Abstract
French hospitals are urged by health authorities to develop antibiotic stewardship (ABS) programmes in order to improve antibiotic use and to decrease their consumption. We performed a longitudinal survey to describe ABS measures implementation and antibiotic use and to study relationships between ABS measures and trends in antibiotic use between 2005 and 2009.
Data on ABS, antibiotic use and activity were retrospectively collected by questionnaires sent to hospitals voluntarily participating in the south-western France network. ABS measures covered organization, resources, restrictive and persuasive actions. Antibiotic use was retrieved from pharmacy records and expressed as the number of defined daily doses/1000 patient-days according to national and WHO guidelines using 2009 defined daily dose values to monitor trends. Relationships between ABS measures and antibiotic use were studied by multivariate logistic regression.
Between 2005 and 2009, the degree of implementation of ABS increased in the 74 participating hospitals. Antibiotic use remained stable, with variations according to hospital groups and antibiotic classes. In hospitals with more ABS measures, antibiotic use in general and fluoroquinolone use tended to remain stable or to decrease. Educational activities were associated with a decrease in fluoroquinolone use in the univariate analysis. In the multivariate analysis, practice audits and time dedicated by the antibiotic advisor were significantly associated with a decrease in total antibiotic use and fluoroquinolone use, respectively.
This first longitudinal study, in 74 hospitals, showed that human resources and persuasive ABS measures, in the context of a multidisciplinary approach, are helpful in controlling total antibiotic and fluoroquinolone use.
Introduction
To tackle the high level of antibiotic use and antimicrobial resistance (AMR), after having implemented a national programme to decrease healthcare-associated infections and AMR, in 2001 France launched a national strategy aimed at improving the use of antibiotics. Several official texts and guidance focused on hospitals, disseminating requirements for antibiotic stewardship (ABS) programmes. The ultimate aim of ABS is to optimize antibiotic use, i.e. to use the right antibiotic for the right indication, at the correct dose and duration, only when necessary, in order to improve patient outcomes, ensure cost-effectiveness and lessen the risk of adverse effects, including antimicrobial resistance.1 Antibiotic use in French hospitals has long been known to be higher than in other European countries.2,3 Therefore ABS programmes aim not only at improving antibiotic use and controlling the use of broad-spectrum antibiotics, but also at decreasing consumption.
Enforcement of mandatory reporting and public disclosure of the structure and process indicator reflecting ABS programmes (‘ICATB,’ composite indicator on antibiotic prudent use) has enhanced professionals’ commitment to improve antibiotic use.4 For more than a decade, the regional coordinating centre for nosocomial infection control (CCLIN Sud-Ouest) has managed epidemiological surveillance networks on antimicrobial resistance and antibiotic use in hospitals.5,6 These networks are based on the voluntary participation of hospitals. Since 2005, methods for surveillance have not changed, and have been consistent with national guidelines and requirements. We performed a 5 year longitudinal survey at the region-wide level in order to describe ABS programme development and antibiotic usage in hospitals, and to study relationships between measures to improve antibiotic management and trends in antibiotic use over the years.
Methods
Data collection
We used information from five retrospective annual surveys conducted from 2005 to 2009 in the network of south-western French hospitals. ABS measures, antibiotic use, hospital type and activity were collected by auto-questionnaires filled in by pharmacists, microbiologists and infection control teams. ABS implementation was assessed by questions regarding organization (existence of an antibiotic committee and number of yearly meetings), human and information technology (IT) resources, restrictive actions and persuasive/educative actions. Besides measures required by health authorities for the annual reporting of the ICATB indicator, the questionnaire explored additional features on the practical implementation of measures, as previously described (the questionnaire is available as Supplementary data at JAC Online).4
Data on antibiotic use were retrieved from pharmacy records using an Excel spreadsheet, in accordance with the French national surveillance method.7 Antibiotics were classified according to WHO recommendations using the Anatomical Therapeutic Chemical classification and the defined daily doses (DDD) system, using 2009 DDD values to assess trends over the years. Antibacterials for systemic use (J01), rifampicin (J04AB02) and oral imidazoles derivatives (P01AB) were considered. The number of units of each antibiotic dispensed was collected for the whole year for the whole hospital for inpatients. Emergency departments, day-care wards and outpatients were excluded. The number of patient-days (PD) was collected for the whole hospital. Antibiotic usage was expressed as number of DDD/1000 PD.
Analysis
Descriptive analysis of ABS and antibiotic use was performed according to hospital type. Trends in antibiotic use were assessed using the Spearman test.
