Between 2002 and 2007, we observed an overall 26.5% reduction (adjusted for FLS fluctuations) of winter antibiotic prescriptions in France; the reduction was consistent across all age groups and all 22 regions, as well as across the most frequently used therapeutic classes. The initial objective of the “Keep Antibiotics Working” program to reduce antibiotic use in the community by 25% was reached in 5 years.
The impact of educational campaigns aimed at decreasing antibiotic prescriptions by preventing antibiotic misuse at the population level has rarely been evaluated. In Europe the European Surveillance of Antimicrobial Consumption (ESAC, http://www.esac.ua.ac.be
) program collects data on antimicrobial consumption in ambulatory care and hospital settings from 34 European countries' national surveillance systems and publishes a yearly report 
. In the United States, investigations on antibiotic consumption trends are based on data recorded for a population sample as provided by the National Ambulatory Medical Care Survey 
. Other published North American studies evaluated interventional campaigns but are based on data from only a few medical practices 
, Medicaid 
, or other medical care providers 
. Few other interventional studies based on national data are available: Belgium 
, Sweden 
, and Australia 
. Prescription data in North America have also been analyzed in several areas (British Columbia, Wisconsin, and Minnesota) 
In the European Union, several countries have initiated national campaigns to optimize antibiotic use 
. Our results are consistent with the recently published article from the ESAC program; this article similarly reported a reduction of >15% of antimicrobial drug use for penicillins, cephalosporins, and macrolides (similarly, no reduction was noted for quinolones) during the 2000–2004 period 
. The overall decline we observed can be compared to those observed in three other national studies for which comparable data are available: Belgium, Sweden, and Australia. In Belgium, a significant decrease (after controlling for FLS variation) was obtained after the first year but not after the second 
; a longer follow-up might show a more pronounced and long-term decrease 
. In Sweden, a country that had one of the lowest antibiotic consumption rates in Europe 
, an intervention was launched in 1995 
. This intervention did not involve a public campaign and was reported as having led to a −20% change in overall antibiotic sales between 1995 and 2004 
. The Australian investigation focused mainly on consumer awareness and physician behavior changes 
In France, comprehensive coverage of the population by the NHI and drug reimbursement for outpatient care offer a unique opportunity for in-depth analysis of drug use data, in particular use of antibiotics among outpatients for almost the entire population. To the best of our knowledge, this is the largest body of data—with nearly half a billion data entries and the longest time-series of individual and weekly data on antibiotic prescriptions—to evaluate antibiotic use in the community ever analyzed.
Our data show that the primary objective of the French national campaign was largely achieved, with a 30.1% decrease in antibiotic use in children <6 y old. This result is very encouraging, because a substantial proportion of antibiotic prescriptions for young children are unnecessary because of the viral origins of their infection 
. The most important decrease in antibiotic use for children <5 y was noted after the second campaign. There is no clear explanation for this change, but it suggests that repeated campaigns may be necessary to overcome initial resistance by parents and physicians to reduced antibiotic prescription.
In addition, children aged 6–15 y also had significantly lower antibiotic use (a −35.8% change over the study period). This change may be due to the use of rapid tests to diagnose group A streptococci tonsillitis, which was promoted by the campaign in this age group, which is at a higher risk of group A streptococci throat infections.
Antibiotic consumption changed by −24.1% among young adults (26–35 y), who were initially the biggest antibiotic consumers among adults. Demographic characteristics of this population suggest that most parents of young children fall within this age category; as a result, it is likely that young adults have been specifically affected by the campaign. Finally, prescriptions in the 21–25 y age group remained stable from 2000 to 2007 (as compared to decreases observed in older and younger age groups). Two hypotheses may explain this observation. First, 21- to 25-y-old adults are less likely to live in or interact with collective institutions (daycares, schools, etc.) or be in close contact with children than other age groups; as a result, they are less likely to be exposed to VRIs, which are typically spread by young children. Second, it may be difficult to further decrease antibiotic use in this age group, as they already represent the age group with the lowest antibiotic prescription rate before the campaign.
While crude results did not differ markedly from those adjusted for FLS fluctuations, accounting for the latter was crucial to the interpretation of the changes recorded. Indeed, in many countries, VRIs account for a high share of unnecessary antibiotic prescriptions. Our intervention model highlighted a significant weakening of the association between FLS incidence rates and antibiotic prescriptions after the second yearly campaign. This observation suggests that the fraction of antibiotics prescribed for viral illnesses has significantly decreased, which is a highly encouraging result, as one of the campaign's main objectives was to reduce VRI-associated antibiotic overuse. However, the association between FLS and antibiotic prescriptions persists, at a low level, even after the campaign.
