating immunity in populations is supported by the fact that some in this period believed the test not only detected immunity but also enhanced it. While producing batches of diphtheria antitoxin serum at the Wellcome Physiological Research Laboratories in London, immunologist Alexander Thomas Glenny (1882–1965) discovered that when previously inoculated animals were exposed to diphtheria toxin a second time, they demonstrated a ‘secondary stimulus’ in antitoxin production ‘10 to 100 times that reached after an injection into a normal animal’.Footnote 64 He suggested that ‘the very small amount of toxin used for the Schick test, apart from its action as a diagnostic agent, may also play an active part in immunisation’ by providing such a secondary stimulus.Footnote 65 Dudley agreed that multiple exposures to diphtheria enhanced immunity, arguing in 1928 that his own findings concurred with Glenny’s ‘experiments with diphtheria toxin and guinea-pigs’ and that the ‘dense diphtherial environment’ at the Greenwich school ‘acted as a primary antigenic stimulus’ and the secondary stimulus was ‘a mechanically injected diphtheria antigen’.Footnote 66 Though he believed the immunity induced by diphtheria serums was often not sufficient to create durable immunity, such an ‘artificial immunizing agent […] may act as an efficient secondary stimulus to those who have been sensitized by natural primary stimuli from the environment’, or alternatively, ‘act as a primary sensitizing stimulus to those who are about to receive further stimuli from the natural immunizing influences in a diphtherial environment’.Footnote 67
The notion of an ‘immunity index’ reveals a subtle but important shift in the concept and public health relevance of herd immunity that was directly linked to the Schick test and that points towards later interpretations of it as an elimination threshold that began to surface in the postwar era. Moreover, because it enabled the quantification of the shifting immunological landscape of a population and the identification of remaining susceptibles, the test pushed Dudley to conceive of herd immunity as a guide to intervening with diphtheria serums and vaccines. He argued that ‘the ecological point of view’ revealed the complex interaction of a host population and its parasites in a shared environment. Epidemics were ‘manifestations of a loss of balance between the mutual adjustment of host and parasite’, and public health should operate as a form of ‘applied ecology’.Footnote 68 Analysing the development of tissue culture and isotopic tracers in twentieth-century life sciences, historians of science Angela Creager and Hannah Landecker suggest that ‘technical and conceptual innovation are inextricable, as new methods change the kinds of questions that can be posed’.Footnote 69 The conceptual implications of the array of immunological technologies for testing, treating, and preventing diphtheria illustrate this entanglement of technical and conceptual innovation. It pushed Dudley to understand herd immunity both as an observable feature of ‘natural’ epidemics and as a quality that could be cultivated.
There was, however, considerable uncertainty about the safety and population-level impact of diphtheria serums. Immunisation with toxin-antitoxin serum could take up to six months to produce sufficient immunity, emphasising ‘the importance of arranging for the work of active immunisation in inter-epidemic periods’. The serum could produce ‘considerable local redness and swelling, and occasionally more or less definite constitutional disturbances’ in up to half of adults and older children.Footnote 70 Furthermore, because diphtheria was known to be spread by asymptomatic carriers, some worried that serum treatment could inadvertently increase the carrier rate, elevating the risk to remaining susceptible children.Footnote 71 This was part of the reason, historian Jane Lewis argues, that it was not until the 1930s ‘that the burden of medical opinion in Britain shifted to a wholehearted support of [diphtheria] immunisation’.Footnote 72 By the 1940s, Wilson was exploring the strength of children’s immune responses to variable doses of Alum Precipitated Toxoid (APT) diphtheria vaccine, Schick testing three communities in Berkshire before and after immunisation to compare the efficacy of different doses.Footnote 73 References to a ‘herd immunity threshold’ for purposes of disease control and elimination do not begin until well into the postwar era and are more closely associated with diseases such as measles, mumps, and rubella.Footnote 74 Use of the Schick test and diphtheria serums and vaccines earlier in the century, however, provides a glimpse into how the entanglement of herd immunity with immunological technologies for detecting and enhancing it produced an underlying shift in applications of the concept.Footnote 75
