Examining Diabetes Inequalities Using Individual and Area-level Income in Urban and Rural Saskatchewan
dc.contributor.committeeMember | Neudorf, Dr.Cory | |
dc.contributor.committeeMember | Muhajarine, Dr.Nazeem | |
dc.contributor.committeeMember | Teare, Dr.Gary | |
dc.contributor.committeeMember | Leis, Dr.Anne | |
dc.creator | Thurairasu, Lisa 1988- | |
dc.date.accessioned | 2017-01-03T15:08:44Z | |
dc.date.available | 2017-01-03T15:08:44Z | |
dc.date.created | 2014-10 | |
dc.date.issued | 2014-11-14 | |
dc.date.submitted | October 2014 | |
dc.date.updated | 2017-01-03T15:08:44Z | |
dc.description.abstract | Background and Rationale: Diabetes is a growing health problem in Saskatchewan, disproportionately affecting people in low socioeconomic groups compared to people more well off (1,2). The absence of individual-level socioeconomic status (SES) data in administrative databases necessitates researchers to use area-level SES as a proxy for measuring health inequalities (3). This study compares individual and area-level income for measuring diabetes inequalities in a CCHS study sample and estimated population to determine whether an area-based measure can be used as a proxy for individual-level data in urban and rural Saskatchewan. Methods: Health administrative data containing diabetes cases was linked to the 2007-2008 CCHS combined cycle containing individual-level income, which was merged with the 2006 Canadian Census containing area-level income at the dissemination area (DA)-level. Individual and area-level incomes were compared for the ‘unweighted’ and ‘weighted’ CCHS population. The ‘unweighted’ population was the CCHS study sample of Saskatchewan respondents in which no sampling weights or bootstrap weights were used. Contrarily, the ‘weighted’ population was the estimated Saskatchewan population derived from applying sampling weights and bootstrap weights to the study sample. The statistical methods used in this study included descriptive analyses, bivariate analyses, and multivariable analyses. Odds ratios of the final multivariable models for the ‘unweighted’ and ‘weighted’ population were compared to determine whether area-based income underestimates diabetes inequalities. The software, SAS Enterprise Guide 6.1, was used to analyze the data. Results: There was relatively low agreement between individual and area-level income. Individual and area-level income had varying patterns of influence on diabetes in the study sample and in the estimated population. Overall, income gradients were larger in the ‘unweighted’ population compared to the ‘weighted’ population. The over-representation of older individuals (who have a higher proportion of diabetes than younger people) and the under-representation of younger individuals in the study sample compared to the estimated population and the ‘actual’ Census population as seen in Table 4.1, could have led to the stronger association in the (‘unweighted’) study sample (4,5). However, as individuals generally earn higher income with age (6) and as seniors often earn low income (2), these factors can only partially explain the observed patterns. Of individuals with diabetes, gradients were observed between area-level income and proportion of individuals with diabetes, while a pattern was present for individual-level income, in the study sample and estimated population. Within each income category, the reverse pattern occurred for individual-level income and area-level income in the study sample and estimated population. However, age-adjusted rates revealed clear downward gradients. Based on the final logistic model, the odds of the ‘unweighted’ individual-level income model produced a downward gradient, while statistically insignificant U-shaped curves were present for the ‘unweighted’ area-level income model, ‘weighted’ individual-level income model, and the ‘weighted’ area-level income model. The odds of diabetes was also smaller in rural areas compared to urban areas in the final model, contradictory to literature (7,8), however the results were statistically insignificant. Study Implications: This study provides decision support for the use or disuse of area-level income in measuring diabetes inequalities accurately in Saskatchewan. Future research, especially qualitative research, can examine the mechanisms of individual and area-level income on health. | |
dc.format.mimetype | application/pdf | |
dc.identifier.uri | http://hdl.handle.net/10388/ETD-2014-10-1806 | |
dc.identifier.uri | http://hdl.handle.net/10388/7649 | |
dc.subject | health inequalities | |
dc.subject | diabetes | |
dc.subject | individual versus area-level | |
dc.subject | socioeconomic status | |
dc.subject | income | |
dc.title | Examining Diabetes Inequalities Using Individual and Area-level Income in Urban and Rural Saskatchewan | |
dc.type | Thesis | |
dc.type.material | text | |
thesis.degree.department | Community Health and Epidemiology | |
thesis.degree.discipline | Community and Population Health Science | |
thesis.degree.grantor | University of Saskatchewan | |
thesis.degree.level | Masters | |
thesis.degree.name | Master of Science (M.Sc.) |