Basic science and statistical understanding

[MrsTerryPratchett]

I'm not a scientist. I do have a BSc. but nothing more impressive. I do wince when I hear some assumptions that people make who don't have basic science (and statistical) knowledge. What basic scientific or statistical principles do you wish the general public (and members thereof) knew about?

I'll start. Causation and correlation. A correlation doesn't prove causation. Particularly when that correlation is a correlation of one. "I had the flu vaccine and got the flu really badly" isn't causation.

I wish people had a very basic understanding of risk. My father is over 60 and overweight. He had covid - symptoms in march/april and tested positive for antibodies. DSis is convinced the antibody test was wrong because "he definitely would have died if he'd had it".

minuscule OP : not a term as such but a totally undefined amount that actually reveals a misunderstanding of mortality rates.

I would add a lack of understanding of the context of data and how to actually interpret - I work in research and I've found this to be a major problem not just with the general public but with doctors and with people who should, in their professional life, understand it. For example, not understanding that if you don't define your sample you will be unable to draw conclusions about your results. There was a well-known issue with that in covid reporting when it was discovered that deaths were being reported as covid deaths even if the person tested positive more than 28 days before their death. That was later changed but I know personally of circumstances where people who tested negative for covid and then died were still recorded as covid deaths on the basis that the doctor reckoned they had it. Therefore the definition of a covid death is 'anything that looks like it might be a covid death' which essentially renders the data meaningless.

Any I've missed?

False positive/false negative.

There has been an awful lot of "My friend, who's a nurse, says x, y and z" this year, both on MN and on social media. Nurses are trained to nurse and they mainly do an excellent job, but it doesn't mean they are generally particularly hot on stats.

My mum was one for 20 years, and she has no more idea about statistics and data than your average wo/man on the street.

Added to the above problem is the fact that covid doesn't actually have to be the direct cause of death (hence died with covid), meaning that what's being recorded is any death where the person also tested positive for covid (even if the covid presents no symptoms at all) or any death where the doctor reckons covid was present (don't give a shit about actual cause). What conclusions can be drawn from that data? Well one conclusion is that 'people die and some of them have covid while doing it.' That's about it.

To be clear, I'm not saying at all that covid doesn't cause death. I'm sure it does. It just that we have absolutely no idea of the real extent to which it causes death (rather than happens to be present at the same time as death) because of how shit the data is.

All of these, plus

Who funded the work and why?

Always label everything

What surprised me with the Cambridge Analytics situation was the fact that people didn't seem to understand how completely useless reams and reams of context-free 'data' is. I was glad to see that CA finally admitted their data was just a heap of shit

www.reuters.com/article/us-facebook-privacy-cambridge-analytica-idUSKBN1HV17M

Because a lot of data is just shit. Without very rich context - who is this person, why did they make that choice, what were the conditions surrounding that choice - data points are just noise, useless little blips on a graph. The context has to be extremely specific too, plus the data goes stale really fast. Using data properly is crazy complicated and doesn't just involve a grasp of stats, it also involves and ability to interpret - ie understand what the fuck the whole thing means. To date, I've met two people (one is my DH) who can actually interpret data in a meaningful way. And I work in the industry.

TheDailyCarbuncle that depends on the data source - the death certificate data does give true cause of death, so distinguishes between dying with covid and dying from covid. It is just a bit slower to be collected, because people often don’t register deaths for a week or two.

But the idea that we have no idea who’s died of covid and who hasn’t is unfortunately deliberate misinformation peddled by covid deniers.

And that is a COVID thing because people keep saying, "the average age of people who die from COVID is over the average age people generally die" like that's an indication that we shouldn't care. That very much depends on how those numbers are calculated.

Not only that, but as NHS England sent out the instruction in March/April that if a care home called for an ambulance, none were to be sent; then the death rate for people in care homes is going to greater (because they got no hospital care even oxygen therapy), than any other different age group, who were admitted to hospital and got the treatment they needed to save their life, as far as is humanly possible!

@Changechangychange

TheDailyCarbuncle that depends on the data source - the death certificate data does give true cause of death, so distinguishes between dying with covid and dying from covid. It is just a bit slower to be collected, because people often don’t register deaths for a week or two.

But the idea that we have no idea who’s died of covid and who hasn’t is unfortunately deliberate misinformation peddled by covid deniers.

I'm not sure where you're getting your information but my information is direct from death certs and the ONS. By necessity, recording of COD tends to be somewhat inaccurate - if it's not a suspicious or sudden death, the doctor records the most likely COD without verifying it. That's normal and accurate enough for what's needed. But in a pandemic, COD should be verified, not taken at the word of the certifier. It is absolutely not verified and that is a solid fact, mostly because the staff aren't available to verify it. Therefore COD is recorded as it usually is - many times with a best guess. People with negative covid tests are having their COD recorded as covid - that is also a solid fact, verifiable. It's not like anyone is hiding it - it's standard practice.

