By Paul J. Rosch, M.D.
President, The American Institute of Stress
Clinical Professor of Medicine and Psychiatry
New York Medical College
Originally published in the Health and Stress newsletter of The American Institute of Stress

The Art of Mining Salt Study Statistics

The INTERSALT study seemed to confirm Dahl's findings. However, when the four primitive societies with both extremely low sodium intake and very low blood pressures were excluded no such correlation was found in the other 48 groups.

This was reminiscent of Ancel Keys' famous study where he "cherry picked" seven countries out of 15 around the world and demonstrated a straight-line relationship between animal fat and cholesterol consumption and deaths from coronary heart disease. Had Keys selected data from the eight other countries that were available to him the results would have been exactly the opposite.

The INTERSALT researchers conveniently neglected to mention that the population of the four countries responsible for skewing the total figures to coincide with their preconceived conclusion also had less stress, less obesity, ate far less processed foods and much more fiber from fruits and vegetables. They also tended to die at younger ages from other causes and often too soon to have developed any significant degree of coronary atherosclerosis.

Critics complained that these four societies that distorted the average figures for sodium intake and hypertension were so different from the rest of the groups, especially those in the U.S.A. and U.K., that it was "like comparing apples with stringbeans rather than oranges."

The Yanomami Indians in the rain forests of Brazil had mean blood pressures of 95/61 and equally low urinary sodium levels. These primitive people had no evidence of hypertension, obesity or alcohol consumption and their blood pressures did not rise with age.

When the available data from the other more civilized societies was reviewed, statisticians found that as sodium intake increased there was a decrease in blood pressure, just the opposite of what had been reported. The lowest salt intake seemed to be in a subgroup of Chicago black males despite the fact that their incidence of hypertension was above average. Conversely, high blood pressure was relatively rare in participants from China's Tianjin Province even though this study group had the highest salt intake.

When confronted with these discrepancies, the researchers reanalyzed their data in an attempt to justify their conclusions. However, the only thing they could come up with was that a higher sodium intake could be correlated with a faster rise of blood pressure as people grew older. This is referred to as "mining the data" since a relationship between blood pressure and aging was never a goal of the study. Nor did this observation address the major purpose of determining whether increased dietary sodium was related to higher rates of illness or death for everyone.

While it may be true that "figures don't lie," liars can still figure. The first law of statistics is that if the statistics do not support your theory you obviously need more data. The second is that if you have enough data to choose from, anything can be proven by statistical shenanigans. A good example are the numerous "risk factors" for coronary heart disease like a deep earlobe crease or premature vertex baldness that are really "risk markers." These simply represent statistical associations rather than competent causes. You can't use a statistic to prove another statistic.

However, the anti-salt statisticians had a field day with the data from the 1999 follow-up study of NHANES (National Health and Nutrition Examination Survey) which began tracking 20,729 Americans in 1971. They reported that participants who ate the most salt had 32 percent more strokes, a whopping 89 percent more deaths from stroke, 44 percent more heart-attack deaths, and 39 percent more deaths from all causes.

This finally seemed to prove precisely what the government had been preaching all along. In addition, the study's conclusions were seemingly credible due to the large number of subjects and a 19-year average period of observation, enough time to determine whether people would have increased mortality rates or a higher incidence of illness from consuming too much salt.

As the lead author proudly proclaimed, "Our study is the first to document the presence of a positive and independent relationship between dietary sodium intake and cardiovascular disease risk in adults".

Pouring Salt in Low-Sodium Wounds

However, when independent researchers reanalyzed the data they discovered that dietary sodium intake was associated with higher rates of illness and death only in participants who were overweight. There was no correlation between sodium and increased cardiovascular disease risk in the remainder. Undaunted, another study author continued to claim that the conclusions were valid since statistics showed that more than one in three Americans were overweight and most ate too much salt.

