months. The first and sixth SMFQ measures were not utilised for the current analysis as they were collected concurrently either with baseline covariates or the alcohol outcomes. The item ‘I was very restless’ was omitted from the 13-item SMFQ in our models because preliminary models based on the earlier time points suggested that individuals were uncertain about the meaning of this item—response rate for this item was noticeably lower and unlike the other 12 items, endorsement rates did not change monotonically with time.
