Reader Comments

Post a new comment on this article

Referee Comments: Referee 1

Posted by PLOS_ONE_Group on 12 May 2008 at 22:28 GMT

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.
**********

The spread of resistant pneumococcus in the community and its clinical consequences is a complex issue in which many controversies exist. In this study, the authors present a thoughtful mathematical prediction model of the impact of the replacement of certain antibiotics currently used, but not exclusively, to treat pneumococcal infection, by ketolides (a new antibiotic class, closely related to macrolides).

Ketolides are a class of antibiotics that hasn't yet found its place in the clinical arena. Although, very effective against several multi-drug resistant pneumococcus, it is also associated with severe hepatic failure, carrying very high mortality among those cases. In fact, the rates of liver failure associated with ketolides were higher than the ones associated with grepafloxacin, a quinolone taken out of the market due to similar toxicities. Given that most antibiotics prescribed for presumed "pneumococcal infection" are actually unnecessary (because they are prescribe in the setting of viral infections or bronchitis), any effort to encourage broad ketolide use in the community must be taken VERY CAUTIOUSLY. Furthermore, a wide variety of (less toxic) antibiotics are currently available to treat pneumococcal infections (doxycicline, TMP/SMX, cephalosporins, quinolones, macrolides, penicillins, etc). These facts, among other, motivated the FDA to change the approved indication for ketolides in the U.S to mild, proven pneumonia (treatment of bronchitis, sinusitis, otitis, COPD exacerbations and others, are no longer approved). On the other hand, as the authors acknowledge (but not encourage), the most effective and efficacious intervention to decrease the spread of antibiotic resistance in the community is the cautious and appropriate use of antibiotics (and not the use of new ones).

Although mathematical models are a useful tool to predict future events, they depend entirely on the assumptions included in the model. Although the authors were thoughtful regarding making these assumptions, I am concerned that several might not entirely reflect reality.

1. The authors took into account 3 main classes of antibiotics (PCN, macrolides and ketolides), leaving out many other antibiotics used very commonly in the community (particularly quinolones and doxycicline). I would strongly encourage the authors to discuss the reasons to exclude those other antibiotics from the model (or to support the restriction of the model to the selected antibiotics).

2. The authors used data on the annual use of the 3 antibiotics included in the analysis. However, the indication for the use of that particular antibiotic was not clear (did they use the annual use of antibiotics for any indication?, or did they use the annual use of antibiotics prescribed for pneumococcal-related infections?). In any case, discussion regarding the implications of this selection would be important.

3. The authors do not take into consideration an important feature of the pneumococcal resistance to macrolides. In general, low-level resistance to macrolides is provided by an efflux pump (which accounts for the majority of cases of macrolides-resistant pneumococcus in the US). Despite these low-level resistance, most of the patients infected with these strains and treated with macrolides, achieve total clinical cure. On the other had, changes in the ribosomal structure confer higher levels of resistance and a higher likelihood of clinical failure (which accounts for the vast majority of cases in France). It would be important to discuss why the authors did not distinguish these two different patterns of resistance in their model.

4. Although the authors include into their model the (well-know) fact that most ketolide-resistant pneumococcal strains are already resistant to macrolides, it is not clear to me if they also considered the opposite, and very likely situation, in which ketolide-resistant pneumococcus will also be cross resistant to macrolides (and therefore, considered multi-drug resistant under their definitions). By inducing resistance to ketolides, it is very likely that we will loose both, ketolides and macrolides as alternative options.

5. It is not clear if the authors included the effect of children vaccination on the rates of pneumococcal colonization and/or infection among adults.

Although the mathematical model seems accurate based on the assumptions made by the authors, the results might be misleading if they are not interpreted in the context of the limitations that those same assumptions confer to the model and its predictions. Furthermore, it would be very appropriate to further discuss better alternatives to reduce the spread of resistance, specifically encouraging the rational use of antibiotics instead of the generalized use of new, expensive and potentially very toxic antibiotics.