"If you tell me the Hazard Ratio and the number of events then I can tell you the p value (or pretty close to it)." If you knew only the number of events - both arms, and excluding the smaller trials - I would think you could get fairly close to the p value.