Reposting due to comment loss
ET – First, your hypothetical death. No, that’s not where I’m coming from. I am speaking only of causes of death on death certificates.
‘Is all the data that states that so many people died from car accidents or ovarian cancer or alcoholism or knife crime or whatever else wrong?’
Not at all. If a death certificate states that death occurred as a result of heart failure (direct cause) due to internal bleeding, as a result of being hit by that damn bus again, then the bottom of line 1 will show one of the numerous ICD codes relating to accidental death, specifically the one relating to buses. We may safely add this death to the number of people ‘killed by buses’ or ‘killed in road accidents’ because it is the UNDERLYING cause. We can also use the other information on the certificate to produce other stats which may be useful, such as ‘in all deaths by bus, x percent involved internal bleeding’ or ‘heart failure is the result of internal bleeding in x percent of cases’ or even use the line 2 info to say ‘x percent of people suffering from diabetes are killed by buses’. What we cannot do is simply add up all the occurrences of ‘internal bleeding’ from all death certificates and announce ‘Internal bleeding caused x number of deaths’. Because it would be a rubbish statistic and totally unreliable as a basis for decision making.
I did look at the WHO guidance for death certification and recording. According to this Covid-19 must be at the bottom of line 1, i.e it must be the underlying cause of death, to be counted. They give a number of useful examples where there are co-morbidities. For instance, a death certificate which has 1a Acute respiratory distress syndrome due to 1b covid-19 due to 1c HIV, would be a certification error, as HIV should be on line 2. Another example has 1a heart failure due to 1b myocardial infarction, with covid-19 listed on line 2 i.e. in the ‘other significant factor contributing to death’. In big red letters it states This is NOT a covid-19 death.
The guidance also notes
‘Deaths due to COVID-19 are different from COVID-19-related (or COVID-19–associated) deaths. These may be deaths due to accidental or incidental causes, or natural causes when COVID-19 is not identified as the underlying cause of death according to ICD coding guidance (see Section 4.2)’
On the other hand, the ONS states ‘For overall counts such as in the weekly deaths release, the Office for National Statistics (ONS) uses the concept of “deaths involving COVID-19”, which means any mention of U071 or U072 anywhere on the certificate.
The start of this exceeding tortuous discussion was initiated by Duck’s suggestion that perhaps the lower death rates in Asia might be accounted for by differing methods of recording. Perhaps he was right.