Methodological issues in cohort studies that relate sodium intake to cardiovascular disease outcomes: a science advisory from the American Heart Association

LK Cobb, CAM Anderson, P Elliott, FB Hu, K Liu… - Circulation, 2014 - Am Heart Assoc
LK Cobb, CAM Anderson, P Elliott, FB Hu, K Liu, JD Neaton, PK Whelton, M Woodward…
Circulation, 2014Am Heart Assoc
Background—The results of cohort studies relating sodium (Na) intake to blood pressure–
related cardiovascular disease (CVD) are inconsistent. To understand whether
methodological issues account for the inconsistency, we reviewed the quality of these
studies. Methods and Results—We reviewed cohort studies that examined the association
between Na and CVD. We then identified methodological issues with greatest potential to
alter the direction of association (reverse causality, systematic error in Na assessment) …
Background
The results of cohort studies relating sodium (Na) intake to blood pressure–related cardiovascular disease (CVD) are inconsistent. To understand whether methodological issues account for the inconsistency, we reviewed the quality of these studies.
Methods and Results
We reviewed cohort studies that examined the association between Na and CVD. We then identified methodological issues with greatest potential to alter the direction of association (reverse causality, systematic error in Na assessment), some potential to alter the direction of association (residual confounding, inadequate follow-up), and the potential to yield false null results (random error in Na assessment, insufficient power). We included 26 studies with 31 independent analyses. Of these, 13 found direct associations between Na and CVD, 8 found inverse associations, 2 found J-shaped associations, and 8 found null associations only. On average there were 3 to 4 methodological issues per study. Issues with greater potential to alter the direction of association were present in all but 1 of the 26 studies (systematic error, 22; reverse causality, 16). Issues with lesser potential to alter the direction of association were present in 18 studies, whereas those with potential to yield false null results were present in 23.
Conclusions
Methodological issues may account for the inconsistent findings in currently available observational studies relating Na to CVD. Until well-designed cohort studies in the general population are available, it remains appropriate to base Na guidelines on the robust body of evidence linking Na with elevated blood pressure and the few existing general population trials of the effects of Na reduction on CVD.
Am Heart Assoc