Understanding consecutive epidemic waves via the Spanish flu pandemic in Madrid and Spain

Author: Laura Cilek (CSIC) 

As the coronavirus COVID-19 pandemic continues to rage across the world, many geographies are beginning to experience new waves of a once-under control virus.  Second, third, and even fourth epidemic waves within a single pandemic event are not a new phenomenon to researchers.  In fact, even before the peak of the first wave of the current pandemic, scientists began to warn of new successive outbreaks.  But what can history teach us about the dangers and impact of these later waves?  Today, we look at the 1918 Spanish Influenza pandemic, particularly in Spain and Madrid, in order to understand how individual waves contributed to the overall evolution and outcome of the pandemic.


While the exact origin of the 1918 Spanish Influenza pandemic is uncertain, two things are believed to be true: 1) the initial strains of H1N1 influenza, which led to the deaths of more than 50 million people, were likely circulating before 1918, and 2) the pandemic did not start in Spain.  While initial outbreaks were recorded in the United States and Europe during the spring of 1918, one theory of the virus’s origin places it as early as 1916 (others, even earlier) in British and French military encampments, where living conditions were poor, and soldiers and animals lived and interacted in close proximity.  The first widely reported outbreaks (and where our story starts) came from Spain in May of 1918, where the virus became known as “The Naples Soldier,” a song from a popular opera, because the song, as the virus, was catchy.’  As a neutral country in WWI, and due to press embargoes that disallowed reporting of the virus on the front lines in other countries, the rest of the world, upon hearing the news from Spain about these new outbreaks, christened the virus “The Spanish Flu,” thereby establishing the nomenclature by which we still know the pandemic today.


Many places in both Spain and the rest of the world did not experience a first wave at all.  In other locations where one did occur, especially major cities, the herald wave (in comparison of what was to come) was often hardly noticeable in mortality trends. In this regard, the city of Madrid is different.  Perhaps fueled by large crowds from all across the city intermingling at the San Isidro (patron saint of the city) festivals in mid May, transmission skyrocketed, leading to a city full of infected individuals (including the king himself!), and while most individuals recovered without any problems, the sheer amount of the population infected led to a very noticeable influenza-related mortality peak in June, with an estimated R in the municipality of around 3 and almost 1500 excess deaths.


The fall of 1918 brought about a new wave of epidemic influenza, featuring a strain of virus ten times more deadly that seasonal flu, which wreaked havoc on the world, Spain, and Madrid. This time, the virus entered Spain’s northeast corner via France, and the progression of the pandemic can clearly be seen spreading west and south through the country, leaving no province unaffected.  In Spain, the age structure of excess mortality, while unique to the Spanish flu, followed a similar pattern to that observed in the rest of the world.  In normal flu seasons, the most affected are the oldest and weakest age groups.  However, the 1918-19 pandemic created vast increases in the mortality of young adults, particularly those aged 25-30.  This unique pattern and the mechanisms behind its manifestation are still studied by researchers today, but some evidence points towards exposure to the virus that caused the pandemic beginning in 1889 during a critical development period impacting the ability of individuals to effectively fight the virus.  This in turn would have left those weakened from the flu defenseless in a battle against subsequent bronchitis and pneumonia, which was actually the primary cause of death during the pandemic. Ultimately, while the eldest, youngest, and weakest all suffered heightened mortality, this distinct age specific mortality pattern, in which the “strongest” age groups suffered the highest relative excess, remains unique to the Spanish flu.  In the fall wave, there is a clear increase in excess mortality (both absolute and excess) among these young adults.  However, this peak is much smaller than observed in other geographic areas.  Moreover, the overall mortality peak in the city is smaller than in nearly all other locations in Spain.  Could these peculiarities be the result of the strong herald wave a few months before?

Following the robust fall wave in Spain, a third wave again ravaged much—but not all— of the country, including Madrid.  Generally speaking, in areas that experienced a winter wave of epidemic influenza, excess mortality reached extreme levels, and in some cases were higher than that of the fall wave. However, once again, the mortality peak in Madrid never reached the levels in other affected provinces. Instead, the apex of the mortality peak, similar to the levels observed during the fall wave, reached roughly the same levels of that of the normal seasonal peak. That is to say, during the period of September to April 1918-19, there are three peaks in the time series of mortality of roughly the same height (unadjusted for seasonal variation). Thus, in Madrid, the fall and winter waves, which represent the bulk of overall mortality in the pandemic throughout the city, Spain, and the world, are often studied as a single excess mortality event.   Again, the city diverges from what may be perceived as characteristic of the Spanish influenza pandemic in terms of both  the ages and levels of peak excess mortality.   Why?

