Hello,

I think you can start with the link (meta-analysis of the regression slopes).

If you want to combine slopes and intercepts simultaneously, you will probably need  co-variance matrix from each study. I think that  slope, intercept and corrlation between x and y is insufficient, but look at materials provided for the R package metafor (2nd link).

I hope it will help.

With regards,

P.Holko

https://arxiv.org/pdf/0801.4442.pdf

http://www.metafor-project.org/doku.php/analyses

I also recommend using the metafor package for R. It is very versatile for meta-regressions. 

Are you using 'meaningful' measures that are the same across studies, or do the metrics for the outcome- and predictor measures vary? 

Thanks Jan, have been looking at the metafor package.

The outcome and predictor measures are the same across studies although there are potential moderators which differ from study to study.

OK - it makes things easier if you have the same metrics. Then you can use the beta-component and its SE as effect size and weight. 

Not sure, but this might be helpful: 

http://stats.stackexchange.com/questions/214107/linear-regression-from-two-meta-analysis-outcomes-in-r-metafor

Use the HLM software from Scientifc Software (free student version on the web) and construct a meta-anslysis using hierarchical linear modeling.