The range of immunological tools available for testing, treating, and preventing diphtheria during the interwar period pushed researchers to quantify and cultivate herd immunity. That relatively few such technologies existed for many other infectious diseases, however, meant alternative formulations of the concept continued to flourish. In the 1950s, British virologist and Chief of WHO’s World Influenza Center, Christopher Andrewes, invoked herd immunity to explain the periodicity of influenza. While influenza’s ‘emergence would be hindered by a high level of herd immunity’ in one location, at ‘the fringe of its earlier exploits […] herd immunity would be lower’ and the ‘reappearance’ of the disease would be more likely.Footnote 76 Andrewes related the virus’ ability to alter its antigenic structure to the accumulation of herd immunity. ‘Over a period of years’, he wrote in 1953, ‘variations may be played upon one antigenic theme, but after some time the possibilities will be exhausted (the herd will be generally resistant to closely related variants), and the introduction of a new motif will be necessary to keep things alive’.Footnote 77
Though not directly referring to herd immunity, Australian virologist and disease ecologist, Macfarlane Burnet, similarly suggested that influenza viruses ‘undergo mutation to serological types which are not readily susceptible to characteristic antibody of the community’. Fortunately, a rule of ‘exceptionally virulent’ influenza epidemics was a rapid return ‘to a tolerable equilibrium between host and parasite’.Footnote 78 Commenting more generally on infectious diseases of humans, microbiologist René Dubos wrote in 1955 that the newborn entering ‘the human herd’ confronts ‘certain microbes which are not yet fully integrated in human life by evolutionary forces and with which he as an individual has not had any experience’.Footnote 79 Historian Warwick Anderson suggests that disease ecologists, including Burnet and Dubos, ‘sought a means to relate microbiological processes to larger environmental or biological forces, a way to describe the interactive, dynamic relationships between host and parasite and physical milieu’.Footnote 80 While the technologies surrounding diphtheria drove efforts to enhance a population’s immunity and control disease, many continued to invoke herd immunity as an explanatory mechanism for the rise and fall of endemic infections like influenza.
My suggestion that the development and application of immunological technologies drove transformations in the concept and policy implications of herd immunity is supported by the postwar expansion of mass vaccination and the concomitant rise of disease elimination strategies in that period. As historian Nancy Leys Stepan has discussed, efforts to eradicate diseases such as smallpox, malaria, and yellow fever were defining features of the international health landscape of the postwar era, culminating in the WHO’s 1974 Expanded Programme on Immunization and its declaration of smallpox eradication in 1980.Footnote 81 If diphtheria had been the quintessential disease for delineating the public health implications of herd immunity in the interwar period, in the postwar period, the focus shifted to polio.Footnote 82 In the late 1950s, as Jonas Salk promoted his inactivated polio vaccine and Albert Sabin competed to advance his live attenuated vaccine, tentative prospects of polio elimination were raised. Speaking before the Royal Society of Health in May 1959, Salk was confident his vaccine was but the first of many that would ‘make it possible to bring under effective control, through the use of killed-virus vaccines, many of the viral pathogens’.Footnote 83 A few years later, he wondered whether the ‘herd effect’ produced by mass immunisation would reduce transmission to ‘a point approaching the conditions for extinction’.Footnote 84 Wilson, now in his mid-sixties, was sceptical, as revealed by draft notes he prepared on the subject.
