Latest salt Cochrane review including non-blood pressure surrogates

A new Cochrane review on dietary salt interventions was published last week by Graudal et. al (1), an update to a 2003 review. The meta-analyses examined the effects of randomized interventions with high or low dietary salt on blood pressure (BP), hormones and lipids. The trials included consisted of subjects with normal or elevated BP, and sodium intake estimated via 24 hour sodium excretion. In total, 167 studies were included that compared dietary sodium intakes <120 mmol with >150 mmol. In studies on subjects with elevated BP, the median trial duration was 28 days,and the average 24 hour sodium excretion in the trials were 71 mmol in the low-salt intake groups and 196 mmol in the high salt groups. In studies on subjects with normal BP, the median trial duration was 7 days, and the average 24 hour sodium excretion in the trials were 50 mmol in the low-salt intake groups and 201 mmol in the high salt groups.

Outcomes were systolic (SBP), diastolic (DSP), and mean BP (MBP), renin, aldosterone, adrenaline, noradrenaline, triglyceride, cholesterol, LDL, and HDL. The results were also analyzed by race.

Here is a summary of each meta-analysis: (note the full text has nice tables if you prefer reading results that way) Read results as low salt diet compared to high salt:

  • BP in normotensive Caucasians (62 trials, 71 comparisons): SBP: -1.27 mmHg, DBP: -0.05 mmHg. A sub-analysis of trials of more than 4 weeks (16 trials, 19 comparisons) found: SBP: -1.29 mmHg, DBP: -0.45 mmHg. For trials that also measured MBP (28 trials, 31 comparisons), MPB: 0.00 mmHg. Trials only measuring MBP (13 trials, 14 comparisons) found +1.01 mmHg.
  • BP in hypertensive Caucasians (74 trials, 76 comparisons): SBP: -5.48 mmHg, DBP: -2.75 mmHg. A sub-analysis of trials of more than 4 weeks (47 trials, 49 comparisons) found: SBP: -5.18 mmHg, and DBP: -2.59 mmHg. For trials that also measured MBP (21 trials, 23 comparisons), MPB: -3.56 mmHg. Trials only measuring MBP (5 trials, 7 comparisons) found -2.03 mmHg.
  • BP in normotensive blacks (6 trials, 7 comparisons): SBP: -4.02 mmHg, DBP: -2.01 mmHg (nonsignificant).
  • BP in hypertensive blacks (8 trials, 9 comparisons): SBP: -6.44 mmHg, DBP: -2.40 mmHg.
  • BP in normotensive asians (3 trials, 3 comparisons): SBP: -1.27 mmHg, DBP: -1.68 mmHg.
  • BP in hypertensive asians (7 trials, 7 comparisons): SBP: -10.21 mmHg, DBP: -2.60 mmHg.
  • Renin (70 crossover trials, 81 comparisons): standardized mean difference (SMD) of sodium reduction was 1.15. In comparisons of more than 2 weeks (29), the SMD was 0.67, and of more than 4 weeks (14) the SMD was 0.47.
  • Aldosterone (51 crossover trials, 59 comparisons): SMD: 1.36. In comparisons of more than 2 weeks (20), SMD was 0.99, and of more than 4 weeks (9) the SMD was 0.70.
  • Noradrenaline (31 crossover trials, 39 comparisons): SMD: 0.52. In comparisons of more than 2 weeks (12), SMD was 0.17, and of more than 4 weeks (6) the SMD was 0.06 (nonsignificant).
  • Adrenaline (14 crossover trials, 16 comparisons): SMD: 0.30. In comparisons of more than 2 weeks (8), SMD was 0.21, and of more than 4 weeks (5) the SMD was 0.24 (nonsignificant).
  • Cholesterol (24 crossover trials, 25 comparisons): cholesterol concentration: +5.76 mg/dl. In comparisons of more than 2 weeks (13), cholesterol concentration +2.48 mg/dl (nonsignificant), and of more than 4 weeks (9), cholesterol concentration was +3.21 mg/dl (nonsignificant).
  • A subanalysis of trials with moderate sodium reduction (median 81 mmol/24h) with 15 studies found an increase in cholesterol concentration of +3.86 mg/dl (nonsignificant).
  • LDL (15 crossover trials, 16 comparisons): LDL concentration: +2.88 mg/dl (nonsignificant). In comparisons of more than 2 weeks (8), LDL concentration +2.45 mg/dl (nonsignificant), and of more than 4 weeks (6), LDL concentration was +3.72 mg/dl (nonsignificant).
  • HDL (17 crossover trials, 18 comparisons): HDL concentration: +0.09 mg/dl (nonsignificant). In comparisons of more than 2 weeks (10), HDL concentration -0.61 mg/dl (nonsignificant), and of more than 4 weeks (8), HDL concentration was -0.14 mg/dl (nonsignificant).
  • Subanalyses of trials with moderate sodium reduction found no significant differences in HDL and LDL.
  • Triglyceride (18 crossover trials, 19 comparisons): TG concentration: +6.78 mg/dl. In comparisons of more than 2 weeks (11), TG concentration +7.78 mg/dl, and of more than 4 weeks (7), TG concentration was +8.37 mg/dl (nonsignificant; P = 0.09).
  • A subanalysis of trials with moderate sodium reduction (median 81 mmol/24h) with 12 studies found an increase in triglyceride concentration of +8.03 mg/dl.

So when salt is reduced, there are increases in several risk factors that may diminish (or theoretically eliminate) the reduced risk for cardiovascular disease predicted from the BP reduction. The authors note that in previous analyses these were not significant, but there are with the added data. The surrogates that were significant in the trials of at least 4 weeks were renin and aldosterone, whereas adrenaline, and triglyceride were “borderline significant” as they put it at 4 weeks with P = 0.09 and P = 0.10 respectively. The renin-aldosterone system (and adrenaline may also contribute) is what compensates upon sodium reduction and attempts to maintain blood pressure, explaining why the actual reduction in BP is smaller than what might be expected.

