Intended for healthcare professionals

Papers

Analysis of trends in premature mortality by Labour voting in the 1997 general election

BMJ 2001; 322 doi: https://doi.org/10.1136/bmj.322.7298.1336 (Published 02 June 2001) Cite this as: BMJ 2001;322:1336
  1. Danny Dorling, professor, quantitative human geographya,
  2. George Davey Smith, professor, clinical epidemiology (george.davey-smith{at}bristol.ac.uk)b,
  3. Mary Shaw, senior research fellowc
  1. a School of Geography, University of Leeds, Leeds LS2 9JT
  2. b Department of Social Medicine, University of Bristol, Bristol BS8 2PR
  3. c School of Geographical Sciences, University of Bristol, Bristol BS8 1SS
  1. Correspondence to: G Davey Smith
  • Accepted 23 May 2001

Mortality relates to voting patterns within areas: mortality is higher the greater the proportion of the electorate who vote Labour or abstain and the converse is the case with regard to the percentage of the electorate who vote Conservative.1 This reflects the socioeconomic characteristics of individuals who vote for these parties, with Labour being identified with the working class and the Conservatives with the middle class. In the 1997 election, Labour was returned to office after 18 years in opposition. The government has released targets for reducing health inequalities and made it clear that such a reduction is a principal policy aim.2 These targets may be difficult to meet for two reasons. Firstly, factors influencing inequalities in adult health act from an early age onwards and may not respond rapidly to social change3; secondly, there has as yet been no reduction in social inequality (as indexed by income inequality) under the Labour government.4 Here we use premature mortality as an indicator of which population groups have fared best under the present government.

Methods and results

The mortality data are from the Office for National Statistics' digital records of all deaths in England and Wales and the equivalent records from the General Register Office for Scotland.1 The full postcode of the usual residence of the deceased was used to assign each death to one of the 641 parliamentary constituencies to reflect where the deceased usually lived. The death data were provided for single years. Standardised mortality ratios and direct standardised mortality for the age range 0-64 years were calculated using rates for England and Wales.

Because there was no census at the end of the 1990s, population by age group and sex must be estimated. The Office for National Statistics and the General Register Office produced mid-year population estimates for 1999 and earlier years at the local and unitary authority district levels. To maintain a geographical base consistent with previous studies of Britain's health gap, these district level estimates were interpolated to the electoral ward level and then aggregated to parliamentary constituencies. The interpolation was based on population estimates for 1996, which were available at electoral ward level, and was carried out such that for each age-sex group W1999 = W1996 + P1996 x (D1999− D1996), where W and D are the ward and district level population, P is the proportion of D resident in W, and the subscript is the year. The district level population for 1996-9 for each age-sex group is from the mid-year estimates of the Office for National Statistics and the General Register Office.

The table shows the standardised mortality ratios for two periods according to the percentage of the vote for Labour in 1997. Standardised mortality ratios rose by 0.8% reflecting the relatively smaller fall in mortality in Scotland compared with that in England and Wales. The absolute change in mortality nationally fell by 1.8% when mortality for all of Great Britain was directly standardised by age and sex to the population in England and Wales.

Standardised mortality ratios and change in these ratios for people aged 0-64 years by percentage of Labour vote in 1997. Parliamentary constituencies were ranked by percentage of Labour vote and divided into 10 equal population groups

View this table:

In absolute terms, mortality has improved for all but one of the tenths, although mortality has tended to improve most in areas with the fewest Labour voters. However the absolute mortality of people living in the tenth with the second highest percentage of Labour voters has actually risen over this period. In relative terms, mortality worsened in eight of the tenths, and it worsened most in the areas with higher proportions of Labour voters (with the exception of the tenth with the highest percentage of Labour voters). The correlation between the percentage of the Labour vote and (directly age-sex standardised) absolute change in mortality is 0.13 (P=0.002). The equivalent correlation with the change in standardised mortality ratio is also 0.13 (P=0.001).

Comment

Labour's slogan during the 1997 campaign was: “Things can only get better.” We have shown that in absolute terms things got better for most areas, but improvement was smaller in areas with a higher percentage of Labour voters. In relative terms things got worse for people in constituencies in which a high proportion of people voted Labour, while things got better for people in constituencies where people generally voted Conservative. This mirrors trends in income inequality, which has increased throughout the period of the Labour government4: the Gini coefficient for the distribution of adjusted post-tax income increased from 38 in 1997-8, to 39 in 1998-9, and 40 in 1999-2000.4 Where Labour has improved the life chances of poorer people in Britain they have tended to only just move these people above various “poverty lines.”5 It is possible that this trend accounts for the absolute rise in mortality for those younger than age 65 experienced by people living in the tenth with the second highest percentage of Labour voters. Time, and the continued monitoring of the performance of the government through statistics, will tell.

Acknowledgments

The authors would like to thank Dr Richard Mitchell for his assistance in calculating the population denominators.

Contributors: DD was responsible for data management and statistical analysis and contributed to developing the core ideas. GDS initiated the research and contributed to the discussion of the core ideas. MS coordinated the research and contributed to the statistical analysis. All authors contributed to interpreting the findings and writing the paper. All authors are guarantors.

Footnotes

  • Funding MS is funded by ESRC Fellowship R000271045.

  • Competing interests None declared.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.