Trials of Salk’s vaccine began in the UK in late 1956, but several issues remained unresolved. How might the vaccine interact with maternal antibodies protecting newborns? Would multiple doses be required to ensure lasting immunity? If so, at what ages should they be given? Wilson was uncertain about the practicality and prudence of attempting to banish polio. On the one hand, if the virus remained widespread in the community ‘vaccinated children will be subjected to frequent sporadic infections, which may serve to reinforce their immunity’. On the other hand, ‘if, as in diphtheria, the infecting agent virtually disappears, then the immunity conferred by Salk vaccination may very well wane to a point at which it will be insufficient to protect against an infection to which the subject is exposed some years later’. Might mass vaccination paradoxically increase the burden of the disease years after having brought it under control? In a community in which many individuals had been immunised with Salk’s vaccine ‘the virus in circulation may well undergo a diminution’, Wilson wrote. However, next to this typed paragraph, he scrawled: ‘Extermination of the virus is nonsense.’Footnote 85 Though he thought it possible that if enough people were vaccinated ‘a degree of herd immunity might be established that would succeed in reducing the incidence of poliomyelitis to minimal proportions’, he doubted that public uptake would be sufficient. Ultimately, he believed ‘the future lies with the attenuated vaccine given by mouth’.Footnote 86 Troublingly, this was not only because of its ease of administration and the intestinal immunity it produced, but also because he thought it possible that the ‘vaccine virus’ might inadvertantly spread to those choosing not to be vaccinated. Perhaps then ‘the wild virus should be virtually excluded from propagation and the chain of infection should be broken’.Footnote 87
Perhaps the aging Wilson’s concerns over what he referred to a few years later as ‘the hazards of immunisation’ reflected his coming of age in an era when many more infectious diseases were unavoidable realities of everyday life.Footnote 88 In lecture notes prepared just a few years before his passing, he recalled that over the course of his life he had ‘suffered from all the common infectious diseases of childhood and, later, from tuberculosis, malaria, dysentery and infectious hepatitis’. Despite this lifetime of infection, he remained ‘physically and mentally active’ and, at nearly ninety years of age, weighed ‘the same as I did in my late teens’.Footnote 89 Over sixty years had passed since a much younger man, recently returned from service in India and recovering from dysentery and malaria, had first ‘hitched his waggon’ to Topley’s star and begun observing epidemics in mice. Over those decades, however, the conceptual and policy implications of herd immunity had become increasingly entwined, and he seemed uneasy with the growing war on microbes that was in part a result of his own work. Towards the end of his life, the concept he had helped develop through rigorous observations of human and animal epidemics was becoming central to policies of disease elimination.
Predators and prey: modelling elimination
Just five years after Wilson passed away, British ecologist turned epidemiologist, Roy Anderson, delivered the eighth biennial Tansley Lecture before the British Ecological Society. Opening his talk, Anderson noted that in the 1960s and 1970s research in parasite ecology had emphasised ‘the prevailing concepts and fashions in the discipline of parasitology, as opposed to those in ecology’. In the 1970s, however, scientists more influenced by ecological modes of thought ‘began to influence thinking about the parasitic mode of life’. This resulted in a ‘convergence in the concepts employed by ecologists in thinking about the transmission and persistence of infectious agents in natural or managed plant and animal communities, and those employed by epidemiologists concerned with the study of infection and disease in human communities’. This convergence was important for the study and management of infectious diseases of humans, and the nature of ‘the parasitic mode of life’ suggested that ‘pathogens and hosts coevolve in a very dynamic way’. Discussing diseases from rabies in red foxes to measles and HIV among humans, Anderson considered influenza most illustrative of this dynamic. Echoing Andrewes and Burnet four decades earlier, he suggested that antigenic shift enabled influenza’s ‘persistence in a population with a high degree of herd immunity to earlier antigenic variants’.Footnote 90 Closing his talk, he reiterated that epidemiology had ‘much to learn from ecological research’ and that its hitherto ‘rather sterile statistical approach’ often precluded attending to ‘the dynamic interplay between populations of hosts and infectious agents’.Footnote 91
Born in 1947, Anderson completed his bachelor’s and PhD at Imperial College London, submitting a doctoral thesis on the ecology of helminth parasites of bream in 1971.Footnote 92 Afterwards, he began a postdoctoral fellowship in the Department of Biomathematics at Oxford where he worked with statistician Maurice Bartlett who explored the critical community size necessary for disease outbreaks.Footnote 93 During this time, he met ecologists on both sides of the Atlantic and began using computers to disentangle the complex factors driving equilibrium in ‘predator-prey’ relations.Footnote 94 In 1984, he became head of Imperial’s Department of Pure and Applied Biology, a position he held until leaving to take over the zoology department at Oxford University before subsequent events resulted in his return to Imperial in 2000.
Historian Hannah Gay places Anderson’s intellectual development in what she terms the ‘Silwood Circle’, an ‘all-male fraternity’ of ecologists coalescing at the Silwood Park campus of Imperial College who combined ‘experiment and field observation with a mathematically informed theoretical approach’.Footnote 95 During his time in the Silwood Circle, Anderson met Australian theoretical physicist turned mathematical ecologist, Robert May (1936–2020). Born and raised in Sydney, May, or ‘Bob’ as he was known to friends and colleagues, obtained the first graduate degree in theoretical physics in Australia, graduating in 1959.Footnote 96 Following posts at Harvard and Sydney, he spent fifteen years as professor of zoology at Princeton before relocating to Oxford’s Department of Zoology in 1988, where he held a joint Royal Society Professorship at Imperial. In 1995, he became Chief Scientific Advisor to the government of the UK and, in 2000, president of the Royal Society.Footnote 97 Anderson and May first met at a workshop on mathematical ecology held in York in the summer of 1973.Footnote 98 By the end of the decade, their growing friendship resulted in a lasting professional collaboration that reshaped the mathematical modelling of infectious diseases.
