Retrieval of missing data for meta-analysis: a practical example

Int J Technol Assess Health Care. Summer 2004;20(3):296-9. doi: 10.1017/s0266462304001114.


Objectives: To examine the feasibility of retrieving missing outcome data for summary meta-analyses using an example dealing with the effects of aerobic exercise on lipids and lipoproteins in adults.

Methods: Missing lipid and/or lipoprotein data from a currently developed meta-analytic data base were requested by means of electronic mail from 39 of 174 (22.4 percent) eligible studies. Binary logistic regression was used to examine whether year of publication and country were significant predictors for whether data would be provided.

Results: Of the thirty-nine studies from which data were requested, usable data were received for thirteen (33.3 percent) of the studies. The addition of these previously missing data decreased the percentage of eligible studies that would have had to be excluded by 33.5 percent (from 22.4 percent to 14.9 percent). Neither year of publication nor country in which the study was conducted (United States versus other) were significant predictors of whether missing data would be provided or not (p > .05).

Conclusions: Moderate success was achieved in the acquisition of missing outcome data dealing with the effects of aerobic exercise on lipids and lipoproteins in adults. However, whether this level of response is true in other areas of research needs to be determined by additional research.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Clinical Trials as Topic / statistics & numerical data*
  • Exercise
  • Humans
  • Lipids / blood
  • Meta-Analysis as Topic*
  • Reproducibility of Results
  • Research Design


  • Lipids