Editorials

Salt and blood pressure revisited

BMJ 1996; 312 doi: http://dx.doi.org/10.1136/bmj.312.7041.1240 (Published 18 May 1996) Cite this as: BMJ 1996;312:1240
  1. Dag S Thelle
  1. Professor Centre for Epidemiologic Research, Institute of Community Medicine, University of Oslo, Oslo N-0318, Norway

    How much more evidence do we need?

    The relation between salt intake and blood pressure is no news to the food industry, nor to the expert committees in the United States, Norway, and elsewhere recommending reductions in daily intake of salt to about 100 mmol sodium or less.1 2 However, with three quarters of the presently consumed salt well hidden in processed food, there is little that people can do to influence their intake. Thus, any attempts to influence the amount of salt in food must be directed at the food industry.

    Hard data are now accumulating to give substance to the debate, most notably the Intersalt study, the first report of which was published in the BMJ in 1988.3 This cross sectional study of 10 074 men and women with a broad age span was designed to describe the association between urinary excretion of sodium chloride (as a measure of salt intake) and blood pressure. After adjustments for body mass index, alcohol intake, sex, and age, it showed that a reduction in sodium intake of 100 mmol/day would reduce systolic and diastolic blood pressures by 2.2 mm Hg and 0.1 mm Hg respectively. This was based on individual data and was lower than expected from previous studies.4 But there was more to Intersalt than this. The study also had an ecological design that allowed the slope of the blood pressure curve to be estimated at different ages and different levels of sodium intake. This showed that increasing intake of sodium chloride by 100 mmol/day would increase systolic blood pressure by 10 mm Hg 30 years later. Was this true? Did Intersalt reflect the real relation between salt and blood pressure, and why were the individual results so much weaker than the ecological findings?

    In this issue of the BMJ (p 1249) the Intersalt researchers present updated results for the relation between sodium excretion and blood pressure.5 These results are more robust than those in their first report. A striking finding is that the association between sodium excretion and blood pressure is stronger when body mass index is not adjusted for. The most likely explanation for this finding is that body mass index, which correlates with sodium excretion, is measured more accurately than sodium excretion and will therefore emerge as the strongest explanatory variable in a multiple regression analysis.6 That sodium excretion is the critical factor is also strongly suggested by data from the three Chinese Intersalt collaborating centres, which reported low body mass indexes but some of the strongest associations between sodium excretion and blood pressure, and some of the highest rises in blood pressure.

    Causal relation is difficult to demonstrate

    The magnitude of the effect of sodium excretion on blood pressure in this updated analysis is similar in the analyses within and across populations. This is comforting, even if the lack of effect on diastolic blood pressure in the population analysis when body mass index is adjusted for remains unexplained. A major reason for the stronger association between sodium excretion and blood pressure in the updated analysis than in the first report is a more complete correction for regression dilution bias, a correction which is warranted when variables are measured with error.

    The design and updated results of Intersalt may still fail to convince sceptics of a causal relation between salt intake and blood pressure, and some of the difficulties in demonstrating an association should therefore be emphasised. These include the measurement of salt intake, which is notoriously inaccurate for individuals, and the range of variation, which may be too narrow within a population compared with the large variation between individuals. The Intersalt study does not have the perfect design to overcome these difficulties, and on its own it cannot answer the question as to whether high salt intake causes high blood pressure. But until someone sets up a 30 year longitudinal study to monitor sodium chloride intake and blood pressure prospectively in a sufficiently large population, this hybrid cross sectional, within population and cross population, ecological study is likely to be the only feasible epidemiological design.

    The updated version of Intersalt provides robust results that are in concert with other studies, including experiments on animals and clinical trials.7 A recent study on chimpanzees showed that adding 100 mmol of sodium a day to their food increased their systolic blood pressure by 12 mm Hg. Blood pressure rose further with further increases in sodium intake and fell when sodium supplementation was stopped.8 The Intersalt results must be viewed in the context of such existing evidence suggesting a causal relation between salt intake and blood pressure. Whether the evidence is strong enough to warrant the reductions in salt recommended by the authors is, as always, a question of judgment. But useful clinical and public health actions have been undertaken on much weaker evidence.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.