Covid-19, herd immunity and harvesting effect?

Conclusions in short

Herd immunity from covid-19 infections significantly reduced further covid-19 deaths.  Where 45 – 60 % of the population had been infected the reduction exceeded 90%.  This result, however, applies in conditions where physical contacts are restricted, like they were in Italy in October – December 2020, and no new resistant variants played a role.

Signs of the decay in herd immunity could be seen at 10 – 12 months.

Herd immunity via infecting 60 % of the population would cause 600 – 800 excess deaths per 100 000, plus a severe burden of disease for the population and society.  No democratic society would tolerate nor could its existing health care service manage the consequences.  Luckily, this is a theoretical scenario only, because vaccination will provide a more reliable, affordable and acceptable path to herd immunity.

Harvesting effect, assumption that covid-19 mostly kills old and frail individuals, who would otherwise die only weeks or months later, gets zero support from the Italian mortality statistics.  The 50…500 % excess deaths in March – May were not followed by any mortality dip in the summer, when the covid-19 cases and deaths remained low.

 

Prelude – How did we get here

A year ago evidence had already spread from Wuhan, China, that covid-19 is a deadly disease [albeit with two orders of magnitude disagreement about how lethal] and that its unrestricted expansion into a pandemic would exert worldwide health, social and economic disruption beyond anything experienced in the recent decades.  The virus, SARS-CoV-2 had already made inroads to dozens of countries in all continents, and the first European city to feel its wrath was Bergamo in Lombardy, Italy.

Many countries in East-Asia decided to suppress the epidemic with the aim of eliminating it like SARS in 2002 – 3 [East-Asian approach].  In contrast the default policy in the most lax or self-confident Western countries was to prep up the annual flu epidemic practices, reserve capacity in the hospitals and in the absence of any vaccine essentially let the epidemic happen without too much control.  In the more stringent European countries the plan was to flatten the curve, i.e., manage the spread of the virus within the population and to channel it around the more vulnerable populations so that the health care facilities could tackle the inevitable load and the epidemic would generate herd immunity within the population before autumn or winter [European approach].

Critical voices were sounded not only about the probability of success in managing an infection that had already shown >10 % lethality for large segments of the population, but also about the very ethics of it.  Instead, application of the East-Asian suppression/elimination approach, already in effect in China, South-Korea, Taiwan and Singapore, were demanded.  Those policies, again, were criticized as being impossible to maintain, inevitably leading to closed countries and regions – large prison camps with no exit in sight – because SARS-CoV-2 had already spread too far to be eliminated from the face of the Earth like SARS-CoV-1.  Before last Autumn little promise existed for capabilities to suppress/eliminate the covid-19 pandemic by effective vaccination within the next few years, or to treat it affordably and with a comforting degree of success.

Although much new knowledge, social and individual experience has accumulated in the past 12 months, only one true game changer has emerged, the vaccines.  The human and social cost of not preventing the virus from spreading through the population turned intolerable and has led to humanitarian disasters in, e.g., Bergamo/Italy, New York/USA, Manaus in Amazonas and now all of Brazil, Peru, and countless smaller and less publicized epicentres.  The pandemic has already taken and will take much longer to get under control than assumed – not too different from the Spanish flu pandemic of 1918-19.

The East-Asian approach was adopted also in Australia and New Zealand and compared to the European and American approaches it has resulted in only 1/100 of the death and 1/10 of the economic consequences.  These countries have on one hand isolated themselves with strict border controls and mandatory quarantines for all immigrating individuals.  On the other hand, within their borders they have enjoyed much shorter, more localised lockdowns and fewer, if any social and business restrictions for most of the time since March 2020.  In a way these countries have developed into the predicted “prison camps”.  Note, however, that (i) getting into, not out of them is restricted, (ii) the restrictions concerning personal freedoms, social and business life in Taiwan, New Zealand etc. are minimal in comparison to most European countries, which are merely speculating the possibilities of lifting or strenghening some of their current lockdown measures, and (iii) the vaccines will help out the East-Asian, European and American countries just the same.  New Zealand and Australia have now agreed to start on 19 April a Trans-Tasman travel bubble between the two countries.

The Fifth Amendment of the US Constitution defines life, liberty and property as unalienable human rights.  This is a well prioritised list and hard to disagree with.  From what we know today, the East-Asian approach to the pandemic has protected each of these three human rights incredibly much better than the European approach, or American for that matter.

