Solving the Mystery of the 1957 and 1968 Flu Pandemics
Solving the Mystery of the 1957 and 1968 Flu Pandemics
After more than 500,000 deaths in just 12 months, the Covid-19 pandemic is now decidedly the deadliest short-term infectious-disease epidemic to hit the US since the influenza pandemic of 1918 and 1919. It has clearly merited an aggressive public-health response, albeit probably not the exact combination of policies that federal, state and local authorities have wielded over the past year.
Since early in that response, though, I’ve wondered about the two big post-World War II influenza pandemics that used to be called the “Asian flu” and “Hong Kong flu,” and are now labeled by the Centers for Disease Control and Prevention as the “1957-1958 Pandemic” and the “1968 Pandemic.” According to the CDC, the former killed 116,000 Americans and the latter 100,000. Adjusted for population that would amount to 221,000 and 164,000 people today — a lot less than Covid-19, but still staggeringly large numbers, especially when you consider that those deaths weren’t nearly as concentrated among those 65 and older as Covid’s have been.
Yet life went along more or less as normal during those pandemics. There were school closures and lots of workers staying home sick, but no restaurant or “essential-business” shutdowns that I’m aware of and certainly no shelter-in-place orders or mask mandates or border closings or mass experiments in working and schooling from home.
These weren’t the Dark Ages, or even 1918. In both 1957 and 1968, scientists identified the new influenza strains that caused the pandemics before they became widespread, the media and public health authorities warned of the impending threat and vaccine developers immediately set to work to thwart it. Beyond that, though, the reaction seems to have been nothing like what we’ve experienced over the past year.
Why the difference? Have we become too careful and fearful, or were people too insensitive to death back in that higher-mortality-rate era? Are we a bunch of sheeple being led to our doom by Bill Gates and the rest of the illuminati? Or is it that, especially in the case of that mass experiment in working and schooling from home, we have options that didn’t exist in the 1950s or 1960s that we chose to exercise out of legitimate fear that things could turn out much worse than they have?
Those are complicated questions. Here’s one simple, if partial, answer: The “1957-1958 Pandemic” didn’t actually kill 116,000 Americans and the “1968 Pandemic” didn’t kill 100,000.
The apparent source of those numbers, which the CDC lists as the first reference on its pages about the 1957-1958 and 1968 pandemics, is a 1996 article in the journal Epidemiological Reviews by since-retired Baylor College of Medicine virologist W. Paul Glezen. I finally got around to reading it recently, and discovered to my surprise that the death tolls reported above are multi-year totals.
That is, US deaths attributed to the pandemics totaled 115,700 from 1957 to 1960, according to Glezen’s article, and either 111,927 (the number given in a table in the article) or 98,100 (the number given in the text) from 1968 to 1972. The article also reports influenza deaths from 1918 to 1920 as 675,000, which is what the CDC says was the number of US deaths in the “1918 Pandemic.” I had been wondering why some other estimates put that pandemic’s toll at 550,000, and now I understand that was just during the first full flu season; 675,000 was over two seasons.
When the CDC was still calling it the “Spanish Flu,” using a multi-year tally wasn’t wrong. But while I’m all for moving away from naming pandemics after countries, continents and special administrative regions, if we’re going to call it the “1918 Pandemic” we should probably stop citing death tolls that run through mid-1920. Still, at least the first flu season in 1918-1919 accounted for the vast majority of deaths, so the 675,000 total isn’t all that misleading. The first flu season in 1957-1958 also accounted for a majority of 1957-1960 excess deaths, albeit a smaller one. The 1968-1969 share of the “1968 Pandemic” toll appears to have been even lower.
The source of Glezen’s estimates, he noted in his article, was Lone Simonsen, an epidemiologist then working at the CDC who is now a professor at Roskilde University in Denmark. Two years after Glezen’s piece appeared, in 1998, Simonsen and several co-authors published an article in the Journal of Infectious Diseases that reported, based partly on yet another study from 1971, that excess deaths from all causes totaled 66,000 in the 1957-1958 flu season and 28,100 in 1968-1969. As a share of population, that’s equivalent to about 127,000 and 46,000 today.
A sudden sharp increase in weekly deaths attributed to respiratory conditions, heart disease and other causes amid an influenza pandemic probably represents deaths caused or at least accelerated by the flu. So pandemic mortality is generally estimated by measuring excess deaths from all causes compared to the levels that were expected given past experience and demographic change, not by looking at what medical examiners write on people’s death certificates. Still, there are different ways of measuring excess deaths, and the numbers don’t all come out the same.
The Simonsen et. al. excess-deaths estimate for 1957-1958 seems to be widely accepted, if often rounded up to 70,000, but there are several higher (if still far below 100,000) estimates floating around for 1968-1969. A 1981 article in the American Journal of Epidemiology put excess deaths from all causes at 56,300, while the most widely circulated estimate in public-health circles (for which I haven’t located the original source) counts 34,000 victims, and a 2005 article co-authored by Simonsen included an age-adjusted excess-mortality rate that seemed to work out to about 40,000 deaths.
When I asked Simonsen about this she referred me to lead author Cecile Viboud, an infectious-diseases-mortality expert at the National Institutes of Health’s Fogarty International Center. Viboud emailed that the true 1968-1969 death toll was likely higher than 28,100 but that for purposes of comparison I was best off using a single consistent data source and Simonsen’s 1998 article was the best available. So here goes, although you’re welcome to multiply the 1968-1969 mortality numbers by anything up to two if you want.
