How bad it has been

JuneHello. You are about to spend six days inside a biosecurity course, and we will walk the readings with you. I am June. That is Theo. We are two AI voices, and between us we have read everything this course assigns.

TheoWhich means you can listen on the way somewhere and still be able to argue about it in the afternoon.

JuneThis is the first episode, and it answers the first question the course asks. How bad have pandemics actually been?

TheoAnd underneath that, the question we will spend most of the half hour on. How does anybody know?

JuneSo here is the shape of the week. The first unit is four steps. How bad pandemics have been. Then that some of it was done on purpose. Then what artificial intelligence and synthetic biology change. Then what a protected world would look like.

TheoWe are on step one. The baseline. Everything after this is measured against it, which is why it is worth being careful here.

JuneBecause at some point this week somebody is going to say that the Black Death killed half of Europe. The course's own introduction says it. And you want to know where the number came from.

TheoNot to catch anybody out. To know how much weight it can carry.

JuneSo here is what we will defend today. The death tolls in these readings are not just large. They are uncertain in a way that matters. Because everything we know about pandemics before living memory comes from four different ways of counting, and not one of them is clean.

TheoAnd we should say what this episode is. It is a list of the dead, most of them without names. We will say the numbers exactly as the readings give them, and we will not decorate them.

JuneAgreed. The spine of this episode is a piece by Saloni Dattani at Our World in Data, written in twenty twenty three, called what were the death tolls from pandemics in history.

TheoAnd her answer is a list of numbers?

JuneHer answer is a method. The piece is as much about how you count as about what happened. She opens by saying COVID nineteen brought pandemics to the front of public consciousness, but that they have afflicted humanity for millennia, and she lists five of them going back centuries.

TheoAnd then the line that sets up everything. Dattani's point is that our ancestors were largely powerless against these diseases, and had no way to evaluate what they were really costing.

JuneNo way to evaluate the cost. Not just no way to stop it. Without good record keeping of cases and deaths, she says, the impact of an outbreak was underrecognised, or even forgotten.

TheoSo her claim is that we underestimate pandemics. Both how often they happen and how bad they are.

JuneBoth. And she names two gaps. One, we have records from some countries and not others, even when the geographic reach of the disease was very wide. Two, we often do not know which pathogen caused an outbreak at all.

TheoWait. So we cannot always say whether something was one pandemic?

JuneThat is exactly her point. If you do not know the pathogen, you cannot say whether a historical event was one pandemic or several outbreaks of different diseases running at the same time in the same place. So let us do the four methods first, because everything else hangs off them. Then we take one number apart in front of you.

JuneMethod one. Excess deaths. Dattani defines it as the extra deaths during a pandemic, set against the number you would have expected in a typical year.

TheoOh, so you mean you do not ask what people died of. You ask how many more people died than should have.

JuneThat is it. And the reason it exists is practical. It is useful, she says, even when records from death certificates are not available. You do not need to know the cause. You need a normal year to compare against. Method two. Net population reduction. The difference in population size before and after. She says it is often used for severe events where a significant fraction of the population died, and her example is the Columbian Exchange.

TheoSo you count the people, you come back a century later, you count again, and the gap is your answer.

JuneThe gap is your answer. Which means the gap contains everything else that happened in that century too. Hold that thought, because it comes back.

TheoMethod three.

JuneEpidemiological modelling. Dattani builds it from knowledge of how the disease transmits, its geographic spread, its fatality rate, access to treatment, and other data.

TheoAnd fatality rate there means the share of people affected who die from it. That is her own gloss. It sounds like a fixed property of a disease and it is not. It moves with treatment, and with how well you are counting cases. And four.

JuneRecorded deaths. Also called confirmed deaths. The number of deaths officially reported with the disease as the cause.

TheoWhich sounds like the most honest one. It is the only one that is actually counting bodies.

JuneAnd it is the one Dattani warns about hardest. She says this method may vastly underestimate a pandemic's real toll, because comprehensive historical records are lacking. And then she says the thing I think is the most important sentence in the reading. Even today, in many parts of the world, cause of death registration is missing.

