The overall response rates for Canada and the US were 65.5% and 50.2% [18
]. Given that the JCUSH was conducted using random digit dialing, invalid telephone numbers created difficulties in the conduct of the survey and in the computation of the response rate. Given the small number of telephone companies in Canada, Statistics Canada was able to validate telephone numbers for working residential telephones selected for dialing in Canada, while for the US sample, Statistics Canada was unable to validate telephone numbers [18
]. Using validated telephone numbers in Canada, an overall household response rate of 72% was obtained. One respondent was selected for each of the responding households, with an overall person-level response rate of 90.9%, yielding an overall response rate of 65.5%.
For the US, because of the large number of unvalidated telephone numbers, the response rate was calculated differently. In the US sample, the resolution rate represents the proportion of sample telephone numbers that could be positively identified as residential or nonresidential; the resolution rate was 80.4%. The majority of the unresolved telephone numbers reached persons or machines that hung up before identifying themselves or rang with no answer. The cooperation rate measures the proportion of known households within which an interview was completed; the cooperation rate was 62.4%. The overall response rate is the product of the resolution and cooperation rates, 50.2% [18
Table presents basic descriptive information on JCUSH respondents. Although the two populations are similar, obesity appears to be more prevalent in the US, while the proportion with more than a high school-level education is lower in Canada.
Sociodemographic Characteristics of the Samples
Table compares the demographic characteristics of the US and Canadian JCUSH surveys to contemporary surveys of the non-institutionalized population in both countries: the 2003 US Behavioral Risk Factor Surveillance System Survey [44
] and 2003 Canadian Community Health Survey [45
]. (Comparisons between the JCUSH and census data are found in [27
], and [47
].) The age distributions in the JCUSH surveys match the age distributions in the contemporary population health surveys in both countries. In both countries, those who were married were overrepresented in the JCUSH, and those with less than a high school education were underrepresented.
Comparison of JCUSH to 2003 US Behavioral Risk Factor Surveillance System Survey and 2003 Canadian Community Health Survey
The prevalence of a number of chronic conditions in the two countries based on results from the JCUSH is reported in Table . The prevalence of a number of chronic conditions appears to be higher in the US than in Canada.
Comparison of Prevalence of Chronic Conditions, US and Canada, in % (18+)
Three different methods were used to compare the health of those residing in Canada and the US. First, Table shows unadjusted results and adjusted results based on a linear regression model that included gender, race, education, income, smoking status, and BMI. For both unadjusted and adjusted results using established standards for interpreting HUI3 scores [48
], mild morbidity burdens were observed for respondents 18 through 39 years regardless of country. For the 40+ group, Canadians appear to experience higher levels of HRQL than US residents. In the adjusted results, the differences in mean HUI3 scores for the 40-64 age groups and 65+ age groups are clearly quantitatively important (≥0.03), statistically significant for the 40-64 age groups, and almost statistically significant for the 65+ age groups. In the unadjusted results, the differences are smaller; for the 40-64 group, the difference of 0.01 is statistically significant and sufficiently large in the context of population health to be meaningful. Results of comparisons of health in the white-only Canadian and US populations, both unadjusted and adjusted, are similar to the results for comparisons of the entire sample.
Unadjusted and Adjusted Mean Health Utilities Index Mark 3 Scores by Age and Country
A comparison of the distribution of levels within each of the eight HUI3 attributes between Canada and the US (data not shown) reveals that the prevalence of moderate and severe disability is systematically higher in the US than in Canada.
Second, mortality rates are lower in Canada [4
]. The effects of the differences in mortality rates between the countries are reflected in the estimates of the additional life expectancy of 19-year-olds (Table ). Life expectancy at birth is also higher (and the infant mortality rate is lower) in Canada than in the US.
Comparisons of Health, Poverty, and Income Inequality in Canada and the United States
Third, HALE estimates reflect differences in both morbidity and mortality (Table ), with a difference of 2.7 more years of "perfect health" in Canada compared to the US. The 95% confidence interval around the estimate of HALE for Canada is 51.5 to 52.5 years; for the US, the 95% confidence interval is 48.9 to 49.7 years. The difference between the estimates of HALE for Canada and the US is statistically significant with p < 0.0001. The 95% confidence interval around the estimated difference of 2.7 years is 2.0 to 3.4.
Counterfactual #1 shows what the HALE in Canada would have been if Canada had experienced US mortality rates and Canadian morbidity; it differs from the actual estimate by 1.8 years of perfect health. Counterfactual #2 shows what HALE in Canada would have been if Canada had Canadian mortality rates but US morbidity. The difference between the actual and the counterfactual is 0.9 years of perfect health. Thus, differences both in mortality and in morbidity are important in accounting for the higher HALE in Canada. The results also indicate that mortality differences are quantitatively more important than morbidity differences in accounting for the difference in HALE.
Given that differences in poverty and social and economic inequality may be important in accounting for differences in HALE in Canada and the US, Table includes data on poverty rates and income inequality based on data from the Organisation for Economic Co-operation and Development 2008 [55
]. The poverty rate is defined as the proportion of individuals with incomes below 50% of the median. Income inequality is measured using the Gini coefficient that reflects the difference between the actual distribution of household income and an ideal distribution of complete equality. Gini coefficients range from 0.00 (complete inequality) to 1.00 (complete equality). Clearly, poverty is more prevalent in the US than in Canada, especially among the elderly. Further, income inequality is substantially higher in the US than in Canada.