Getting to Know the GDD - Live Webinar (Recording)
Recorded 9:00-10:15 AM EST Thursday, June 11, 2020
00:00-02:06 - Welcome/housekeeping
02:07-11:59 - An overview of our history and goals
12:00-22:00 - How we collect, standardize, and aggregate data
22:01-30:46 - An overview of our dietary intake prediction model
30:47-39:37 - Published findings from the GDD team
39:48-48:13 - The various aspects of our website and data tools
48:14-51:22 - Our strategic partnerships with global organizations
51:23-58:00 - FoodEx2 harmonization of food data
58:01-1:00:55 - Closing remarks
1:00:56-1:14:31 - Question and answer session
Q&A session transcript
This is an abridged version of the 15-minute question and answer session from this webinar.
When do you think the data would be available publicly for other researchers like what GBD does?
Draft estimates for 12 of our dietary factors are currently available. Our plan is to make estimates for all other dietary factors available this summer. Primary survey data will be made available via the site on a rolling basis as we obtain data sharing agreements from our data owners.
The self-reported dietary intake data is commonly under-reported. Could GDD's method correct the potential under-reporting?
Yes. We consider multiple 24-hour dietary recalls to be the 'gold standard' assessment method, which the model uses as a reference when assessing inherent bias for various survey methods. For example, if FFQs systematically under-report intake compared to multiple 24hR, the model accounts for this bias in its estimates.
Are data expressed in raw edible amounts to fully compare food intakes?
We ask data owners to contribute their data in as-eaten amounts.
What method was used to exclude or remove dietary record with implausible intake?
We used various resources to determine both individual and group plausibility ranges for adults and children for all of our diet factors. Any values which exceeded this range were investigated further. These plausibility guidelines and our workflow for determining the biological plausibility of data will be published later this year.
Any plan to publish the pregnant group data?
Primary survey data on pregnant women will be made available via our site on a rolling basis as we obtain data sharing agreements from data owners. We are also working on an expansion of our model to estimate intake by pregnancy status but do not anticipate it will be published this summer along with our other estimates.
Can you discuss where the data on diet-related adverse health outcomes come from or how they are incorporated into the model and how they are related to the data on food and nutrient intake from the surveys in GDD?
We do not incorporate diet-related adverse health outcomes into the GDD model. Publications on diet-disease relationships from the GDD team utilize a cumulative risk assessment approach to analyze the relationship between GDD dietary estimates and GBD health outcome data.
In absence of country nutrient specific analysis and food composition database (e.g Trans fatty acid), how do you estimate at the country level?
We do not derive nutrient intake from food data. Any nutrient data held by the GDD has been generously contributed by our data owners after utilizing their local food composition tables. If countries or populations do not have nutrient intake data, our model's nested hierarchical structure allows us to borrow information from surrounding countries/populations and covariate data sources to estimate nutrient intake.