**You Are Here:** U.S. Preventive Services Task Force > Topic Index > Vitamin D Supplementation > Evidence Update

#### Vitamin D With or Without Calcium Supplementation for Prevention of Cancer and Fractures (continued)

## Appendix

A total of 33 publications met our eligibility criteria for KQ 2 (the associations between vitamin D status and the clinical outcomes of cancer), but 5 publications were excluded because they were superseded by later publications in the same cohort with more cancer cases (70-73). Many cohorts had multiple publications reporting different cancer outcomes of interest. There was no overlap in study populations in each cancer outcome in our systematic review and in each dose–response meta-regression.

We performed linear dose–response meta-regressions to examine the associations between blood 25-(OH)D concentrations and the risk for prostate and colorectal cancers by using a mixed-effect logistic regression model. Specifically, we fitted a mixed-effects meta-regression (with fixed intercepts and random slopes) using the exact binomial likelihood, which explicitly models between-study variability in the strength of the dose–response relationship. For each study, we back-calculated the “effective counts” of events in each category of 25-(OH)D concentration based on the pertinent adjusted log odds ratios (vs. a reference exposure category), their variance, and the total number of participants per exposure category and by solving a set of nonlinear equations (74). The effective counts of events are such that when used in a logistic regression with the exposure categories as the sole predictors, they result in the same log odds ratios (coefficients), variances, and covariances as those from the original adjusted model. The mean value per exposure category of 25-(OH)D concentration is also needed for dose–response meta-regressions. When it was not reported, the midpoint between exposure category thresholds was selected, and for the open categories, we imputed a mean intake 20% lower for the lowest quintile threshold or 20% higher for the highest quintile threshold, respectively.

To show the individual study results in Figure 3, we calculate the adjusted probability of cancer (odds/[1 + odds]) by using the effective numbers of case-patients and controls and plotted against the mean value of each exposure category of 25-(OH)D concentration.

*AHRQ Publication No. 12-05163-EF-3
Current as of December 2011*

**Internet Citation:**

Chung M, Lee J, Terasawa T, Lau J, Trikalinos TA. *Vitamin D With or Without Calcium Supplementation for Prevention of Cancer and Fractures: An Updated Meta-Analysis for the U.S. Preventive Services Task Force*. AHRQ Publication No. 12-05163-EF-3. December 2011. http://www.uspreventiveservicestaskforce.org/uspstf12/vitamind/vitdeart.htm