Referee 1's Review:

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication, the manuscript has been revised in light of these comments and to address other editorial requirements.
*********

This is a nice contribution to the literature. A discussion of mortality differentials by a proxy of SES (education) is needed, and the additional decomposition by cause of death is interesting and justified. There are two points worth clarifying:

1) It is unclear how death rates were generated when deaths were removed from the numerator, but the population at risk was left unaltered. I understand how this was done for entire States, but for the remaining States this made no sense. It would have been better to distribute the unknowns in the numerator to existing education/race/sex categories based on their observed percent distribution by year. As now calculated, I would question the validity of the death rates.

2) The transition from the 9th to 10th ICD certainly created issues with cause-specific death rates, but the NCHS publishes comparability ratios. If these were used, it should be mentioned. If not used, why not? I know this is mentioned in the list of weaknesses, but you don't explain HOW this transition might have influenced the rates.