Friday, January 25, 2019

2015 CCHS-Nutrition PUMFs and Master Files

I have a researcher who is interested in knowing the fruit and vegetable consumption of Canadians and wonders if either the PUMF's or Master Files from the 2015 Canadian Community Health Survey (CCHS)-Nutrition will provide such data. The researcher would ideally like to know the amount consumed (e.g. number of apples), the variety (e.g. what kind of apple?), and whether the food item was organic.  

I have reviewed my notes from the Jan. 30, 2018 webinar on the 2015 CCHS-Nutrition​ and scanned the documentation for the eight files in ODESI. I have also had a quick look at the PUMF's in ODESI. It is clearly a very complex survey! As far as I can tell from the documentation, this type of data and level of detail is not available from the PUMF's. Is that correct? 

​​Slide # 23 from the webinar notes that questions about fruit and vegetable consumption were dropped from the Health Component module of the 2015 CCHS-Nutrition. I wasn't sure how to interpret the following information from the Reference Guide to Understanding and Using the Data:  2015 Canadian Community Health Survey -- Nutrition (June 2017):

"The PUMF for 2015 CCHS-Nutrition will include data on nutrients from foods, a summary of vitamin/mineral supplement use, and the health questionnaire data. Additional data at more detailed levels such as food, ingredient, and recipes along with Canada Food Guide tiers may be included depending upon the file structure and the results from a mandatory confidentiality review.” (p. 41)

If the information is not available from the PUMF's, would it be available through the Master Files which are in the RDC's? Unfortunately, metadata for these Master Files is not yet posted on Statcan's NESSTAR server. 

We’ve received the following response from subject matter:

“The CCHS-Annual continues to have the Fruit and Vegetable consumption question module (FVC) as part of their core content asked every year.  The module asks about the frequency of eating fruit in the past month, but doesn’t get into the detail of what fruit.

The 2015 CCHS-Nutrition asked detail about what specific foods are eaten, but only for one day, a 24-hour dietary recall.  The goal of this is to get data on what nutrients are consumed – not to get estimates of what exact foods provide those nutrients.  However, since apples are frequently consumed throughout the population, it is possible to use the survey data to estimate how many apples are eaten on any given day, and the characteristics of people who eat them.  This is not recommended for less frequently consumed foods, say lasagna.

The survey coded each food reported by respondents with Health Canada’s Canadian Nutrient File (CNF) codes, a subset of which are included in the survey documentation known as the FDC file.  This file contains the name description of a food, ingredient or recipe, a code specific to it, and a string of nutrient values for one gram of that food.  The CNF is also available on Health Canada’s website, so the researcher can look at that until the Nesstar metadata are available.  The data are limited to the level of detail available there.  The CNF/FDC file does not have separate entries for any organic versus non-organic foods because there is little difference in the nutrient content. 

For apples, the following information is relevant to the researcher’s questions:

  • There are data related to apples:
    • a general code for apples (summary Bureau of Nutritional Science food group variable FDC_FGR = 40B),
    • various detailed codes (variable FID_FID ) but these only differentiate when the nutrient content is different (e.g. fresh versus canned, dehydrated, frozen).
    • There are no data related to specific varieties of apples (e.g. Spartan versus Gala, McIntosh, Pink Lady) because the differences in nutrients per gram of apple are not statistically important (i.e. the data for all fresh apples would be coded to one code)
    • The code variable FID-FID has to be used in conjunction with the label attached to the code, which are located in two other variables on the FDC file:  FDC_DEN (Food name – CNF – English) and FDC_DFR (Food name – CNF – French).  For example: FID_FID = 1487 FDC_DEN = “Apple, canned, sweetened, sliced, heated” 
  • All of these variables are available in both the master file and the PUMF.  The analysis using the PUMF might be a little easier because the PUMF HS file includes derived variables that aggregate the gram amount of food eaten that day for certain foods, and apple is one of those foods (the variable is BNSD40B – “Gram Weight – Apple”).”

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.