Meeting of the Healthy Eating & Active Living (HEAL) Sub-Core Indicator Group
Teleconference
Thursday, November 27, 2007; 1:30-3:00 a.m.
DRAFT MINUTES
Present:Shanna Hoetmer, Peggy Patterson, Deborah Moore, Catherine Shea, Wendy Young
Regrets: Elsa Ho, Krystina Nickerson, Mamdouh Shubair
Chair:Shanna Hoetmer
Minutes:Wendy Young
Distributed via email prior to meeting:
-Minutes
1.0Approval of Minutes of September 20, 2007
Minutes approved
2.0Approval of Agenda
Agenda approved
3.0Update from the CIWG meeting
3.1Indicator comments are provided to help with interpretation
3.2National indicators should be restricted to Statistics Canada and CIHI
3.3Timelines and tasks will be defined based on grant requirements.
3.4The HEAL Subgroup can recommend removing the Physical Activity Frequency
4.0Small group indicator update and questions
4.1Physical Activity Index
Recommendations related to physical activity index are attached. The number of “not stated” respondents (coded as 9) in CCHS cycle 3.1 needs to be added from share files that PHU have.
Action: Add derived variable for 3.1
Action: Clarify the difference between the corresponding national indicator leisure-time physical activity and physical activity
4.2Physical Activity Frequency
The small group can circulate a recommendation to remove the Physical Activity Frequency with the rationale. Physical Activity Frequency is used only to calculate the Physical Activity Index. Physical Activity Frequency is no longer used in the field.
Action: Small group circulate recommendation with rationale
4.3Fruit and Vegetable Consumption
Correspondence from Brenda Wannell of Statistics Canada related to “not stated” and proxy reported is attached to the minutes
The description of this indicator is: “Percentage aged 12 and over that have consumed five or more servings of fruits and vegetables daily”. The survey asks the respondent how often they usually eat or drink the food (for example, once a day, three times a week, twice a month). At this time there is no validation that servings and time are equivalent. The note “The National Health Indicator presents the information as number of times fruits and vegetables are consumed daily rather than the number of servings consumed daily” can be kept.
4.4Food Insecurity now called Food Security
Action: Small group to distribute a summary email
4.5Cost of a Nutritious Food Basket
The costing protocol is being updated to accompany new Public Health standards. No further work will be completed on this indicator until the new protocol has been circulated
4.6BMI
The small group has just started to revise it.
4.7BMI for Adolescents
Small group is trying to make the indicator comments more concise
The method of calculation has changed
BMI is based on self report. Women tend to underestimate weight and men tend to overestimate height. For population, BMI is the only measure available. Waist circumference and waist/hip ratio are not available
5.0Other
5.1Membership
The physical activity group will examine chronic disease indicators. Each group must have an epidemiologist.
6.0Date of next meeting
Monday January 28 at 1:30
Correspondence related to proxy reporting and not stated
______
From: Hoetmer, Shanna
Sent:Thursday, October 04, 2007 11:50 AM
To:'Brenda Wannell ()'
Cc:Shea, Catherine; Peggy Patterson
Subject:Fruit & Vegetable Consumption
Hi Brenda,
As I mentioned at the Core Indicators for Public Health meeting last Friday, our group has some questions regarding the treatment of "not stated" for the derived variable for fruit and vegetable consumption(i.e. FVCEGTOT in cycle 3.1).
The percent of respondents "not stated" for Ontario was 4.5% in 2005 (using the Ontario Share File), but this ranges from about 3 to 7% depending on the public health unit. Some of these are due to proxy interviews and the rest are due to missing responses in the questions that are used to derive the variable. For example, for Ontario 1.7% of the 4.5% "not stated" were proxy interviews.
How are these responses treated at Statistics Canada? In some cases we have recommended excluding or including "not stated" at some threshold level, and I know that in the previous iteration of this indicator, exclusion of the "not stated" was the recommendation. I know that there are issues either way. If we include the not stated from the denominator, it reduces our estimates, and the proportion of not stated varies over place and time (or cycle). If we exclude the not stated when the proportion is quite high, it brings to question the representativeness of the estimates. If a threshold is used, from year-to-year or place-to-place the inclusion/exclusion may be different, which would make trends and comparisons difficult. And then how should respondents "not stated" because of proxy interview be treated? Differently than the other "not stated" respondents?
Please let me know if you or any of your colleagues can offer any insight into this issue, specific to the fruit & vegetable consumption module.
I really appreciate your help,
Shanna
Shanna Hoetmer, MHSc, Epidemiologist
Epidemiology and Research, Public Health
York Region Community and Health Services Department
465 Davis Drive, Tannery Suite 318-09
Newmarket, ON, L3Y 2P1
Tel: 905-830-4444 x 4501
Fax: 905-954-4002
Email: From: [mailto:
Sent: Friday, November 16, 2007 2:21 PM
To: Hoetmer, Shanna
Cc: ;
Subject: RE: Fruit & Vegetable Consumption
Hi Shanna -
You have raised very good points, in factyour questions/commentsinspiredusto work on anote for ourusers about the treatment of not stated, since this question keeps coming up. This text is still in the works.As a result, I neglected answering your specific question.I'm very sorry to keep you waiting for a reply!
It is difficult to make a recommendation since thistreatmentof not stated isa matter of preference.In general, there seems to be a divided opinion on whether not stated should be included or excluded from the denominator, especiallywhen there is a larger proportion of non-response.As you know, within the Health Indicators project, weusea consistent approach. To inform users about missing values, we publish estimates for thenon responseas a separate column inour detailed tables, mind you theseare often suppressed due to small counts. We canappreciate that for specific variables(where the NS is relatively high)one may wish toremove the non response, for the reasons you mentioned,however by not including them in the denominator, you are assuming that the missing values are random, and this is not always the case.
To respond to the specific question onhow we handleproxy reporting,I consultedSylvain Tremblaywho's responsible for CCHS .1 content, and learned the following: Proxy are treated as non-response- i.e.,like all not stated (NS).We don't make the distinction between NS proxy and other NS because of the confusion this may end up causing and also because we believe that in generalthe weak proportion of proxy will have a negligible effect on the estimates. The NS for proxy are clearlynot random. For most proxy interviews, assumingthe proxy rules are correctly applied by interviewers,we are talking abouta clearly unhealthiersegment of the population.
I've copied Sylvainon thismessage, in case he has additionalinsights to offer, relevant to the fruits and vegetable consumption module, that may be helpful.
Again, my apologies for taking this long to get back to you,and have a great weekend,
Brenda
Brenda Wannell
Project manager/ Gestionnaire de projet
Health Indicators and CCHS .1 Dissemination/ Indicateurs de la santé et Diffusion de l'ESCC .1
Health Statistics Division / Division de la statistique de la santé
Statistics Canada / Statistique Canada
613-951-8554
Page 1 of 5
MINUTES: Healthy Eating Active Living Sub-Group Meeting November 27, 2007