Cross-trial comparing reta vs tirz vs sema — methodologically how dirty is this really?

I keep seeing people stack the headline weight loss numbers side by side: 24.2% reta, 20.9% tirz, 14.9% sema. Different trials, different durations, different populations. How much can we actually conclude?

My take: the magnitude differences are large enough that the direction is probably right but the exact spacing is unreliable. We need direct head-to-head, which only exists for tirz vs sema right now (SURPASS-2 and the recent SURMOUNT-5).