Three of these figures carry a percent sign and only two of them are measuring the same kind of thing. That is worth a minute before anyone decides the inflammation result is the impressive one — it is the same units problem as any relative figure quoted without its denominator.
What was pooled
Nine randomized trials of subcutaneous semaglutide against placebo, covering 6,239 adults with overweight or obesity and without diabetes. [1] Weight change came out at a mean difference of −12.04% of body weight, 95% CI −13.08 to −11.00, and waist circumference at −9.36 cm, 95% CI −10.27 to −8.45. Both at p < 0.00001.
The number with a different denominator
C-reactive protein fell by a mean difference of 40.90%, 95% CI 35.42 to 46.37. Put beside a 12.04% weight change, that looks like the larger effect by a factor of three.
It is not, because the two percentages are shares of different things. 12.04% is a proportion of the person. 40.90% is a proportion of a blood test result, and which blood test result matters enormously: C-reactive protein varies by an order of magnitude between people, swings with any infection or injury, and has a distribution skewed hard to the low end. A large relative drop from a small starting value is a small absolute change.
The safety line is not a clean bill
Serious adverse events came out at a risk ratio of 1.12, 95% CI 0.76 to 1.65, p = 0.55. No significant difference.
That interval reaches 1.65. Nine randomized trials and 6,239 participants still leave open the possibility of serious events being 65% commoner on the drug, and equally the possibility of them being less common. It is a genuine unknown rather than a reassurance, and the honest reading is the same one every safety signal on this site gets.
One note on the conclusion
The authors point to the STEP UP trial of semaglutide 7.2 mg as showing greater weight reduction than previously available data. That is a higher dose than the pooled trials used, so it should sit next to the −12.04% rather than inside it.
On this roster a dose change is usually a price change too, which is the part a pooled efficacy figure never carries and whether a seller holds its price as the dose climbs answers instead.