We performed univariate analysis to identify relationships between individual ABS measures and changes in antibiotic use between 2005 and 2009, expressed as a percentage of the use in 2005 = (no. DDD/1000 PD in 2009 − no. DDD/1000 PD in 2005)/no. DDD/1000 PD in 2005 × 100. ABS measures implemented between 2005 and 2008 were taken into account. We focused on specific measures, reflecting a high level of involvement in the French context, i.e. measures requiring resources and time and that were not in place in all hospitals: antibiotic committee holding at least three meetings per year, use of prescriptions with stop-orders, computerized systems for prescription or pharmaceutical analysis, educational activities, practice audits, time spent by the pharmacist (≥median; i.e. 2 h per week for at least 2 years) and by the antibiotic advisor (≥median; i.e. 0.5 days per week for at least 2 years). According to national guidelines, the antibiotic advisor should be a physician, well experienced in antimicrobial therapy, who should not only act as an advisor for colleagues but should also play a leading role in the ABS programme. The change in antibiotic use was compared between two groups of hospitals defined by the implementation of the studied measure the preceding year, using non-parametric tests (Wilcoxon or Kruskal–Wallis test). This comparison was made for total antibiotic use and for non-urinary fluoroquinolone (i.e. all fluoroquinolones except norfloxacin) use, third- and fourth-generation cephalosporin use, glycopeptide use and the proportion of some broad-spectrum antibiotics (fluoroquinolones, glycopeptides, carbapenems and third- and fourth-generation cephalosporins) among total antibiotic use. These antibiotic classes were chosen because of their potential impact on bacterial resistance and because the control or the decrease in their use can be achieved by ABS interventions.8–10
In addition, a logistic regression analysis was performed to identify factors associated with a decrease or stability in antibiotic use between 2005 and 2009. The dependent variable was dichotomous: change in antibiotic use ≤0% or >0%. ABS measures identified as predictors of achieving a decrease in antibiotic use at the 0.25 P-value were included in the backward multivariate logistic regression analysis. The type of hospital was kept in each model. Analyses were performed using Excel (Microsoft Excel 2003; Microsoft Corp., Redmond, WA, USA) and SAS (version 9.1; SAS, Cary, NC, USA) software.
Results
Participants in the survey
Among 260 hospitals involved in the CCLIN Sud-Ouest network in 2009, 74 had provided information on ABS and on antibiotic use each year since 2005. Their activity—in terms of percentage of PD in acute care among total PD and in terms of admissions—had not significantly changed throughout the study period. A global trend toward a steady and slight decrease in both PD and admissions was observed. These hospitals were 22 secondary/tertiary non-teaching public hospitals, 26 acute care private hospitals, 14 rehabilitation centres, 7 local hospitals and 5 psychiatric hospitals, accounting for 17% of hospital beds in the region.
Implementation of ABS programmes
Implementation of ABS programmes increased over the years of the study in all hospitals, with variations in the average number of measures in place according to the type of hospitals. In 2009 the most comprehensive ABS programmes were seen in acute care private hospitals; local hospitals had fewer measures in place than other types of hospitals. Almost all participating hospitals had implemented an antibiotic committee, a formulary including a list of restricted antibiotics, guidelines for surgical antibiotic prophylaxis and monitored AMR and antibiotic consumption. Between 2005 and 2009, major progress was achieved in the number of yearly meetings held by the antibiotic committee, in the development of IT support for prescription and pharmaceutical analysis, in the appointment of an antibiotic advisor, in the implementation of a multidisciplinary team for antibiotic management and in education (Figure 1). Despite these improvements, IT resources, prescriptions with stop-order or prior approval and educational activities remained among the least common measures in 2009. Regarding human resources, only 39% of hospitals were able to provide an estimation of the time dedicated by the antibiotic advisor to the prudent use of antibiotics.
Implementation of ABS measures between 2005 and 2009 in 74 hospitals. AB, antibiotic.
Trends in antibiotic use
Globally, antibiotic usage in number of DDD/1000 PD remained stable over the years: +2.5%, with variations according to hospital type (Figure 2). Total use of fluoroquinolones was stable, and even decreased in rehabilitation centres (−24%) and acute care private hospitals (−10%). Among fluoroquinolones, the use of levofloxacin doubled between 2005 and 2009 and the use of ciprofloxacin showed a decrease (Figure 3). Conversely, the use of third- and fourth-generation cephalosporins increased, mainly due to a major increase in ceftriaxone use (Figure 4). Glycopeptide use remained relatively stable. None of the variations was statistically significant.