TheoEven today. Not in the fourteenth century. Now.

JuneAnd that is her stated reason why the confirmed count for COVID nineteen sits so far below the real total.

TheoFour methods. Excess deaths, how many more than normal. Net population reduction, the size of the gap. Epidemiological modelling, built up from how the disease behaves. And recorded deaths, what was written down.

JuneAnd none of them is clean. Each is the least bad option for the records that happen to survive.

TheoCan I ask the thing I got stuck on? What makes something a pandemic rather than a very bad outbreak?

JuneThere is no universally accepted definition. That is Dattani's own first clause. She says diseases called pandemics share several characteristics. A vast geographic range, a continent or several. Highly infectious. Affecting large numbers of people. Rapidly growing in range. And caused by a novel pathogen against which there is little or no pre existing immunity.

TheoBut no official line anywhere?

JuneIn her endnotes, the World Health Organization defines pandemics only for influenza. A novel influenza virus emerges, it can sustain human to human transmission, it causes disease, and the population has little or no immunity. For diseases in general there is no official definition. What it does have is a declaration. A Public Health Emergency of International Concern. Its highest outbreak response, for an extraordinary event that is a public health risk to other states through international spread and may need a coordinated international response.

TheoAnd does that one do anything?

JuneIt carries a legal obligation on countries to respond promptly. One more thing before the worked number. The centre of Dattani's article is a timeline. Every pandemic is a circle, and the size of the circle is its estimated death toll.

TheoAnd the ones with no estimate?

JuneThose are drawn as triangles. Pandemics without a known death toll are depicted with triangles.

TheoSo on the chart of history's worst events, some of them have no size at all.

JuneYou could say the triangles are the real subject of this episode. Everything we are about to count is a circle, and the triangles are the part of the past that never got counted.

JuneLet us walk through one number slowly. Out loud. The Black Death, from half of Europe to a real figure.

TheoAnd we start where the course starts. Its own introduction page says it in five words. The Black Death killed half of Europe.

JuneThat is the sentence you will hear in the room. So what is underneath it. Dattani's figure. The Black Death, thirteen forty six to thirteen fifty three, killed around fifty to sixty percent of Europe's population, approximately fifty million people, in just six years.

TheoAround fifty to sixty percent. Approximately fifty million. Six years.

JuneAnd the cause is now settled. Careful examination of the sources, she says, both revised the toll upwards and confirmed the bacterial cause. Yersinia pestis.

TheoA bacterium. And the reason she names it is that nobody in the fourteenth century knew it existed, or how it moved. Germ theory arrives in the late nineteenth century. Her line is that without that knowledge they had little understanding of how to protect themselves.

JuneSo. Step one. Where does fifty to sixty percent come from? Because there was no census.

TheoNo population censuses were conducted then. So the understanding of the Black Death's impact in Europe comes from tax and rent registers, parish records, and archaeological remains.

JuneTax registers. Somebody's rent stops arriving, and seven hundred years later that is how we know they died. Step two. Dattani says plainly where the weakness is. Those records come from a limited number of European regions, and then get extrapolated across the whole continent.

TheoWhich regions?

JuneIn her endnote, parts of Spain, Italy, France and England. Hundreds of communities inside those areas, from hamlets up to cities. And from those, a figure for a continent. Step three, and this is the part I want you to have in your pocket. There is not one estimate. There are four, and they do not agree. Benedictow estimates that upwards of sixty percent of Europe's population died. Aberth estimates around fifty one to fifty eight percent. Christakos and colleagues put it at forty percent. And Jedwab, Johnson and Koyama estimate thirty nine percent of Western Europe's population.

TheoThirty nine at one end. Upwards of sixty at the other. That is not a rounding difference. That is a different event.

JuneAnd Dattani tells you what she did with that. She gives a rounded estimate of fifty to sixty percent, using the recent and comprehensive studies by Aberth and Benedictow. Her stated judgement, with the others shown so you can disagree.

TheoStep four. Turn the percentages into people. And I want to say clearly that this next bit is our arithmetic on her two numbers. It is not a figure she publishes.

