BMJ No 7041 Volume 312 Saturday 18 May 1996

Commentary: Sodium and blood pressure in the INTERSALT Study and other studies: In reply to the Salt Institute

---

This INTERSALT reply to the invited Salt Institute piece in this BMJ has as its focus: 1. Salt Institute errors -- of omission and commission -- on the research findings concerning dietary salt and blood pressure (BP), especially (but not exclusively) in regard to INTERSALT findings;

2. Salt Institute misrepresentations about availability of INTERSALT data;

3. Salt Institute efforts to overturn public health recommendations for lower salt intake by the population to improve health.

---

1. Salt Institute Errors on the Research Findings on Salt and BP

1.B. The Salt Institute piece omits any mention of the INTERSALT within-population findings showing a positive independent relationship for its 10,000+ participants between 24-hour urinary sodium excretion of these individual men and women and their systolic BP, as reported in the BMJ in 1988 and here (13, 14).

1.C. In addition, the Salt Institute piece omits any mention that INTERSALT recorded significant positive results in 9 of 10 analyses testing its prior cross-population hypotheses -- on relationship across its 52 population samples of sample median 24-hour sodium excretion to sample median systolic BP, median diastolic BP, prevalence of hypertension, slope of SBP with age, slope of DBP with age (age-sex standardized and age-sex-BMI-alcohol standardized) (13, 14). It completely misrepresented the INTERSALT findings by its erroneous statement, ". . . the primary INTERSALT hypotheses were largely negative . . .". The results it did cite were from truncated 48-sample analyses carried out post-hoc, with exclusion of the four low salt samples.

1.D. The Salt Institute piece also omits any mention of results of the special, additional cross-population analyses -- requested by the Salt Institute -- on slope of BP with age: In all these analyses, population sample median (or mean) 24-hour sodium excretion was consistently, strongly, independently, significantly related to difference in population sample SBP with age (e.g., age 55 minus age 25), across 52 and 48 samples, with use of all methods proposed by the Salt Institute to estimate BP difference with age (14). These included analyses controlled for young adult BP as a possible confounder (the point articulated in the Salt Institute piece -- see below) (see INTERSALT figure) (14).

1.E. The Salt Institute piece further errs by commission in asserting that only longitudinal prospective data can be used to examine properly relationships across populations between their average salt intake and their average BP levels at different ages (e.g., age 55 compared with age 25). Most cross-sectional population studies -- current and past -- show considerably higher average BPs among their middle-aged and older strata than their young adult strata. It is a reasonable inference that these higher average BP levels with age reflect rise in BP with age for each age-specific cohort, especially since long-term follow-up data on cohorts -- from life insurance companies and prospective epidemiologic studies -- uniformly show that BP of most individuals rises from young adulthood through middle age (14). Therefore, it is reasonable to carry out cross-sectional cross-population analyses on relationship of such exposures as high salt intake (as well as high dietary Na/K, excess alcohol consumption, overweight) to difference in BP with age. And it is sound to include results of such analyses as components of the totality of data on salt and BP used to assess the etiologic relationship. Decades of fruitful research of this kind on many exposures and diseases show that the Salt Institute assertion is in error on this matter.

1.F. The Salt Institute piece makes further errors of commission in regard to INTERSALT cross-population analyses on sample median sodium excretion and sample slope of BP with age. It implies incorrectly that in the linear regression analyses estimating slope of BP with age for each sample, the equations were ". . . arbitrarily constructed with an intercept of zero . . .". No such procedure was done, since it makes no sense to force the slopes to pass through BP zero at age zero. Such equations would yield poor fits for the age-BP relationship in adults, hence they cannot be used to estimate the association of BP with age from age 20 through age 59.

1.G. The Salt Institute piece also errs in arguing that it is critical to adjust for age 20 BP in analyzing relationships between sample median sodium excretion and BP slope with age. There is no valid scientific rationale for this adjustment, and neither peer reviewers nor other scientific colleagues have ever suggested this adjustment. Nonetheless, further INTERSALT analyses were performed, including such an adjustment, as requested by the Institute. As expected, inclusion of this adjustment for young adult BP had no effect. With adjustment for sample average young adult BP, sample median sodium excretion remains strongly, independently, significantly related to BP difference with age: 24-hour sodium excretion higher by 100 mmol on average was associated with SBP/DBP at age 55 (compared to age 25) higher by 10-12/6-7 mmHg (see INTERSALT figure). These are findings similar to all others from INTERSALT analyses on this matter (including the original 1988 findings and those stemming from other Salt Institute requests). The Salt Institute has had these data since 1994, but chose to omit them from its BMJ piece.

