#12: Everything great you missed since November, part one
This week: Declines in cancer, the significance of hospitals, climate & religion, bird maps, gender pay gaps, intelligence & personality, the pyramids of Giza, fun history podcasts, and more.
This is my twelfth post of Scientific Discovery, a weekly newsletter where I’ll share great new scientific research that you may have missed.
I’m back, finally! I handed in my PhD thesis the other week, and now I’m on a much needed break before my viva. So I’ll catch you up on what you’ve missed while I was away.
I started this newsletter for many reasons, but a major one is to highlight how much great scientific research is regularly being done; how it helps to understand the world around us and how it can make a material difference to our lives.
This is a long post – feel free to scroll through until you spot something that interests you. If you like this, I hope you subscribe! And if you spot any errors, I’d like to know so I can fix them.
Why have some cancers declined?
Study: Cancer statistics, 2023 (Rebecca L Siegel et al., 2023)
You may not know that death rates from several cancers have been declining.
Here’s an overview of why, with data from the United States.1 You can explore these trends for other countries using the WHO’s Global Cancer Observatory website here.
Cancer death rates grew over the 20th century. But since then, between 2001 and 2020, the cancer death rate declined by 27%. In other words, it’s become 27% less likely since then that people of the same age die of cancer, on average.
Much of this decline comes from the great decline in smoking, which had massively raised the risk of lung cancer, which became the most common type of cancer deaths. Smoking also raises the risk of many other cancers, including cancers of the throat, oesophagus, stomach, colon, rectum, liver, kidney, bladder and others. So, declines in smoking have also reduced the rates of other cancers.
But smoking isn’t the only cause of the decline.
The death rates of several cancers caused by infections are also declining:
Stomach cancers and intestinal cancers, which are largely caused by the bacteria Helicobacter pylori, have declined with sanitation, more screening and treatment, including ‘eradication therapy’.
Cervical cancer (of which Human Papillomavirus (HPV) causes likely all cases) has been declining. This is due to HPV vaccination and more early screening and treatment.
You can find out here how HPV vaccination rates have risen among teenagers across states in the US, over time. Cervical cancer is the second most common cause of cancer deaths among young women in the US, so it’s really important to speed up vaccination uptake there and worldwide.
In many countries, there is no national program at all to vaccinate people against HPV.
Liver cancers have declined. This is partly due to hepatitis B vaccination. Globally, the hepatitis B and C viruses cause around 80% of hepatocellular carcinomas, which account for around 90% of liver cancers. A vaccine for hepatitis C doesn’t exist yet.
Which other pathogens cause cancers, and how many cancer cases do they cause? You can find out on the WHO’s Global Cancer Observatory here, or the underlying study here.
Some cancers – especially breast cancer, kidney cancer, melanoma and chronic myeloid leukaemia – have been declining due to better screening and treatment.
Our knowledge of the causes of cancers – whether that’s smoking, asbestos, bacteria, viruses, radiation or other environmental factors – has helped us tackle them at the source and prevent many cancers entirely, even aside from improvements in screening and treatment.
In case you didn’t know, cancers have a moderate to low heritability: the differences between people’s risks of developing cancer are usually down to differences in the environment (including smoking, infections, pollution, random chance, etc.), but this varies between different types of cancers.
Unfortunately, not all cancers have declined. For example, the death rates from pancreatic cancer, uterine corpus cancer and oral cavity cancers continue to rise. I plan to write much more about trends for specific cancers, their causes, prevention and treatment, in the future.
Once again, you can explore these trends in other countries using the WHO’s Global Cancer Observatory website here. You can also find global vaccination rates for HPV and HepB on Our World in Data from WHO data, here and here.
Where can you find birds?
Interactive dataset: eBird. (The eBird team, 2022).
This is a question I’m sure everyone wants answered.
Find out in the explorer here. This big dataset is compiled from data from birdwatchers, who enter information about where and when they have spotted birds. You can contribute too. The authors have made adjustments for how bird sightings are affected by the weather, travel, and the behaviour of different users.
