Imagine that you have access to a collection of studies in which an age range, a sample size and an "effect" (say, cigarettes per day) is reported. What is the best way to construct a meta-regression of cigarettes per day by age?

The hard part here is to correctly interpolate across these studies with different sample sizes and different age ranges. First, the distribution of ages won't be reported. Second, I only have the effect mean and variance for the entire study. In the limit, it seems reasonable that enough studies with enough age ranges will give me an appropriate estimate of the relationship between age and cigarettes per day. Do you know if this sort of analysis has been done before, and if the appropriate statistical tests have been derived?


1 Answer 1


There is a Stanford study asked a similar question to your study.

Ohayon MM, et al. Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep. 2004;27:1255-73.

However, it is unclear how they represent the age range for each primary study as a covariate of sleep parameters. (fig 1a-d)

In my study, the purpose is to tell if the effect sizes for each study are related to age. Therefore, I categorized the age range for each study into "preschoolers", "children", "Adults" and "Seniors" and conduct sensitivity analysis/meta-regression.

Meta-regression is only useful to study the possible sources of heterogeneity in systematic review of observational studies. It could be useful to tell if packs smoked per day is related to age, but it might not be appropriate to estimate the packs smoked per day by age.

Egger M, et al. Meta-analysis Spurious precision? Meta-analysis of observational studies. BMJ 1998;316:140

  • $\begingroup$ Thanks for the suggestions. I suspect that there is some underlying statistics that can help address this general problem, still...shouldn't we be able to construct confidence intervals for packs smoked per day by age? $\endgroup$ Jan 13, 2012 at 21:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.