They also note that 6 other meta-analyses using different inclusion criteria on sodium and BP RCTs have found similar effect size estimates, suggesting the amount of sodium difference or duration of intervention doesn’t seem to have a significant effect on how much BP is lowered.

This struck me as odd from one of their last sentences:

Furthermore, sodium reduction resulted in a significant increase in plasma cholesterol (2.5%) and plasma triglyceride (7%), which expressed in percentage, was numerically larger than the decrease in BP.

Noting that there are bigger numbers of different variables doesn’t seem appropriate or relevant in a scientific paper.

And, when the trials were analyzed by duration of at least 2 and 4 weeks, the results were nonsignificant anyway, so it seems the effect is acute (for cholesterol).

They conclude with the following:

The increase in triglyceride was numerically unchanged in studies with a duration of at least 2 weeks and in studies with sodium reduction to moderate levels of sodium intake. Due to the relatively small effects and due to the antagonistic nature of the effects (decrease in BP, increase in hormones and lipids), these results do not support that sodium reduction may have net beneficial effects in a population of Caucasians. In Caucasians with elevated BP, short-term sodium reduction decreases BP by ~2–2.5%, indicating that sodium reduction may be used as a supplementary treatment for hypertension. In Asians and blacks, the effect of sodium reduction was greater, but at present too few studies have been carried out to conclude different from that above.

While this may be the case, there is no attempt to mathematically estimate the net effect (maybe not possible). They seem to think for hypertension it is worth reducing salt intake (which would be ~30% of Americans, and an additional ~37% have prehypertension), but don’t worry so much about normotensives. How do we know the changes in triglycerides will negate the benefits in reduction in BP? Additionally, the meta-analysis on triglycerides found a significance when all 19 comparisons were included and with the 11 trials of at least 2 weeks, but the 7 comparisons of at least 4 weeks duration were not statistically significant (P = 0.09). But they state that this is “borderline significant”. Is this an appropriate conclusion? Is it possible that this reflects an acute effect and more studies/data would further increase the P value? They write:

The present meta-analysis indicates that the adverse effect on lipids, especially triglyceride, is not just an acute effect as previously assumed,16 but may be persistent also in longer-term studies.

It is certainly a point that triglyceride was close to significance, but it still might be a relatively acute effect, since the trials of more than 4 weeks were trending away from significance.

It is also unfortunate that there isn’t (enough?) data on normotensives vs hypertensives on hormones and lipids to see if there are greater responses by the measured hormones and lipids in the hypertensive population.

Some Context

I recommend reading their Discussion. They cite some of the recent controversies like the Strazzulo et al. (2009) meta-analysis on prospective studies that found a 20% increased risk of cerebrovascular death on a high-salt diet vs low-salt (which seems to have some limitations), and the Stolarz-Skrzypek et al. (2011) cross-sectional and longitudinal results suggesting an inverse association with sodium excretion and cardiovascular events that was criticized immediately by the HSPH. They also note (importantly!) the recent studies suggesting sodium reduction could have a negative effect in patients with heart insufficiency and type 1 and 2 diabetes (1,2,3,4). Considering hypertension often goes hand in hand with these diseases, further research is needed immediately. The latest Dietary Guidelines recommend only 1,500 mg for those 50 or older, blacks, and people with hypertension, diabetes, and chronic kidney disease, which applies to 47.6% of Americans according to 2005-2008 NHANES, but 98.5% (99.4% over age 17) of this population consumes more per day anyway, on average at least 2,300 mg per day.

Net Effect?

Earlier this year there was another Cochrane review by Taylor et al. that combined 7 trials with at least 6 month follow up that included the surrogate BP but also directly measured cardiovascular outcomes. The reductions in BP in normotensive and hypertensive subjects were similar to the results here, but they were left with insufficient statistical power to estimate effects of salt reduction on cardiovascular morbidity or mortality. The poorly worded Plain Language Summary and media reporting created lots of confusion which was covered by Scott Gavura at Science Based Medicine and by myself here. But I noted in my post a letter submitted to the Lancet by He and MacGregor who combined the normotensive and hypertensive data for more statistical power and dropped one of the trials in their analysis and wrote:

“Our results show that there is now a significant reduction in cardiovascular events by 20% (p<0.05) and a non-significant reduction in all-cause mortality (5-7%), despite the small reduction in salt intake of 2.0-2.3 g per day.”

Taylor et al. noted in their discussion that such trials (6 months+) only modestly impact sodium excretion and BP. Sodium is in everything and reducing it long term is difficult. But it seems that even small reductions can have significant positive cardiovascular effects, at least according to that data. It would also be expected that with enough data on hypertensives the effect would be greater.

I am not an expert in this area. But as of now I perceive the evidence to be in favor of salt reduction, and especially for hypertensives and possibly ethnic groups other than Caucasians. The most effective way to do this at a population level would be to reduce added sodium to processed foods, which constitute the majority (~75%) of sodium intake. In my opinion, at the individual level, there would likely be more benefit switching the focus off reducing a single nutrient and onto adding foods to the diet that have similar or additive effects on blood pressure reduction. But the academic controversy sure is interesting!

References

1. Graudal, N., Hubeck-Graudal, T., & Jürgens, G. (2011). Effects of Low-Sodium Diet vs. High-Sodium Diet on Blood Pressure, Renin, Aldosterone, Catecholamines, Cholesterol, and Triglyceride (Cochrane Review) American Journal of Hypertension DOI: 10.1038/ajh.2011.210