The methodologic issues involved in evaluating trends in mortality by Census Division (CD) are explored. The three trend measures investigated included Poisson regression, ordinary least squares (OLS), and a measure based on Raveh's measure of monotone association for time series. / Assumptions required for the parametric methods of Poisson regression and OLS were not satisfied. The non-parametric method based on the ranking of the measure of monotone association did not require such assumptions. Significant spatial correlation in the measure of monotone association was observed for all-cause mortality. / Using deaths among persons 35-69 years for 3 causes, comparisons were carried out by cross-tabulations, Pearson rank correlation coefficients, and an investigation of discrepancies. Minor discrepancies resulted from differences in trends of age-specific death rates and from CDs with small populations. Rather than distinguishing further between the three trend measures, guidelines should be developed for minimum populations at risk required for mapping as well as methods for combining regions.
Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMM.59637 |
Date | January 1990 |
Creators | Semenciw, Robert |
Publisher | McGill University |
Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
Language | English |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Format | application/pdf |
Coverage | Master of Science (Department of Epidemiology and Biostatistics.) |
Rights | All items in eScholarship@McGill are protected by copyright with all rights reserved unless otherwise indicated. |
Relation | alephsysno: 001116131, proquestno: AAIMM66376, Theses scanned by UMI/ProQuest. |
Page generated in 0.0016 seconds