1.H. The Salt Institute, in its Figures 1 and 2 and text comments on them, makes further errors of both omission and commission. It omits to mention that what it labels ". . . initial systolic blood pressure (as predicted from the regressions) . . ." is, in fact, not observed data but rather the Institute's inappropriate statistic, estimated SBP at age zero. INTERSALT participants were all adults. Accordingly, regression slopes in INTERSALT reflect differences in BP across the age span 20 through 59. Textbooks on regression analysis warn that regression slopes across a series of observed points (in INTERSALT, for persons ages 20-59) should not be extrapolated to predict values far beyond these points (15-17). "Predicting" BP at age zero from the adult slope of BP with age is unsound biologically as well as statistically: patterns of change in BP from birth to young adulthood are not the same as patterns of BP change from age 20 to age 59, as is clear from comparisons of data on children and adults. The irrationality of this prediction is well illustrated by specific data points in Salt Institute Figure 1: e.g., for the Papua New Guinea sample, one of the four remote low salt samples in INTERSALT, with low average SBP/DBP at ages 20-59, slope of SBP with age was observed to be downward (inverse) so that SBP at age 55 (105.8 mmHg) was lower by 4.2 mmHg compared to age 25 (110.0 mmHg). The Salt Institute's inappropriate extrapolation of this inverse slope to age zero (the intercept) gives an age zero (birth) BP of 113.5 mmHg, 7.7 mmHg higher than at age 55! Obviously, this erroneous statistical maneuver produces estimates making no biological sense.

This Salt Institute error of commission is then compounded, by relating this estimated age zero BP (the intercept) to slope of BP with age (Salt Institute Figure 1): Since the intercept is a function of the slope, it is illogical and incorrect to turn it around and use it to "explain" cross-sample differences in slope.

The Salt Institute piece attributes biologic meaning to the inverse relationship displayed in its Figure 1. This "finding" -- it argues -- supports the idea that slope of BP with age depends primarily on initial BP rather than salt intake; the higher the estimated age zero BP, according to the Salt Institute, the less the BP rise during adulthood; BP rise with age is just a "catching up" in BP at older ages by those populations starting with lower BPs. But if this were the case, then most INTERSALT population samples should have had similar BPs at age 50-59, due to this "catching up". But that is not the finding. As INTERSALT data published in 1989 show (18), sample average SBPs at ages 50-59 range from 94.1 to 155.1 mmHg for all 52 samples (from 116.1 to 155.1 mmHg without the four remote low salt samples). Variability across samples in average SBP is greater at ages 50-59 than at ages 20-29, not less (as predicted by the Salt Institute "catch up" notion).

Figure 2 perpetuates the error of commission in Figure 1, this time by regressing estimated age zero systolic BP on 24-hour median sodium excretion. The Papua New Guinea data again serve to illustrate the absurdity of this "analysis": For the Papuans, the erroneous extrapolation to age zero of the inverse slope of BP with age purports to show that they have the highest age zero SBP, despite their low sodium excretion of <10 mmol/day and their low adult BP throughout the whole 20-59 age span. At the other end of the graph, the Chinese Tianjin sample, with a large upward slope of BP with age (6.4 mmHg/10 years) is purported to have a low age zero SBP, despite the highest 24-hour sodium excretion of all INTERSALT samples (242 mmol/day). Based on this wrong extrapolation to age zero of the INTERSALT findings on BP slope with age, the Salt Institute goes on to use this "zero BP" and an undefined "initial urinary sodium excretion" to suggest that ". . . the higher a society's initial urinary sodium excretion, the lower its mean systolic blood pressure." INTERSALT had no data on "initial urinary sodium excretion", and there was no observed "zero BP". This Institute speculation ignores valid, observed findings from the INTERSALT cross-population data demonstrating strong, independent, significant relationships between sodium excretion and slope of blood pressure with age.

1.I. Salt Institute Figure 3 extends the errors of commission: Again, it is incorrect to use sample estimated age zero SBP to adjust slope of BP with age, since estimated age zero SBP is a flawed statistic.

