Estimating Regional Variation in Cancer Patient Survival
Dickman, P.W. Estimating Regional Variation in Cancer Patient Survival. Ph.D. thesis. The University of Newcastle, Australia, 1997.
Abstract
One of the main targets of the World Health Organisation's "Health for All"
program is the reduction of health inequities between and within countries. Patient
survival is the principal outcome measure of the effectiveness of cancer patient care. As
such, information on the extent of regional variation in cancer patient survival is of
interest to health authorities working to obtain equitable health care systems. The broad
aim of this research is to develop and apply methodologies to estimate the level of
systematic regional variation in cancer patient survival. Specifically, the aim is to
identify those cancer sites which exhibit high levels of systematic regional variation in
survival. A high level of systematic regional variation in patient survival suggests
possible inequity in cancer patient care and can highlight cancer sites or regions where
there is potential for cancer control programs to be effective.
A comprehensive analysis of regional variation in survival has not previously been
conducted, so appropriate statistical methodologies must be developed. By modeling the
excess hazard due to cancer, it is a simple matter to estimate the relative survival rate
for each region in a country, while controlling for potential confounding factors such as
age, sex, and year of diagnosis. It is not clear, however, how these region-specific
estimates should be utilised to quantify the extent of systematic (as opposed to random)
regional variation in survival. It is possible to test the statistical significance of the
regional term in the model, although this is almost always significant for the common
cancers due to the large sample size. Statistical significance is dependent on sample
size, so cannot be used to rank cancer sites according to the extent of systematic
regional variation. A method for estimating the magnitude of systematic regional variation
in patient survival, after correcting for the expected random variation, has been
developed as a part of this thesis. The resulting measure provides a single estimate of
the level of systematic regional variation in survival for each site (or any patient
group).
The methodology was applied to data from the Nordic countries (Denmark, Finland, Norway,
and Sweden). It was estimated that 2.5% of cancer deaths during 2008--2012 in 12 major
cancer sites could be prevented by eliminating regional variation in patient survival. The
percentage of potentially savable deaths did not depend on country or sex but it did
depend on the cancer site. Both the level of patient survival and the degree of regional
variation in survival varied between the four countries. The sites in which the biggest
percentage savings could be achieved were melanoma (11%) and cervix uteri (6%). The sites
in which the highest number of deaths could be saved were prostate, colon, melanoma and
breast. Sites with high levels of regional variation were identified and are potential
targets for cancer control programs.
Copies of my thesis are, unfortunately, no longer available, although the main results of the thesis are contained in the fololowing article:
Dickman, P.W., Gibberd, R.W., Hakulinen, T. Estimating Potential Savings in Cancer Deaths by Eliminating Regional and Social Class Variation in Cancer Survival in the Nordic Countries, Journal of Epidemiology and Community Health, 51: 289-298, 1997. [Medline]