JuneFifty million people is fifty to sixty percent of Europe. Work it backwards. If fifty million is half the continent, the continent is about one hundred million. If fifty million is sixty percent, the continent is about eighty three million.

TheoSo a Europe of somewhere between eighty three and one hundred million. Now take Christakos and colleagues at forty percent, on that same Europe. Forty percent of eighty three million is about thirty three million. Forty percent of a hundred million is forty million.

JuneAnd at the top, Benedictow, upwards of sixty percent of a hundred million. Sixty million or more.

TheoSo the honest answer to how many people the Black Death killed in Europe is somewhere between about thirty three million and sixty million or more, depending entirely on which historian you are reading.

JuneNow put a different source next to it. The Bulletin of the Atomic Scientists ran an illustrated history of the deadliest epidemics, by Erik English, also in twenty twenty three. For the Black Death it gives more than twenty five million dead, between thirteen forty seven and thirteen fifty one.

TheoTwenty five million. Against Dattani's approximately fifty million.

JuneDifferent window, for one. English's number covers four years, hers covers six. We cannot tell you which is right, and neither piece claims to settle it.

TheoStep five. Dattani explains why the twentieth century estimates were too low. Partly because sources have been unarchived more recently. But mostly because the earlier records came disproportionately from the nobility.

JuneAnd the nobility died less?

TheoBetter nutrition, better hygiene, and likely lower fatality rates. Her words. So for decades the toll was estimated from the survival of the people best placed to survive.

JuneThe record of the worst thing that happened to Europe was written in the handwriting of the people it hurt least.

TheoAnd she names two problems that are still there. One, the method measures the net change in mortality during the pandemic. So it can absorb other events of that time, and it does not account for declines in other causes of death.

JuneThere it is. The gap contains everything else that happened.

TheoTwo, the records are limited to parts of Europe. And she says separately that researchers know large outbreaks also occurred in Western Asia, the Middle East and North Africa, but that no comprehensive estimate of the global toll is available.

JuneSo there is no global figure for the Black Death at all. Only a European one, built from four countries.

TheoAnd the two readings disagree even about how it moved. Dattani says Yersinia pestis was transmitted from rat fleas to humans. The illustrated history says rat fleas were thought to be the main transmitter, then notes that recent theories argue human body lice were the more likely vector.

JuneSo. Half of Europe. Underneath it, four historians between thirty nine and sixty percent, a toll between about thirty three and sixty million, a continent estimated from four countries, records weighted toward the people who survived best, no global figure, and an open argument about the insect.

TheoAnd that is the best documented pandemic before the modern era. The one Dattani calls one of the earliest with a methodically estimated death toll.

JuneLet us take the second method now and watch it do something the first one cannot. Net population reduction.

TheoThe Columbian Exchange. Dattani's timeline shows an estimated forty eight million deaths.

JuneAnd she defines the term. The Columbian Exchange is the period after Columbus's fourteen ninety two voyage, when populations, ideas and crops moved between the Americas and the rest of the world. It also involved extensive war, conquest and slavery, and the spread of several deadly diseases.

TheoShe names six of them, smallpox first, and says there were others. All of which had already killed many people in Europe, and were more severe for Native Americans, who had been isolated from them and had no immunity.

JuneSo here is the net population reduction, done out loud. The Native American population is estimated at around fifty four million before Columbus arrived. Over the following century around forty eight million died. The population in sixteen hundred is five point six million. A reduction of about ninety percent.

TheoFifty four million to five point six million.

JuneAnd both of those figures come from a mix. Archaeological records, tribal records, censuses, epidemiological modelling, and land and crop use.

TheoAnd Dattani flags in her endnote that pre Columbian population estimates are generally uncertain. Partly because disease may have spread faster and further than Europeans travelled. And partly on assumptions as specific as how many houses in a settlement were occupied at a given time. And she says recent archaeological work on land use has revised the initial population upwards. Which, because this is a subtraction, revises the toll upwards too.

JuneThat is twice. Look harder, the number gets worse.

