DRAFT
time since their last mammogram was less than 2 years were categorized as ‘regular
screeners’ and those indicating that more than two years had elapsed since their last
mammogram were categorized as ‘irregular screeners’.
Inclusion/Exclusion criteria
We excluded males and only included women aged 50 to 69 in our analysis. We also
excluded women who indicated that their last mammogram was for diagnostic purposes4.
The resulting sample comprised of 2,671 observations across the years 2002 and 2004.
Analysis
A multivariate multinomial logit (MNL) model was used for this analysis. The MNL
model estimates the importance of a woman’s individual characteristics on her screening
behaviour. Predicted MNL coefficients indicate how a unit change in the explanatory
variable, relative to the reference case will impact on the logit index, measured in units of
log odds. The results are more easily interpreted in terms of relative risk ratios which are
calculated by taking the exponent of the estimated coefficient.
Explanatory variables
The aim of this study was to test for systematic variation in screening behaviour,
including an investigation of the similarities and differences between never screeners and
irregular screeners.
Age was inputted as a continuous variable with an age-square component in the model to
account for any non-linear dynamics associated with increasing age. Age acts as a control
variable because of potential confounding with income and because the Breast Screening
program’s direct recruitment strategy commences when a woman turns fifty years of age.
4 Women who undergo a mammogram for diagnostic purposes are eligible for subsidies under the Medicare
but not the BreastScreen Australia program. Mammograms were classified as diagnostic if the respondent
stated their last mammogram was because of (1) history of breast cancer, and/or (2) breast problems or
symptoms at the time the mammogram is taken.