This is a clearly written and well described retrospective cohort analysis. The methodology is clearly described. The Database used could be felt to represent practice at the higher quality end of UK practices.

The paper states that database has been demonstrated to be representative of the national population. It would be useful if this sentence were expanded to explain what this means and whether this covers patient age, ethnicity, and practice issues such as size, geography (i.e. representative of all areas of UK) etc.

The results and analyses are clearly described. The conclusions are well described and fit the findings. The strengths and limitations are fairly stated.

I found a couple of "typo's" (a) In table one under gender the number of males is missing in my downloaded copy (b) ref 24 contains additional info about author competing interests

This authors have modified the paper in line with the reviewers comments. This has improved the understanding and readability of the paper