Below, you can see when and where you can find Blue-tufted Starthroats, for example. They are mainly found in Brazil, Argentina and Paraguay. I have never seen one. I picked it simply because it looked cute in the little icon.
The data also shows you where bird populations have been increasing or decreasing, and where they migrate to and from. Pretty neat!
There is also another bird breeding map focusing on Europe here.
Modern hospitals matter more than you think
This detailed study looks at the impact of modern hospitals on infant mortality rates in the 20th century.
They focus on a large program that aimed to modernise hospitals in North Carolina, called the Duke Endowment program, which began in 1926. It involved several parts: expanding, improving or building hospitals with new technology, attracting young and qualified physicians to the area, and improving the management practices in those hospitals.
Although it took place during the Jim Crow era, the program placed an emphasis on racial equality and the majority of funds went to hospitals that accepted both black and white patients.
They found that the hospital modernisation program reduced infant mortality in those areas by around 10%, with larger reductions for black infants than white infants (16% vs 7%); I think this was because black infants had much higher mortality rates beforehand.
The benefits lasted throughout the lives of people in those areas; even when they reached the age of 56–65, their mortality rates were around 9% lower than they would have been otherwise.
The program also led to better responses to new medical technology. After sulfa drugs were introduced in 1937, the hospitals that received support and were in communities with high rates of infectious disease had twice the reduction in infant mortality rates. (Sulfa drugs are a class of mainly anti-bacterial drugs which are known to have had large effects on overall mortality rates.)
I really liked this paper – it’s very thorough and was also easy to read. You can read it in full here.
Intelligence and personality
Study: Personality and Intelligence: A Meta-Analysis (Jeromy Anglim et al., 2022)
How is intelligence related to personality?
This large meta-analysis answers this question with data from people’s performance on tests of general intelligence, and their responses to personality questionnaires. The authors looked at correlations between intelligence and: 1) responses to particular questions (items) on these questionnaires, 2) small sets of questions (facets), and 3) broader themes of questions (domains).
Overall, intelligence is positively correlated with the domain ‘openness’, and slightly negatively correlated with ‘neuroticism’. But some facets within those domains also have interesting correlations, which the authors cover in much more detail in the paper.
Below are some of the correlations they found, with the NEO personality questionnaire, for example.
I expect you will now wonder: Is this because intelligence affects personality? Or personality affects intelligence? Or a secret third thing affects them both? But sadly, I don’t know the answer.
Climate shocks and religion
How do climate shocks affect religious behaviour?
This study looks at the impact of droughts (which have become more frequent due to climate change) on religious behaviour in Afghanistan. Agriculture is a major part of the economy in Afghanistan (23% of their GDP), so droughts may make a big difference to people’s livelihoods and their religious behaviour.
The authors estimate that major droughts increase religious adherence. The size of this increase is around 24% as much as the effect that the Taliban has on religious behaviour, when it contests or controls a district.
I found the methods of this paper really interesting, so here’s some more detail: How do you measure religious behaviour?
You could use surveys, but it’s often difficult to do surveys with many people over a large period of time, particularly in poor countries, and people may not remember or report their behaviour accurately.
But you can also measure religious behaviour in other ways. In Islam, people are expected to pray each day several times, including during the ‘Maghrib prayer window’, which is a period of time just after sunset. During this daily prayer time, there is a strong norm against talking to others, and hence the volume of phone calls dips; this dip can be used as a measure of the level of religious adherence.
You can see this in the chart below, which shows phone call volume at different times of day, across two years. The dip is the lighter wavy line. This is what the authors used to measure religious adherence; they looked at the impact of major droughts on the size of the dip.
Sources of the gender pay gap
Study: Within-job gender pay inequality in 15 countries (Andrew M Penner et al., 2022)
There has been a long debate on where the gender pay gap comes from. Is it from men and women being paid differently for the same jobs? Or is it from them having different careers, working hours and career progressions? (Note that both can be affected by discrimination, policy, and social conventions.) How much do each of the two sources matter?
