Referee 2'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.
**********

Review of the original submission:
The article by Opatowski et al. provides an interesting and valuable assessment on the hypothetical role of ketolides, a relatively new antibiotic class, on the dissemination of antibiotic-resistant pneumococcal strains. This is a very well written manuscript that reports on a simulation study based on data from France and other countries. The methodology addresses the most important determinants of antibiotic resistance and the authors concluded that the widespread use of ketolides could slow the diffusion of antibiotic-resistant pneumococcal strains.

Currently, ketolides are not recommended for use in children in the US and the safety of these drugs has been questioned after reports of rare but serious adverse events. The information provided by this article would be valuable in the considerations for the judicious use of these novel drugs and on the ongoing discussions of potential benefits-adverse events related to the use of Ketolides.

Some specific comments on the manuscript

Abstract
I would suggest rewording the presentation of the study aims, using terms such as "examine" or "assess the potential..." might be better suited for simulation studies.
Introduction
Reference 8 should be accompanied with the article that specifically examined the effect of PCV7 on antibiotic resistance (Kyaw et al. NEJM).
As in the abstract I would suggest caution in the wording of the study aims.
Methods
One initial short paragraph summarizing the study methods could be helpful. It should include a definition of the population taken as reference and from which study parameters where obtained (i.e. France) and state clearly the study outcome, colonization with antibiotic-resistant S. pneumoniae.
Please clarify that parameters in the first paragraph are described in detail in a specific parameter section (alternatively, reorganize these sections in the manuscript)
In table 1, infectious contact rates between groups are constant and unidirectional for all age groups. The respective text in the manuscript "contact rates were fixed at equilibrium values from the best fitting model of carriage-age distribution" is unclear. Please elaborate on how rates were calculated and provide a reference.
Given the apparent importance of the booster dose in nasopharyngeal colonization, it is necessary to clarify that "vaccinated" children in the models refer to those fully immunized (3+1 doses).

I am not sure there is a precise estimate for the duration of pcv7-derived immunity. Providing the rationale for the assumption of "13 years on average" would be necessary. It is unclear whether this immunity is assumed to be fixed during those 13 years or to decrease over time, this should be discussed.

In the paragraph describing the antibiotic-exposure parameters, the time frame for the data obtained should be presented. During the last years several initiatives to promote the judicious use of antibiotics have been started and prescription patterns have likely been changing over time. For instance, several US studies have described a reduction in overall antibiotic prescription rates during the last years, whereas prescriptions of broad spectrum antibiotics, including macrolides, have been increasing. Concurrent increases in macrolide resistant pneumococcal infections have been reported. Clarify which age groups were included in the ketolides exposure assessment.

The description of the sensitivity analysis is unclear, one or two sentences clarifying the applied approach would be helpful. Are the "outcomes", the parameters that the authors used? Within what ranges were these parameters allowed to vary?

Results
The model simulations for macrolide and penicillin resistance were well reproduced using data from 1992-1997, before introduction of PCV7. Were data available enough to perform a similar assessment using most recent data (i.e. 2003-2006) including PCV7?
The scenarios described in the simulations use fixed percentages of antibiotic prescriptions replaced by ketolides. In real life, these should represent different looks of one continuous process - ketolide introduction. In other words, ketolide prescriptions are not expected to remain constant throughout the study years but to increase progressively from 0% to a certain % following a specific rate. Have the authors evaluated these alternative scenarios, i.e. ketolide prescriptions increasing a 2% per year, etc...

Discussion
Aside information from modeling studies, population-based studies have reported similar changes in disease rates caused by macrolide resistant strains (add references for Kyaw et al. NEJM and Stephens et al. The Lancet).

The sensitivity analysis suggested that when the mean time until treatment acts decreased, resistant would increase. Please comment on this finding.
Children account for a large proportion of antibiotic use and several results of this article were observed when children were allowed in the study population. However, ketek, for instance is approved for use in adults but not in children in the US, and safety concerns related to the use of the drug remain (see Ross and Soreth et al NEJM). Further discussion on this regard would be informative for the readers.

Review of the first revised manuscript:
The authors have appropriately addressed the comments/questions raised in the former review. I feel this revised version is substantially improved and I have no additional remarks regarding the article.