Intended for healthcare professionals

Education And Debate

Commentary: Sodium and blood pressure in the Intersalt study and other studies—in reply to the Salt Institute

BMJ 1996; 312 doi: (Published 18 May 1996) Cite this as: BMJ 1996;312:1285
  1. Jeremiah Stamler, professor emeritusa,
  2. Paul Elliott, professorb,
  3. Alan R Dyer, professora,
  4. Rose Stamler, professor emeritusa,
  5. Hugo Kesteloot, professorc,
  6. Michael Marmot, professord

    for the Intersalt Steering and Editorial Committee

  1. a Department of Preventive Medicine, Northwestern University Medical School, Chicago, IL 60611-4402, USA
  2. b Department of Epidemiology and Public Health, Imperial College School of Medicine at St Mary's, London W2 1PG
  3. c Department of Epidemiology, St Raphael University Hospital, Leuven, Belgium
  4. d Department of Epidemiology and Public Health, University College London Medical School, London
  1. Correspondence to: Professor Stamler.

    The Salt Institute continues to misrepresent the findings of Intersalt (and other studies) on salt and blood pressure, and the availability of Intersalt data. It not only incorrectly attempts to discredit the Intersalt finding that average population salt intake is positively related to slope of blood pressure with age1; it also ignores or misrepresents two other important results—the positive relations between sodium excretion of individuals and their blood pressure (found in over 10 000 adults), and the positive association of average population sodium intake with average blood pressure and prevalence of hypertension across the 52 populations in the Intersalt study. The institute's statement that “the primary Intersalt hypotheses were largely negative” is incorrect.

    Intersalt analyses and findings

    The Salt Institute was clearly not happy with the implications of the original Intersalt findings on salt intake and blood pressure slope with age—an additional 9 mm Hg higher systolic blood pressure over 30 years (for example, from age 25 to 55) for an additional 100 mmol population sodium intake. It therefore requested that the Intersalt group perform several additional analyses on the slope of blood pressure with age, using statistical methods suggested by the institute.

    These were done, and all analyses again showed a significant relation of average population salt intake (as assessed by average 24 hour urinary sodium excretion) and population blood pressure slope with age.2 The analyses included those that had controlled for pressure in young adults (ages 20-29). Findings were similar to the original results: the higher the population salt intake, the greater the upward slope of blood pressure with age—for all 52 population samples and also when the four low sodium samples were excluded (fig 1). These findings were made available to the Salt Institute in November 1994. The institute has never cited any result of these analyses upholding our original estimates. They are published in this BMJ.2

    Fig 1
    Fig 1

    Plots of difference in systolic blood pressure (mm Hg) over 30 years (age 55 minus age 25) in relation to median urinary sodium excretion (mmol/24 h) across (left) 52 samples and (right) 48 samples (excluding the four low sodium samples). Standardised for age and sex and adjusted for blood pressure at age 20-29, median body mass index, percentage alcohol drinkers, and median alcohol intake among drinkers. Difference in systolic blood pressure calculated as shown; b (SE) given in mm Hg greater difference per 100 mmol higher urinary sodium excretion in 24 hours

    The Salt Institute, setting aside results of the requested analyses, substituted other “analyses” yielding the figures attached to their article here. In these figures, they use methods so illogical, so contrary to biology and accepted statistical methodology, that much space would be required to evaluate them fully. Here we note four of these errors.

    Firstly, it is incorrect—statistically3 4 5 and biologically—to take a relation observed over the ages 20-59 and extrapolate the resulting regression equation down, to “estimate” blood pressure at age zero (birth). It is known that the slope of blood pressure from birth to young adulthood differs from the slope over adult years.

    Secondly, it is incorrect—statistically—to correlate these erroneous blood pressures at age zero with slope of blood pressure with age (from which the blood pressures at age zero were calculated) and then to use them to “adjust” slope of blood pressure with age, or to correlate these blood pressures at age zero with urinary sodium concentration.