Anderson and May introduced an array of mathematical formulas for modelling human diseases, complicating the role of the host population in the evolutionary and epidemiological dynamics of disease from the perspective of the ‘predator-prey’ framing of ecology.Footnote 99 In 1978 and 1979, they coauthored a series of articles in Journal of Animal Ecology and Nature, laying the groundwork for further collaboration and introducing complex non-linear mathematical equations to understand the dynamic interplay between parasites and host populations as ‘a particular manifestation of the general predator-prey interaction’.Footnote 100 From this framework, the relation between host and parasite became one of two interacting populations, the predator-pathogen using the host population or ‘prey’ as a resource to sustain itself. They demonstrated how parasitic infections, for example, ‘can influence the growth rate of their host populations’.Footnote 101 Their analyses highlighted that stable states were not guaranteed outcomes of ecological systems and that the equilibrium of any predator-prey interaction might fluctuate according to a variety of parameters unique to each ecological niche. One major distinction applicable to infectious diseases was whether an infection was directly or indirectly transmitted. Directly transmitted infections such as respiratory viruses, for instance, often ‘require high host densities in order to persist’ and might therefore ‘be more commonly associated with animals that exhibit herd or shoaling behaviour’. Indirectly transmitted parasites such as malaria, however, could persist in low-population densities because of the parasite’s high rate of reproduction. Evidence in favour of this distinction came from ‘the abundance of directly transmitted viral and bacterial infections within modern human societies, large herds of ungulates, breeding colonies of sea birds, and the social insects’.Footnote 102 In the real world beyond what they often regarded as simplified and inflexible epidemiological models, host–parasite interactions were shaped by a multitude of variables and were constantly ‘in tension between the stabilizing and destabilizing elements’.Footnote 103
The pair soon began applying these theoretical insights to vaccination strategies, with the goal of driving pathogen populations below the densities required to sustain themselves in human communities. Despite the recent eradication of smallpox and substantial reductions in the burden of polio and diphtheria, in 1982, they expressed concern that ‘airborne infectious diseases […] remain endemic throughout most of the developed world’, raising fundamental questions about vaccination policy.Footnote 104 Part of the problem was that classical models informing immunisation strategies assumed a homogenously mixing population with ‘susceptible and infected individuals mingling like the molecules in an ideal gas’.Footnote 105 In reality, transmission dynamics were not entirely random, and much of their work sought to delineate the key factors in any specific host-parasite interaction in order to tip the balance in favour of the human population. One such variable was the age-structure of a population, which differed widely between high- and low-income nations. If individuals were vaccinated at an age above the average age of infection in that community, the disease would persist, irrespective of very high quantities of herd immunity.Footnote 106 Population density and rates of intermixing were of comparable importance, and Anderson and May believed that ‘in many low density rural areas, infections such as measles could not persist endemically […] without repeated introductions from nearby urban centers’.Footnote 107 Though they often wrote about herd immunity in relation to vaccination, they did not regard the concept as synonymous with disease elimination and were careful to refer to ‘sufficient levels of herd immunity for elimination’ and ‘very high levels of artificially induced herd immunity are required to eradicate diseases’.Footnote 108
At the centre of Anderson and May’s contributions was a rethinking of the nature of a host population. Counterintuitively, assuming a homogenous population could both overestimate the levels of herd immunity required to achieve elimination, while leaving pockets of susceptibles among whom the pathogen would continue to circulate. Assuming such a population as part of a measles elimination programme, for example, would require over 95% of that population’s members to be immune. This simplification, however, overlooked basic differences such as the disproportionate role of urban centres versus more sporadic transmission in less densely populated towns and rural areas. They doubted that measles could persist endemically in the second type of community without continual reintroductions from denser urban areas and, accordingly, optimal vaccination targets would vary across communities.Footnote 109 At the same time, in a study of the minority Bedouins of Israel, who they described as having a ‘relatively high young-old mixing rate, due to the special nature of their communal life’, they demonstrated that lower vaccination rates for measles sustained the pathogen.Footnote 110 In general, however, recognising the spatial heterogeneity of populations meant that an ‘optimal vaccination policy’ could be developed that would require lower overall rates of immunisation while distributing immunity sufficiently to achieve elimination.
