– or –
Please login or register to participate.
critter Oct 11, 2017 11:35 AM
Looking at the latest AirQuality2017-15-29 pdf report and more specifically at the table 10.1. for Belgium. I see annual estimated premature death rates for PM2.5 of 6860 and NO2 of 6470. This makes for an annual combined rate somewhat between 60 and 120/100000 depending dependency of both risk factors. On the other hand, when I go to http://ihmeuw.org/47yq, I find an estimated death rate attributed to all air pollution of 46/100000. Can anyone explain the discrepancy to me?
Replies (5)
EEA Oct 12, 2017 01:00 PM
You should focus on PM2.5 Co=0, that is, 8340 deaths; and NO2 C0=20; that is 1870 deaths, as these are the counterfactual concentrations used in previous years.
The difference with the study you mentioned is probably due to the use of a different counterfactual concentration. See for instance footnote 42.
Kind regards.
critter Oct 13, 2017 09:07 AM
Thanks, this is a bit what I thought.
I was trying to find back the corresponding C0 number for PM2.5 in the GBD calculation. The EEA states that the recent GBD threshold is in the range of 2.5 to 5. I supposed that it was therefore comparable to the EEA report. But if I calculate back the EEA numbers of premature deaths or YLL from 2.5 to 5micrograms/m^3, I get to premature death rates and YLL which are roughly in agreement with GBD. The problem is that it's the C0=0 number which is twice as bit which gets into the press ... and they like BIG SCARY NUMBERS.
We have to put the mortality and morbidity risk of air pollution in relation to other risk factors. Here we are talking around 5% of mortality and morbidity compared to more than double for smoking or dietary ...
critter Oct 19, 2017 09:05 AM
Coming back to my earlier post. In the meantime, I received an answer to a parallel question I emailed to the contact address of the GBD. Here it is (unfortunately not sufficiently specific):
Dear Dr. Ritter,


Thank you for your message; I apologize for the delay in responding. The discrepancy is likely due to the sources being used to underlie the estimates. GBD uses a wide variety of data sources and methodological approaches to produce the most accurate and comprehensive estimates possible. If you are interested in why GBD differs from official country estimates, the exact reason will depend on the cause and the country in question, but there are several possible explanations:

· Official data sources assign a large proportion of deaths to non-specific causes (i.e. garbage codes, or causes that are not official causes of death).

- Official statistics generally do not report these deaths as being associated with a particular cause.

- GBD systematically evaluates the so-called “garbage codes” and “redistributes” them by age, sex, and location to likely underlying causes.

- This sometimes results in GBD showing different numbers than official statistics.

- This also sometimes results in GBD showing higher numbers than studies of single causes or groups of causes, as such analyses also usually do not evaluate the contribution of non-specifically coded deaths.

· The sum of cause-specific deaths in each year must equal the total annual deaths in each population (e.g. each age/sex/location).

- This is only possible when all causes are analyzed in a mutually-exclusive, collectively-exhaustive fashion and this is not a feature of studies on individual diseases or groups of diseases.

- In the GBD, this is accomplished by independently estimating the levels and trends in all-cause mortality for each age/sex/location and independently estimating the levels and trends in mortality due to each cause of death for each age/sex/location. The GBD ensures that for any age/sex/location, the sum of number of deaths for each cause does not exceed the total number of deaths as calculated by the all-cause mortality analysis.

· The official data sources were not available for incorporation into GBD estimates. The reasons could be lack of centralization, lack of sharing, or long delays (sometimes multiple years) in countries releasing their data.


To view what sources were incorporated in GBD 2016 for air pollution, please refer to the GBD 2016 Data Input Sources Tool: http://ghdx.healthdata.org/[…]?components=6&risks=85. Specific countries can be chosen using the ‘Locations’ dropdown menu.


Please let me know if you have any further questions.

EEA Oct 20, 2017 10:27 AM
Thank you for this additional information. As pointed out by the GBD expert and also in our report, results are not fully comparable for a number of reasons. And just to add another one to the already mentioned ones: how concentrations are calculated/estimated.
Kind regards.