Commentary: A reply to Ken Judge: mistaken criticisms ignore overwhelming evidenceBMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7015.1285 (Published 11 November 1995) Cite this as: BMJ 1995;311:1285
- Richard G Wilkinson, senior research fellowa
- aTrafford Centre for Medical Research, University of Sussex, Brighton BN1 9RY
Despite Judge's highly personal focus, mine is far from being the only evidence of an association between national mortality rates and income distribution. The criticisms that Judge directs at two of my five demonstrations of the relation would therefore, even if they had been accurate, leave the bulk of the evidence unscathed. As well as summarising the evidence which Judge ignores, I shall also show why we should not be surprised at a relation between mortality and income distribution, and then deal with methodological problems that he inadvertently raises.
The association between mortality and income distribution was first reported by Rodgers using data from around 1965 from 56 developed and less developed countries.1 Since then it has been found by many others. Flegg found that income distribution was related to national infant mortality among a group of 59 developing countries,2 and Le Grand reported that it was related to average age of death in a group of 17 developed countries.3 Analysing data from 70 rich and poor countries, Waldmann found that, if the real incomes of the poorest 20% were statistically held constant, increases in the incomes of the richest 5% were associated with rising rates of infant mortality.4 Among developed countries Wennemo has shown close relations between infant mortality and measures of income distribution and relative poverty.5 I have shown relations in developed countries between income distribution and life expectancy on two sets of cross sectional data and three sets of data on changes over time.6 7 8 Most recently, Kaplan et al at Berkeley and Kennedy et al at Harvard have independently found income distribution and life expectancy to be closely associated in 50 states of the United States.9 10 Lastly, using mid-century data for 20 countries at different stages of …