PDF
Version (1,377 KB)
Influential Observations in Weighted Analyses: Examples
from the National Longitudinal Survey of Children and Youth
(NLSCY) - View
Abstract
Jennifer J Macnab, JJ Koval, KN Speechley and MK Campbell
The estimation of heritability for twin data based on
concordances of sex and disease - View Abstract
Hongzhuan Tan, Mark Walker, France Gagnon and Shi Wu Wen
Breast cancer trends in Manitoba: 40 years of
follow-up - View
Abstract
Alain A Demers, Donna Turner, Daojun Mo and Erich V Kliewer
Smoker preference for “elastic cigarettes”
in the Canadian cigarette market - View Abstract
Michael O Chaiton, Neil E Collishaw and Aaron J Callard
Trends in mortality from diabetes mellitus in Canada,
1986–2000 - View
Abstract
Jinfu Hu, Glenn Robbins, Anne-Marie Ugnat and Chris Waters
Chronic Diseases in Canada (CDIC) is a quarterly scientific journal focusing on current evidence relevant to the control and prevention of chronic (i.e., non-communicable) diseases and injuries in Canada. The journal publishes a unique blend of peer-reviewed feature articles by authors from the public and private sectors that may include research from such fields as epidemiology, public/community health, biostatistics, behavioural sciences and health services. Authors retain responsibility for the contents of their papers, and opinions expressed are not necessarily those of the CDIC Editorial Committee nor of Health Canada nor of the Public Health Agency of Canada.
Authors: Jennifer J Macnab, JJ Koval, KN Speechley, MK Campbell
Abstract:
This paper highlights the impact of survey weights on model fit in
multiple linear regression with specific reference to the National
Longitudinal Survey of Children and Youth (NLSCY) and provides
recommendations for the treatment of influential observations.
Multiple linear regression was used to estimate the association
between child and family factors in the preschool years and
vocabulary development at school age. Analyses were performed with
and without survey weights. The model fit was assessed by examining
the distribution of the studentized residuals and the change in the
regression coefficients that would occur if an observation were
removed. Two summary measures of influence, Dffits and Cook's D
are reported. The models were refit excluding influential
observations. Weighting of the linear model resulted in previously
non-influential observations having an undue influence on the
estimation of the regression parameters in the weighted model. The
influential observations were driven primarily by the size of the
survey weight as opposed to unusual values of x and y. Researchers
working with large national health surveys such as the NLSCY and
the National Population Health Survey (NPHS) are advised to include
a detailed influence analysis before any final conclusions are
made.
Authors: Hongzhuan Tan, Mark Walker, France Gagnon and Shi Wu Wen
Abstract:
Heritability is an important measure in chronic disease
epidemiology. Almost all developed ?methods of heritability
estimation for dichotomous outcomes in twin data are based on
concordance of monozygous (MZ) and dizygous (DZ) twins. However,
most existing twin registries, which provide a unique and efficient
opportunity to assess the putative genetic basis of diseases, do
not have zygosity information. We developed a method that can be
used to estimate the heritability for twin data with no information
on zygosity. The only conditions on using this method are that the
studied disease incidence is not strongly related to sex, and the
distribution of zygosity is in accordance withWeinberg's rule.
Using asthma twin data which has histological confirmed zygosity,
we compared the results of our method with Holzinger's formula.
The heritability of asthma was 24.88% (95%CI 21.98% - 27.78%) and
29.83% (95%CI 22.28% - 37.38%) estimated by ourmethod and
Holzinger's formula respectively.We conclude that our new
method can be used in the estimation of heritability with large
twin register data with no zygosity information available.
Authors: Alain A Demers, Donna Turner, Daojun Mo and Erich V Kliewer
Abstract:
This study reports a comprehensive array of breast cancer
statistics for Manitoba for a 40-year period. Data from the
Manitoba Cancer Registry were combined with the provincial
population-based registration file to determine trends in breast
cancer incidence, prevalence and mortality rates, as well as
survival and the probability of being diagnosed with breast cancer
in the next 10 years. The age-standardized incidence rate of breast
cancer increased by 0.99/100,000 women per year over the 40 years
of follow-up (69.6/100,000 women in 1960, 109.9/100,000 women in
1999).Mortality rates peaked in 1986 (35.7/100,000 women), while
the 1999 mortality rate (26.0/100,000 women) was almost comparable
to the 1960 rate (22.4/100,000 women). No significant trend in
mortality rate was observed over the 40-year period. The 5-year
prevalence rate of breast cancer increased by 8.6/100,000 women
per. Between 1960-64 and 1995-99, 5-year survival increased from
0.62 to 0.86. The probability of being diagnosed with breast cancer
in the next 10 years increased the most for women 60 years of age.
The breast cancer burden in Manitoba is rapidly evolving mainly
because of the increasing incidence and the better survival of
cases.
Authors: Michael O Chaiton, Neil E Collishaw, and Aaron J Callard
Abstract:
Elastic cigarettes are characterized by yields of constituents that
increase proportionally faster than smoke volume as cigarettes are
smoked more intensely. Elasticity may function to overcome physical
limitations in increasing puff volume during nicotine-seeking
behaviour. The purpose of this study was to determine if there are
elastic cigarettes in the Canadian cigarette market, and to
determine smoker preference for elastic cigarettes. Elasticity was
calculated for 115 brands in the Canadian filtered cigarette market
for puff volumes of 44 and 56 ml. Puff volumes, nicotine and tar
deliveries were obtained from earlier published documents.1 Sales
data were used as a proxy for smoker preference. Ordinary least
squares regression was used to determine the association of sales
and elasticity in the Canadian cigarette market. The cigarette
brands ranged from a mean elasticity value of 1.21 to 0.67. Of the
115 Canadian cigarette brands tested, 23 brands had a mean
elasticity value significantly over 1.00, making them elastic.
After adjusting for brand, the average elastic cigarette sold an
average of 361 million cigarettes while an inelastic cigarette sold
89.5 million cigarettes (p < 0.0001). The difference in sales
between elastic and inelastic cigarettes was independent of tar
yield and filter type. Elasticity was not associated with tar yield
(p = 0.2734). There are elastic cigarettes in the Canadian
cigarette market and the results suggest a possible smoker
preference for elastic cigarettes. Utilizing elasticity may be
valuable in the development of future harm reduction
strategies.
Authors: Jinfu Hu, Glenn Robbins, Anne-Marie Ugnat, and Chris Waters
Abstract:
The purpose of this study was to examine trends in diabetes
mellitus (DM) mortality rates in Canada, including analysis at the
provincial level, during the period 1986–2000. The study
population included Canadians aged 35 and over. Age-standardized
mortality rates (ASMRs) were computed. Linear regression was used
to calculate the average annual percentage change (AAPC) by age,
sex and province. The results showed a substantial increase in DM
mortality rates among those aged 35 and over, particularly for men;
the AAPC indicated an increase of 2.4% for men and 0.7% for women.
When the mortality rates were plotted for three time periods, the
rates increased with each successive age group and period for both
sexes. Mortality from DM increased significantly in both sexes in
Canada between 1986 and 2000, particularly in men.