Most research on the Spanish influenza pandemic covers the three waves outlined above, focusing on differences in excess mortality by age, sex, geographic, socioeconomic, and occupational groups, then attempting to better understand the mechanisms behind these differences. Much less attention is paid to the final echo wave(s), which occurred in the years following the last winter wave, though not always at the same time.  For example, while Spain experienced a powerful echo wave in the winter of 1919-20 (and another, more subdued one the following winter), the corresponding fourth peak in Scotland happened one year later, in 1920-21.  These echo waves differed greatly from their antecedents in that the age structure of mortality no longer reflected the unique peaks in young adults but rather, excess deaths were concentrated in the youngest ages.  That is, those without previous exposure—children born following the last wave of winter 1919—lacked the antibodies that the rest of the population had acquired over the last two years.

So, having recounted the general narrative of the Spanish flu pandemic in Spain and Madrid, we can now contextualize the consequences of successive waves within the overall mortality impact of the pandemic.  The strong herald wave in Madrid appears to perhaps have dampened some of the potential subsequent mortality impact in the protracted fall and winter wave from September, especially compared to other regions of the country which did not experience a wave.  Yet, while perhaps the herald wave left some parts of the population protected, continued exposure over months, as in the case of the cumulative time period of the herald, fall, and winter waves in Madrid, can also lead to increased deaths overall, especially in particularly vulnerable groups.  For example, in the city of Madrid, neighborhoods with the highest mortality rates in the pre-pandemic period had the highest excess mortality rates in each wave of the pandemic.  That is to say, the virus had a prolonged impact on the mortality rates in the city, and those in areas with poor health and high mortality before the pandemic bore the greatest burden during it as well, even as a relative measure to non-pandemic times.  

It is natural to try and compare the mortality patterns of Spanish Influenza with that of the COVID-19 outbreaks.  These analyses can help contextualize the current events with what has happened in the past so that we can better understand how the immediate future of the outbreaks may evolve.  To facilitate this, mortality patterns according to age and sex for all of Spain in 2020 are pasted below next to those of the city of Madrid in the Spanish flu pandemic.

Here, we see clearly divergent age-specific mortality patterns in the pandemic of 1918 and the first wave of 2020.  The waves depicting Spanish influenza pandemic encompass the city of Madrid only, and show, in all waves, increased mortality in all age groups, especially young and middle-aged adults.  When considering absolute excess, there is a slight bump during the fall waves in adults, but this bump is slightly higher in the ages concentrated around 25-30.  Why do the patterns of relative and absolute mortality differ?  This is simple—the underlying mortality rates of the eldest ages are higher, so it takes a much larger amount of overall excess mortality to increase the relative rate.  The opposite pattern hold true for the youngest ages—lower underlying rates means that lower amounts of excess deaths can still equal higher relative excess. 

The first wave of the 2020 pandemic (presented here as all of Spain, not only the city of Madrid) reveals starkly different mortality patterns that that of the Spanish flu pandemic.  Of note, the overall mortality excess rates are low—lower, in fact, than any of the Spanish flu waves.  This is due to not only an increase in medical technology and capacity, but additional preventative methods taken an differences in lifestyles between 1918 and 2020.  It is also possible the overall virulence of the virus also plays a role in these mortality differences.  Unsurprisingly, the excess mortality is concentrated among the oldest—those about 70.  Nonetheless, there is slight excess in middle- and late-middle aged adults; while this is not analyzed here, we speculate that these may be related to higher risk individuals with  preexisting conditions.  Children and young adults, especially males, have lower mortality relative to normal levels—this may be a relic of the strong lockdown measures put in place, which thereby eliminated the risk of deaths to accidents and other external causes, the largest killers in these ages.

Overall, evidence from the Spanish flu pandemic shows the virulence of and exposure to the virus may well play a role in the overall mortality outcome, perhaps providing some population level protection in later, more virulent waves.  Yet the evidence to date from the first wave of the COVID-19 pandemic in Spain indicated that medical advances and prevention techniques play a much larger role in tempering potential excess mortality. In addition,  individual- and group- level dynamics and susceptibilities must also be recognized and considered when analyzing the impact and future prevention strategies of a pandemic so that the most vulnerable may also be protected.  This is evident among the highest risk individuals (especially the elderly and those with preexisting conditions) during the current pandemic, who continue to be hospitalized and die at much higher rates than those outside the risk zones.  As the second (European) wave of the current pandemic continues to rise, it is imperative to remember how past consecutive waves affected overall excess mortality in order to mitigate the potential mortality impact in the current pandemic.

Figure caption:

Age-specific relative (observed deaths/expected) (L) and absolute (observed – expected) (R) excess rates for the city of Madrid during the 1918-29 pandemic compared to that of all of Spain during the first wave of the COVID-19 pandemic.  Madrid data comes from death records between 1917 and 1920, and Spain-wide data from the Instituto Nacional de Estadística (INE).