Phone +44(0) 141-649-9387

russell@ecob-consulting.co.uk

Watt, G. & Ecob, R. A comparison of mortality trends in Edinburgh and Glasgow: are the age-cohort predictions of mortality differences supported on the 1991 data? J. Public Health Medicine. 2000, 22 (3) pp 330-336.

A golden opportunity was presented, with the availability of data from 1991 census and death registration data for 1989-93, to compare the reality of deaths, both overall rates and differences in rates between cities, with predictions obtainable from the age-cohort model on deaths around 1961-1981 censuses shown in our previous paper (Watt and Ecob, 1992, see below). All cause mortality rates fell between 1979-83 and 1989-93 by a larger amount in Edinburgh than Glasgow. Differences in life expectancy between the cities at age 35 increased by 44% to 4.7 years in men and by 19% to 2.5 years in women. Mortality rates in both cities fell by a larger amount than predicted, by 10% in men and by 6% in women. The widening of differences in life expectancy between Glasgow and Edinburgh is partly due to a historical trend of longevity increasing more quickly in Edinburgh. Moreover reductions in mortality are generally significantly more than expected from the age-cohort model is shown by the fact that in 9 out of the 16 possible age/gender combinations the rates are outside and below the 95% confidence interval band of the predicted rates. Seven of these are for men (all Glasgow age groups and all but youngest age group for Edinburgh) and 2 are for women (oldest age groups in both Edinburgh and in Glasgow).

Ecob, R. The relationship of population density, postcode sector hectarage and area deprivation to Mortality in four Scottish Cities. Chapter for 'The geography of health inequalities in the developed world. Ed. Boyle, Curtis, Gatrell, Graham, Moore. Blackwell. Oxford. To appear 2000.

Through merging of data, at a postcode sector level on mortality, deprivation and population size, from 1981 census (and available as Appendices to Carstairs and Morris's, 1991, book 'Mortality and Deprivation in Scotland') with data on area hectarage of these sectors obtained from 1991 Small Area Statistics, it is possible to examine in some detail the possible influence of population density and area size (hectarage) of postcode sector on mortality in addition to the known effects of area deprivation. Though the four major cities were found to vary in the extent of mortality given deprivation (Glasgow having the highest mortality and Aberdeen the lowest), no differences were found between cities in the relationship to deprivation. Population density was related in a non-linear way to mortality, particularly high mortality being found in areas with very low population densities. An additional inverse effect of area size on mortality was found, area size being a proxy for distance from city centre. The relationship to population density (given area size)was found to be stronger in Glasgow and Aberdeen than in Dundee and was absent in Edinburgh. Causes are postulated in the manner and extent of population movement, particularly in Glasgow, in connection with mainly post-war urban development. In order to gain further insights into the mechanisms involved, the need for prospective individual level data on social factors and mortality is stressed.

Ecob, R and Der, G. An iterative method for the detection, elimination or possible imputation of outliers in longitudinal data using multilevel models. In Reise,S and Duan, N eds 'Multilevel Modeling:Methodological Advances, Issues, and Applications'.Lawrence Earlbaum, London. To appear 2000

An iterative method of detecting outliers in longitudinal data comprising repeated measures on individuals over time is proposed. This is based on the supposition that an outlier - an incorrect observation - will distort a statistical model fitted to the data thereby biasing the evaluation of the correctness or otherwise of the remaining observations.  This method proceeds by identifying outliers on the basis of a fitted model, eleiminating these and refitting the model, proceding iteratively until convergence. Residuals are considered at the lowest level only – however this method can be easily extended to accommodate residuals at higher levels (individuals and above).This method is assessed on two data sets on heights of a cohort of 15 year old males and females subsequently followed up at ages 16,18 and 23 from a study examining the social patterning of health over time (West of Scotland Twenty-07 study) and on data on weight of calves on 11 occasions for two groups with different treatments. An initial assessment of this method indicates that convergence is usually fast.

Ecob, R and Macintyre, S. Small area variations in health related behaviours; do these depend on the behaviour itself, its measurement, or on personal characteristics. Health and Place. 200, 6 pp 261-274.

In this paper we examine the area patterning of four health related behaviours (smoking, alcohol consumption, diet, and exercise) in the West of Scotland, after controlling for a range of individual/household characteristics, using multilevel models.  Smoking and drinking were measured both as binary and as continuous variables, and diet and exercise were each measured in two ways: ‘good’ (health promoting) and ‘bad’ (health damaging).  ‘Area effects’ (unattributed random variation by post code sector) were found for ‘bad’ diet  and for smoking consumption and both 'good' and 'bad' diet, ‘bad’ exercise patterns and current smoking were associated with postcode sector deprivation. For bad diet this effect was found only for individuals in more affluent households, and for ‘good’ exercise and current smoking the association with area deprivation differed between adolescents and adults. We conclude that the influence of area on health related behaviours varies according to the behaviour and the way it is measured and that the influence of are deprivation and/or of area can vary by age and household deprivation.