@MrsTerryPratchett

Maybe we should start a list of terms that people bandy aroud:

PCR
RNA
Exponential
R0

Any I've missed?

Peer reviewed. Mumsnetters love dismissing things on the basis they haven’t been peer reviewed (while anything that has been takes on the status of absolute truth).

@Changechangychange

A direct quote from the guidance on certifying deaths during the pandemic says:

^For example, if before death the patient had symptoms typical of COVID19 infection, but the test result has not been received, it would be satisfactory to give ‘COVID-19’ as the cause of death, tick Box B and then share the test result when it becomes available. In the circumstances of there being no swab, it is satisfactory to
apply clinical judgement.^

ie the doctor does not have to have any particular evidence of covid to certify the COD as covid. That's the official, government guidance.

Ok all you clever clogs - please explain life expectancy to me! I keep hearing it's 82 point something. So if I were 60 tomorrow - can I reasonably expect another 22 point something years?

I enjoyed reading through this thread, I have a BSc so can dissect some data but don't claim to be an expert in the slightest.
I do know a huge amount of people who just don't get it. They think they are clever spouting off "statistical data" but we both know they couldn't even manage GCSE maths. The media has a lot to answer for I think, but if I had to chose one thing I learnt this year it would be how completely stupid some people are.

@JudyShakes

Ok all you clever clogs - please explain life expectancy to me! I keep hearing it's 82 point something. So if I were 60 tomorrow - can I reasonably expect another 22 point something years?

No, you can expect more than that because you already didn't die at some of the risky points, e.g. you didn't die in the first year of your life. You didn't die in your 50s so the risk of dying in your 50s is now irrelevant to you. You would expect longer than that, if you're of average health (you'd have to consult an acturial table to see how much longer!)

It's easiest to think about if you think about the extreme end. If you're 85 and average life expectancy is 82, clearly you don't 'expect' to die three years ago; that didn't happen so your own life expectancy is clearly longer.

@JudyShakes

Ok all you clever clogs - please explain life expectancy to me! I keep hearing it's 82 point something. So if I were 60 tomorrow - can I reasonably expect another 22 point something years?

Well that's average life expectancy isn't it. So it's a generalisation. You can bet more specific estimates depending on you health, sex, ethnicity, wealth, where you live and a comprehensive list of other factors.

But in average people live to 82. Whatever ... my nana died last week aged 93 my husbands bets friend died 4 years ago aged 39. But 82 is where the bell curve peaks I guess.

TheDailyCarbuncle thanks, I’ve filled in plenty of death certificates in my time and am aware of how to complete them.

If somebody dies in an RTA, or of metastatic breast cancer, you cannot put “Covid pneumonitis” in 1a just because they had a positive swab a few weeks ago.

Those instructions date to the first wave, where people did not have easy access to covid swabs outside of secondary care. You still have to believe that, medically, Covid was the primary cause of death. It is not a way of hiding cancer deaths and strokes behind Covid.

But given the high false negative rate on most Covid tests, if someone has very characteristic Covid symptoms, has been in close contact with people who did test positive for Covid, and dies suddenly with no alternative clear cause then Bayes’ theorem dictates that yes you should put Covid on the death certificate, regardless of a single negative test.

Place marking because I have things to do atm but when time allows I have many things to say.

For those of you who understand the death certification process, can you explain if this is right or not

Someone dies of breast cancer
You put that in 1a?

Then if they have a positive CV test, then that is put in the second box.

But this death is recorded as both breast cancer and covid - is that right?

• by recorded, I mean it classes as both a breast cancer death and a Covid death?

what's being recorded is any death where the person also tested positive for covid (even if the covid presents no symptoms at all) or any death where the doctor reckons covid was present (don't give a shit about actual cause)

we have absolutely no idea of the real extent to which it causes death (rather than happens to be present at the same time as death

This is what I am taking issue with. Death certificates do not have “Covid pneumonitis” as main cause of death (part 1A diagnosis) if you had a positive swab but no symptoms, or if you tested positive but clearly died of something else. But you know that very well.

hamstersarse

1A is the main cause, 1B and 1C are things that caused 1A (ie 1A E. coli septicaemia, 1B ascending cholangitis. You don’t need 1B or C unless they directly caused 1A.

Part 2 is any other contributor, even if it didn’t directly cause death. So in my example above, you could put ischaemic heart disease and diabetes in part 2 if you thought having those made the patient more likely to die of sepsis.

You might put covid in part 2 if you thought having had covid had weakened them. You wouldn't put an asymptomatic positive swab in there. And no, it wouldn’t be counted as a “covid death” if it was in part 2, in the same way that my example patient wouldn’t be classed as a death from heart disease.

I am still not clear @Changechangychange

When we say we have had '500 deaths today' - how do we truly understand what the situation is?

Say someone had terminal breast cancer and is admitted to hospital for something related to the deterioration of the disease but then is tested positive in hospital (as it seems is happening a lot - hospital infections) - would this person be counted on the Covid death statistics that the media report?