As to Salt Institute "findings" for 48 samples (without the four low salt samples), in INTERSALT 48-sample analyses with control for young adult BP (see INTERSALT figure), 100 mmol/day higher sodium excretion is estimated to be associated with 30-year SBP/DBP difference (e.g., age 55 minus age 25) greater by 9-11/3-4 mmHg, not the "flat" relation of Salt Institute Figure 3. Erroneous use of statistics by the Salt Institute once again produced erroneous results.

1.J. Salt Institute Figure 4 also displays meaningless data, and the text on this graph is fraught with errors. It omits mention of what the R2 statistic is in this case. In fact, it is an estimate for each population sample of how well individual age (x) "explains" individual BP (y), from linear regression analysis of BP on age (y = a + bx) used to calculate slope of BP with age (b).*

The Salt Institute text about its Figure 4 "findings" says, "For the low salt centers, the relation between blood pressure and age breaks down . . ." (our emphasis). No, not breaks down, is flat. BP is not higher with age, probably because -- as data from INTERSALT and many other studies indicate -- habitual low salt intake from birth and weaning on in these remote populations is a key factor helping to maintain optimal low-normal BP throughout life (2, 12, 13, 18, 19). This is a reasonable inference repeatedly noted by researchers and independent expert groups. It is entirely ignored in the Salt Institute piece.

The Salt Institute piece goes on to say about the four INTERSALT low salt samples, ". . . these are the centers that are the most influential in determining the relation between slope and urinary sodium excretion". This is a further error of commission on scientific questions in the Salt Institute piece: The 48-sample analyses (omitting the low salt samples) on the relation of sample 24-hour sodium excretion to sample slope of SBP with age yield positive strong significant findings similar to 52-sample analyses (see INTERSALT figure). Exclusion of the four low salt samples influences the result only slightly.

In summary on the research findings concerning dietary salt and blood pressure, the Salt Institute piece is fatally flawed due to multiple errors, of omission and commission. These include persistent silence and ignoring of strong significant positive INTERSALT results, including those generated in response to Salt Institute requests and made available to the Salt Institute in November 1994. They also include persistent silence and ignoring of written criticisms as transmitted to the U.S Dietary Guidelines Advisory Committee in February 1995 (20), detailing the scientifically unsound nature of Salt Institute graphic displays and Salt Institute statements about them.

In view of these multiple errors in the Salt Institute viewpoint on the research findings on salt and blood pressure, it is understandable that no scientist from the academic community is a co-signator of this piece, or any other recent Salt Institute piece, including its communication to the U.S. Dietary Guidelines Advisory Committee on January 31, 1995. These documents have as their author only the President of the Salt Institute, trade organization of the commercial salt producers.

It is also understandable -- given the fatal flaws in Salt Institute statements challenging current guidelines -- that the Dietary Guidelines for Americans, Fourth Edition, 1995 reiterated the recommendation to the whole population that salt and sodium be consumed only in moderation, e.g., no more than 6 grams of salt per day (21). This was in accordance with the judgment of the Dietary Guidelines Advisory Committee (6), based on its review of the scientific evidence, including the INTERSALT findings -- and despite concerted efforts by the Salt Institute to discredit or ignore INTERSALT (and other) findings in order to eliminate the guideline on salt. The 1994 U.K. Report on Nutritional Aspects of Cardiovascular Disease made recommendations similar to those of the U.S. Dietary Guidelines Advisory Committee, again based on review of the research evidence by a diverse group of experts (5).

2. Salt Institute Misrepresentations about Availability of INTERSALT Data

2.B. The Salt Institute piece articulates a ". . . concern, evident at the time of publication [as to] whether or not the slopes of blood pressure with age had been adjusted for the blood pressure intercepts in each of the 52 individual regression equations", and alleges that ". . . the INTERSALT investigators were not willing to disclose these values". Suffice it here to make two points: While the Salt Institute piece claims a concern from 1988 on about intercept data, no communication on this matter was forthcoming from the Salt Institute to the INTERSALT investigators until several years later (see below). Moreover, no request was really needed for intercept data, since these were essentially calculable from data published by INTERSALT in the BMJ in 1988 and the J Human Hyp in 1989 (13, 18). (In 1. above we have dealt with the scientifically erroneous use of age zero intercept data by the Salt Institute in its graphs.) Further, based on the analyses by INTERSALT in response to Salt Institute requests in 1993-4 (see below), detailed tabulations were sent to the Salt Institute of regression data including intercepts for each of the 52 samples, per the methods given in our paper in this BMJ (14).

