A sidebar in the report tells us about the data:
The data showed people in Scotland and Northern Ireland ate more saturated fat and salt, and fewer fruit and vegetables, every day than people in England, while differences between England and Wales were smaller.and:
Changing the diet to a typical English one would save about 11,000 of these lives - or just under 4,000 a year - with the biggest impact in Wales and Northern Ireland.That's odd, how can the biggest impact be in Wales if that's where the smallest dietary differences are? The answer is in Table 2 of the paper: it calculates that 53% of the "mortality gap" would be closed in 2009, 43% in 2007, and 120% in 2008. It's not obvious that these numbers should be quoted as a percentage of the mortality gap in that year, since presumably there's a time lag between eating imprudently and dying from it, but the big difference in 2008 is not in mortality but in the diet data, shown in Table 1, which shows the Welsh eating more of everything in 2008 than in the other two years.
The Table 1 data seem to come from this spreadsheet giving household nutrient intake and this spreadsheet for eating out. It's the household intake that has the anomalous data in 2008. I suppose that the nutrient data will have been calculated from the food purchasing data given here, which show that the Welsh bought more carcase meat (especially), cheese, fruit and vegetables, and biscuits in 2008 than in any other year from 2001 to 2009. I am deeply sceptical that that represents a genuine change in consumption rather than a data-gathering glitch.
I'm putting this one tentatively in the GIGO pile, pending some enquiries into DEFRA's data gathering.
Update: DEFRA does give some information about its methodology. This pdf is about sampling: the data for 2009 come from surveying 5825 households across the UK. I wonder if one of the Welsh households in 2008 held a barbecue for the whole village the week they got asked.