Trends in total antibiotic use in no. DDD/1000 PD, 2005–09, in 74 hospitals, south-western France. hosp., hospital; rehab, rehabilitation. Δ = change between 2005 and 2009.
Trends in fluoroquinolone use in no. DDD/1000 PD, 2005–09 in 74 hospitals, south-western France.
Trends in third- and fourth-generation cephalosporin (3rd–4th GC) use in no. DDD/1000 PD, 2005–09, in 74 hospitals, south-western France.
Relationship between ABS programmes and antibiotic use
The univariate analyses identified antibiotic policy measures associated with a significant difference in antibiotic use trends between 2005 and 2009. On the whole, the lowest percentage increase in total antibiotic and non-urinary fluoroquinolone use was achieved in the hospitals that had implemented the ABS measures (Table 1). The trend was similar for the change in broad-spectrum antibiotic proportion among total use. Contrasting relationships were seen for third- and fourth-generation cephalosporins and glycopeptides. Interestingly, in hospitals having carried out education sessions between 2005 and 2009, the use of fluoroquinolones remained stable, whereas it increased in others. The growth in the use of broad-spectrum antibiotics was significantly lower in hospitals that had implemented prescriptions with stop-orders.
Trends in antibiotic use between 2005 and 2009 according to hospital type and ABS measures in place in preceding years, expressed as a percentage of change in the number of DDD/1000 PD (N = 74)
| All antibiotics (%) | Pa | Fluoroquinolonesb (%) | Pa | Third- and fourth-generation cephalosporins (%) | Pa | Broad-spectrum antibioticsc (%) | Pa | |
|---|---|---|---|---|---|---|---|---|
| Hospital type | ||||||||
| public hospitals (n = 22) | +7.9 | +7.9 | +59.5 | +9.9 | ||||
| private hospitals (n = 26) | +0.7 | +8.6 | +14.3 | +9.0 | ||||
| rehabilitation centres (n = 14) | −1.5 | −12.9 | +67.1 | +2.2 | ||||
| psychiatric hospitals (n = 5) | +70.4 | +88.3 | +63.3 | +116.4 | ||||
| local hospitals (n = 7) | −15.1 | 0.15 | −6.1 | 0.16 | −11.5 | 0.03 | +12.3 | 0.09 |
| ABS measures | ||||||||
| antibiotic committee holding ≥3 meetings/year | ||||||||
| yes (n = 55) | +3.1 | +6.4 | +41.4 | +18.9 | ||||
| no (n = 19) | +12.8 | 0.08 | +13.8 | 0.42 | +30.5 | 0.85 | +5.9 | 0.70 |
| prescription support with stop-orders | ||||||||
| yes (n = 56) | +2.5 | +2.8 | +43.9 | +8.6 | ||||
| no (n = 18) | +15.2 | 0.61 | +25.5 | 0.35 | +22.0 | 0.52 | +37.4 | 0.05 |
| IT support for prescribing or for pharmaceutical analysis | ||||||||
| yes (n = 39) | +0.7 | +0.3 | +35.9 | +6.2 | ||||
| no (n = 35) | +11.0 | 0.80 | +17.3 | 0.53 | +41.6 | 0.20 | +26.0 | 0.12 |
| educational activities | ||||||||
| yes (n = 40) | −4.0 | −3.3 | +37.9 | +10.1 | ||||
| no (n = 34) | +16.9 | 0.27 | +22.0 | 0.03 | +39.4 | 0.47 | +22.0 | 0.66 |
