Am I the only one who is bothered that these threads don’t start on Monday anymore?
Posting a request from a past open thread again: Does anyone have a table of probabilities for major (negative) life events, like divorse or being in a car accident? I ask this to have a priority list of events to be prepared for, either physically or mentally.
The lifetime risk of developing cancer is 44 % in males and 38 % in females. The lifetime risk of dying from cancer is 23 % in males and 19 % in females. It’s worth mentioning that the methods for gathering medical mortality statistics are pretty biased, if not completely bonkers.
ETA: Apparently a new WHO recommendation for filling death certificates was introduced in 2005-2006 and this caused a significant drop in pneumonia mortality in Finland.
I’m not entirely sure if it works this way in the whole EU, but it probably does. It’s more complicated than what I explain below, but it’s the big picture that matters.
The most common way to record mortality statistics is that the doctor who was treating the patient fills a death certificate. There are three types of causes of death that can be recorded in a death certificate. There are immediate causes of death and there are underlying causes of death. There are also intermediate causes of death, but nobody really cares about those because recording them is optional. The statistics department in Finland is interested in recording only the underlying causes of death and that’s what gets published as mortality statistics. Only one cause of death per patient gets recorded.
If someone with advanced cancer gets pneumonia and dies, a doctor fills the death certificate saying that the underlying cause of death was cancer and the immediate cause of death was pneumonia. Cancer gets recorded as the one and only cause of death by the statistics department. Depending on the patient, possible underlying causes of death could also be alcoholism, coronary heart disease or alzheimer’s disease or whatever is accepted by a department that checks these certificates.
The doctor’s opinion of whether it was the pneumonia or the chronic disease that killed the patient doesn’t really matter. If he also fills the underlying cause of death as pneumonia, he gets a scolding letter and has to fill it again until he gets it right.
What if the patient has several chronic diseases that could have been underlying causes of death? Well, you only get to pick one, and only that one gets recorded as the cause of death. You can list the other diseases too as contributory causes of death, but this doesn’t really effect any statistics. I guess it would be less biased to flip a coin or something, but I think most doctors just pick something fitting.
A colleague of mine once tried to record pneumonia as the underlying cause of death, the patient was an alcoholic (not sure how bad it was). He got a letter saying he should fix the certificate and that people in developed countries don’t die of pneumonia anymore. Wonder why that is...
in USA they can fill in 20 secondary causes on the death certificates and all
the anonymized death certificates since 1959 are available online from NCHS
in computer-readable form to check/search for conditions.
Irregularities usually appear when there is a switch from one ICD-Code to a new one,
so in 1969,1979,1999.
Other irregularities are often checked, compared with other states,countries,conditions
and the reason discovered
What if the patient has several chronic diseases that could have been underlying causes of death? Well, you only get to pick one, and only that one gets recorded as the cause of death. You can list the other diseases too, but not as causes of death.
It seems I miscommunicated here. What I meant to say that listing these other diseases has no meaningful impact on the mortality statistics, although technically speaking they are causes of death. If the point is to gather accurate statistics, listing them feels like a consolation prize, because statisticians don’t seem to be interested in them.
In Finland a direct translation for these would be “contributory causes of death”. That’s probably the same thing as secondary causes of death. The problem is, it’s difficult for someone who makes these into statistics to know how important they were. Almost anything the patient has can be listed as a contributory cause of death.
Even a bigger problem is that listing them is completely optional. If almost nobody fills them in properly (because they usually have better things to do), that is another good reason for a statistician not to use them.
Is filling in the secondary causes mandatory in US? Are there clear restrictions for what can be listed? If not, I’m not sure if they provide all that useful information, statistically speaking. Are they really used in meaningful way in any statistics?
Irregularities usually appear when there is a switch from one ICD-Code to a new one, so in 1969,1979,1999.
I suppose WHO recommendations for filling these certificates impact the US too.
