There are not many places on earth where we have detailed cause-of-death data from before the era of widespread vaccination.
Massachusetts is one of those places.
From 1842-1877, 70% of all deaths were from diseases which we today have vaccines to prevent.
cc @RichardHanania this feels like it's up your alley
huge pain in the butt to hand-copy all these historic vital stats, but I did it a few years back and have never regretted it!
For example, here's typhoid. Vaccine available 1896. You could try to say there was a pre-vaccine decline, but it's hard to know for sure. Certainly absolutely no shot of falling to <500/yr pre-vaccine.
Here's scarlet fever deaths. Vaccine invented 1924, widespread availability by the early 1930s.
Scarlet fever pretty much only kills kids, so the figure falls over time partly because because Massachusetts births were falling. But you can see deaths drop to ~zero after vaccine.
I could go on.
Vaccines obviously aren't the only factor improving health. Sanitation efforts and improved treatments had big effects too. But you don't go from vaccinatable deaths at 70% to 3% without vaccines mattering!
Okay, one more.
TB deaths were already declining when the vaccine was rolled out. And they didn't drop to super-low levels until antibiotics. But it's obvious vaccination helped a ton!
Measles is an interesting one.
Vaccine isn't available until 1963. Measles is viral, so antibiotics 1930 and later aren't a plausible explanation either.
But we can see that there's no chance of "staying at zero" without a vaccine.
I don't think vaccines are the entire cause of falling death rates from conditions for which we have vaccines.
But they're obviously part of it!
Influenza is a weird one. First, the data starts in 1890 because it wasn't widely recognized as a discrete disease in MA's classification system before then. Second, there's the huge pandemic bump. I'm gonna ignore that. Third, vaccine development was a slow roll 1931-1950.
But by 1950-55 at the latest, influenza vaccination was relatively common if far from universal.
Again, I'm not saying vaccines caused 100% of improvements. But they're part of the story!
When you're moving fast you make mistakes. Here are some of mine:
1) First graph has an error. I double-counted tuberculosis deaths in many early years. Fixed. Trends are identical, but 1842-1877 average is now 50%, not 70%. Note that the actual graph of tuberculosis deaths IS correct.
2) Scarlet Fever turns out to be complex. 1924 vaccine was of limited effectiveness, discontinued 1944. It appears it was somewhat effective, but sulfanomides are probably the bigger story there.
3) A lot of people responding pointing to antibiotics. Yes I agree antibiotics are a HUGE part of the story. But also, basically 100% of people who say "We use too many vaccines" will also say "We use too many antibiotics," so this isn't much of a rebuttal.
As an aside, this isn't my data, it's from @OurWorldInData , but I've added the labels for smallpox variolation and vaccination.
The other diseases all reflect vaccines developed during the period of modern improvements in sanitation and health.
Smallpox was much earlier:
How here's the same graph, but with Massachusetts smallpox data added in.
Massachusetts was an early adopter of variolation AND of vaccination for smallpox. Even so, in both series you can see the improvement in the post-1880s vaccine in both time series!
Here's whooping cough deaths in Massachusetts.
As an aside, whooping cough cases are once again on the rise, largely due to unvaccinated kids.
Cholera is an interesting case. Several cholera vaccines were developed 1870-1900. None got widespread usage. I'm actually not sure if cholera vaccination ever became common in America.
Moreover, cholera mostly kills children, with similar symptoms as other diseases.
As a result 1) it's hard to know how reliable old stats are, 2) it's not clear how many people ever even got vaccinated in Massachusetts.
Anyways, here's all deaths of all cholera-or-maybe-like-cholera.