| practice audits | ||||||||
| yes (n = 60) | −1.9 | +0.7 | +31.9 | +9.5 | ||||
| no (n = 14) | +37.7 | 0.06 | +40.8 | 0.14 | +67.4 | 0.13 | +41.7 | 0.42 |
| time spent by the pharmacist (for at least 2 years) | ||||||||
| ≥median (n = 46) | −0.3 | +4.3 | +40.3 | +14.6 | ||||
| <median or no data (n = 28) | +15.3 | 0.73 | +15.0 | 0.60 | +35.9 | 0.28 | +17.2 | 0.22 |
| time spent by the antibiotic advisor (for at least 2 years) | ||||||||
| ≥median (n = 19) | −2.1 | −5.2 | +74.2 | +11.0 | ||||
| <median or no data (n = 55) | +8.3 | 0.76 | +13.0 | 0.42 | +26.3 | 0.06 | +17.1 | 0.50 |
| All antibiotics (%) | Pa | Fluoroquinolonesb (%) | Pa | Third- and fourth-generation cephalosporins (%) | Pa | Broad-spectrum antibioticsc (%) | Pa | |
|---|---|---|---|---|---|---|---|---|
| Hospital type | ||||||||
| public hospitals (n = 22) | +7.9 | +7.9 | +59.5 | +9.9 | ||||
| private hospitals (n = 26) | +0.7 | +8.6 | +14.3 | +9.0 | ||||
| rehabilitation centres (n = 14) | −1.5 | −12.9 | +67.1 | +2.2 | ||||
| psychiatric hospitals (n = 5) | +70.4 | +88.3 | +63.3 | +116.4 | ||||
| local hospitals (n = 7) | −15.1 | 0.15 | −6.1 | 0.16 | −11.5 | 0.03 | +12.3 | 0.09 |
| ABS measures | ||||||||
| antibiotic committee holding ≥3 meetings/year | ||||||||
| yes (n = 55) | +3.1 | +6.4 | +41.4 | +18.9 | ||||
| no (n = 19) | +12.8 | 0.08 | +13.8 | 0.42 | +30.5 | 0.85 | +5.9 | 0.70 |
| prescription support with stop-orders | ||||||||
| yes (n = 56) | +2.5 | +2.8 | +43.9 | +8.6 | ||||
| no (n = 18) | +15.2 | 0.61 | +25.5 | 0.35 | +22.0 | 0.52 | +37.4 | 0.05 |
| IT support for prescribing or for pharmaceutical analysis | ||||||||
| yes (n = 39) | +0.7 | +0.3 | +35.9 | +6.2 | ||||
| no (n = 35) | +11.0 | 0.80 | +17.3 | 0.53 | +41.6 | 0.20 | +26.0 | 0.12 |
| educational activities | ||||||||
| yes (n = 40) | −4.0 | −3.3 | +37.9 | +10.1 | ||||
| no (n = 34) | +16.9 | 0.27 | +22.0 | 0.03 | +39.4 | 0.47 | +22.0 | 0.66 |
| practice audits | ||||||||
| yes (n = 60) | −1.9 | +0.7 | +31.9 | +9.5 | ||||
| no (n = 14) | +37.7 | 0.06 | +40.8 | 0.14 | +67.4 | 0.13 | +41.7 | 0.42 |
| time spent by the pharmacist (for at least 2 years) | ||||||||
| ≥median (n = 46) | −0.3 | +4.3 | +40.3 | +14.6 | ||||
| <median or no data (n = 28) | +15.3 | 0.73 | +15.0 | 0.60 | +35.9 | 0.28 | +17.2 | 0.22 |
| time spent by the antibiotic advisor (for at least 2 years) | ||||||||
| ≥median (n = 19) | −2.1 | −5.2 | +74.2 | +11.0 | ||||
| <median or no data (n = 55) | +8.3 | 0.76 | +13.0 | 0.42 | +26.3 | 0.06 | +17.1 | 0.50 |
aNon-parametric tests (Wilcoxon or Kruskal–Wallis).
bFluoroquinolones except norfloxacin.
cBroad-spectrum antibiotics: proportion of third- and fourth-generation cephalosporins, carbapenems, fluoroquinolones and glycopeptides among total antibiotic use.