I have a pet interest—carefully looking at how standard, universally-accepted, real-life, empirical data is collected and produced and whether it actually represents what everyone blindly assumes it does. In the field of economics, for example, closely examining how, say, the GDP or the inflation numbers are calculated is… illuminating.
The problem is that the problems aren’t summarizeable in a neat half a page list. And it’s not like the calculations are wrong, rather they are right under a certain set of assumptions and boundary conditions—and the issue is that people forget about these assumptions and conditions and just assume they’re right unconditionally.
For an introduction take a look at e.g. Shadowstats. I don’t necessarily agree with everything there, but it’s a useful starting point.
I twitch when changes in GDP are reported to a tenth of a percent—it seems to me that it couldn’t be measured with such precision. Do you think I’m being reasonable?
My own (uninformed) intuition is that GDP changes would be much more accurate than absolute GDP values, just because systematic errors could largely cancel out.
I’m assuming that the GDP is some sort of measure of the health of the economy—that’s why people are concerned with it. The health of the economy seems to me like rather an approximate sort of thing.
GDP—Gross Domestic Product—basically means the sum of the value (in the economic sense) of all goods produced domestically during a given period, e.g. a year.
If you want to measure the “health of the economy”, that’s quite different. You’ll have to define what do you mean by that expression and then decide which measurements do you want to consider. For example, some people might consider the unemployment rate to be one those measurements, or, say, the Gini index, or the median income, or… the possibilities are endless.
Why do people measure the value of all the goods produced domestically during a year?
From Wikipedia: “GDP was first developed by Simon Kuznets for a US Congress report in 1934. … After the Bretton Woods conference in 1944, GDP became the main tool for measuring a country’s economy.”
Yes, the GDP number is, of course, imprecise. By itself it’s not a problem—most of our measurements are imprecise.
I am not sure what are you getting at. Do you think that GDP is useless or cannot be measured or what?
Am I the only one who is bothered that these threads don’t start on monday anymore?
I like it for purely selfish reasons. I can’t easily post between Sunday bedtime and Tuesday evening. If I want to post and the thread starts on Monday, my post will be less visible.
Am I the only one who is bothered that these threads don’t start on Monday anymore?
Posting a request from a past open thread again: Does anyone have a table of probabilities for major (negative) life events, like divorse or being in a car accident? I ask this to have a priority list of events to be prepared for, either physically or mentally.
The lifetime risk of developing cancer is 44 % in males and 38 % in females. The lifetime risk of dying from cancer is 23 % in males and 19 % in females. It’s worth mentioning that the methods for gathering medical mortality statistics are pretty biased, if not completely bonkers.
Would you be willing to expand on this?
ETA: Apparently a new WHO recommendation for filling death certificates was introduced in 2005-2006 and this caused a significant drop in pneumonia mortality in Finland.
I’m not entirely sure if it works this way in the whole EU, but it probably does. It’s more complicated than what I explain below, but it’s the big picture that matters.
The most common way to record mortality statistics is that the doctor who was treating the patient fills a death certificate. There are three types of causes of death that can be recorded in a death certificate. There are immediate causes of death and there are underlying causes of death. There are also intermediate causes of death, but nobody really cares about those because recording them is optional. The statistics department in Finland is interested in recording only the underlying causes of death and that’s what gets published as mortality statistics. Only one cause of death per patient gets recorded.
If someone with advanced cancer gets pneumonia and dies, a doctor fills the death certificate saying that the underlying cause of death was cancer and the immediate cause of death was pneumonia. Cancer gets recorded as the one and only cause of death by the statistics department. Depending on the patient, possible underlying causes of death could also be alcoholism, coronary heart disease or alzheimer’s disease or whatever is accepted by a department that checks these certificates.
The doctor’s opinion of whether it was the pneumonia or the chronic disease that killed the patient doesn’t really matter. If he also fills the underlying cause of death as pneumonia, he gets a scolding letter and has to fill it again until he gets it right.
What if the patient has several chronic diseases that could have been underlying causes of death? Well, you only get to pick one, and only that one gets recorded as the cause of death. You can list the other diseases too as contributory causes of death, but this doesn’t really effect any statistics. I guess it would be less biased to flip a coin or something, but I think most doctors just pick something fitting.