The Salt Institute allegation that the INTERSALT investigators were not willing to report intercept data is false.

2.C. The Salt Institute piece states that the Institute obtained the additional data it sought "through a lengthy process involving negotiations between lawyers for the Salt Institute and legal counsel representing the INTERSALT colleagues". This vague account omits relevant information about the actual sequence of events: During the five years from publication of INTERSALT findings in BMJ in July 1988 to June 1993, the INTERSALT investigators received no request from the Salt Institute for further data or data analyses. The first communication was a pejorative letter dated June 18, 1993, from the Washington law firm representing the Salt Institute, not from the Institute itself or from any researchers acting for it.

This letter -- addressed to the INTERSALT International Co-Principal Investigator in the U.S. -- requested that the INTERSALT "study data underlying the findings" be made available to the Salt Institute. Based on consultation with all members of the INTERSALT Steering and Editorial Committee, with INTERSALT local Principal Investigators at the 52 centers around the world (the collectors and owners with their institutions of the INTERSALT data), and University and personal attorneys, this request from Salt Institute lawyers for the INTERSALT raw data was rejected, for several important stipulated reasons, summarized in the INTERSALT paper in this BMJ (14).

At the same time, the INTERSALT investigators indicated their willingness to do any scientifically sound and practically feasible further analyses that the Salt Institute proposed, and to make the detailed findings available to the Salt Institute. Since the Salt Institute communication had come from its lawyers, the INTERSALT International Co-PI in the U.S. was advised by University and personal attorneys to respond through them, and other INTERSALT PIs in the U.S., the U.K., and other countries agreed.

Hence the "process involving negotiations between lawyers" was a consequence of Salt Institute approaches, not the choice of INTERSALT researchers. These exchanges led to agreement that INTERSALT would do further analyses on pattern of BP with age for the 52 samples worldwide. These were done, as were further cross-population analyses relating sample median sodium excretion to sample difference in BP in middle age compared to young adulthood. Detailed data were made available to the Salt Institute in November 1994, after the Salt Institute agreed through its attorneys to adhere to the standard professional procedure of treating the data as confidential pending their publication. The data were written up, the paper was submitted to the BMJ, underwent peer review, was revised accordingly, and is in this BMJ (14).

In its use of these data (generated at its request), the Salt Institute -- both before the U.S. Dietary Guidelines Advisory Committee in January 1995 and in its piece in this BMJ -- has omitted any mention of the findings made available to it on the relationship of salt to BP. The INTERSALT findings were apparently not to its liking: there were consistent, strong, independent, significant relationships of sample median sodium excretion to sample difference in BP in middle age compared to young adulthood. These findings were concordant with original cross-population INTERSALT analyses reported in 1988, concordant also with within-population INTERSALT findings on sodium excretion of individuals and their systolic pressures (reported in 1988 and updated here), and with extensive findings from other research using every investigative method. All these findings supported the judgment that high salt intake, common in populations worldwide, is an important adverse exposure leading etiologically to population adverse blood pressure levels and impairment of health.

3. Salt Institute Efforts to Overturn Public Health Recommendations for Lower Salt Intake by the Population

*As is clear from its formula, R2 for each sample is a function of slope of BP with age: R2 = b2 * (Sx2 /Sy2 ), where b is sample regression slope of BP on age, Sx2 is sample variance of age, and Sy2 is sample variance of BP. As the formula shows, the larger the slope, the larger the R2. For samples with BP not higher with age, or actually lower with age, both slope (b) and R2 are near zero; age, unrelated to BP, does not "explain" BP.

Department of Preventive Medicine, Northwestern University Medical School, Chicago, IL 60611-4402, USA Jeremiah Stamler, professor emeritus Alan R. Dyer, professor Rose Stamler, professor emeritus

Department of Epidemiology and Public Health, Imperial College School of Medicine at St. Mary's, London W2 1PG Paul Elliott, professor Department of Epidemiology, St. Raphael University Hospital, Leuven, Belgium Hugo Kesteloot, professor

Department of Epidemiology and Public Health, University College, London, Medical School, London Michael Marmot, professor

References

1. U.S. Department of Health and Human Services. The Surgeon General's Report on Nutrition and Health. Washington: US Government Printing Office, 1988.

