I have 3 papers suitable for inclusion in my systematic review looking at high versus low platelet to red call ratio in TBI, and want advice as to whether I can combine their estimates of effect in a meta-analysis.

One RCT which provides an unadjusted odds ratio and adjusted odds ratio of 28-day mortality for two groups (one intervention (high ratio) and one control (low ratio), adjusted for differences in baseline characteristics).

One retrospective cohort study which provides absolute unadjusted 28-day mortality data for two groups (one exposed to high ratio, and another exposed to a low ratio). They have also performed a sophisticated propensity analysis to adjust for the few differences between the groups and multivariate cox regression to adjust for factors associated with mortality, and presented hazard ratios.

Finally, a post-hoc analysis of a RCT, which compares outcomes for participants grouped according to presence/absence of haemorrhagic shock (HS) and TBI. This generates 4 groups - neither HS nor TBI, HS only, TBI only and TBI + HS. I am interested in the latter two as they included patients with TBI. One group was exposed to a high ratio, whereas the other a lower ratio. The authors provided unadjusted mortality data for all groups, and they adjust for differences in admission characteristics, to generate odds ratio of 28-day mortality. However, they present these adjusted odds ratios of death at 28days for the HS only, TBI only and TBI + HS groups compared to the neither TBI nor HS group, not to each other.

I could analyse unadjusted mortality in a meta-analysis, but want to know if I can combine all or some of the adjusted outcome measures, I have described instead? Any help greatly appreciated.