In their mathematical framing, ‘equilibrium’ denoted the point at which the host-parasite association reached ‘unity’, or put differently, when the rate of transmission was stable – each infectious individual, on average, infecting one other person, that is, ‘R = 1’. Far from disregarding the distribution of infection-acquired immunity in a population, it was both the mechanism to be emulated and itself a contributing variable, as any host population already in contact with an infectious agent would have acquired some degree of immunity, reducing its rate of transmission – ‘R’– below its maximum – ‘R 0’. Their most important contribution was modelling the role of population heterogeneity. They demonstrated that assuming a homogenous population in which all individuals were equally likely to interact with one another like ‘molecules in an ideal gas’ hindered efforts to undermine the resources of a pathogen population.Footnote 111 While Topley had referred to the role of a population’s ‘herd structure’ in shaping transmission dynamics, the subsequent rise of computers and more advanced knowledge about the immune system gave the concept of spatial heterogeneity direct application to immunisation strategies.Footnote 112 In 1985, Anderson and May drew attention to the need for elimination programmes to account for such factors as
the precise course of infection within an individual, the demography of the host population, the duration of acquired immunity and maternally derived protection, age-related changes in the degree and intimacy of contacts among people, and the prevailing levels of genetic, spatial and behavioural heterogeneity in susceptibility/resistance to infection.Footnote 113
During the 1980s and 1990s, the global expansion of vaccination pushed mathematical epidemiologists to create models that could inform disease elimination strategies.Footnote 114 By the early 1990s, vaccination rates for measles, mumps, and rubella were increasing in the UK, heralding the prospect of eliminating these diseases. The Daily Mail reported in May 1990, for instance, that measles notifications had fallen from 13,930 for the same period the previous year to 5,753, suggesting that ‘[m]easles could be a disease of the past’.Footnote 115
The difficulty of translating these insights into immunisation policy was made apparent in 1982, when Anderson and May weighed in on contrasting vaccination strategies against rubella between the USA and the UK. In the USA, both boys and girls were vaccinated at preschool age, and the incidence of rubella had declined. In the UK at the time, however, ‘the adopted policy is to vaccinate girls (and only girls) between 11 and 15 years of age’ alongside ‘selective postpartum vaccination in women not found to have antibodies during antenatal care’.Footnote 116 The policy reflected different approaches to maintaining herd immunity. The strategy in the USA was dependent upon maintaining 80%–85% vaccination uptake among boys and girls while in the UK, Australia, and much of Europe only girls were vaccinated at a later age, resulting in ‘natural immunity among children less than 12 years old and continued circulation of the virus in the community, allowing immunity to be boosted’. ‘This policy should be more effective at preventing [Congenital Rubella Syndrome] if the proportion of persons in the population who have been vaccinated is low (less than 60 percent)’, noted an epidemiology textbook in 1976.Footnote 117
As historian Leslie Reagan points out, universal rubella vaccination was in part controversial because the vaccine was the first to be given to people (boys) who were not themselves at risk of the disease.Footnote 118 The policy in the UK, Anderson and May noted, ‘protected the individuals most at risk’ but seemed to ‘have little impact on the overall incidence of rubella in Britain’. Furthermore, elimination was ‘simply impossible’ if the average age of vaccination exceeded the average age of infection, or, in their notation, if ‘V > A’. Where elimination was the goal, an optimal policy should aim to keep ‘V as low as possible’ while also recognising ‘the duration of protection provided by maternal antibodies’.Footnote 119 Negotiating the shifting dynamics of vaccination and herd immunity could be tricky, however, as Anderson soon discovered when a reporter at The Guardian newspaper quoted him as describing the policy in the UK of ‘vaccinating at an age later than the age at which they get it’ as ‘appalling’.Footnote 120 Writing to May, at that time still at Princeton, Anderson was ‘horrified to discover’ that the reporter used his words ‘to suggest that the current Department of Health policy is incorrect’. He was ‘extremely irritated with the Guardian man’, noting ‘how bloody careful one has to be in talking to such people’.Footnote 121 There was ‘no point in getting upset about such things’, May wrote back, as it was his ‘experience that newspapers always get everything wrong, in such a way as to make one look a fool’.Footnote 122 The episode captured the growing entanglement of herd immunity in competing scientific and public health practices for managing infectious diseases.Footnote 123
