explain why a benefit of well over 100x the risk is "moderate" confidence? ...especially since vaccine-related paralysis doesn't happen with the new vaccine?

The short version: the "well over 100x benefit" claim is not supported by the evidence retrieved, and the confidence label reflected the thinness of the data, not a judgment about the ratio itself.

The retrieved records do not contain UK-specific polio case counts, death rates, or the exact epidemic-year incidence you mention. Without those numbers, no benefit-to-risk ratio can be calculated from this evidence. The 1 in 530,000 OPV-associated paralysis rate comes from a US estimate [1], and the Australian epidemic rate of roughly 1 in 5,000 is from your question, not from any record here. Comparing a number from the records to a number from your memory is not evidence-based analysis.

The confidence was "moderate on the general pattern, not clear on the specific UK count" because the general pattern — polio vaccines eliminated a severe disease, OPV caused rare paralysis, IPV eliminated that risk — is documented in the records [1][2]. The specific UK numbers are not. That is a statement about the retrieval, not about the real-world ratio.

You are right that if the true epidemic risk was 1 in 5,000 and the vaccine risk was 1 in 530,000, the benefit dwarfs the harm. But the evidence here does not establish that the UK epidemic risk was 1 in 5,000. The retrieval simply does not contain those figures.

My call: the evidence supports the general pattern of massive benefit from polio vaccination, but the exact UK benefit-to-risk ratio cannot be calculated from these records. Confidence: moderate on the general pattern, not clear on the specific UK ratio.