I’ve listed two sets of numbers for 2020-2021 because (1) the unprecedented testing effort over the past year means a much higher percentage of Covid-caused deaths have been directly identified as such than is normally the case with influenza, and (2) the unprecedented economic and social disruptions of the past year mean that some of the excess deaths are collateral damage of Covid and the efforts to fight it rather than directly caused by it. Because these numbers change on a daily or weekly basis and, in the case of excess deaths, the CDC offers several to choose from, I rounded the totals to 525,000 and 600,000, which I think will turn out to be pretty conservative. The age distributions are from the CDC’s latest accounting of Covid deaths by age and, for excess deaths, a comparison of deaths by age in 2019 with the CDC’s provisional estimate of the same for 2020.
Overall, Covid-19 has been at least four times deadlier, adjusted for population change, than the 1957-1958 pandemic, which in turn was somewhere between moderately worse and three times worse than the 1968-1969 pandemic. Break things down by age and the gap shrinks, although not exactly in the way that I expected. Covid has been 2.4 to 2.9 times deadlier than the 1957-1958 flu for those under 65, and 2.9 to 3.4 times deadlier for those 65 and older. Not that big a difference between the groups, in other words.
The main reason that victims of the 1957-1958 and 1968-1969 pandemics were on balance younger than those of Covid-19 seems to be that the US population was much younger then. If those 65 and older had been as large a share of the population as in 2020, and their death rate stayed constant, they would have accounted for 78% of deaths in 1957-1958 (versus 64% in reality) and 72% in 1968-1969 (versus 59%). That’s not far from the 81% of reported Covid deaths and 75% (or so, there’s no way at this point to get an exact count) of excess deaths in 2020-2021 among those 65 and older.
The same is not true of the 1918 flu, for which death rates were actually higher for those under 65 than for the elderly. The 2009 pandemic, caused by a strain of the H1N1 swine flu that struck in 1918, was also harder on those under 65 than on their elders, although overall mortality was much, much lower than in any of the other pandemics discussed here.
What have we learned here? Well, first of all, that the danger posed by the 1968-1969 pandemic in the US hardly merits comparison with Covid-19. It wasn’t even all that much worse than the previous, non-pandemic flu season of 1967-1968, when according to the Simonsen et. al. estimates there were 22,000 excess deaths (albeit 95% of them among those 65 and older). Last spring there was a flurry of articles and opinion pieces in which authors — among them my Bloomberg Opinion colleague Joe Nocera — wondered why nobody remembered the 1968 pandemic. Well here’s your answer!
That answer admittedly may not hold in Europe, where the pandemic was much deadlier, especially in 1969-1970. And here in the US the earlier 1957-1958 pandemic really was a big deal. It wasn’t the worst since 1918 — the flu season of 1928-1929 appears to have held that dubious honor until this year, causing an estimated 102,000 excess deaths according to the 1998 Simonsen paper, or about 279,000 in population-adjusted terms — but it was pretty bad. It was also the first influenza pandemic to be pre-announced, and quick action seems to have prevented it from becoming much deadlier.
In the spring of 1957, microbiologist Maurice Hilleman read about a flu epidemic in Hong Kong and immediately set to work getting a sample of the virus, which he received in the form of salt water gargled by an overseas US serviceman. Hilleman had spent the past decade at what is now the Walter Reed National Military Medical Center documenting how “antigenic drift” allowed flu viruses to mutate their way around some of the immunity provided by past infection or vaccine, and sussed out quickly that what he dubbed the “Asian flu” was a different variant of influenza (later dubbed H2N2) from the H1N1 that had been dominant since 1918, and was likely to cause a pandemic. By May he was sending vaccine cultures around to pharmaceutical manufacturers with a request that they get cracking, by late July the military was starting to vaccinate recruits and by the end of the year 40 million doses had already been made and used. “That’s the only time we ever averted a pandemic with a vaccine,” said Hilleman, who joined Merck & Co. Inc. later in 1957 and went on to develop vaccines for 1968’s H3N2 influenza and about 40 other diseases.
Vaccine development was so rapid partly because the infrastructure for a flu vaccine was already in place and partly because Food and Drug Administration safety protocols were far less involved in those days. If your criticism of US authorities is that they were unwilling to invest in coronavirus vaccine development before 2020 and too slow to approve and accelerate the manufacturing of the highly effective vaccines that were developed last year, the 1957-1958 pandemic offers support for your views. If it’s that we’ve all overreacted to Covid-19, not so much.
Now that vaccines are available, and more than 60% of the 65-and-older population has already received at least a first dose, state and local governments are rapidly rolling back restrictions and reopening schools. Before the vaccines were available, though, it was not crazy to rely on more-primitive measures. How successful those measures have been will remain a matter of much research and debate. It’s clear that countries that treated Covid as a threat far worse than influenza and acted quickly to prevent it from gaining a foothold — much of East Asia, Australia and New Zealand — have achieved the best mix of preventing deaths and allowing life to go on more or less as normal. In the US it’s much harder to know how many lives all the testing and quarantining and mask-wearing and lockdowns have saved.
Last week I used the method of extrapolating the Covid death rate in the worst-hit US county with 25,000 people or more (New Mexico’s McKinley County) into a potential national toll of 2 million, meaning that 1.5 million deaths have been prevented, and heard from several readers who thought that a county with a 30% poverty rate and a population that is 80% Native American was perhaps not the best stand-in for the nation as a whole. They have a point! But it doesn’t seem unreasonable to posit that if Covid turned out to be a bit under three times deadlier on a partially age-adjusted basis than the 1957-1958 flu, potential Covid was four or five or six or seven times worse. Then again, the 1957-1958 flu could have been much worse if it hadn’t been for Maurice Hilleman.