Anderson and May’s work had a profound impact on a generation of infectious disease epidemiologists and mathematical modellers. In 1991, they published their influential textbook, Infectious Diseases of Humans: Dynamics and Control. Footnote 124 Around this time, they also procured funding from the Wellcome Trust to establish the Wellcome Trust Centre for the Epidemiology of Infectious Disease (WTCEID) at Oxford University.Footnote 125 Opening in 1993, the centre would train dozens of researchers who soon dispersed around the world, including several prominent commentators and government advisors on the COVID-19 pandemic.Footnote 126 Only seven years after it opened, however, following personal conflicts for which he apologised and accusations of financial conflicts of interest, Anderson resigned from Oxford, relocating with many of his staff and students to head Imperial’s Department of Infectious Disease Epidemiology.Footnote 127 As the government of the UK responded to outbreaks of foot-and-mouth disease, bird flu, and the 2009 influenza pandemic, Anderson and colleagues produced models to inform the public health responses.Footnote 128 When fears about the severity of a novel strain of H1N1 influenza intensified in 2009, epidemiologists from Imperial and elsewhere scrambled to predict the scale of the outbreak. Fortunately, their estimates of a fatality rate between 0.3% and 1.8% turned out to be greatly inflated, subsequent revisions lowering the figure to 0.091%.Footnote 129 As it turned out, their models had failed to account for the populations’ pre-existing immunity.Footnote 130
In recent decades, references to herd immunity as the percentage of immune members of a population required for disease elimination have become frequent, at times blurring the distinction between an observable mechanism undergirding the fluctuating incidence of an infection and a public health target to be cultivated through vaccination in pursuit of disease elimination, an ambiguity that Anderson and May carefully avoided.Footnote 131 Nonetheless, by 1993 epidemiologist Paul Fine, who had been a colleague of Graham Wilson in the early 1980s, commented that along with ‘the growth of interest in herd immunity’ which had accompanied growing vaccination coverage ‘there has been a proliferation of views of what it means’.Footnote 132 This article has argued that this proliferation of views reflects the growing entanglement of the concept with immunological technologies for inducing immunity and ultimately for attempting to eliminate certain infectious diseases.
Conclusion: do we have herd immunity to COVID-19?
Following the fraught discussions of infection-acquired herd immunity early in the COVID-19 pandemic, many came to regard the term as largely irrelevant in the absence of a vaccine. In October 2020, however, the debate resurfaced after three epidemiologists published an online document known as the Great Barrington Declaration (GBD). Motivated by ‘grave concerns about the damaging physical and mental health impacts of the prevailing COVID-19 policies’ they argued that ‘all populations will eventually reach herd immunity – i.e. the point at which the rate of new infections is stable – and that this can be assisted by (but is not dependent upon) a vaccine’. Criticising society wide attempts to suppress transmission, they argued in favour of an alternative strategy of ‘focused protection’ that ‘balances the risks and benefits of reaching herd immunity’ by allowing ‘those who are at minimal risk of death to live their lives normally to build up immunity to the virus through natural infection, while better protecting those who are at highest risk’.Footnote 133 Other scientists, however, soon dismissed the proposal as ‘a dangerous fallacy unsupported by scientific evidence’ and argued that ‘measures that suppress and control transmission need to be implemented widely’ to ‘effectively suppress SARS-CoV-2 infections to low levels that allow rapid detection of localised outbreaks’. They were adamant that ‘[a]ny pandemic management strategy relying upon immunity from natural infections for COVID-19 is flawed’.Footnote 134 One scientist insisted that ‘[h]erd immunity has never been achieved through naturally acquired infections and is only possible at global population scale through mass immunization’.Footnote 135 Yet, others argued that it was mistaken to claim ‘that we have never naturally achieved herd immunity to any pathogen’ and that this belief arises from ‘the incorrect assumption that herd immunity implies elimination’.Footnote 136 Even the World Health Organization seemed unsure about the precise meaning of the term, changing the definition on its website to exclude reference to infection-acquired immunity, only to reinstate it some weeks later while emphasising that the organisation ‘supports achieving ‘herd immunity’ through vaccination’.Footnote 137
