Article Text

Download PDFPDF

Trends in CVD
Stroke mortality trends for England: modelling the past to predict the future
  1. C. J. Sutton,
  2. J. Marsden,
  3. C. L. Watkins,
  4. M. J. Leathley,
  5. P. Dey
  1. University of Central Lancashire, Preston, UK

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


    In the UK, overall stroke mortality has declined. It is important to planners of health service provision to understand how trends in stroke mortality are changing and, given changes in population structure, how trends in stroke mortality will affect the future burden on the nation. We therefore aimed to investigate recent stroke mortality trends amongst those aged over 40 years in England and to make predictions of stroke mortality to 2015.


    We obtained annual aggregated stroke death (ICD-10 I60 to I69) and population data for England from the Office of National Statistics for the years 1979–2005 and used these data to investigate time trends in gender-specific mortality rates for adults aged over 40 years. We applied log-linear modelling to identify effects attributable to age, linear “drift” over time, non-linear period and birth-cohort effects; discontinuities in “drift” were also considered. Different forms of model were compared for relative goodness-of-fit using Akaike’s Information Criterion and residual analysis was used to identify specific aspects of lack of fit of individual models. Predictions to 2015 were made for the best-fitting models.


    Stroke mortality predictions were not particularly sensitive to many of the aspects of the underlying choice of model, with, for example, the more complex models including different age and period effects for males and females providing similar predictions to simpler models. However, predictions were highly sensitive to varying assumptions about potential discontinuities in the “drift”. Our predictions are that stroke deaths are likely to continue to decrease up to at least 2015. For 2015, our predictions of deaths from stroke varied from 34 429 to 46 538 (compared with 47 213 deaths in 2005) under the different plausible model scenarios. We also noted that there were some potentially interesting trends in mortality amongst recent birth cohorts. In particular, mortality was higher in older age groups, but the difference between the older and younger age groups appears to have decreased over time for both sexes. Modelling of the subgroup aged 40–69 suggested a relative rate increase in mortality amongst those born since the mid-1940s compared with earlier cohorts; this trend appears to have been sustained in men.


    Downward trends in stroke mortality are set to continue overall to at least 2015. However, these trends appear to be levelling off in middle-aged people, particularly men. This raises concerns that stroke mortality might increase as these cohorts grow older.