Is "Wakefield cluster" a chance finding or a telltale clue to suspect research?

This page is research from an investigation by Brian Deer for the UK's Channel 4 Television and The Sunday Times of London into a campaign linking the MMR children's vaccine with autism. | Go to part I: The Lancet scandal | Go to part II: The Wakefield factor

One mystery in Andrew Wakefield's research concerns a cluster of parents apparently blaming MMR. Details of the first 30 children enrolled for the research were published first in the Lancet (only 12 cases), then in an abstract in the journal Gut (the first 12, plus the next 18). Of the first 12 cases, 8 apparently associated the MMR vaccine with autism: two out of three. But in the next 18 cases, only 3 apparently made this association: one out of six. How could this happen if the children were all seen under the same protocol?

Scenario 1: even spacing of MMR-autism reports as series continues

Scenario 2: bunching continues in line with reported Lancet cluster

These charts give two illustrative scenarios. In maroon are cases where parents apparently made the MMR-autism association. For the Lancet 12, they're distributed in accordance with the details given in the paper. The Gut abstract doesn't identify individual cases, so two scenarios are presented: first in which the continuation of the series into Gut involves an evenly spaced reporting of an MMR-autism link; second in which the bunching reported in the Lancet paper continues. Either assumption, or anything between, reveals an extraordinary cluster at the start of the research - research which began with a contract between Wakefield and lawyers.

Both scenarios assume that the data given in the Gut abstract are accurate, and haven't been toned-down - avoiding a transparently unbelievable figure for the cumulative number of parents alleged to be blaming MMR [reported in the abstract as 11/30]. In an earlier version of the same Lancet paper, based on the same 12 children, the parents of 9 are reported to have blamed MMR, which would make the above charts even more extraordinary. Analysis suggests that it was probably the ninth child in the series [in darker grey above] whose status changed between versions. Why the number of parents recorded in medical records making an association with MMR would go down hasn't yet been explained. Subsequently, some parents of the Lancet 12 children changed their minds, and the number alleged in court documents to be damaged by MMR went up from 8 [or 9, depending on the version] to 11 [the other was a US citizen, flown in for tests].

Curiously, the hospital's ethics committee ruled that only children enrolled after December 18 1996 were to be included in the trial - a stipulation accepted in writing by the investigators. But if all the children admitted to the hospital in this series prior to that date had been excluded [a total of 7, possibly 8, or even 9 in the earlier version, who must inevitably be from the far left of the diagrams], it's not clear how many allegations of links with MMR would have been left to publish in the Lancet and used to launch the worldwide vaccine scare.