    Thirdly, Intersalt data show the error in the Salt Institute's statement that observed higher blood pressure at older ages seen in populations with lower calculated blood pressure at age zero is merely a matter of “catching up” with populations starting at higher blood pressure levels. If that were so then, by ages 50-59, blood pressures should be similar across the 52 samples. This was not the case in the Intersalt study. As reported in 1989, systolic blood pressure at ages 50-59 ranged from 94.1 mm Hg to 155.1 mm Hg for all 52 samples, and from 116.1 mm Hg to 155.1 mm Hg when the four low sodium samples were excluded.6

    Fourthly, the Salt Institute, by incorrect use of the regression equation observed for blood pressure slope with age, created a false see-saw. Inevitably, incorrect extension of the steepest observed upward slopes (Portugal, Tianjin) down to the intercept (blood pressure at age zero) results in low age zero blood pressures for such samples, while for samples with an observed downward slope of blood pressure with age (for example, Papua New Guinea) or very small upward slopes (for example, Xingu Indians) blood pressure at age zero comes out high.

    The Salt Institute adds further error when it seeks to use this “finding” to distort the whole salt-blood pressure slope with age relation. It says that “the higher a society's initial sodium excretion, the lower its mean systolic blood pressure.” Initial (age zero) sodium excretion (and age zero blood pressure) were never measured; they are artefacts of the Salt Institute's extrapolations far beyond the observed data.

    Results of other research

    Not only does the Salt Institute misrepresent or omit Intersalt findings showing that higher salt intake is related to higher individual blood pressure and greater upward slope of blood pressure with age; it also chooses to ignore data—accumulated over decades by countless other studies, with every method of investigation—that have found the same relation. A cornerstone of biomedical science is that the totality of research results must be considered to arrive at sound judgment as to whether an exposure causes greater risk to health. There is extensive evidence—from animal experimentation, clinical investigation, trials, population studies, anthropology—that underlines the conclusion of a causal relation between high salt intake and adverse patterns of blood pressure in populations.1 6 7 8 9 10 11 12 13 14 15 16 17 18 For example, among thousands of animal experiments, a recent one adds important data on results of salt feeding: when chimpanzees, originally with optimal blood pressure levels on a diet with no salt added, had incremental additions of dietary salt, systolic blood pressure rose 12 mm Hg with an increase of 5 g/day, rose 26 mm Hg with 15 g/day, and was restored to optimal when animals were returned to the original diet with no salt added.19

    There are similar impressive results from many studies, but none are cited by the Salt Institute, leaving the false impression that the recommendations from expert groups and public health agencies for salt reduction by the population (in the United States, the United Kingdom, and elsewhere) are based solely on Intersalt findings. The findings of Intersalt agree with a whole body of concordant evidence.

    Availability of data

    Our final point concerns availability of data from Intersalt. We are aware of few studies that have published so much detailed, peer reviewed data as has Intersalt. The original article in the BMJ in 1988 included three tables in an appendix giving detailed information on 27 variables (descriptive and analytical).1 One year later, in a special issue of the Journal of Human Hypertension, detailed age and sex specific Intersalt data were given in 38 appendix tables, with 20 columns of data each.6

    This widespread dissemination of data was based on an early decision by the steering and editorial committee to share data with fellow scientists and public health agencies. As is customary in scientific investigation, raw data on individuals remain the confidential property of local investigators, in this case the 52 investigators in 32 countries. The most recent Intersalt paper2 further described reasons for our declining the request to supply the study's raw data to the Salt Institute, the trade organisation of salt producers.

    The first request from the Salt Institute came through its attorneys, six years after the original data were published. The Intersalt group's reply stated our willingness to perform additional reasonable scientific analyses. These requests were received, the analyses were performed, and the data were sent to the Salt Institute. They have not been cited anywhere by the institute.


    Repeated recommendations to the population for salt reduction,7 8 9 10 11 12 13 14 15 18 such as those made by the US Departments of Agriculture and of Health and Human Services in 199512 18 and by the report of the Cardiovascular Review Group of Britain's Committee on Medical Aspects of Food Policy in 1994,11 rest on a strong research base. They need to be implemented. Since 75% of salt consumed in the United States, the United Kingdom, and other countries comes from salt added in food production,2 8 the food industry—by reducing salt added in food processing—has an important role in implementing these public health recommendations.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    16. 16.
    17. 17.
    18. 18.
    19. 19.
    View Abstract