A colleague of mine once tried to record pneumonia as the underlying cause of death, the patient was an alcoholic (not sure how bad it was). He got a letter saying he should fix the certificate and that people in developed countries don’t die of pneumonia anymore. Wonder why that is...
in USA they can fill in 20 secondary causes on the death certificates and all the anonymized death certificates since 1959 are available online from NCHS in computer-readable form to check/search for conditions. Irregularities usually appear when there is a switch from one ICD-Code to a new one, so in 1969,1979,1999. Other irregularities are often checked, compared with other states,countries,conditions and the reason discovered
It seems I miscommunicated here. What I meant to say that listing these other diseases has no meaningful impact on the mortality statistics, although technically speaking they are causes of death. If the point is to gather accurate statistics, listing them feels like a consolation prize, because statisticians don’t seem to be interested in them.
In Finland a direct translation for these would be “contributory causes of death”. That’s probably the same thing as secondary causes of death. The problem is, it’s difficult for someone who makes these into statistics to know how important they were. Almost anything the patient has can be listed as a contributory cause of death.
Even a bigger problem is that listing them is completely optional. If almost nobody fills them in properly (because they usually have better things to do), that is another good reason for a statistician not to use them.
Is filling in the secondary causes mandatory in US? Are there clear restrictions for what can be listed? If not, I’m not sure if they provide all that useful information, statistically speaking. Are they really used in meaningful way in any statistics?
I suppose WHO recommendations for filling these certificates impact the US too.
Very interesting, thank you.
I have a pet interest—carefully looking at how standard, universally-accepted, real-life, empirical data is collected and produced and whether it actually represents what everyone blindly assumes it does. In the field of economics, for example, closely examining how, say, the GDP or the inflation numbers are calculated is… illuminating.
Details?
The problem is that the problems aren’t summarizeable in a neat half a page list. And it’s not like the calculations are wrong, rather they are right under a certain set of assumptions and boundary conditions—and the issue is that people forget about these assumptions and conditions and just assume they’re right unconditionally.
For an introduction take a look at e.g. Shadowstats. I don’t necessarily agree with everything there, but it’s a useful starting point.
Thanks.
I twitch when changes in GDP are reported to a tenth of a percent—it seems to me that it couldn’t be measured with such precision. Do you think I’m being reasonable?
My own (uninformed) intuition is that GDP changes would be much more accurate than absolute GDP values, just because systematic errors could largely cancel out.
GDP as reported is the product of a particular well-defined calculation. That product can easily be calculated to whatever precision you feel like.
When you say “it couldn’t be measured with such precision”, how do you define the Gross Domestic Product that couldn’t be measured precisely?
I’m assuming that the GDP is some sort of measure of the health of the economy—that’s why people are concerned with it. The health of the economy seems to me like rather an approximate sort of thing.
GDP—Gross Domestic Product—basically means the sum of the value (in the economic sense) of all goods produced domestically during a given period, e.g. a year.
If you want to measure the “health of the economy”, that’s quite different. You’ll have to define what do you mean by that expression and then decide which measurements do you want to consider. For example, some people might consider the unemployment rate to be one those measurements, or, say, the Gini index, or the median income, or… the possibilities are endless.
Why do people measure the value of all the goods produced domestically during a year?
If nothing else, there has to be a fudge factor because some of the economy is underground.
From Wikipedia: “GDP was first developed by Simon Kuznets for a US Congress report in 1934. … After the Bretton Woods conference in 1944, GDP became the main tool for measuring a country’s economy.”
Yes, the GDP number is, of course, imprecise. By itself it’s not a problem—most of our measurements are imprecise.
I am not sure what are you getting at. Do you think that GDP is useless or cannot be measured or what?
I like it for purely selfish reasons. I can’t easily post between Sunday bedtime and Tuesday evening. If I want to post and the thread starts on Monday, my post will be less visible.