TheoThe illustrated history tells the same century in pieces. The fifteen twenty Mexico smallpox epidemic, five to eight million dead. Then a disease called cocoliztli, the Nahuatl word for pestilence, in fifteen forty five, millions of deaths. Then cocoliztli again in fifteen seventy six, a further two to two and a half million. And by sixteen hundred, Mexico's population had fallen to only two million, from fifteen to thirty million at contact.

JuneAnd the cocoliztli pathogen has never been identified. The leading theories are salmonella, or a viral haemorrhagic fever tied to drought and to living conditions under Spanish rule.

TheoThe piece also uses a term worth having. Virgin soil epidemic. An outbreak in a population with no prior exposure, and so no immunity. It uses it for Mexico in fifteen twenty, and for Japan in seven thirty five.

JuneAnd Japan is the one I promised you. Because of how the number was made.

TheoThe Japanese smallpox epidemic of seven thirty five killed a third of Japan's population by seven thirty seven. And that toll is derived from records of defaults on loans taken out to buy rice, where a default is assumed to mean the borrower died, set against the estimated population. The account calls it a conservative estimate.

JuneSomebody borrowed money for rice and never paid it back, and that is the record we have of a third of a country dying.

TheoThere is one more like that. The Italian plague of sixteen twenty nine to sixteen thirty one, the Great Plague of Milan, roughly one million dead in northern Italy. And the account says estimates are hard to come by and rest partly on Alessandro Manzoni's novel The Betrothed, which described the plague in detail.

JuneA novel.

TheoA novel. Written two centuries after the event, and used because better sources are not there.

JuneSo add that to the list of ways we know. Tax registers. Parish records. Bones. Unpaid rice loans. And a novel.

TheoEvery one of those is somebody's best effort with what survived. None of them is a count.

JuneTwo more figures from that piece. The Plague of Justinian, from five forty one, an estimated one hundred million dead by five ninety, with Constantinople seeing ten thousand deaths a day and losing roughly one fifth of its citizens.

TheoAnd the account itself says some scholars consider those estimates and their consequences exaggerated. Which matters, because Dattani covers the same pandemic and gives it no death toll at all. She says the first known bubonic plague pandemic began in five forty one with recurrent outbreaks until the mid eighth century, and the most severe was the Justinianic Plague of five forty one to five forty nine.

JuneOne reading says one hundred million. The other declines to give a number. Same pandemic.

TheoAnd the illustrated history adds that plague has killed more than two hundred million people worldwide across all of history.

JuneNow the two pandemics where you can see the method change the headline. Nineteen eighteen, and COVID nineteen.

TheoA word first, because Dattani explains why flu keeps coming back rather than happening once. Antigenic shift. Different flu strains combine into a novel strain, suddenly, and the new one can be more infectious and more lethal than what came before.

JuneShe also says comprehensive flu death tolls only exist for the pandemics of the last hundred and forty years.

TheoThe largest is nineteen eighteen, the so called Spanish flu. Dattani's estimated death toll is fifty to one hundred million, compiled from recorded death tolls and regional excess death estimates.

JuneAnd she makes a point about the names. The names of flu pandemics describe where they were first reported. Not necessarily where they began.

TheoWhich the illustrated history backs up. It puts the first recorded cases at Fort Riley in Kansas, and says the Spanish flu name stuck because neutral Spain's papers covered it from the start while other countries were censoring flu news to protect morale.

JuneAnd that piece gives the toll as between forty million and fifty million deaths, from nineteen eighteen to nineteen twenty. More, it says, than the total deaths in the First World War.

TheoForty to fifty million in one reading. Fifty to one hundred million in the other.

JuneAnd in Dattani's endnote the spread is wider than either. Johnson and Mueller arrived at fifty million and stated the true toll may be up to one hundred million. Earlier historians ranged from thirty million globally, Patterson and Pyle in nineteen ninety one, to eighteen million in India alone, Mills in nineteen eighty six. And a twenty eighteen reassessment by Spreeuwenberg and colleagues estimated around seventeen million.

TheoAnd Dattani says she thinks that seventeen million figure is likely itself an underestimate, because poorer countries saw higher flu mortality rates. That is her view, stated as her view.

JuneSeventeen million to one hundred million. For an event inside living memory for some families.

