Operational and Methodological Lessons Learned from the 2003 Joint Canada/U.S. Survey of Health: Operational and Methodological Lessons Learned from the 2003 Joint Canada/U.S. Survey of Health Catherine Simile National Center for Health Statistics Ed Rama Statistics Canada


Slide2: The findings and conclusions in this presentation are those of the author(s) and do not necessarily represent the views of the Centers for Disease Control and Prevention.


Overview: Overview Background Lessons learned Questionnaire Design Sampling and Data Collections Data Processing and Release


Slide4: BACKGROUND


How It Started: How It Started Both NCHS and STC were involved in international efforts to improve cross national comparisons of health data NCHS/STC yearly interchanges to discuss common interests Idea for a joint survey launched at the September 2000 Interchange


Objectives: Objectives Produce highly comparable data on the Canadian and American populations unaffected by difference in data collection methodology on the following core indicators: Health care Functional status Health status Risk factors


Objectives (cont.): Objectives (cont.) Influence content of the each country’s ongoing, national health surveys to enhance comparability and data quality Develop a model for successful collaboration towards standardizing concepts


Anticipated Results: Anticipated Results Greater understanding of the implications of survey standardization Generation of hypotheses for observed differences, if any


Organization: Organization Project Team –Responsible for day to day operations Steering Committee – project oversight All interviews conducted from Statistics Canada’s Regional Offices using RDD and CATI


Survey tasks: Survey tasks Questionnaire design - questions taken from ongoing surveys in both countries -the Canadian Community Health Survey and the National Health Interview Survey Average duration of questionnaire – 25 minutes Editing specifications specific to questions Interviews conducted in English, Spanish (US), and French (Canada)


Population Covered: Population Covered Residents of Canada and the US aged 18 and older living in private dwellings with telephones Canadian samples stratified by province US samples stratified by 4 regions Sample designed to produce reliable national estimates for 3 age groups (18-44, 45-64, 65 and over) by gender


Sample Size: Sample Size Canada: 3,505 US: 5,183


Timeline: Timeline 2001 – Questionnaire design and Cognitive testing June 2002 – Final content decided July – Oct. 2002 – Design and testing of collection applications Oct. 2002 – Interviewer training Nov. 4, 2002 – Data collection starts


Timeline (cont.): Timeline (cont.) Mid-November 2002 – U.S. sample introduction letters sent February 2003 – U.S. sample re-screened to try to identify out-of-scopes among the ring no answers March 2003 – Conversion letter sent


Timeline (cont.) : Timeline (cont.) March 31, 2003 – End of collection in Canada April 2003 - Data collection in US from Toronto office extended targeting specific cases follow-up letters sent to answering machines and ring no answer June 2003 collection resumed in Vancouver to convert refusal cases July 14, 2003 - collection officially ended September – official response rates calculated


Slide16: LESSONS LEARNED


Cognitive Testing: Window of Opportunity : Cognitive Testing: Window of Opportunity Normal practice: U.S.: individual one-on-one English interviews in Washington D. C. agency office Canada: focus groups in English in Ottawa and in French in Montreal JCUSH: Combined approach: one-on-one interviews and focus groups for both U.S. and Canada in agency labs and off-site


Cognitive Testing, (cont’d): Cognitive Testing, (cont’d) Forced us to think about comparability differently: More difference between subsamples in country than between countries Led to a new way of doing cognitive testing in U.S.


Translation: Implementation of New Guidelines: Translation: Implementation of New Guidelines Languages used in this survey English (both countries) French (Canada only, required by law) Spanish (U.S. only, customary)


Steps from the Guidelines:: Steps from the Guidelines: Pre-translation preparation--NO Selection of contractor--YES Completion of translation from final text--NO Review of translation—YES Bilingual review used survey and topic experts, U. S. Census interviewers, translators, and French speaker from Canada Adjudication--YES Development of survey instrument --YES Pretest of survey, including translators--NO Selection of interviewers YES Training of Interviewers--NO Incorporation of feedback from the field--YES


Data Collection: Plan for Assumptions: Data Collection: Plan for Assumptions “Clean sample” and resolving cases Cooperativeness of Canadian and U.S. population different Differences in implementing legal requirements impacted discretion and authority of data collection staff


DATA COLLECTION RESULTS: Resolution, Cooperation and Final Response Rates, JCUSH: DATA COLLECTION RESULTS: Resolution, Cooperation and Final Response Rates, JCUSH


Resolving Cases: Resolving Cases Sampling methodologist for each country determined number of telephone lines necessary to reach intended sample sizes # of lines selected Targeted Sample size United States 32,009 5,000 Canada 10,334 3,500


Resolving Cases (Cont’d): Resolving Cases (Cont’d) Definition of clean sample different GENYSIS removed 1/3 of original U.S. sample The remaining 2/3 sent to Statistics Canada assumed to be “clean” Working residential numbers cannot be verified in the U.S. like they can in Canada


Cooperation: Resolution: Cooperation: Resolution Evident in 20% U.S. unresolved cases: Ring no answer 31% Unresolved answering machine 9% Hang up during introduction (likely) 8%


Cooperation: Refusals: Cooperation: Refusals Percentage of all eligible cases: U.S. Canada Refusals 21% 14% Breakoffs 11% <.01% TOTAL 33% 14%


Collection Monitoring: Collection Monitoring Not necessary to monitor unresolved cases where 100% of the cases can be relatively easily resolved The monitoring system used works well for a Canadian sample, but not for a U.S. sample where much of resolution work is done by field staff Throughout the data collection period, the staff was always uncertain as to what was happening, and so had difficulty ascertaining how to allocate resources Did not help that U.S. staff could not easily travel to Canada


Discretion and Authority of Data Collection Staff: Discretion and Authority of Data Collection Staff Restrictions and delays to convert non response (all communications approved by NCHS Institutional Review Board)


Data Release: Data Release Followed Statistics Canada’s usual practice. Preparation for release included both: editing and review of microdata for public release collaborative analytical report released at the same time Collaborative analysis hampered by legal restrictions on data access


Summary: Summary Windows of Opportunity Provided New ways of doing cognitive testing First opportunity to implement new translation guidelines Communication and Clarifying Assumptions Crucial “Clean sample” and resolving cases Differences in cooperativeness of Canadian and US population require different monitoring and collection strategies Difference in implementing legal requirements not insignificant in their impacts impacted the discretion and authority of interviewers difficult to collaborate in analysis


Slide31: Dream lofty dreams, and as you dream, so shall you become. James Allen Failure is the opportunity to begin again more intelligently. Henry Ford 1863-1947, American Industrialist


Slide32: Access Data and Reports NCHS website at www.cdc.gov/nchs/nhis.htm Statistics Canada website at www.statcan.ca


Contact Information: Contact Information Catherine M. Simile, Ph.D. National Center for Health Statistics Division of Health Interview Statistics 3311 Toledo Road, Room 2115 Hyattsville, MD 20782 Phone: (301) 458-4499 Email: cus4@cdc.gov