How does Cronometer handle changing nutrition values for packaged foods?

Citizen_Snips

Cronometer maintains a huge database (CRDB) of packaged/labeled foods. How does Cronometer handle it when one of these items changes nutrition values?

Suppose I've added an item from the CRDB to my nutrition log in the past and it changes nutritional information, either because the actual nutritional content changed or the package and/or serving size changed. How does Cronometer and the CRDB handle this?

If the single entry in the CRDB is updated, all of my past entries of this food will update automatically and be incorrect. This seems like a bad option.

Multiple versions of the same item could be added to reflect changes in serving size and/or nutritional content. This seems like the best option.

comanchesue

Oh man, I surely would not want to have to multiple options for the same food! I think this would be a spectacularly bad idea in practice. It is easy enough to edit the serving size when we enter the food.

Karen_Cronometer

@Citizen_Snips when we find a new version of a product, we retire the old version to preserve your past data and then replace it with a new version that you can add to your diary with the updated nutrition information.

In the case where foods have been updated with more accurate or new data, then we would apply these changes to the current version of the food and this will update your past food diaries to increase the accuracy of your data.

Lector

@Karen_Cronometer as a software engineer, I agree this is the correct approach.

But how does the Cronometer team distinguish between more accurate data, and new versions of the product? What if the product label looks the same in user-submitted issues?

Karen_Cronometer

We know it's more accurate data when our lab-analyzed data sources are updated or new information is added.

For packaged foods, we always assume it's a new product formulation if there are any changes in the nutrition information. We also keep track of when a food was added to our database, when it was last updated, and whether the label is the older American style or the newer format (American 2016). We can also compare to the manufacturer's website, if we don't have enough information to go on.

Citizen_Snips

After testing Cronometer, it seems I've found the answer to my original question - packaged food/supplements are added to the database with seemingly no regard to whether or not there is an existing entry in the database. For example, suppose I want to add Optimum Nutrition Gold Standard 100% Whey Protein Powder, French Vanilla Creme to my Diary. Typing "optimum french" into the search bar turns up 4 entries that all refer to the same product, each with slightly differing nutritional information (there are technically 7 results, but the other 3 refer to a casein protein powder that I assume is actually a different product).

Do these 4 entries refer to the same product added to the CRDB at different points in time, with each having the correct nutritional information at the time it was added? Or do one or more of these entries contain a typo? There's really no way to know with the way the CRDB is organized.

I realize that with such a long product name the risk of duplicate entries increases, but I think the main takeway from this is that if you really value having accurate nutritional information in your diary, the best practice is to avoid "crowd-sourced" information and just rely on your own custom foods for anything that doesn't exist in the USDA database.

Karen_Cronometer

The duplicate entries are usually a result of manufacturers sourcing slightly different ingredients or ratios of ingredients depending on supply and price and these factors can depend on your location.

They make a different label with each batch they produce where the nutrition information differs. You can try searching by barcode number to narrow it down.

Another option is to use a generic version of whey protein from NCCDB to estimate the nutrition you're getting.

Lector

Thanks Karen, that might explain why I'm seeing slightly off data in over 80% of the barcodes I scan. Apparently the manufacturers generate new labels, but not new barcodes when they tweak the formulation?

when we find a new version of a product, we retire the old version to preserve your past data and then replace it with a new version that you can add to your diary with the updated nutrition information.

Does that process affect only my entries, or the entire CRDB? What if I have an old stock of a food with long shelf life, and someone else scans the barcode of an updated version of it that just came out? I'll be eating my old stock but with the mismatched information from the new version?

Karen_Cronometer

When a food is retired, it is removed from the food search. So if you search for a food that was retired (by barcode or by name) it will bring up the new one instead for all users.

The retired food still exists though. For example, if you copy and paste it from a past entry in your diary, if will past the old version. You could also save it as a custom food this way too. Open the older version in your diary and use the menu to save a copy, if you like.