TheoAnd she says why the range is that wide. Missing death registration. Missing records. Misdiagnosis. And under reporting in rural and native populations.

JuneThe rest of her flu timeline, exactly as she gives it. Eighteen eighty nine, an estimated four million deaths. Nineteen fifty seven, two million. Nineteen sixty eight, two million. And two thousand nine, between one hundred thousand and one point nine million. All from excess mortality against the years immediately before and after. And the illustrated history gives eighteen eighty nine to eighteen ninety an estimated one million, not four. The same piece records St Petersburg losing one hundred and eighty thousand of its one million inhabitants within five weeks of November eighteen eighty nine. COVID nineteen. And here the choice of method is not a historical curiosity. It is a live argument.

TheoThe official confirmed count, as the illustrated history reports it, is six point seven million.

JuneAnd Dattani's timeline shows around twenty seven million excess deaths between January twenty twenty and November twenty twenty three.

TheoSix point seven million. Twenty seven million. Same pandemic, same years, two methods.

JuneAnd she is explicit about why she uses excess mortality. Global confirmed deaths are certainly much lower than the real total, because testing was very limited in many countries and cause of death registration is still lacking in many. And excess mortality also captures the deaths caused indirectly.

TheoAnd she says where the twenty seven million comes from and why. It is The Economist's estimate, chosen because it is continuously updated and the methodology is well documented. She calls that a judgement about updating and documentation, not a finding.

JuneAnd she shows her work. The World Health Organization and the Institute for Health Metrics and Evaluation have also estimated excess deaths, but only to the end of twenty twenty one. For that period all three are similar. Eighteen point two million from the Institute for Health Metrics and Evaluation. Fourteen point eight million from the World Health Organization. Seventeen point eight million from The Economist.

TheoAnd she shows a wide uncertainty band around the excess figure, while making the point that even its lowest estimate is much higher than the confirmed count.

JuneAnd here is the tension I want to leave you holding. Only one of those two COVID numbers can be compared with nineteen eighteen.

TheoRight. Nineteen eighteen's fifty to one hundred million is built from recorded deaths plus excess mortality estimates. So to put COVID next to it you have to use the twenty seven million. Use six point seven million and you are comparing two different kinds of thing.

JuneAnd the other half. Every historical toll in this piece has been revised upward as records were digitised. The Black Death, upwards. The pre Columbian population, upwards. Dattani thinks the lowest nineteen eighteen estimate is too low.

TheoSo if the past keeps getting worse the harder we look, what does that do to the base rates the rest of this course is going to use?

JuneNeither reading answers that. But it is the question I would take into the discussion, because every argument later this week about how often this happens is standing on these numbers.

TheoOne last pair, on the fatality rate we flagged earlier. Dattani says that during nineteenth century outbreaks, between thirty and sixty percent of people diagnosed with cholera died, in many major cities including New York and Berlin. Today the case fatality rate ranges between zero and ten percent globally.

JuneCase fatality rate being the share of people diagnosed who die of it. And that fall is not one discovery. Dattani lays out a chain. Clean water and sanitation. Identifying the bacterium. Antibiotics. And simple rehydration treatment.

TheoAnd she notes those figures are limited by weak surveillance and by countries using different case definitions. So even the good news number carries its warning.

JuneAnd the toll. Seven cholera pandemics in two centuries. Between eighteen sixty five and nineteen forty seven, at least twenty three million people died of cholera in India alone.

TheoWith no global total for the earlier waves. The illustrated history says no global estimate exists for the third cholera pandemic, the deadliest of the seven, though more than one million are believed to have died in Russia alone.

JuneThere is a third reading on this page, and it is a different kind of document. Let us finish with it, because it tells you what to do with everything we have just said.

TheoIt is the opening chapter of the United States Army's Medical Aspects of Biological Warfare. A historical overview, from poisoned darts to pan hazard preparedness, by James Martin, George Christopher and Edward Eitzen.

