The authors need to explain where they got the incidence data, how they chose the date ranges, specify when they are using relative versus absolute incidence, and provide raw data as supporting material. I describe my concerns in more detail in the specific comments.
The incidence data are poorly explained. In the first paragraph of the Epidemiological consequences of antigenic drift section, the authors use relative incidence. In the next paragraph, are they switching to absolute incidence? This is not clear. If they are switching to absolute incidence, I don't think they ever explain in the Methods where that data comes from. In either case, what are the justifications for using absolute versus relative incidence? Also, why do they not analyze incidence prior to 1998/1999? Finally, because each influenza season spans two calendar years, why do they not break up the data by season rather than year?
In revising this section in the manuscript, we have made the source of the incidence data and its construction much more clear. In addition, we have removed the relative vs absolute incidence confusion, sticking instead with one measure: ILI x proportion of viral isolates attributable to a lineage.
The 1998/1999 season is the earliest we have that properly distinguishes B/Vic and B/Yam in the CDC data.
This would be much preferable. However, many virus isolates used in this analysis lack temporal resolution beyond the year of sampling. Because of this, we decided to stick with just year of isolation for the present analysis.