52 significant hazard-ratio estimates from vitamin D RCTs — most cluster below 1, but they are not 52 independent findings
Asked:
“Can you now take significant result and make a chart. Group rows by class, and for each class show CI ranges in different studies.” (Follow-up to: “Find me all RCTs for vitamin D. In the table, report study name, paper name, outcome measure, HR, CI.”)
Below are 52 deduplicated hazard-ratio estimates from vitamin D randomized controlled trials whose extracted 95% confidence interval excludes 1.0, drawn from 12 study labels and grouped into seven outcome classes. Falls, fractures, and frailty carries 22 estimates; cancer 15; diabetes and metabolic disease 5; autoimmune and inflammatory disease 4; cardiovascular disease 4; mortality and other clinical outcomes 1 each.
Read before interpreting
“Significant” here means only that the extracted CI excludes 1.0 — no multiplicity-adjusted p-values were used. Estimates mix primary, secondary, subgroup, sensitivity, post-intervention, exploratory, and ancillary analyses; multiple rows can come from the same trial or paper, and some are alternate labels or analytic variants of the same endpoint. Do not pool these values or rank effectiveness from them. HR < 1 favors fewer events only when the event is adverse; for outcomes such as medication persistence, interpretation differs. Data come from a broad automated web search and extraction, so labels and values reflect the source pages as found.
Forest plot of significant estimates
Point = hazard ratio; line = 95% CI on a log scale; vertical line marks HR = 1. Colour identifies the study within each class. Hover a row for the paper and source.
The widest intervals sit in falls, fractures, and frailty, where CI bounds span 0.05 to 6.0 across 3 study labels — several are small-subgroup or variant analyses of the same endpoints. agsjournals.onlinelibrary.wiley.com
Cancer contributes 15 significant estimates from 5 study labels, including two above 1 — invasive cancers (HR 1.22) and breast cancer (HR 1.34) in high vitamin D intake quartiles of the Women's Health Initiative. pmc.ncbi.nlm.nih.gov
All 4 autoimmune and inflammatory estimates come from a single study label (VITAL), with CI bounds between 0.38 and 0.99 — variants of overlapping autoimmune endpoints, not four trials. doi 10.1002/art.42811
Mortality and “other clinical outcomes” each rest on a single estimate; the lone other-outcomes row has HR above 1 and its direction is not necessarily unfavorable. sciencedirect.com
All 52 significant estimates
Class
Study
Outcome
HR
Lower
Upper
Paper
Source
Method: broad web systematic search of vitamin D randomized controlled trials with automated extraction of hazard ratios and confidence intervals; 52 deduplicated estimates whose extracted 95% CI excludes 1.0, grouped into seven outcome classes. Counts are of reported significant estimates, not trials, participants, or independent findings; non-significant estimates and effect measures other than HR were cut for space. Some source links are DOI identifiers or secondary summary pages as found by the crawl.