Trends in antibiotic use between 2005 and 2009 according to hospital type and ABS measures in place in preceding years, expressed as a percentage of change in the number of DDD/1000 PD (N = 74)
| All antibiotics (%) | Pa | Fluoroquinolonesb (%) | Pa | Third- and fourth-generation cephalosporins (%) | Pa | Broad-spectrum antibioticsc (%) | Pa | |
|---|---|---|---|---|---|---|---|---|
| Hospital type | ||||||||
| public hospitals (n = 22) | +7.9 | +7.9 | +59.5 | +9.9 | ||||
| private hospitals (n = 26) | +0.7 | +8.6 | +14.3 | +9.0 | ||||
| rehabilitation centres (n = 14) | −1.5 | −12.9 | +67.1 | +2.2 | ||||
| psychiatric hospitals (n = 5) | +70.4 | +88.3 | +63.3 | +116.4 | ||||
| local hospitals (n = 7) | −15.1 | 0.15 | −6.1 | 0.16 | −11.5 | 0.03 | +12.3 | 0.09 |
| ABS measures | ||||||||
| antibiotic committee holding ≥3 meetings/year | ||||||||
| yes (n = 55) | +3.1 | +6.4 | +41.4 | +18.9 | ||||
| no (n = 19) | +12.8 | 0.08 | +13.8 | 0.42 | +30.5 | 0.85 | +5.9 | 0.70 |
| prescription support with stop-orders | ||||||||
| yes (n = 56) | +2.5 | +2.8 | +43.9 | +8.6 | ||||
| no (n = 18) | +15.2 | 0.61 | +25.5 | 0.35 | +22.0 | 0.52 | +37.4 | 0.05 |
| IT support for prescribing or for pharmaceutical analysis | ||||||||
| yes (n = 39) | +0.7 | +0.3 | +35.9 | +6.2 | ||||
| no (n = 35) | +11.0 | 0.80 | +17.3 | 0.53 | +41.6 | 0.20 | +26.0 | 0.12 |
| educational activities | ||||||||
| yes (n = 40) | −4.0 | −3.3 | +37.9 | +10.1 | ||||
| no (n = 34) | +16.9 | 0.27 | +22.0 | 0.03 | +39.4 | 0.47 | +22.0 | 0.66 |
| practice audits | ||||||||
| yes (n = 60) | −1.9 | +0.7 | +31.9 | +9.5 | ||||
| no (n = 14) | +37.7 | 0.06 | +40.8 | 0.14 | +67.4 | 0.13 | +41.7 | 0.42 |
| time spent by the pharmacist (for at least 2 years) | ||||||||
| ≥median (n = 46) | −0.3 | +4.3 | +40.3 | +14.6 | ||||
| <median or no data (n = 28) | +15.3 | 0.73 | +15.0 | 0.60 | +35.9 | 0.28 | +17.2 | 0.22 |
| time spent by the antibiotic advisor (for at least 2 years) | ||||||||
| ≥median (n = 19) | −2.1 | −5.2 | +74.2 | +11.0 | ||||
| <median or no data (n = 55) | +8.3 | 0.76 | +13.0 | 0.42 | +26.3 | 0.06 | +17.1 | 0.50 |
| All antibiotics (%) | Pa | Fluoroquinolonesb (%) | Pa | Third- and fourth-generation cephalosporins (%) | Pa | Broad-spectrum antibioticsc (%) | Pa | |
|---|---|---|---|---|---|---|---|---|
| Hospital type | ||||||||
| public hospitals (n = 22) | +7.9 | +7.9 | +59.5 | +9.9 | ||||
| private hospitals (n = 26) | +0.7 | +8.6 | +14.3 | +9.0 | ||||
| rehabilitation centres (n = 14) | −1.5 | −12.9 | +67.1 | +2.2 | ||||
| psychiatric hospitals (n = 5) | +70.4 | +88.3 | +63.3 | +116.4 | ||||
| local hospitals (n = 7) | −15.1 | 0.15 | −6.1 | 0.16 | −11.5 | 0.03 | +12.3 | 0.09 |
| ABS measures | ||||||||
| antibiotic committee holding ≥3 meetings/year | ||||||||
| yes (n = 55) | +3.1 | +6.4 | +41.4 | +18.9 | ||||
| no (n = 19) | +12.8 | 0.08 | +13.8 | 0.42 | +30.5 | 0.85 | +5.9 | 0.70 |
| prescription support with stop-orders | ||||||||
| yes (n = 56) | +2.5 | +2.8 | +43.9 | +8.6 | ||||
| no (n = 18) | +15.2 | 0.61 | +25.5 | 0.35 | +22.0 | 0.52 | +37.4 | 0.05 |
| IT support for prescribing or for pharmaceutical analysis | ||||||||
| yes (n = 39) | +0.7 | +0.3 | +35.9 | +6.2 | ||||
| no (n = 35) | +11.0 | 0.80 | +17.3 | 0.53 | +41.6 | 0.20 | +26.0 | 0.12 |
| educational activities | ||||||||
| yes (n = 40) | −4.0 | −3.3 | +37.9 | +10.1 | ||||
| no (n = 34) | +16.9 | 0.27 | +22.0 | 0.03 | +39.4 | 0.47 | +22.0 | 0.66 |
| practice audits | ||||||||
| yes (n = 60) | −1.9 | +0.7 | +31.9 | +9.5 | ||||
| no (n = 14) | +37.7 | 0.06 | +40.8 | 0.14 | +67.4 | 0.13 | +41.7 | 0.42 |
| time spent by the pharmacist (for at least 2 years) | ||||||||
| ≥median (n = 46) | −0.3 | +4.3 | +40.3 | +14.6 | ||||
| <median or no data (n = 28) | +15.3 | 0.73 | +15.0 | 0.60 | +35.9 | 0.28 | +17.2 | 0.22 |
| time spent by the antibiotic advisor (for at least 2 years) | ||||||||
| ≥median (n = 19) | −2.1 | −5.2 | +74.2 | +11.0 | ||||
| <median or no data (n = 55) | +8.3 | 0.76 | +13.0 | 0.42 | +26.3 | 0.06 | +17.1 | 0.50 |
aNon-parametric tests (Wilcoxon or Kruskal–Wallis).
