Study finds connections between birth month and health
newsroom.cumc.columbia.edu
newsroom.cumc.columbia.edu
A phrase later in the articles shows that further study may find a connection; implying that they don't have one already.
A more accurate HN link title is "Study could help scientists uncover new disease risk factors pertaining to birth date".
It's a complicated universe so full causal explanations are pretty much always unattainable in the real world, but I'd hope for some identification and discussion of a mechanism.
The connections that MAY be found might be things like: - November babies getting exposed to extreme cold during the first month of development - Spring babies exposed to high pollen
The most important information in the article.
I'd be far more surprised if there were no link!
I know it's not new knowledge – that's kind of why I'm unsure about the skepticism. We know that these effects already exists, and this is a wider-scale analysis of potential instances of those effects.
It's certainly possible that this bias is more related to parents behaviour. Some parents would conceive in August resulting in their children being born in May.... Other parents would conceive in January resulting in children being born in October. People which conceive during different seasons might be different both genetically and behaviourally which could influence offspring's health by means of genetic factors but also habits learned from their parents during childhood (dietary choices for example).
Or not.
Obviously astrology is bunk in that planets don't affect you. But it's conceivable that some claims in astrology are based on evidence that is actually correlated to birth month and seasons. We would expect to see inverse correlations for these factors in the southern hemisphere.
"The spacing of human births in a large population is described by a Poisson distribution, which shows that every year in a city of one million people about 4,000 pairs of time twins are born 5 minutes apart or less."
"Unlike the Roberts and Greengrass approach (count each time twin against every other time twin within one day, which would have produced nearly 600,000 pairs), serial correlation counts each time twin once only, thus minimizing the risk of artifacts. So the test conditions could hardly have been more conducive to success. But the results are uniformly negative, see Table 2. The effect size due to astrology is 0.00 ± 0.03."
A 'connection', to me at least, implies causality. No causal relationship has been found here.
I never feel like a god of statistics. My complaint here wasn't even about statistics. It's not even a specialty of mine. Note my complaint was about words.
The meaning and usage of words, particularly the impressions and messages they convey to people--the ideas, beliefs, as well as the certainties they establish and invoke when spoken, written, heard, and read--on the other hand, most certainly is something I care very much about. There are many areas outside a probabilistically determined statistical model in which properly understanding, specifying, and communicating correlation and causality are useful and important.
A similar environmental effect was observed with anyone born during the Dutch hunger winter being significantly more susceptible to certain diseases:
This motivated Columbia’s scientists to compare 1,688 diseases against the birth dates and medical histories of 1.7 million patients treated at NewYork-Presbyterian Hospital/CUMC between 1985 and 2013.
I immediately thought "hmm, tread carefully there fellas."
But you're right, they didn't do full cross validation. They performed a smaller scale validation where they trained on 80% and tested on 20%. Their writing is a bit muddled here, but it looks like 12 of the 16 novel associations (those not previously reported in literature) were validated (as in passed the above test). They did not perform validation on previously reported associations that they confirmed - presumably because they drew from an independent dataset from the original publication, and so this test constitutes an independent validation of the original study.
That's a very important point. The effect (if it were to be proved) could be reversed in the southern hemisphere, and completely absent in non-city environments. I'm plucking those hypotheses out of thin air, but it seems obvious that the risk goes up because of some underlying reason.
Perhaps being born in New York in October increases the chance that you'll spend most of your first few months of life in a damp-infested flat without much natural light. Perhaps it's because of increased stressed levels at school associated with being one of the oldest in your peer group.
Without knowing the underlying causes, this is just an interesting tidbit.
It would be interesting to see a similar study for the southern hemisphere.
They also compare some of their data to data from Denmark. So it's all Northern hemisphere, mid-latitudes.
[1] http://jamia.oxfordjournals.org/content/jaminfo/early/2015/0...
Interesting stuff to think about. Seems to be what this article is getting at. Imagine how magnified it was when agriculture played a bigger role.
Spurious correlation achieved by misusing statistics. Not to mention publication bias.