Comments to the Author There is a great need for studies that thoroughly analyse intervention efforts to reduce antibiotic resistance and the increase in healthcare associated pathogens like Clostridium difficile. The authors submit the present paper as a successor to multiple earlier excellent publications relating to this important topic and there is no question in my mind that the paper merits publication.

However, there are several concerns that need to be addressed:

Major points Although the study is appreciated as being a methodological study, focusing on the utility of a quasi-experimental design to improve and target antibiotic stewardship, I would find it appropriate if the authors would extend the amount of background information to help readers evaluate the feasibility of the interventions used:
1. Antibiotic policy. Specifics are lacking in the Materials and Methods section about the changes in antibiotic policy that were implemented to minimize the use of high-risk antibiotics (recommendations for treating major infections). The brief explanation in lines 164-8, with reference to publication 19, only clarify the nature of the audits that were undertaken. I cannot find any information in earlier publications that specifies what regimens (from the low-risk antibiotics) that were recommended e.g. for sepsis, severe pneumonia, intraabdominal infections, febrile neutropenia, meningitis, or surgical prophylaxis. The authors own work (reference 15) in Infect Control Hosp Epidemiol 2011 only mention a restriction in the use of fluoroquinolones. As an example of reporting changes in treatment recommendations which the authors may consider, I would point to the paper from the SHSC Trust that is referred to in lines 328-31 (reference 39).

1. Local antibiotic resistance and low-risk antibiotics. One reason why the antibiotic policy issue, above, seems important, is that the low-risk group consists of antibiotics which seem less adequate for treating serious infections, at least in a UK hospital with a assumedly moderate-to-high prevalence of multiresistant bacteria (e.g. MRSA). It would therefore seem adequate with a brief clarification of the resistance epidemiology at the Causeway Hospital.
2. ESBL-infections were probably in part the cause of a four-fold rise in the use of carbapenems?
3. "Penicillin" in the low-risk group does include Piperacillin/tazobactam?

4. Cephalexin is denoted low-risk (line 155) but medium-risk (line 150).

5. Cl. difficile infections. a) The impressive reduction in use of high-risk antibiotics, especially cephalosporins, is reported in detail both in the text and Table 1. I find it considerable more difficult to appreciate the magnitude of reduction in Cl. difficile infections from the pre- to the post-intervention period. In Table 2, the coefficient for the intervention is somewhat unclear for a non-statistician: would minus 0.042713 mean that a simultaneous reduction by 1 (one) DDD/100 bed-days for all high-risk antibiotic agents reduce the CDI-incidence by 0.043 after 2 months? This is hard to grasp when the average monthly incidence is said to be 0.08/100 bed-days. Possibly I have misunderstood, but I might not be the only ignorant reader, so I suggest a more thorough explanation would be appropriate. And why not (additionally) report the average annual or 6-monthly number of CDI episodes both for the pre- and post-intervention period? It is not easy to appreciate the reduction from the figures 1 - 3. For a 30-bed ward, however, the average incidence would represent about 1 CDI case every 2. month. b) There is no mention of the severity of the CD infections among the 320 cases. They may in other words represent any spectre from mainly mild diarrhoea to a considerable number of patients with fulminant pseudomembraneous colitis. Were any differences in distribution of severe cases before and after intervention noted? What about in-hospital mortality, both Cl. difficile related and due to all other types of infections? In line 333-4 (Discussion) patient clinical outcomes are referred to as subject for further studies, but I propose that the inclusion of some clinical aspect and end-points, which should not be hard to come by, would add substantially to the quality of the study. If left out, I suggest that the lack of such information may be considered as a limitation of the study.

6. Disposition. The article is overall well-written. However, I find the repetition of the aims of the study as well as - maybe - the summary of main findings in the first sections of the Discussion to be superfluous. Also, the Discussion could in my opinion be improved if some of the aforementioned clinical aspects as well as comments on the antibiotic regimens could be elaborated somewhat upon. Did Causeway have an extraordinary high level of use of e.g. cephalosporins at the outset? Is there any concern that a four-fold rise in the use of carbapenems may partly counteract the benefits of a (small?) decrease in (mild?) CDIs, when one considers a broader perspective of the problems relating to antimicrobial resistance?

Minor points 1. What is an "ecological study design"? The terms "ecological" and "design" are straightforward, but the combination seems a bit awkward since one may ask, as opposed to what: An economical design? Clinical design? The purpose of the study, to limit the untoward consequences of irrational antibiotic use, is clearly stated from the introduction. So what characterizes the ecological design, that clinical data are omitted?

1. The incidence of CDI is stated as a monthly average of 0.08/100 bed-days over 6.5 years (78 months). With 320 CDI cases overall, this calculates to a total of 400 000 bed-days for the 242 bed study hospital. With a 100% occupancy rate, 242 beds amounts to 242 x 365 x 6.5 = 574 145 bed-days during the study period. From this simple math the actual average occupancy of 69.7% (400000/574145) seems very low and would certainly be an argument why cross infections may play a minor role in propagating CDI. But is this very low average rate correct? Could there be peculiarities associated with the Causeway hospital that would explain this? A short description of the hospitals services, under the initial "setting" section, would then seem appropriate.