He admitted that the NHANES research "was not specifically designed to answer" the question of sodium and health - in other words, more mining of the data. In addition, the entire study depended on just one 24-hour recall of sodium intake. When questioned about the dubious value of such information he was forced to concede that "At best, the estimate for sodium is imperfect". He also agreed that measuring the concentration of sodium in a 24-hour urine specimen would have provided more accurate information about dietary habits and excess consumption.

Statistics are somewhat like expert witnesses in that they can be used to testify for either side depending on what you want to prove. When Michael Alderman, a highly regarded epidemiologist and past president of The American Society of Hypertension scrutinized the same data in patients who were not overweight he reported that "the more salt you eat, the less likely you are to die."--(from heart disease or anything else).

Alderman has long been critical of the government's low sodium diet advice for large populations and their focus on sodium intake as it relates to blood pressure rather than to the overall health, quality and length of life of individuals. He examined the relationship between sodium intake and health effects in 3,000 patients with mild to moderate hypertension. In addition, his group measured sodium excretion, which is much more accurate than estimating dietary intake. At the end of four years, they found that those who consumed the least sodium had the most myocardial infarctions and other cardiovascular complications.

The reason for this is that when you restrict vital nutrients like salt (or cholesterol) all sorts of strange things can result. Low sodium diets can increase levels of renin, LDL and insulin resistance, reduce sexual activity in men and cause cognitive difficulties and anorexia in the elderly. Tasteless and dull low sodium diets can cause other nutritional deficiencies. Lowering sodium with diuretics to treat hypertension can cause similar problems.

Renin is possibly the most powerful and dangerous blood pressure raising substance known. Indeed, the study done by Alderman's group found that for every two percent increase in pretreatment plasma renin activity there was a 25 percent increase in heart attacks. No such correlation was found with increased sodium intake.

There are no research reports that justify putting everyone on a low-sodium diet. A meta-analysis of 83 published studies that included people who had been randomly assigned to follow a high or low sodium diet found that in those with elevated blood pressures, a low sodium diet was able to lower systolic pressure 3.9 mm Hg and diastolic pressure by 1.9 mm Hg.

However, in others with normal pressures, cutting salt intake reduced blood pressure by only 1.2 mm systolic and 0.26 mm diastolic. I don't know how many of you have ever taken a blood pressure but it is almost impossible to detect such minute differences. If you use the standard method and take repeated blood pressures over a few minutes each reading often varies by 5 mm. or more and it is extremely difficult to detect a diastolic measurement difference of 2 mm.

These figures were arrived at because meta-analysis is a technique that allows statisticians to look at studies that may have been designed for different reasons but contain data on specific items that can be combined and averaged for whatever purpose you choose.

I have never been a great fan of meta-analysis, since it often illustrates that "statistics are a highly logical and precise method for saying a half-truth inaccurately." Low sodium diets may be helpful for some hypertensive patients by reducing their need for drugs but there is no proof to support official recommendations that they are good for everybody.

Slipping Through Some Legal Loop-holes

As previously noted, low salt diets may not be as entirely harmless as proponents often claim. In the meta-analysis survey, which was published in the Journal of the American Medical Association a few years ago, researchers reported that cholesterol and LDL "bad" cholesterol increased with sodium reduction. More importantly, blood levels of renin and aldosterone also rose in proportion to the degree of sodium reduction.

This compensatory response to increase blood volume would tend to raise blood pressure and possibly the likelihood of cardiovascular complications. Since the government began promoting sodium restriction and diuretics three decades ago, the incidence of hypertension and strokes has increased and the previous declining rate of heart attacks has leveled off.

Investigators from the Salt Institute also wondered why there would be any dramatic rise with age if population blood pressures showed no association with dietary sodium intake. Because this was the only positive finding of the INTERSALT study they asked if an independent expert could analyze all the data, especially since this was a research project that had been funded by taxpayer money.

The study authors refused claiming proprietary ownership and that this was only the first in a series of papers. It would also reveal confidential information about the study participants which, under INTERSALT's policies and alleged federal regulations, they were "obligated to protect from disclosure."

The NIH, which funded the study, was also petitioned but said that the financial arrangement had been structured specifically to exclude them from access to the raw data. This seemed strange. Sensing that some significant information was being withheld and mindful of the old saying that "the devil is in the data", the Salt Institute refused to be stymied.