In this case it's pretty clear the existence of the cholera vaccine was irrelevant. Basically nobody used it in MA, and the real solution was water treatment and sewage infrastructure.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Annual individual consumption * Years Child Remains Dependent
in industrialized countries, AIC can be proxied using something like GDP per capita, and years-dependent is now approximately 20-30. call it 24.
note that the approximation in the formula is based on more careful calculations from the small list of countries we have data from on actual parental spending, government spending on kids, parental time use, motherhood earnings penalties, etc
it's an all-in cost
actual family budgetary costs are a lot lower.
in the US, I estimate that the all-in cost of rearing a child to economic independence is about $2 million in total private and public costs in both money, opportunity cost, and time.
It's clear that @propublica 's strategy is to spam stories of alleged deaths due to abortion bans, and never actually engage with any of the arguments about how they're actually running a cover operation for medical negligence.
From the latest one.
They want to blame Texas' abortion ban for a hospital sending away an actively miscarrying women WHO ALREADY TESTED POSITIVE FOR SEPSIS.
everybody agrees this was a case of the hospital failing to provide basic, obvious standard of care.
nobody has evidence this failure was caused by the abortion law.
To begin with, some basic facts: Finland's total fertility rate was around 1.87 children/woman as recently as 2010. It did NOT decline during the "great recession" after 2007, but actually ROSE.
Since 2019, Finland's fertility has bounced around a lot, but the decline 2019-2024 was just 0.08 children per woman, vs. the decline from 2014-2019 of 0.36. So clearly the pace of decline has slowed, even if not stopped entirely.
But you may wonder: what drove Finland's decline? Did big families get rarer, or did people stop having families at all, or what was it?
Here's parity-specific birth rate indicators:
You can see they all decline after 2010. Here's each indicator, its 2022 value expressed as a ratio of its 2010 value:
You can see that 3rd births rates fell the most, down almost 30%, then 1st birth rates, down about 27%, then 5th, then 4th, then 2nd, down about 15%.
But they're all down. Finnish women became less likely to have an extra birth at every single parity.
What does this look like in terms of total birth count?
Well, it looks like appreciable declines for every birth order. And indeed, births fell 25-33% at every parity.
So did Finland's fertility decline because of a broad-based shift away from kids across all families? Perhaps!
But now let's ask this another way:
Comparing 2010 to 2022 births, what share of the decline in births was 1st vs. 2nd vs. 3rd, etc?
37% of the decline is due to lost first births, 36% second births, 17% third births, 5% fouthr, and 5% 5th+.
So more than a third of the total decline was due to a drop in first births, and more than half was due to a drop in first or second births. Low-parity births accounted for the lion's share of decline.
I think some open questions in the Finnish case are: 1) Why was Finland so resilient to the Great Recession? 2) Why the drop then at 2010? 3) Why was the drop so broadly shared across parities?
For the curious, here's Korea.
Korea's big drop post-2010 coincided with NO CHANGE in rates of progression to higher-parity births! People with 2 did NOT become less likely to go on to 3 (or 4, or 5).
The entire decline was falling rates of progression to 1 and 2.
UN, IHME, VID, all produce population forecasts- and they always seem too optimistic. Human population will start declining much earlier than the UN expects.
This has been obvious for a long time. @jburnmurdoch is right to highlight it--
But why does the error persist?
🧵🧵
You can actually see a defense of the UN's method here:
TL;DR-- the UN's method really is the best-performing forecast method in historic data compared vs. other structural forecasting methods.pnas.org/doi/10.1073/pn…
Let's talk about what a structural forecast is.
Basically, we're talking about forecasts which: 1) Do not incorporate any explicit assumptions about GDP or education 2) Use the same method for all countries and periods with no special sauce
Just finished talking on a panel for a nice @BrookingsInst event about REMOTE WORK AND FERTILITY.
Here's a finding I haven't published anywhere on remote work and fertility across 8.6 MILLION employed women in the ACS. Remote-working women have WAY higher birth rates!!!
Now, note that this effect is plausibly reverse causality: it's very plausible that women work remotely because they had a baby in the past year! Unfortunately ACS is going to be entirely "current work and retrospective fertility."