Intended for healthcare professionals

Letters

Why heart disease mortality is low in France

BMJ 1999; 319 doi: https://doi.org/10.1136/bmj.319.7204.255 (Published 24 July 1999) Cite this as: BMJ 1999;319:255

Miscoding may explain Japan's low mortality from coronary heart disease

  1. Akira Sekikawa, Fellow (akira+{at}pitt.edu),
  2. Lewis H Kuller, Professor
  1. Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, 15261, USA
  2. Råbygatan 2, Lund, Sweden

    EDITOR—We agree with Law and Wald that one must consider the cohort effect—the time lag—in investigating the association of levels of risk factors with mortality from coronary heart disease.1 Analysis of mortality from coronary heart disease in birth cohorts since the second world war is therefore important.

    We evaluated mortality from coronary heart disease in the United States, Japan, South Korea, and other countries, as well as within the United States by state, for men aged 35-44. We found that the very low mortality from coronary heart disease in Japan (5.5/100 000, compared with 11.4/100 000 in South Korea and 26.4/100 000 among American white men in 1992) might be an artefact.

    A substantial proportion of mortality from coronary heart disease among men aged 35-44 may be miscoded as heart failure (ICD-9, code 428) because in Japan more than 60% of mortality from diseases of the heart (codes 390-429) was coded as heart failure. This proportion is usually very low in this age group—1% among American white men. If half the mortality from heart disease was due to coronary heart disease in Japan there would be only a twofold difference in mortality from coronary heart disease between the United States and Japan: 26.4/100 000 among American white men versus 13.5/100 000 among Japanese men in 1992.

    In some states in the United States we found that mortality from coronary heart disease among white men aged 35-44 is similar to that in France (14.7/100 000): 13.8/100 000 in Washington, 14.7/100 000 in Connecticut, 15.4/100 000 in Kansas, 16.4/100 000 in Colorado, and 17.1/100 000 in California. The low rates in these states are not due to heavy alcohol consumption compared with that in other states; rather, they are strongly related to educational attainments of the population and the prevalence of cigarette smoking and perhaps other risk factors.

    Studies of recent birth cohorts by country provide better evaluation of causes of death and measurements of atherosclerosis. Focusing on smaller geographic areas as opposed to whole countries may improve understanding of geographic variation in mortality from coronary heart disease. Alcohol may be an important attribute for a lower rate of coronary heart disease in France but is not necessarily the only key factor.

    Atherosclerosis has a long incubation period. Diet measured 30 years before reported death may better predict the extent of atherosclerosis and mortality from coronary heartdisease.

    References

    1. 1.

    Authors' hypothesis is wrong

    1. Uffe Ravnskov, Independent researcher (uffe.ravnskov{at}swipnet.se)
    1. Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA, 15261, USA
    2. Råbygatan 2, Lund, Sweden

      EDITOR—The idea of a time lag between increases in consumption of saturated fatty acids and an increase in mortality from coronary heart disease, presented by Law and Wald,1 was shown to be wrong many years ago.

      A host of confounding variables may explain an association between a population's consumption of saturated fatty acids at a certain time and mortality from heart disease.2 Secular trends are more reliable. If the amount of saturated fatty acids in the diet is important, changes in their consumption should eventually be followed by similar changes in mortality from heart disease even if the initial correlation is false.

      In a recent review I presented the results of the four most comprehensive studies of secular trends, including 103 time periods in 36 countries.3 In 30 time periods consumption of saturated fatty acids had increased, as had mortality from coronary heart disease. But in 33 periods in which consumption had increased, mortality from coronary heart disease had remained unchanged in 10 and had decreased in 23.

      Of particular interest for the Law-Wald hypothesis is one of the studies, in which the changes in mortality from coronary heart disease were correlated with the changes in consumption of saturated fatty acids over 24 years. In that study an increase in mortality from coronary heart disease had followed an increase in consumption of saturated fatty acids in seven countries. In 11 countries, however, where consumption had increased by between 15% and 190% (mean 54%), mortality from coronary heart disease had been unchanged in three countries and had decreased by between 6% and 27% (mean 15%) in eight.

      This should not be a surprise because the combined results from eight ecological, 41 cross sectional, 25 cohort, and six case-control studies as well as a meta-analysis of nine controlled randomised trials also contradict the suggestion that consumption of dietary saturated fatty acids has any influence on atherosclerosis or coronary heart disease.3 There is no French paradox either. Anyone who plots the data from the MONICA (monitoring trends and determinants in cardiovascular disease) study will find that the French data just happen to be situated at the extreme corner of a scatter diagram that includes most combinations of consumption of saturated fatty acids or serum cholesterol and mortality from coronary heart disease.

      References

      1. 1.
      2. 2.
      3. 3.
      View Abstract

      Log in

      Log in through your institution

      Subscribe

      * For online subscription