They asked the ORI (Office of Research Integrity) to determine whether the authors' findings had been fairly reported. ORI claimed they could only proceed if it was claimed that the authors had committed fraud--a Catch-22 situation, since it was impossible to make such an accusation without access to the raw data.

The Salt Institute then sought legal relief. The law requires that all federal guidelines affecting the public must be written and promulgated according to the Government Code. This mandates open meetings and discussions and that the final rules or guidelines must be published in the Federal Register.

It took three years for their attorneys to finally obtain the raw data dealing with just one of several specific questions that had been posed. This was enough to bring down the house of cards. A detailed explanation of how the data had been manipulated to support predetermined conclusions was published in the British Medical Journal in 1996 and was subsequently endorsed by various authorities.

The NIH has consistently circumvented the Government Code with its cholesterol and hypertension guidelines by claiming they were written by outside experts not subject to these regulations, even though they are presented as official policy. The National Heart, Lung and Blood Institute, Department of Health and Human Services and U.S. Department of Agriculture have repeatedly referenced the INTERSALT study as justifying sodium restriction.

The FDA even authorized a "sodium and hypertension" food label health warning that states, "The INTERSALT study reported a statistically significant relationship between sodium intake and the slope of systolic and diastolic blood pressure with age." How can anyone claim that this is not official policy?

In 1998, Congress mandated that federal agencies make available to the public all such data by broadening the Freedom of Information Act. It also included other provisions for the Office of Management and Budget to require all federal agencies to adhere to this new access-to-data standard. Unfortunately, this is not retroactive. Fifteen years later we still do not have access to all the INTERSALT data and hundreds of studies started prior to 1998 are also exempt.

Last month, a congressional bill was introduced mandating that the results of the more than $45 billion spent annually for research should be freely available to taxpayers. It would also prohibit all scientists who receive federal funding from holding copyright to their research. Don't hold your breath waiting for this bill to become law.

The DASH Study-Déja Vu All Over Again?

The NIH funded DASH (Dietary Approaches to Stop Hypertension) study reported in 1997 that blood pressure could be significantly reduced by eating a diet rich in fruits, vegetables and low-fat dairy products. This DASH combination diet was more effective than a typical American high fat, low fiber, low mineral diet and even one of fruits and vegetables, particularly in people with elevated blood pressures. All three diets had the same sodium content and there was no attempt to restrict salt. Government officials were anxious to show that restricting sodium would lower blood pressure even more.

This seemed to be confirmed in a follow-up DASH-Sodium study in 412 subjects with elevated and normal blood pressures that were randomly assigned to follow the DASH diet or a control typical American diet. The two groups were further divided into three categories: those who ate 3.3 grams of sodium/day (the amount in the average American diet); 2.4 grams/per day (the current recommended level); and 1.5 grams/day.

Researchers reported in May 2000 that reducing sodium intake from the high to low levels resulted in an average progressive lowering of systolic blood pressure of 6.7 mm Hg for those on the control diet and drop of 3 mm Hg for Dash Diet subjects. Hypertensive patients showed a greater response to a low sodium diet in both groups, with an impressive 11.5 mm Hg reduction for those on the control diet.

Thus, sodium restriction lowered blood pressure in hypertensive and nonhypertensive men and women regardless of race. The belief that, "the lower the blood pressure the better", prompted the NHLBI director to declare that the four-decade-old controversy was now over. Everyone should adhere to a low sodium diet.

Not everyone agreed. The DASH diet was rich in calcium, potassium, and magnesium, all of which have been found to lower blood pressure. The study group was not representative of the American public and all meals had been prepared rather than selected. The available statistics suggested that for those on the DASH diet with normal blood pressures, cutting salt intake in half had little effect.

Diet was the most important influence and there was no significant additional benefit in hypertensives who also restricted salt. Participants were only followed for a month and prior studies had shown that any blood pressure reductions associated with restricting sodium tend to disappear after six months as compensatory mechanisms kick in.