JuneAnd it opens with a word. The word toxin descends from the ancient Greek for arrow poison. The authors use that to make the point that weapons and biology have always been linked. And their history runs a long way back. Thirteen forty six. The year Dattani's Black Death begins. The chapter records cadavers hurled over the walls of Caffa, after war had broken out in thirteen forty three between the Genoese and the Mongols.

TheoAnd it does not claim that act caused the pandemic. It records it as the earliest kind of thing in its own history.

JuneAnd here is why this chapter belongs in an episode about counting. It names what confounds its own history. Secrecy. The difficulty of confirming allegations. The lack of reliable microbiological and epidemiological data. And allegations used as propaganda, or as hoax.

TheoAnd it says that assessing biological weapons programmes is difficult even for highly skilled professionals.

JuneSo all three readings say the same thing about themselves. Dattani says the records are missing. English says wide uncertainty ranges are the historical norm, and that exact tolls are hard to pin down even in an analytics heavy age. And the Army chapter says secrecy and propaganda sit on top of the gaps.

TheoAnd the chapter has a prescription, which is where the rest of the course is going. It argues for moving away from a static list of specific pathogens, toward what it calls a threat agnostic, capability based approach.

JuneMeaning you do not decide in advance which disease you are preparing for?

TheoMeaning exactly that. Flexible platform technologies that can be adapted to counter novel, unanticipated pathogens. Broad spectrum therapeutics. And response capacities that work against a deliberate attack, a natural epidemic or another mass casualty disaster. They call it pan hazard preparedness.

JuneAnd the chapter ends on Joshua Lederberg, whose view it reports as this. That there is no technical solution to the problem of biological weapons, and that it needs an ethical, human and moral solution if it is going to happen at all.

TheoOne number left, and it is a forecast rather than a finding, so we are going to say that clearly. The illustrated history puts the annual risk of another pandemic on the scale of COVID nineteen at about two percent. And says that estimate is expected to triple, because human activity is raising the risk of zoonotic spillover.

JuneZoonotic spillover being a pathogen crossing from an animal into humans. And the piece's argument is that outbreaks reflect things we are doing, not only things happening to us. It names climate change, conflict and environmental degradation.

TheoAnd we will say it once more. Two percent, expected to triple, is that piece's estimate and that piece's expectation. It is not a measurement.

JuneThe same piece gives the other side of the ledger. Technically, it says, our capacity to address outbreaks is stronger than at any point in human history. COVID nineteen vaccines were approved in record time and saved upwards of twenty million lives in twenty twenty one alone. And then the sting. The circumstances surrounding COVID nineteen's emergence and spread remain largely unchanged.

TheoAnd Dattani's ending is conditional in the same way. She says progress came not only from science but from data collection, public health, healthcare access and cooperation. Death records let researchers work out how cholera spread. Coordination on HIV and AIDS prevented millions of deaths.

JuneAnd then her last clause. With better understanding, resources and effort, the world can respond more swiftly. But without such efforts, she writes, we will continue to face major pandemics as we have so far.

TheoAnd on HIV and AIDS, since that is her example of what cooperation did. When it was identified in the early nineteen eighties it had a fatality rate of one hundred percent, and a median survival of about one year after diagnosis. Her timeline shows an estimated thirty three million deaths between nineteen eighty one and twenty twenty two, with around one point five million deaths averted annually now by antiretroviral therapy.

JuneAnd the illustrated history gives more than forty million dead as of twenty twenty one. Against her thirty three million to twenty twenty two.

TheoTwo readings. Seven million apart. On the pandemic we watched in real time.

JuneSo here is the one line I would take into the room. The death tolls are not just large. They are uncertain in a way that matters, because every one of them comes from one of four ways of counting, and the method you choose changes the headline.

TheoAnd a second half, if you want it. COVID nineteen is six point seven million or twenty seven million depending on the method, and only one of those can be compared with nineteen eighteen.

JuneWhich brings us to the next episode. Because there is something we deliberately left on the table today.

TheoThirteen forty six. Caffa. The cadavers over the wall.

JuneEverything we counted today happened to people. Next time, the part that was done on purpose. State programmes, at scale, and the accidents that are on the record. That is the baseline everything later in this course is measured against.

TheoWe will see you there.