The authors of this very large study used national data from 15 countries to answer this question. Research in the past has been limited – without data from within specific firms, it was challenging to distinguish between the two sources.
They find that both matter a lot. Differences within the same job accounted for around half of the overall gender pay gap across the 15 countries, but this varied a lot between countries; it accounted for around 35% the gap in Israel and 96% of the gap in Hungary! These numbers are higher than I expected.
They also found that the gender pay gap within jobs (and the gender pay gap overall) has been declining slightly in most of the countries, as you can see below.
More stuff
If you’ve ever wanted to go to the Pyramids of Giza, but also relax on your couch at the same time, you can do that. Here’s a very cool virtual walkthrough of the Great Pyramid.
Cash incentives can be very cost-effective in reducing air pollution and saving lives (B Kelsey Jack, Seema Jayachandran, Namrata Kala, Rohini Pande, 2022).
Crop residue burning is a major cause of air pollution in India. The government offers rewards to farmers if they don’t burn crops, but these rewards are conditional, given after they’ve been verified to comply. This might deter farmers who have financial constraints or don’t trust that they’ll be paid.
This new RCT finds that giving them a partial but unconditional cash transfer upfront reduces crop burning by 8–12 percentage points, compared to people who receive conditional cash transfers.
Since air pollution has large costs on health, the authors estimate that this translates to saving a life for <$5000.
When you’re shopping, why is it that you often see entire areas filled with very similar firms – like streets full of shoe shops, or a building full of electronics products? (Anna Vitali, 2022)
For customers, this is because it’s difficult to search and travel for new firms in different areas. For businesses, being located in a place with lots of competitors makes it easier to find customers, but it also means they have more firms to compete with.
How do these factors fit together and affect prices? This is a cool paper that answers these questions.
How many domestic cats have been infected by Covid? (Grace B Tyson et al., 2022)
This paper uses routine samples from vets in the UK to estimate that, at the peak, around autumn 2021, around 5% of domestic cats were infected by Covid.
I included a cat pun here at first but then I deleted it to save you from wincing.
Percentage of domestic cats positive for Covid antibodies, from routine vet samples. Source: Grace B Tyson et al., (2022)
I’ve now lost track of the details of each new Covid variant, but here is a great technical overview by Thomas Peacock on the evolution of the new lineages and how they arose.
Good managers are important because they can reallocate people to roles that suit them better (Virginia Minni, 2023).
This study looks at a large number of employees in a multinational firm. The author finds that, when managers are rotated around teams, good managers help move other employees to other roles within the firm, which boosts their productivity and career progression.
Non-anonymised surveys are underestimating rates of depression (James Reisinger, 2022).
To estimate the rates of depression, many researchers use data from non-anonymised telephone interviews of the general population. But this may lead to underestimates, because it relies on people accurately reporting their symptoms to the interviewer; people might be unwilling to share their experiences on sensitive topics (like their own mental health) to people in these interviews.
This is a really interesting study that looks into how much that matters. It uses data from a large commonly-used survey (the NLSY) where people who were supposed to have anonymised in-person interviews were randomly switched to non-anonymised phone interviews.
It turns out that being randomly assigned to non-anonymised interviews reduced the estimated depression rate by 0.21 standard deviations, which is similar to the difference in depression rates between the highest and lowest quarter of income groups. That seems fairly important for this research.
Note I’ve previously written about the lifetime risk of depression; but those estimates came from anonymised in-person surveys.
Even more
Here’s some other great stuff you might like to read.
An ‘introduction to genomics for engineers’. I’m not an engineer but I had a skim through and it looked pretty good and easy to read for a wider audience.
‘Why everyone should pay more attention to India’ by Shruti Rajagopalan. I found this pretty convincing.
Everything you wanted to know about Respiratory Syncytial Virus (RSV), by Edward Nirenberg.