Since all subjects were fed prepared meals there was over 95 percent compliance, which would be difficult to achieve in a real life setting where people choose the foods they want to eat. Almost 60% of the subjects were African Americans and over 40% were hypertensive. Both of these groups tend to be salt sensitive and are hardly representative of the general population.

David McCarron, a hypertension specialist argued that the figures suggested that no benefits would be seen in white men under the age of 45, but here again, all the data were not available. As in the past, requests to release all the data were denied. McCarron complained about this in a letter to The New England Journal of Medicine and in a January editorial in the American Journal of Hypertension, which stated "critical data from a federally sponsored trial have been withheld." Nothing happened.

On May 15, the Salt Institute and the U.S. Chamber of Commerce sought legal relief by invoking the Data Quality Act that took effect last October. This regulation now mandates that official agencies promulgating "influential" results that affect large groups must provide enough data and methods for a "qualified member of the public" to conduct a reanalysis. Since NHLBI's latest sodium restriction recommendations clearly affect a very large group of people and are based on the DASH-Sodium study, the argument that all subgroup data should be made available seems quite valid.

DASH authors will probably argue that they plan to publish more papers and, as noted in a response to McCarron's editorial, they are concerned that he will "dredge the data" and perform statistical analyses on groups that are too small to be meaningful. NHLBI has 60 days to respond but based on past experience, will likely continue to sidestep federal regulations and stonewall concerned scientists.

Should You Avoid Salt? Which Of Some 100 Blood Pressure Pills is The Best For You?

What's the bottom line? Sodium restriction can benefit certain salt sensitive hypertensive patients and might possibly delay the development of high blood pressure in others.

However, this does not apply to the general population, where no study has ever found an association between low-sodium diets and a reduced incidence of cardiovascular or other diseases. Average results from large study groups are not a useful guide to determine optimal treatment for a particular patient. A low fat diet can elevate cholesterol in some even though a mean decrease may occur in a population. An eight-year study of New York hypertensives found that those on low-salt diets had more than four times as many heart attacks as controls with normal sodium intake.

Unfortunately, there is no simple way to determine whether you are "salt sensitive" other than to go on a high sodium diet for a few weeks and then a low sodium diet to determine whether there is a significant change in blood pressure. The NIH recently invited applications for grants to develop an easily administered screening test for salt sensitivity.

Several molecular markers have been proposed and Tulane researchers received a $6.5 million grant to identify genes that might be associated with salt-sensitive hypertension, but a simple and accurate test seems a long way off. The health consequences of salt sensitivity may not be limited to effects on blood pressure. One study showed a link with increased insulin resistance and another found that salt sensitivity increased mortality rates regardless of whether or not it was associated with hypertension.

There is growing recognition that hypertension is a complex metabolic disorder and that treatment efforts must be personalized and directed towards reducing its complications. This is quite different than simply attempting to lower elevated pressures to an arbitrary value based on large-scale study results. A good example is the ALLHAT trial, which concluded that the normal range for blood pressure should be lowered and a thiazide diuretic should be first line therapy for all hypertensives.

There is good reason to believe that this could increase cardiovascular and other complications like diabetes. Some take the view that since most hypertensives usually require more than one type of medication, a shotgun approach using minimal doses of diuretics, beta-blockers, calcium channel antagonists or drugs that affect the renin-angiotensin-aldosterone system is more practical. In contrast, others believe that 60 percent of hypertensives can be controlled on one drug and most others on two.

John Laragh proposes that there are basically two types of essential hypertension: those that are low renin and salt sensitive (30 percent to 35 percent) that respond to antivolume drugs like diuretics, and renin mediated hypertension (60 percent to 65 percent), which can now be treated with one of several antirenin medications based on renin profiling.

The PRA (plasma renin activity) assay he and Sealey developed decades ago was very sensitive and labor intensive. The "Laragh Method" that now uses an automated and widely available direct renin assay seems to be the most logical approach to treat hypertension and reduce its complications. Stay tuned for more on this!