Slides

Add Presentation to Slide Tray Presentation:

Quantitative Synthesis II Quiz

Authors: Joseph Lau, M.D., and Thomas Trikalinos, M.D., Ph.D.

Slides: 1–7 of 7

Quantitative Synthesis II Quiz. Prepared for: The Agency for Healthcare Research and Quality (AHRQ) Training Modules for Systematic Reviews Methods Guide www.ahrq.gov

Add to tray Remove from tray

Meta-Regressions and the Risk of Ecological Fallacy

$This slide presents an excerpt of text describing “Results and Conclusions” from a scientific article. The text reads: Results/Conclusions: It is projected that raising the minimum year-around serum 25(OH)D level to 40 to 60 ng/mL (100-150nmol/L) would prevent approximately 58,000 new cases of breast cancer and 49,000 new cases of colorectal cancer each year, and three fourths of deaths from these diseases in the United States and Canada, based on observational studies combined with a randomized trial. Such intakes are also expected to reduce case-fatality rates of patients who have breast, colorectal or prostate cancer by half. There are no unreasonable risks from intake of 2000IU per day of vitamin D3 or from a population serum 25(OH)D level of 40 to 60ng/mL. The time has arrived for nationally coordinated action to substantially increase intake of vitamin D and calcium. At the bottom right side of the figure is a regression plot. The X axis is labeled “Mean 25(OH)D concentration.” The scale starts at zero and has major tick marks at 10, 20, 30 and 40. The Y-axis is labeled “Odds ratio breast cancer”. The scale is labeled from 0.1 to 1.0. A horizontal dotted line crosses the graph from a point on the y-axis at approximately 0.425. It is met by a vertical dotted line that starts on the x-axis at approximately 37 ng/mL. Four study estimates are represented on the plot by filled diamonds. The points are located at coordinates (10, 1), (15, 0.7), (25, 0.6) and (38, 0.5). A regression line slopes downward from left to right between these points. A legend reads R-squared = 0.73, p trend < 0.02. A text box within the area bounded by the dotted lines reads.”58% reduction in breast cancer risk associated with 38ng/mL serum 25(OH)D/\.” Based on this example, which of the following statements is more likely to be true? A. These conclusions are based on a well constructed meta-regression (see on the right) and are definitely valid. B. The analyses and conclusions are suspect.$

Add to tray Remove from tray

Assessing Study Conclusions

Summary. Meta-analyses require understanding of covariates and how to analyze them. Patient-level covariates differ across patients in the same study or in the same study arm. Study-level covariates pertain to the whole study and do not vary across patients in the same study. Results of subgroup analyses should be viewed with skepticism, especially when adjustments for multiple testing have not been performed. If an association from a meta-regression on the mean of a patient-level covariate is biologically plausible, it merits further study.

Add to tray Remove from tray

Summary

Authors. This presentation was prepared by Joseph Lau, M.D., and Thomas Trikalinos, M.D., Ph.D., members of the Tufts Medical Center Evidence-based Practice Center. The information in this module is based on Chapter 9 in Version 1.0 of the Methods Guide for Comparative Effectiveness Reviews (available at: http://www.effectivehealthcare.ahrq.gov/repFiles/2007_10DraftMethodsGuide.pdf).

Add to tray Remove from tray

Slides

Quantitative Synthesis II Quiz

Quantitative Synthesis II Quiz

Covariates

Subgroup Analysis

Meta-Regressions and the Risk of Ecological Fallacy

Assessing Study Conclusions

Summary

Authors