The Spanish influenza pandemic seen through the BMJ’s eyes: observations and unanswered questionsBMJ 2009; 339 doi: https://doi.org/10.1136/bmj.b5313 (Published 16 December 2009) Cite this as: BMJ 2009;339:b5313
- Tom Jefferson, researcher,
- Eliana Ferroni, researcher
- Correspondence to: T Jefferson
The great Spanish influenza of 1918-9 is perhaps the best known of all pandemics. A vast number of books have been written on the topic, and “Spanish flu” still represents a rich topic of discussion and research.1 Although many scientific questions posed at the time have been answered by nearly a century of subsequent discoveries, some puzzles remain—such as the reason for the high case fatality rate (>2.5% according to some estimates),2 the high incidence in young adults,3 and the role played by the first world war and its aftermath.4 At the time, the aetiology was not known and the disease’s mode of transmission could only be hypothesised by careful observation. In addition the effects of preventive measures remained largely unevaluated. A high number of descriptions of the event seem to be based on secondary sources—that is, they are not direct eyewitness accounts—although recently efforts to go back to primary sources have been made.4
Reliance on secondary sources is a potential source of selection bias: the reader gets a selected view of events, and some of the contemporary observations and explanations are lost. Interpretation of events and actions of a bygone era from a modern perspective is another perilous activity.
We exploited the opportunity to consult the newly digitised BMJ archives to carry out a review of what was published at the time. We tried to look at the pandemic through the eyes of contemporary BMJ contributors and readers and give them their voice back. We chose the Spanish influenza pandemic because we believed some of the observations and issues raised at the time may still be relevant to the events of today, although some have been forgotten.
We carried out a search of the BMJ archives using the keywords “influenza,” “flu,” “pandemic,” and “Spanish influenza” for the period 1918 to 1924. We read the content of each article, looking for unusual or forgotten observations and still unresolved questions. We excluded well known facts (such as clinical descriptions and presentation of the disease), and we did not carry out a detailed analysis of observations that are explicable with today’s knowledge (such as the contemporary conclusion of the presence of “filterable virus,” which was then invisible and unidentified).
We grouped the findings by broad headings covering the capricious nature of the spread of disease and its relation to climatic conditions, the possible multifactorial nature of the high case fatality rate, the unusual effects of poison gas on munitions workers’ health, and possible causality and preventive measures.
We did not carry out a systematic review but selected items that we consider still highly relevant today but largely forgotten. We make no claims as to the completeness or fairness of our selection, but we have listed our search yield and sources in the appendix on bmj.com to encourage readers to consult the original records.
We found 55 articles of varying nature—including studies, reports, and letters—published between July 1918 and October 1920.
Spread of disease and its relation to climatic conditions
One of the least studied aspects of the pandemic was the appearance of synchronous cases or clusters of cases either apparently unconnected or a great distance apart. This had also been a feature of the 1789 and 1889 pandemics (and would be again in 1957 and 1968).5
Dr Andrew Garvie, a general practitioner in Halifax describes the conundrum: “But why the first case in the household was, on the average, more serious than the sporadic, and why the ‘clumping’ should occur, is difficult to understand. Casual observation might lead one to suppose that the spread was due to actual contagion from one house to another. At first I regarded it simply as due to ‘neighbourliness,’ but later on became convinced that this could only be a partial explanation of the spread. In many of the households affected in a ‘clump,’ a suggestion of being in any of the other affected houses was absolutely denied. It will further be noticed that within one particular ‘clump’ two or three houses commenced on the same date, and further, owing to the general fear of the epidemic, spread by newspaper reports and other methods, if the epidemic was known to be present in a house, the house was usually shunned by neighbours. In many cases the houses were not in direct contact but separated by the breadth of the street or by garden walls ... but why people within small radii of one another, of all ages, of different occupations, not coming in actual contact with one another, should develop synchronous attacks, still remains a mystery to me.”6
Dr Major Greenwood, later first professor of epidemiology and vital statistics at the London School of Hygiene and Tropical Medicine, writing arguably the most complete description of the UK pandemic in the autumn of 1919, had the following explanation for the observation: “A mass attack, indeed, forms an invariable link in the chain of events, but scattered individual cases are antecedent. This observation removes the most formidable objection to a belief that influenza is contagious, and it is easy to understand why we discover no confirmation in historical records. In such epidemic diseases as plague the preceding sporadic cases are recorded because their high fatality leads to illusion in bills of mortality; but in consequence of the very low fatality of primary epidemic influenza early mortality records are wanting. The mass phenomenon strikes the imagination of the recorder and an illusion of suddenness and simultaneity is produced.”7
In other words ascertainment bias was the most likely explanation for the observation.
Greenwood and other BMJ contributors (such as Dr Mercer Watson writing in 19198) dedicated a lot of time to describing the prevailing atmospheric conditions in relation to the subsequent waves of the pandemic. This may be a relic of the old miasma versus contagion debate of previous centuries.
High case fatality rate
A hallmark of the Spanish influenza pandemic was its high case fatality rate. Greenwood described its causes in the spring and summer of 1918 period as “excessive mortality being mainly due to the accident of season, aided by the special circumstances of overcrowding and fuel shortage which are due to the war. In a word, this is not essentially a war epidemic.”7 The role of wartime shortages and troop concentrations and movements has always been a moot point with historians and epidemiologists. Those who refute these as factors cite the example of whole countries or isolated communities not involved in the war being virtually wiped out—that is, being fatal exceptions to the theory.9 However, a closer look at the description of conditions by BMJ contributors shows the importance of studying context in the devastation.
