Statistics Canada National Population Health Surveys (NPHS) Amir Erfani, PhD. Department of Sociology Nipissing University North Bay, Ontario, P1B 8L7 Presentation prepared for the Pre-Conference on Health Over Life Course, October 14, 2009, University of Western Ontario 1
Outlines Objectives of NPHS Methodology – Nature of Survey – Sampling & Sample Size – Attrition & Weights – Questionnaire & Variables Studies that used NPHS Our study Bibliography 2
Objectives of NPHS Providing health measures helping the development of public policy Understanding -factors affecting health -Impact of health care utilization on health status Studying the dynamic process of health and illness over time Link health survey data to administrative data (i.e., vital statistics) 3
Nature of The Survey NPHS is basically a longitudinal panel survey: -Cycle 1 (1994/1995) Cycle 6 (2004/2005) -Cycle 2 (1996/1997) Cycle 7 (2006/2007) -Cycle 3 (1998/1999) Cycle 8 (2008/2009) -Cycle 4 (2000/2001) Cycle 9 (2010/2011) -Cycle 5 (2002/2003)Cycle 10 (2012/2013) Cross-sectional survey at cycles 1-3 Three components: 1.The Households (main component) 2.The Health Institutions (taken over by CCHS) 3.The North components (taken over by CCHS ) This presentation is on the Household component 4
Sampling & Sample Target Population: household residents in the ten Canadian provinces in 1994/1995, excluding persons -living on Indian Reserves and Crown Lands, -residents of health institutions, -full-time members of the Canadian Forces Bases & -some remote areas in Ontario and Quebec. Longitudinal Sample (the panel): -Cycle 1 (1994/95) = 17,276 persons 4+ years old, containing 2024 children 4 years old) -Cycle 7 (2006/07) = 10,992 individuals 11+ years old Full response to all Cycles 1-7 5
The Panel by Province Attrition rate from cycle 1 to cycle 7 = 36.4% 6
Attrition & Weights The main Source of Attrition: Refusal & unable to trace -Cycle Twinning Approach (Pooling Repeated Observations) Complex Sampling Design of NPHS: stratification, clustering, and unequal probabilities: -Use weights to get representative estimates "Longitudinal Full" Weight: - WT6BLF -applies to the subset of respondents who responded to all 7 cycles (10,992 persons) Bootstrap weights: ensure the reliability of variance estimates - B5lngf file for “Longitudinal Full” subset (10,992) 7
Household Component’s Questionnaire The Questionnaire includes 1.Core content: - identical questions over the cycles 2.Focus Content: -changes from cycle to cycle Questionnaire Cycle 7: collects data about -Health of the longitudinal respondents, -determinants of health, -socioeconomic information, -demographic information about all members of the longitudinal respondents’ household 8
Variables in the Questionnaire Cycle 7 Household Recode Variables Demographic variables of household members - Date of birth, age, sex, marital status, education, etc. Dwelling characteristics -Type of dwelling, ownership, number of bedrooms Health Component Variables General Health -Perceived health and stress -Sleep (pattern and troubles) -Height & Weight -Body Image 9
Variables in the Questionnaire Cycle 7 Preventive Health -Blood pressure, pap smear, mammogram, & reproductive health Health Care Utilization: - Use of health or medical professionals and practitioners -Hospital & clinics -Home Care Restrictions of Activities Mental Health 10
Variables in the Questionnaire Cycle 7 Physical Activities UV Exposure Repetitive Strain Injuries Stress - Ongoing problems -Work Stress -Mastery Medication Use Smoking Alcohol 11 Social Support Language Education Employment Income Nutrition (Focus content) – Food choice, – Food supplement use & fruit and vegetable consumption – Soft Drink & Milk Consumption Food insecurity (Focus content) Determinants of Health
Studies that used NPHS See Statistics Canada Health Reports. E.g.: Explanatory Studies -Marital breakdown and subsequent depression -Depression and risk of heart disease -Income and psychological distress: The role of the social environment Health Trends and Estimates -Medication use among senior Canadians -Medication use among pregnant women -The journey to quitting smoking -Trends in weight change among Canadian adults 12
Our Study Andrea Wilson & Amir Erfani. Socioeconomic History and Preventable Disease: A Comparative Analysis of Fundamental Cause Theory Research Question: If the risk factors stratifying health are eliminated, will socioeconomic disparities in health remain? Hypothesis: SES should be more strongly associated with diseases that are more preventable than with less preventable diseases, and SES should have a stronger relationship to health in countries where high economic inequality and no universal health insurance leads to greater competition for resources. 13
Our Study Data: Canada NPHS (Cycles 1-5); the U.S. PSID (Panel Study of Income Dynamics) Method: Latent Class Analysis Why did we use NPHS? -Rich data on SES and risk factors over 5 cycles (now 7 cycles!), allowing us to generate -trajectories of SES (Fundamental cause), -trajectories of alcohol consumption & smoking (risk factors) -and to choose highly and less preventable diseases as outcome (cardiovascular/cancer). -Studying causal relationships is doable -Large sample size -Less modifications in the core longitudinal questions over the cycles 14
Trajectories of Income Levels Trajectories of income levels of Canadians during , taken from Latent Class Analysis of income 15
Bibliography Statistics Canada Statistics Canada National Population Health Survey Household Component Cycle 1 to 7 (1994/1996 to 2006/2007) Longitudinal Document. Statistics Canada National Population Health Survey Household Component Cycle 7 (2006/2007) Questionnaire. Statistics Canada Healthy Today, Healthy Tomorrow? Findings from the National Population Health Survey, Health Reports. Catalogue no.: MWE, vol. 3 no. 1. Fitzmaurice, G.M., Laird, N.M., and Ware, J.H Applied Longitudinal Analysis. New York: Wiley 16