Age period cohort analysis of time trends in regional mortality rates in England, Wales and Scotland

Robertson, C. and Ecob, R. (2001) Age period cohort analysis of time trends in regional mortality rates in England, Wales and Scotland. Journal of Evaluation in Clinical Practice, 7 (3). pp. 299-309. ISSN 1356-1294 (

Full text not available in this repository.Request a copy


Previous studies have demonstrated regional variation in mortality rates, this reducing for younger but not older individuals. However, it has not been clear from such studies whether or not the overall findings can or cannot be attributed to particular regions that may have abnormal patterns of change over time. We analyse mortality rates by region using data from the census years from 1931 until 1991 for adults. Age period cohort models, based on the local 'curvatures', are used to describe the changes in the rates. We also use estimable contrasts to compare the later periods with the earlier periods and to compare pre- and post-war cohorts. There is strong evidence that the changes in the mortality rates are associated with non-linear period and cohort effects. There is no evidence that these curvatures vary over regions. There is evidence that the curvatures are not the same among men and women. Among women there is evidence that there is a change in trend in the periods from 1961 to 1991 compared with the trend in the earlier periods from 1931 to 1951, such that the reduction in the mortality rates with time are slowing down for women. Among men who had higher mortality rates, there is not so much evidence of a slowing down of the reduction. Post-war cohorts of women from 1941 onwards are enjoying a reduction in mortality compared with the cohorts from 1901 to 31, while among men there is no evidence of a change in mortality trends among younger cohorts. Changes in the mortality rates over regions have come about from different long-term temporal trends in the regions rather than abrupt changes taking place at different times in different regions. The general convergence of mortality rates over regions over time with a consequent reduction in regional variance is masked by two regions, Scotland and East Anglia. These regions appear to differ from the regions as a whole in the relationship of mortality to age at a given period and in the reduction of mortality over time (drift), Scotland having a higher mortality, especially at younger ages, and lower downward trend, and East Anglia vice versa.