In many European countries the lockdowns and travel restrictions of March … June 2020, aided by the seasonality of the virus itself, achieved low or near zero covid-19 transmission levels in June – August.  Experts warned about the second wave awaiting in the autumn, but large segments of the population felt that the pandemic is all but over, political leaders did not want to spoil the mood kept on  lifting remaining restrictions, and much of the vacationing and holiday travel resumed.  The pandemic began to grow already in August and when the vacationers went back to work and children back to school the second wave was raising and it grew in many countries bigger than the first one.  The golden opportunity that had been provided by almost choking the pandemic in July, was lost by short sighted and wishful European decision makers and impatient populations.  Few were willing to face the reality.  Decision makers dragged their feet, wishing and promising “It’s going to disappear It is disappearing” without the  unpopular restrictions.  Outspoken citizens were tired of being told what to do and not do, let alone being prevented from moving about and entertaining themselves as they pleased and they began to gather for demonstrations about their assumed rights on the streets.  Alternative realities and conspiracy theories fuelled denialism.  Finally, promising news about the relief waiting just around the corner in the form of new vaccines were all too eagerly absorbed by the more facts oriented individuals.

Before Christmas, just when the autumn wave of the pandemic was getting under control, new, more contagious and lethal virus variants, led by the British B.1.1.7, emerged.  The new circumstances forced more responsible decision makers to accept that no realistic vaccine availability schedule could flatten the present third wave much, at best it might reduce its mortality to some extent.  The B.1.1.7 variant led to incredibly fast covid-19 infection growth rates, first in the UK (8.12.20 – 8.1.21) and Ireland (16.12.20 – 8.1.21), then in Portugal (27.12.20 – 28.1.21), Finland (3.1. – 20.3.), Czechia (30.1. – 3.3.), Estonia (8.2. – ), Hungary (15.2. – 20.3.), etc. and forced new lockdowns before and after Christmas.  The Brazilian P.1 variant apparently bypasses the immunity from earlier SARS-CoV-2 infection, and the Pfizer-BioNTech, Moderna and Astra-Zeneca vaccines are less effective for the South African B.1.351 and the Californian B.1.427/B.1.429 variants.

Now, after thousands of virological, clinical and epidemiological studies and dozens of global and detailed daily accumulating databases on SARS-CoV-2 and covid-19 – all available in the internet – we know much more and have immense amounts of data to deepen our knowledge further.  My aim here is to extract from these data (i) indications for herd immunity vs. the percentage of population infected, (ii) estimation of the covid-19 mortality that precedes via-infection-achieved herd immunity, and (iii) evidence about the suggested harvesting effect.

 

What is meant by herd immunity

The human immune system has evolved for as long as different living organisms have lived in close contact, consuming and infecting each other, defending and clearing space for themselves, forming symbiotic relations, etc.  The immune system has become extremely complex, self-learning at both population and individual level.  It is trying to maximise/optimise between two opposite demands, to safely contact/utilize/consume as many other organisms and species as possible, and to rapidly identify, react to, sustain and eliminate the infinite forms of biohazards that these other organisms may present.  Facing a new challenge the immune system of an individual is not always prepared and successful, and may overreact with harmful, even lethal allergic reactions or, as in the case of SARS-CoV-2, cytokine storm.

Once the human immune system has been challenged by a new micro-organism it develops capability to identify it and defend itself.  Within the social practices of a given community the infection of an average individual results in R0 [= reproduction number] new infections among the other individuals.  If R0 < 1.0, the virus and infections may remain endemic in a population, but there is no epidemic.  Increasing social distances and reducing contacts decrease R0 to Rn.  R0 is also reduced when the proportion of the contacted individuals, who are already immune from previous infection or vaccination, increases.  The infections increase when Rn > 1.0, decrease when Rn < 1.0.  A population has acquired herd immunity when the value of Rn stays below 1.0 in its unrestricted physical and social setting.  Herd immunity prevents the expansion of occasional new cases into an epidemic but it does not zero the possibility of new infections.  By diminishing the frequency of infected contacts herd immunity protects also non-immune individuals.

For many viruses, including the common cold, influenza and corona, immunity and thus herd immunity is not absolute or permanent.  This is the likely outcome for SARS-CoV-2 as well, i.e., once acquired herd immunity fades away as individual immunities grow weaker or new virus variants bypass the immune responses of too many individuals and R0 within the population exceeds 1.0 again.

Therefore the questions concerning SARS-CoV-2 herd immunity are: How high proportion of the population needs to have acquired immunity for Rn<1.0 and what is the decay rate of the herd immunity.  I will also try to answer the question, how many lives would be sacrificed in achieving herd immunity via infections.  The answer is relevant for the discussion about the risks herd immunity via vaccination, because presently its only alternative is immunity via infection.

