As pointed out by James Doidge, the JCVI take estimates for protection of prior infection against symptomatic disease and apply it to estimates of severe disease, thereby overestimating benefits by a lot.