Below are details of a fully-funded ESRC-funded PhD studentship to be
supervised by Prof Paul Boyle (who will be moving to the School of Geography
and Geosciences at the University of St Andrews from 1st September 1999). If
you are interested in applying, please contact Paul by email:
[log in to unmask]
Estimating 'intelligent' small area populations for use in medical
studies: accounting for population migration
It is necessary in disease mapping and analysis, undertaken for
geographical areas, to calculate reliable numerators (disease incidence)
and denominators (base population). Unfortunately, while a great deal of
effort is expended to produce the most accurate disease records possible,
the reliability of this information is, for certain purposes, irrelevant
if the base populations against which the cases are compared are poorly
estimated. We argue that for various reasons standard, small area
population estimates are unreliable, particularly those produced some
time after the decennial censuses. Principally this is because of
population migration and, despite the identification of this problem some
time ago, no work has been undertaken to solve it in relation to medical
studies - academic analysis of this is long overdue. This project will
provide the first set of population estimates for small areas designed
specifically for use in medical studies.
Two specific problems, that hinder the derivation of reliable disease
rates, are caused by population migration. First, the production of
cross-sectional population estimates is difficult because migration
patterns are complex. While births and deaths can be estimated for small
areas quite reliably, the lack of regularly updated migration flow data
for small areas makes the migration component in such accounting models
notoriously unreliable. Second, a related problem in epidemiological
studies is the very reliance on cross-sectional population estimates as
denominators. Not only are these data likely to be less reliable than the
information on disease incidence (as described above), but most of these
studies ignore the simple fact that people move (Boyle and Duke-Williams,
forthcoming). Many epidemiological investigations are guilty of this
as any attempt to relate disease incidence to the geographical
environment (e.g. area-based deprivation scores, or point-based
environmental hazards) makes the implicit assumption that the cases have
lived in the area long enough to be influenced by these environmental
factors. In the absence of detailed, individual-level migration
histories, which are certainly not available in most disease registries,
we rarely know whether the person has lived at the same address all
their life, or whether they moved into the area shortly before the
disease was diagnosed. Simple mapping exercises have demonstrated the
difference that the inclusion or exclusion of migrants makes to our
understanding of disease distributions (e.g. Kliewer 1992), but no study
has solved this migration-related problem in population estimates.
_____________________________________
Paul Boyle
School of Geography
University of Leeds
Leeds LS2 9JT
Tel: +44 (0) 113 233 3325
Fax: +44 (0) 113 233 3308
_____________________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|