 

What is meant by harvesting effect

Its scientific synonym is temporary mortality displacement.  The harvesting effect begins with the fact that each individual will die and only die once – no disagreement about that.  The relevance of harvesting effect depends on a value judgement: a death that occurs weeks before one’s time carries less weight than a death that is years or decades premature.  This is not universally agreed upon.  In a murder trial it makes no difference, in public health care management it should and does.  Preventable, environmentally transmitted diseases and deaths [e.g., from environmental tobacco smoke, urban air pollution, contaminated consumer products, manageable or preventable epidemics, etc.] are usually considered to fall in between.

Harvesting effect is often suggested as an explanation for high death counts associated with, e.g., ambient air pollution, heat wave, or flu epidemic, when most of the individuals who die are, indeed, old, sick and frail, and it is assumed that they would otherwise have died within the next weeks or months at the most.  If that assumption is correct the peak that is attributed to harvesting should be followed by a respective dent during the following weeks or months.

The question is, do the available 2020 mortality statistics give support for the presence or absence of harvesting effect in the case for the covid-19 pandemic.

 

The ISTAT Database

I try to answer these questions utilizing the Italian Instituto Nationale di Statistica (ISTAT) total, excess and covid-19 deaths database.  What makes this database particularly useful is its (i) size, [746 thousand deaths in 2020 / 101 thousand excess deaths compared to 2015 – 19 / 76 thousand official covid-19 deaths / population of 60 million], (ii) level of spatial and temporal detail [107 provinces, 12 months], and (iii) quantitatively wide range of impact.

In the first overbearing covid-months the limited understanding of the disease, its attack and fatality rate, the level asymptomatic but contagious infections and lacking capacity of large scale virus testing led to large and regionally varying underestimations of the actual infection counts [in the worst weeks and locations less than 10% of the cases were tested].  From March to May the daily numbers of reported cases per region, province and city, therefore, are of little value for comparison in time or space.  This was understood already in April and the speculations about the true infection numbers both over- and underestimated reality.  In May serological SARS-CoV-2 antigen analyses of the first population samples began to provide more accurate and often surprising [much below the presumed] estimates of the percentages of the infected in populations.  But representative serological data remain still few and far between, provide little spatial and no temporal resolution.

Therefore, most of my analyses in his presentation are based on the more reliable mortality data.  Still, also the reported covid-deaths are almost always under-estimates.  These data, however, are given reliable calibration plus spatiotemporal resolution via comparison to weekly … monthly, local … national mortality statistics.  The total annual 2020 (ref. 2015 – 19) excess mortality in Italy was 25 % or 100 500, in March it was 48 % in the whole Italy and 580 % (!) in Bergamo.  For the whole of Italy the number of excess deaths in 2020 exceeded the official covid death count by 32 % or 24 600 [a rather typical exceedance level in much of Europe].  Except for the least infected provinces and the months of lowest covid-19 incidence, the excess deaths exceeded the reported covid-19 deaths.  For monthly covid-19 deaths in each province I am using the higher of the two values.  Using excess deaths only would reduce the total death count by only 1.1 % but result in some negative covid-19 death estimates, when more realistic values are found in the covid-19 deaths table.

 

Herd immunity from covid-19

The covid-waves of 2020 treated the 107 Italian provinces very differently.  The total excess death rate varied from < 30#/100 000 [provinces of Catanzaro, Siena, Vibo Valentia and Salerno, population 1,9 million] to 179#/100 000 [Italy, population 60 million] and > 500#/100 000 [provinces of Piacenza, Bergamo and Cremona, population 1,8 million].

The covid-19 spring wave, March-May 2020, left the 107 Italian provinces with the infected, seropositive populations that ranged from below 1% to 57%.  The highest percentages were observed in two surveys conducted in Bergamo, 38.5% of 133 workers who returned to the workplace in two companies after the end of the lockdown on May 5. 2020, and 57% of the 9 965 residents who had blood tests between April 23. and June 3.  I use the latter value, because most of its samples were collected later and its base was larger and broader.

For the whole of Italy the two waves of the pandemic, March – May and October – December, resulted in equal, 32%, excess mortality impacts.  Between the provinces the two waves differed considerably, however.  Herd immunity should be observable by comparing the Spring and Autumn waves: a high level of seropositivity generated by a high Spring wave should lower the respective Autumn wave.

