Dietary Sodiumand Mortality: How Much Do We Really Know?

Demilade A. Adedinsewo; Amy W. Pollak; Rickey E. Carter

Disclosures

Eur Heart J. 2021;42(21):2113-2115. 

Graphical Abstract: The association of dietary sodium and mortality has long been a topic of research. However, the relationship between sodium and all-cause mortality at an individual level remains unclear.

Messerli et al. conducted a detailed examination of the association of sodium intake and mortality, and found an inverse association with mortality, as reported in this issue of the European Heart Journal.[1] The authors' conclusions are that such a finding should call into question the present understanding that higher sodium intake is associated with higher risk for mortality. We applaud the authors on taking on the immense task of evaluating the relationship between sodium consumption and mortality on a global scale. While the effort is laudable, there are several aspects about the study design and analysis that must be carefully considered when reviewing the findings of this article. (i) Ecological fallacy is a major concern when interpreting aggregate or country-level data. Ecological studies are mostly hypothesis generating, and interpreting the results on an individual level may be completely misleading. An extreme case of ecological fallacy was demonstrated by Salkeld et al. where Lyme disease was demonstrated to be negatively correlated with obesity rates and the abundance of fried chicken restaurants.[2] (ii) Estimates of sodium consumption are problematic. The authors correctly state that a single urinary sodium measurement may not reflect average sodium consumption during a person's lifetime. Primary data sources were limited or missing in many countries.[3] As pointed out in a recently published review by O'Donnell et al., a key issue, which becomes particularly relevant in observational population-based studies, is the lack of a valid and reliable method to objectively quantify sodium intake in individuals.[4] (iii) Healthy life expectancy and all-cause mortality are not ideal metrics in this context. Multiple randomized trials evaluating sodium reduction and the DASH diet have clearly demonstrated significant reductions in blood pressure[5] and, more recently, the results of the SPRINT trial confirmed that lowering blood pressure is associated with decreased cardiovascular morbidity and mortality.[6] Very recent articles including a meta-analysis from 20 studies confirmed a dose–response relationship between dietary sodium intake and cardiovascular disease risk,[7] and a clinical review article pointed out that prospective cohort study results have been mostly consistent in demonstrating that high sodium intake (>4.6 g/day) is associated with higher mortality and cardiovascular risk.[4] In contrast to the current study, a systematic analysis from the Global Burden of Disease study which evaluated dietary sodium intake, also using 24-h urinary sodium, found the opposite and showed associations between high sodium intake and cardiovascular mortality.[8] This study was more robust in that it evaluated other dietary contributions to mortality and specifically cardiovascular mortality. This study reports that in the WHO European Region, dietary risks accounted for 22.4% of all deaths and 49.2% of cardiovascular deaths—meaning that dietary risks account for less than a quarter of all-cause mortality. Another meta-analysis of 107 randomized interventions from 66 countries demonstrated that 1.65 million annual deaths from cardiovascular disease were attributed to increased sodium intake.[9] As such, evaluation of cardiovascular mortality would be a more appropriate study outcome for the study by Messerlii et al. The authors evaluated non-communicable disease mortality and found no significant associations. A global prospective study evaluating the association between macronutrients and cardiovascular disease showed that dietary fat intake (total and subtypes) was associated with lower total mortality, but found no associations between dietary fat intake and myocardial infarction or cardiovascular disease mortality.[10] The inverse relationship with mortality is unexpected as with the study in question; however, given no relationship to cardiovascular disease mortality, it is likely that the observed increased mortality among those with lower dietary fat intake was due to non-cardiovascular causes. (iv) The unit of analysis is not a person so the results do not apply to individuals. The unit of analysis is a single country-level estimate of sodium levels, and the person to person variation of sodium intake within countries has not been included in the analysis. The authors state 'unexpected vanishing of benefits due to statistical manipulations further attest to the muddiness of the waters regarding outcomes with sodium restriction', and this cannot be overstated as it also applies to the current study. The relationship between sodium consumption and death due to non-communicable diseases was insignificant, which clearly would be a more appropriate outcome than all-cause mortality as we do not expect sodium consumption and consequent risk of cardiovascular disease to increase the risk of death due to infectious diseases. (v) The impact of this study on sodium intake and population health remains unclear. The authors clearly emphasize that whether sodium intake affects life expectancy or all-cause mortality remains unknown and cannot be inferred from their data; as such, this study does not add more to the literature but emphasizes the uncertainty in drawing meaningful associations with fundamentally flawed studies. The U-shaped or J-shaped curve relationship between sodium and health outcomes has been demonstrated in multiple studies evaluating hyponatraemia and mortality. Malnutrition and cardiac cachexia have been identified as a potential contributor to hyponatraemia and might be responsible for the increased mortality seen in these patients. Patients with hyponatraemia tend to have lower levels of urinary sodium except in salt wasting syndromes; as such, their urinary sodium levels may not be reflective of their dietary sodium intake, and this further confounds the observed U-shaped associations between sodium intake (evaluated using urinary sodium measurements) and mortality. In addition, the most common formulas used to estimate dietary sodium intake from urinary sodium concentration have been reported to alter the relationship between sodium and mortality specifically, and they caution against the use of these biased estimates of sodium intake to refute the beneficial effects of a population-wide reduction in sodium intake.[11]

In summary, cardiovascular disease remains the leading cause of death globally. As such, it is not farfetched that interventions to reduce cardiovascular risk will ultimately reduce mortality attributable to cardiovascular disease. Messerli et al. present global data across multiple countries on the inverse association between dietary sodium intake and mortality.[1] These results need to be cautiously interpreted due to the potential for substantial confounding and ecological considerations (Graphical abstract). As rightly mentioned by the authors, these results cannot be interpreted at the individual level and should not be used to guide patient care. With regard to cardiovascular disease management, the American Heart Association has proposed seven ideal cardiovascular health metrics[12] with well-established health-related outcomes. Targeting just one of the seven may be insufficient to reduce overall cardiovascular risk in an individual, let alone in the general population. In addition, other factors are at play, including salt sensitivity which varies with age, race, sex, and renal disease.[13–15] However, this in no way suggests that focusing on one intervention, in this instance dietary modification, is not worth the trouble. Rather, it means that a multifaceted and multi-interventional approach is required to decrease cardiovascular disease burden and subsequently mortality.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....