2. National Research Council, Committee on Diet and Health, Food and Nutrition Board, Commission on Life Sciences. Diet and Health: Implications for Reducing Chronic Disease. Washington D.C.: National Academy Press, 1989.

3. Nationjl High Blood Pressure Education Program. The Fifth Report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC V). Arch Intern Med 1993;153:154-183.

4. National High Blood Pressure Education Program Working Group. National High Blood Pressure Education Program Working Group Report on Primary Prevention of Hypertension. Arch Intern Med 1993;153:186-208.

5. Department of Health. Nutritional aspects of cardiovascular disease. Report of the Cardiovascular Review Group, Committee on Medical Aspects of Food Policy. London: Her Majesty's Stationary Office, 1994. (Report of Health and Social Subjects #46)

6. U.S. Department of Agriculture, U.S. Department of Health and Human Services. Report of the Dietary Guidelines Advisory Committee on the Dietary Guidelines for Americans, Fourth Edition, 1995. U.S. Government Printing Office, Washington, D.C.

7. WHO Expert Committee on the Prevention of Coronary Heart Disease. Prevention of coronary heart disease. World Health Organ Tech Rep Ser 1982; 678.

8. WHO Expert Committee on the Prevention In Childhood and Youth of Adult Cardiovascular Disease. Prevention in childhood and youth of adult cardiovascular diseases-- time for action: report of a WHO expert committee. World Health Organ Tech Rep Ser 1990; 792.

9. WHO Study Group. Diet, nutrition, and the prevention of chronic diseases. World Health Organ Tech Rep Ser 1990; 797.

10. Neaton JD, Kuller L, Stamler J, Wentworth DN. Chapter 9: Impact of systolic and diastolic blood pressure on cardiovascular mortality. In: Laragh JH, Brenner BM, editors. Hypertension: Physiology, Diagnosis, and Management, Second Edition. New York: Raven Press, 1995:127-144.

11. Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Horan MJ, Labarthe D. Prevalence of hypertension in the adult US population: Results from the Third National Health and Nutrition Examination Survey, 1988-91. Hypertension 1995; 25: 305-313.

12. Denton D, Weisinger R, Mundy NI, Wickings EJ, Dixson A, Moisson P, Pingard AM, Shade R, Carey D, Ardaillou R, Paillard F, Chapman J, Thillet J, Michel JB. The effect of increased salt intake on blood pressure of chimpanzees. Nat Med 1995; 1:10091016.

13. INTERSALT Co-operative Research Group. INTERSALT Study--An international study of electrolyte excretion and blood pressure: Results for 24 hour urinary sodium and potassium excretion. Br Med J 1988; 297: 319-328.

14. Elliott P, Stamler J, Nichols R, Dyer AR, Stamler R, Kesteloot H, Marmot M, for the INTERSALT Cooperative Research Group. INTERSALT revisited: further analyses of 24 hour sodium excretion and blood pressure within and across populations. Br Med J 1996, May 17, 1996.

15. Rosner B. Fundamentals of Biostatistics. Boston: Duxbury Press, 1982, p. 368.

16. Neter J, Wasserman W, Kutner M. Applied Linear Regression Models. Homewood, IL: Richard D. Irwin, Inc., 1983, p. 83.

17. Kleinbaum DG, Kupper LL. Applied Regression Analysis and Other Multivariable Methods. North Scituate, MA: Duxbury Press, 1978, p. 56.

18. INTERSALT Cooperative Research Group (Elliott P, Guest Editor). The INTERSALT Study An international cooperative study of electrolyte excretion and blood pressure further results. J Hum Hypertens 1989;3:279407.

19. Mancilha-Carvalho JJ, Baruzzi RG, Howard PF, et al. Blood pressure in four remote populations in the INTERSALT Study. Hypertension 1989; 238-246.

20. Stamler J. Letter regarding salt dietary guideline to Dr. Doris Calloway, Chair, Dietary Guidelines Advisory Committee, Department of Agriculture and Department of Health and Human Services, in reply to the letter of Salt Institute of January 31, 1995. February 27, 1995.

21. U.S. Department of Agriculture, U.S. Department of Health and Human Services. Dietary Guidelines for Americans, Fourth Edition, 1995. U.S. Government Printing Office, Washington, D.C.

---

Home | Current contents | Past issues | Classified ads | Career Focus | Feedback
Collections | About this site | About the BMJ | BMA | Medline

---