‘Historical political economy of migration’ by Volha Charnysh – a book chapter that summarises the research on migration and its historical effects on politics. This seems like a great review; there’s an ungated version here.
New Works in Progress issue
We had a new issue of Works in Progress out in December, and everything in it is very good! I’d recommend reading it if you didn’t get around to it before. Here’s a condensed summary of each piece:
The story of VaccinateCA – by Patrick McKenzie. An epic piece (literally) on how a small group of volunteers managed to speed up the very slow vaccination rollout in California.
Advancing antivenom – by Mathias Kirk Bonde. Snakebites cause the deaths of tens of thousands to hundreds of thousands of people per year, worldwide. This burden is largely in poorer countries, where effective antivenom is expensive and difficult to store and transport.
Pandemic prevention as fire-fighting – by Richard Williamson. As the title says, this piece is about what pandemic prevention efforts can learn from the history of firefighting.
Developing the science of science – by Heidi Williams and Paul Niehaus. On what meta-science research can learn from the empirical revolution in development economics.
The elements of scientific style – by Étienne Fortier-Dubois. Over time, scientific papers have become increasingly difficult to read, but it doesn’t have to be this way.
AI from Superintelligence to ChatGPT – by Séb Krier. This is a great essay on how AI has advanced over the years and what this means.
History is in the making – by Stephen Davies. The way history is taught today focuses on wars and politics. But technology, ideas and communication have dramatically shaped our lives too, and we should learn much more about their history.
On that note…
Entertaining history podcasts I’ve listened to
‘The astronaut origins of food safety’ on the Stuff you missed in history class podcast. I didn’t know that food safety regulations today trace back to NASA’s development of manned space flights (which needed to have safe food to last the whole journey).
‘How haunted houses became a thing’ on the Ridiculous history podcast. This was all completely new to me, but it was very surprising and interesting.
‘History’s dumbest criminals, part 2’ on the Ridiculous history podcast. This was very funny. I didn’t get around to listening to part 1 (yes, I know).
‘The Olympics used to have an… extraordinarily weird art competition’ on the Ridiculous history podcast. Enjoyable!
‘The history of the toilet’ on the Half-arsed history podcast. (The name of the podcast is a bonus, I suppose.)
I’m always looking for new podcasts to listen to that are both entertaining and educational. If you have recommendations, I’d like to know!
Animal photos
On the weekend, I went to Leipzig Zoo. As a result, I saw many animals. Here are some photos I took:
Well, that was a lot! But most of it came from November, so I still have a lot left to share in the next part before I catch up with my regular posting schedule.
If you liked this and learnt something new, I hope you share it with your friends or on social media. And I hope you subscribe if you haven’t already!
See you next time!
– Saloni
By the way, I’m looking for an intro or intermediate online course on demography; if you know a good one, please let me know. Thank you :)
Correction: The original post said that the Duke Endowment program reduced infant mortality rates in the communities by 10%, with a larger reduction for black infants than white infants (16% vs 10%). This included a typo: it should have said 16% vs 7%, i.e. a 7% reduction for white infants. I’ve corrected this (16/01/2023).
This data comes from the ‘underlying causes of death’ listed in death certificates. The data has been ‘age-standardised’, which means they are adjusted for the population getting older.
Why not look at ‘survival rates’? When people want to know whether cancer treatment has improved, they often look at 5 year survival rates (i.e. the percentage who survive at least 5 years after they are diagnosed with a cancer) and whether they have grown. But they depend heavily on the level of testing. If the level of screening grows a lot, and more ‘benign’ cases are detected, it can look like survival is improving, even if there has been no improvement in treatment. This problem is much more limited when it comes to mortality rates, which depend on the level of testing for severe cases.
Congrats on submitting your PhD! I'm in the final 3 months of writing mine. Enjoyed reading this - thanks
> I included a cat pun here at first but then I deleted it to save you from wincing
Fair enough, but now I want to know what it was