Surgeon Lieutenant Francis Temple Grey in charge of the Samoa relief expedition is cited by the BMJ in 1918 explaining the high native mortality compared with the white population: “The incidence among the natives was 80 per cent. Out of a population of 36,405 the deaths numbered 7,264.” Surgeon Lieutenant Grey attributes the high mortality partly to the fact that natives, although apparently of fine physique, have generally a poor chest expansion, and to their habits. “The native house has a raised floor of coral and lava pebbles, a thatched roof supported on poles, and no walls, but at the beginning of the epidemic, when a native fell ill he lay down in his hut, and his family, having pulled down the blinds, which are usually lowered only in wet weather, lay down with him in sympathy. When the fever was at its height, on the third day, the natives cast off their clothes, pulled the blinds up, and many of the men went into the sea to cool themselves. This was often followed by pneumonia, although, except in children, few cases, even with precautions, escaped bronchopneumonia. At the height of the epidemic many lives were lost owing to want of food consequent on the cessation of its collection. On December 8th, 1918, food collecting was resumed, and the decline of the epidemic was popularly dated from that.… Among the whites the incidence was put at 60 per cent, and the case mortality at 2 per cent.”10
Writing in 1920, Dr A H Macklin, formerly of the Imperial Trans-Antarctic Expedition, helps us to understand some of the reasons behind the often quoted devastation in Lapland: “The Laplanders had a very thorough if unsympathetic way of dealing with their cases. The settlements were composed of wooden huts, small, but generally well made and warm. A common type consisted of but one room used by the family for all purposes. Better class Lapps had better huts, with two or three rooms. In each settlement one of the single-room huts had been set apart, and into this each case of sickness as it arose was unceremoniously pushed; and none were permitted to return to their own huts until completely recovered. Whilst there they received practically no attention, and no healthy person ever entered to attend to their wants. Occasionally a bowl of water or reindeer milk was hastily passed in at the door, or a huge chunk of reindeer meat thrown in, uncooked and uncarved… Constipation was a constant factor, and many cases had not had their bowels opened since the onset of their illness—in some cases seven to ten days. Others had voided urine and faeces just as they lay. In some huts those of the patients who had passed the worst stages of their illness assisted and looked after those more acutely ill.”11
The effects of poison gas on munitions workers’ health
By 1918 the use of poison gas, first introduced in 1915, was routine when bombarding enemy trenches. A great number of workers were assigned by the Ministry of Munitions to produce shells filled with the noxious substances. Numerous BMJ authors remark on the apparent protection afforded by all types of gas production except phosgene (table 1⇓)
Dr Gregor, writing in 1919, informs us that it was apparently “custom to take children suffering from whooping cough to the nearest gasworks and expose them to the fumes emanating from the oxide of the iron purifiers during the process of cleansing, and the parents of these children firmly believe that by doing so the attack is much mitigated”.14 He goes on to report a comparative cohort analysis of the incidence of influenza in populations with different exposure levels to gases (table 2⇓). Assuming that uniformed personnel were the fittest population, the difference in incidence is striking.
Causality and prevention
By 1919 Dr Benjafield of the British Expeditionary Force in Egypt thought that “In this case it is more than possible that the primary etiological agent is ultramicroscopic and is a filterable virus, which renders the individuals infected hypersensitive to organisms which formerly only possessed a low degree of pathogenicity.”15 This was also the conclusion of Rose Bradford et al of the British armies in France.16
If this hypothesis were true, experimental vaccines against a collection of bacteria would be expected to achieve at least partial effectiveness. This is exactly what Sir William Leishman experimented with, using it on troops during the closing stages of the war.17 The vaccine contained a mixture of Bacillus influenzae (Haemophilus influenzae type b), streptococci, and pneumococci. Leishman’s BMJ paper does not report details of methods, but his observation of effectiveness against influenza, pulmonary complications, and deaths is upheld by our reanalyses. For example, the odds ratio for pulmonary complications was 0.12 (95% confidence interval 0.08 to 0.18) (table 3⇓).
The views on the effectiveness of other interventions such as masks and distancing were varied.
We found several articles reporting unusual observations. Even today many of these have no explanation. We feel that the observations should first be corroborated and perhaps explored further. The idea that exposure to gaseous fumes (even potentially toxic) prevented or ameliorated respiratory disease seemed established at the time. Today, however, that is flatly contradicted by unreferenced statements in publicly available documents.18
The causes of the high case fatality rate are still unclear, but modern research suggests that the pandemic was a lot more than just a “one germ-one disease” affair 19 This view is supported by the apparent success of antibacterial vaccines against influenza and its complications (with all the probable methodological shortcomings of Leishman’s study).
Agents other than the influenza virus probably played a part. Above all, the environmental explanations of the high Samoan and Lapp mortality rates indicate the peril of generalising across contexts and simplifying causation models. The origin and spread of the pandemic are also far from clear, but the repeated reporting of multiple synchronous foci should be investigated in today’s pandemic. This may give us an indication of the means of seeding, trigger, and possible spread of respiratory viruses.
Cite this as: BMJ 2009;339:b5313
We acknowledge the keen interest and accurate reporting of our predecessors, who kept their nerve in the midst of an ugly pandemic and the carnage of war.
Competing interests: None declared.