The science of multivitamins

The arrow shows the direction of the measure: down when the measure fell, for example blood pressure. It turns green when the change is an improvement. When the size was assessed, 1, 2 or 3 filled arrows mean a small, moderate or large effect. The = sign means no clear difference.

Few studies: only 1 or 2 controlled trials measured this result. Mixed results: the trials disagree.

When sorting by effect, the result with the largest magnitude in each topic counts. In a tie, the better grade of that result counts first, then alphabetical order. “Largest improvement with these blends” considers benefits in comparisons where the effect of the blend can be isolated. “Largest measured effect” also includes markers and combined treatments.

The grade says how well it works: A, it works and is well studied; B, good results; C, few studies (only 1 or 2 controlled trials measured this result) or mixed results (the trials disagree); D, no clear effect. For safety, A and B mean it does not increase risk, C that there are doubts and F that it increases risk. The grade comes from fixed rules: the size of the effect in the main review, the number of trials and people, and the agreement between trials.

The three arrows point in the direction of the measure. For example, a shorter walking time points down. When the record documents the magnitude, one, two or three dark arrows correspond to a small, moderate or large effect. For other changes, one dark arrow shows only the direction. For standardized mean differences, the descriptive benchmarks are 0.2, 0.5 and 0.8 in absolute value.

The = sign means the test did not detect a difference. “Varied results” means the results diverged between studies, tests or time points. For case reports, the table states what was observed. The counts are of publications; one trial can give rise to several articles.

The records summarize each published article, some from the full text and others from the abstract only. This is a targeted selection of up to 150 publications, not a new systematic review. Of the 162 study entries in the results (one study can appear in more than one), 151 come from fixed rules: the largest trials for each result, one publication per trial, and trials with 1,000 or more people. The other 11 are recorded exceptions: 6 studies cited in the summary and 5 editorial choices.