It's 96% effective against the original covid strain, 86% effective against UK variant B.1.1.7, but only 49% effective against the South African variant B.1.351...
That 49% has two confounding factors, 1: it includes a significant portion of hiv positive individuals, the efficacy was 60% if you exclude them. 2: N is not large enough to give you reasonable confidence in either of those numbers the ci was between 19%-80%, which is almost useless, not quite, since it seems like there is evidence that it is at least a little less effective, but by how much is still anyone’s guess at this point.
> only 49% effective against the South African variant
The South African data is a much smaller study without much statistical power and with an extremely wide 95% confidence interval. I’d take that figure with a grain of salt. Kudos to Novavax for including it nonetheless.
I am not an expert, but I think I read that once you acquire immunity to a virus it might be more difficult to acquire same or better level of immunity to a similar virus.
That's why you can't take vaccine for all strains of influenza and if prediction of which strain is going to be prevalent is wrong and you take wrong vaccine you can't do much about it.
Now, the Covid-19 vaccines are different mechanism so I don't know if or how this translates. For now what I read about the way the mechanism is built on some of the vaccines is that these should really be effective even if there are large changes to the virus. But there are of course always specific mechanisms and sequences that the immune system requires to recognize and act on virus and so there can never be universal vaccine.