Several limitations of this study should be noted. First, due to the quasiexperimental design (i.e., absence of a control group) and limited preintervention data, a cause–effect relationship between the campaign “Antibiotics are not automatic” and decreased antibiotic use cannot be proved. For example, the influence of the antibiotic campaigns in other geographically proximate European countries such as Belgium cannot be excluded. Some southern European countries also conducted campaigns during the 2002–2007 period (Greece: mass media campaign in 2001–2003; Spain: mass media campaign since 2006; Portugal: radio campaign in 2004–2007) and observed evolution of outpatient antibiotic use 
. Thus, effects of targeted campaigns versus the spontaneous decrease of antibiotic use should be evaluated in Europe. Very few new antibiotics have been launched in the past decade; as a result, the promotion of antibiotic prescription by the pharmaceutical industry has probably decreased. This might be a confounding factor for the observed reduction of antibiotic use.
Second, we did not have access to information about the pathology for which antibiotics had been prescribed. In France, no information system exists that provides easy access to data linking drug use to a clinical condition.
Third, we did not account for the introduction of the 7-PCV in our analysis. The 7-PCV initially received marketing authorization in France in 2002, specifically for children presenting specific risk factors. It became widely used only at the end of 2006, once it had been recommended for all children <2 y old (unpublished data). It is thus unlikely that its market introduction could explain much of the decreased antibiotic use observed over the 5-year investigation period.
Fourth, FLS surveillance data do not account for other VRIs such as infections due to respiratory syncytial virus, which generally occur several weeks before a flu outbreak. It was not feasible to account for these viral infections, as surveillance data of these infections are not available in France. As a result, it is likely that we underestimated the association between community viral infection and antibiotic prescription.
Fifth, other local initiatives were promoted in France since 2000, such as the campaign “Antibiotics Only When Necessary” promoted in a county in southeastern France (see http://www.gepie.org/
). The added value of such initiatives was not specifically investigated.
Sixth, some authors have reported that antibiotics may substantially reduce the risk of pneumonia after chest infection 
. Therefore, adverse effects of reduced antibiotic use (e.g., increase in certain severe infections) is questionable. We do not address this question here, which remains to be investigated.
Reasons for reduced antibiotic prescriptions, e.g., fewer consultations or improved prescribing, were not evaluated here, the data did not provide the necessary information to address this question. It has been reported that mass media campaigns play a role in influencing antimicrobial prescription practice in the UK 
. Due to the multifaceted approach and targeting of the general public and physicians in parallel, the individual effect of each approach could not be evaluated. We believe that the success of the intervention was in fact a result of the combined approach, e.g., face-to-face peer education and widespread public campaigns, allowing both the practitioner not to prescribe and the patient not to ask for antibiotic therapy.
The results of the present report are highly promising in terms of bacterial resistance control. A recent European study confirms the ecological relationship between antibiotic consumption and rate of MRP at the national level 
. This study strongly underlines the responsibility of countries with higher levels of antibiotic use and recommends that they urgently undertake campaigns devoted to the control of MRP, including the promotion of prudent antibiotic use.
In France from 2001 to 2006, a decreasing trend was observed in the rate of pneumococci resistant to penicillin (47% to 32% of isolates) and the rate of pneumococci resistant to macrolides (49% to 36%) in France (see http://www.rivm.nl/earss
). Because our campaign did not target any specific therapeutic class, it may have prevented interclass switching. Nevertheless, a slight increase of quinolone prescriptions occurred. Although this was a moderate increase compared to large decreases recorded for all other classes, this trends points to the need for careful monitoring of quinolone-resistant bacteria in the community 
Despite the sharp reduction of antibiotic prescriptions observed, France remains a high user of antibiotics 
. Nevertheless, the impact of the decrease in antibiotic use on the prevalence of infections caused by antimicrobial-susceptible and antimicrobial-resistant strains must be investigated. Future studies should combine the assessment of 7-PCV vaccination and antibiotic-reduction policies, and evaluate their respective role in the evolution of S. pneumoniae
invasive infections, according to strain susceptibility.