At times, these debates were directly connected to the legacy of Anderson and May. One of their former students and colleagues, Oxford University epidemiologist Sunetra Gupta, was an author of the GBD. Gupta first met May while she was a sophomore in physics at Princeton University in the 1980s and later pursued postgraduate studies with the pair in the UK.Footnote 138 Perhaps revealing the influence of their ecological thought, she viewed the COVID-19 pandemic as ‘an ecological relationship that we have to manage between ourselves and the virus’, and believed herd immunity offered ‘a way of living with this virus’ like we do with influenza such that ‘through herd immunity, the levels of infection are kept to as low a level as we can get’. Echoing Topley’s reflections almost a century earlier, she suggested that while ‘international travel facilitates the entrance of contagion’ it also ‘brings immunity’ and in trying to contain COVID-19 – hopelessly, in her opinion – we were ‘closing ourselves off not just to the disease, but to other aspects of being human’. For Gupta and many others, herd immunity was not a call for sacrificing those most at risk and did not imply permanent or impenetrable protection against infection. Rather, it offered an epidemiological mechanism for thinking ‘at a communitarian level’.Footnote 139 Her early work modelling the pandemic sought to account for pre-existing immunity arising from infection with endemic coronaviruses, a factor she and her colleagues believed indicated that populations already had some degree of immunity.Footnote 140 Another former student of Anderson and May, however, Harvard epidemiologist Marc Lipsitch, coauthored the response to the GBD that dismissed it as ‘a dangerous fallacy unsupported by scientific evidence’.Footnote 141 He argued that the contribution of any pre-existing immunity was ‘already “baked in” to the data’ and therefore ‘should not change our estimates of R 0 or the herd immunity threshold in any given population’.Footnote 142 Despite their training with Anderson and May, Gupta and Lipsitch strongly disagreed over the role of pre-existing immunity, the feasibility and harms of lockdowns, and the wisdom of leveraging infection-acquired herd immunity to manage the pandemic.
During the pandemic, much of the complexity of these debates was lost and many came to assume that calls for relying upon herd immunity in the absence of vaccination were due to ignorance or even callous disregard on the part of some scientists.Footnote 143 In contrast, I have argued that the debate reflects a long-running tension between observing herd immunity as an inherent feature of infectious diseases and cultivating it to control them. Over the course of its history, herd immunity has often captured an impulse to reach an accommodation with infectious diseases – to attain an ‘unstable equilibrium’ between human hosts and their pathogens.Footnote 144 I pursued this thinking through Topley and Wilson’s early observations of experimental mouse epidemics, the mid-century reflections of disease ecologists such as Burnet and Dubos, and an aging Wilson’s worries about the possible unintended impacts of attempting to banish polio. Just as it might imply accommodating ourselves to the persistent presence of infectious diseases, the growing ability to cultivate herd immunity through a variety of immunological technologies has advanced a different interpretation of the concept: as a public health target to be reached and maintained through immunisation. Already by the mid-1920s, Dudley’s reference to an ‘immunity index’ as the ratio of immunes to susceptibles that might be enhanced through serums and vaccines revealed how such technologies informed alternative visions of herd immunity.Footnote 145 As immunisation expanded in the postwar era, the concept came to have what epidemiologist Paul Fine referred to as a ‘special aura’ in its ‘apparent provision of a means to eliminate totally some infectious diseases’.Footnote 146 As it became widely realised in early 2021 that neither infection nor vaccination created lasting immunity against (re)infection with SARS-CoV-2, many came to dismiss herd immunity as ‘probably impossible’.Footnote 147 Yet as another former colleague of Anderson and May, epidemiologist Mark Woolhouse, notes: ‘believing’ in herd immunity is ‘like being asked if you believe in tides. Like the tide, herd immunity happens whether you believe in it or not’.Footnote 148
Acknowledgements
The author thanks Oxford University’s Centre for the History of Science, Medicine and Technology and Erica Charters in particular for comments on an earlier draft of the article. He also wishes to thank the organisers – Lukas Engelmann and John Nott – and participants of the workshop ‘Data and Disease in Historical Perspective’ hosted at the University of Edinburgh in November 2023.
Competing interests
None.
Funding
The author gratefully acknowledges funding from the Swiss National Science Foundation (SNSF) through a Postdoc. Mobility Fellowship (grant number 214512).