Health inequalities under New LabourBMJ 2005; 330 doi: https://doi.org/10.1136/bmj.330.7506.1507-a (Published 23 June 2005) Cite this as: BMJ 2005;330:1507
Difference in findings is due to geographical coverage
- Madhavi Bajekal, head, Morbidity and Healthcare Team (, )
- Allan Baker, senior researcher
EDITOR—Shaw et al argue that any study that does not use the recently released, final revised population estimates will produce erroneous results, citing a paper by one of us (MB), which showed a narrowing of the life expectancy gap in England.1 2
How much do revisions to population estimates influence the conclusions and can this alone account for the apparently different conclusions in the two papers?
Bajekal used the unrevised population estimates from the 1991 census rolled forward, but this mainly produces a difference in scale rather than direction of trends from the use of revised figures. We found that the conclusions of Shaw et al would still have differed from those of Bajekal had they used unrevised populations.
Both studies examined differential trends in life expectancy between 10ths of populations grouped by deprivation and covering almost the same period. We propose that the reason for the discrepancy in the two findings is due to geographical coverage (England v Great Britain) and the spatial level of analysis (ward v local authority).
Using the same data as Shaw et al, but for England alone, we see no change in the life expectancy gap between the local authorities with the highest and lowest life expectancy for men, and a relatively small increase in the gap for women (0.6 years) over the 1990s (table). Given that the width of the 95% confidence interval at local authority level is about one year, these results show that the life expectancy gap in England remained stable over the period.
Even within England, results from an analysis of wards will differ from those of local authorities. Differences in life expectancy and social deprivation between wards within most local authorities are considerable, and the socioeconomic composition of wards also changes over time. This will tend to make differences in life expectancy between local authorities both narrower and more stable over time than those for wards.
Competing interests None declared.