In the first figure, therefore, I plotted the excess mortality of October – December against the excess mortality of March – May for the 107 provinces.  Most of the provinces appear in the left of the figure where the Autumn wave was broadly similar to the Spring wave.  In the provinces that experienced the highest Spring wave, Cremona and Bergamo the Autumn wave was much lower, i.e., a significant level of herd immunity had obviously developed.  In these provinces 45-60 % of the population had been infected in the Spring wave, and new deaths 6…10 months later in the Autumn wave appear to have been quenched by 90 % or more.  Keep in mind, though, that these results apply in conditions where contacts within the population remained restricted, i.e., Italy in October – December 2020, and no new resistant SARS-CoV-2 variants played a role.  In the absence of contact restrictions higher percentages of the infected would obviously be needed to have the same protective effect.

In the second figure I have plotted the March – May excess deaths per 100 000 people against the estimated percentage of the infected by June 2020 in the populations of the 107 provinces.  From the figure you can see that the human cost of achieving herd immunity via infecting 60 % of the population would be extremely high, in the order of 600 … 800 deaths per 100 000 plus a heavy and in part long term burden of disease for the population, society and health care [For comparison: 5 cases of narcolepsy per 100 000 from the 2009-2010 swine flu vaccinations were considered unacceptable].  The death count and burden of disease could be reduced to some extent via improved medication and care methodologies and by targeting a higher proportion of the infections to less vulnerable subpopulations.  Yet, the required social engineering for targeting the infections would be unthinkable and, besides, it is questionable that any society would tolerate, or its existing health care service could survive the consequences.

Note, however, that this is the most indirect and weakest of the analyses in this article, but this does not affect its general message.

Luckily this is presently a theoretical scenario only.  There is every likelihood that vaccination, improved and updated as necessary, will provide a more reliable, infinitely more affordable and socially acceptable path to herd immunity.

In the third figure I have plotted the month-by-month excess mortality in Bergamo, in the 11 worst hit provinces combined, and in the whole of Italy.  It supports the finding from the first figure that Bergamo, which was hit the worst in the Spring, was best protected in the Autumn vs. the 11 worst hit provinces, and also that the 11 worst hit provinces combined were better protected than the whole of Italy.

Nine to twelve months after the Spring wave also the first signs of decay in the herd immunity in Bergamo begin to appear in the covid-19 deaths of Bergamo from December 2020 to March 2021 [prediction] which are, again, increasing relative to the whole of Italy.

Harvesting effect in covid-19

As explained earlier, harvesting effect means that, e.g., covid-19 mostly kills old, sick and frail individuals, who would otherwise have died only weeks or a few months later.  If harvesting effect would be real in Bergamo, the 576 % excess mortality in March and 129 % in April [6240 excess deaths in March – May] should have been followed by a considerable dent in the mortality of June – September, when only 44 covid-19 deaths occurred.  The absence of harvesting is obvious in the third figure.  There was an overall mortality dip in June – September, but it was only 85 deaths, too small to stand out in the figure and two orders of magnituge below the level that would support harvesting.  Relative to 2015-19 the mortality statistics of 2020 for the whole of Italy show 51 000 excess deaths in March – May, and a dip of 390 deaths in July [none in the other summer months].

Plotting the Summer (y) excess mortality percentages of all 107 provinces against the respective Spring (x) percentages should, in the case of harvesting, produce a clear correlation; heavy harvesting, i.e., high excess mortality in March – April, should result in respective negative excess mortality in June – July, while low excess mortality in the Spring should have minimal or no impact on the Summer mortality.  The fourth figure presents the result.  The excess mortality of June – July is totally unaffected by the excess mortality of March – April, the correlation between them R2 = 10-7.  There is zero harvesting effect. Q.E.D.

The Italian mortality statistics provide zero support for the harvesting effect – mortality displacement – for covid-19.  The death counts of June – July were completely unaffected by the death counts of March – April, i.e., no mortality had been displaced by 1 … 5 months from the Summer to the Spring. 

This finding agrees with the British-Italian study which concluded that a man dying from covid-19 in Italy loses in average 13 years and woman 12 years of life.

While harvesting of the old, sick and frail is often given as the ovious, self-evident explanation – needing no further proof – for the observed mortality impacts from air pollution, heat waves, epidemics, etc., any supporting evidence is rarely given.  Studies on, e.g., the infamous smog episode in London 1951, urban ambient air pollution in European Cities 2021, and the West European heat episode of 2003, have specifically ruled out any significant role of harvesting in the observed excess mortality.

2 vastausta artikkeliin “Covid-19, herd immunity and harvesting effect?

Vastaa