bFluoroquinolones except norfloxacin.
cBroad-spectrum antibiotics: proportion of third- and fourth-generation cephalosporins, carbapenems, fluoroquinolones and glycopeptides among total antibiotic use.
In the multivariate logistic regression analysis controlling for the type of hospital (Table 2), the following ABS measures remained significantly and independently associated with a decrease in antibiotic use: (i) hospitals that carried out practice audits were more likely to achieve a decrease in total antibiotic use; and (ii) hospitals with an antibiotic advisor that spent at least 0.5 day per week on prudent use of antibiotics for at least 2 years between 2005 and 2008 were more likely to achieve a decrease in non-urinary fluoroquinolone use.
ABS measures associated with a decrease in total antibiotic use and in fluoroquinolone (except norfloxacin) use between 2005 and 2009 (multivariate logistic regression controlled for hospital type)
| Decrease in total antibiotic use | Decrease in fluoroquinolone use | |||
|---|---|---|---|---|
| ABS measure | odds ratio (95% CI) | P | odds ratio (95% CI) | P |
| Practice audits | 6.1 (1.2–32.4) | 0.03 | ||
| Time devoted by the antibiotic advisor | 5.7 (1.4–22.9) | 0.01 | ||
| Decrease in total antibiotic use | Decrease in fluoroquinolone use | |||
|---|---|---|---|---|
| ABS measure | odds ratio (95% CI) | P | odds ratio (95% CI) | P |
| Practice audits | 6.1 (1.2–32.4) | 0.03 | ||
| Time devoted by the antibiotic advisor | 5.7 (1.4–22.9) | 0.01 | ||
CI, confidence interval.
ABS measures associated with a decrease in total antibiotic use and in fluoroquinolone (except norfloxacin) use between 2005 and 2009 (multivariate logistic regression controlled for hospital type)
| Decrease in total antibiotic use | Decrease in fluoroquinolone use | |||
|---|---|---|---|---|
| ABS measure | odds ratio (95% CI) | P | odds ratio (95% CI) | P |
| Practice audits | 6.1 (1.2–32.4) | 0.03 | ||
| Time devoted by the antibiotic advisor | 5.7 (1.4–22.9) | 0.01 | ||
| Decrease in total antibiotic use | Decrease in fluoroquinolone use | |||
|---|---|---|---|---|
| ABS measure | odds ratio (95% CI) | P | odds ratio (95% CI) | P |
| Practice audits | 6.1 (1.2–32.4) | 0.03 | ||
| Time devoted by the antibiotic advisor | 5.7 (1.4–22.9) | 0.01 | ||
CI, confidence interval.
In addition, holding at least three meetings per year of the antibiotic committee seemed to be a predictor of a decrease in total antibiotic use, and having performed educational activities was associated with a decrease in fluoroquinolone use, but neither of these relationships was statistically significant.
Discussion
This longitudinal multicentre survey was the first to gather data on both ABS programmes and antibiotic use in a large sample of 74 hospitals during 5 consecutive years. Overall, it showed improvements in the development of ABS programmes and a relative stability in the total use of antibiotics. This survey confirms that, even in hospitals that participated for 5 consecutive years, some measures remained difficult to implement, namely IT resources for prescribing and education. Moreover, gaps seem to arise in practical modalities of antibiotic management, such as the suboptimal use of audit results—i.e. feedback to committees was more frequent than active feedback to prescribers.
It seems encouraging that the antibiotic use in hospitals involved in our yearly surveys remained stable, whereas hospital antibiotic use in France recently increased, when expressed as the number of DDD/1000 PD.11 As described in Dutch, Swedish and US hospitals,12–14 fluoroquinolone use was stable in our survey, whereas surveillance in other countries suggested a trend towards an increase in fluoroquinolone use.15–17 Conversely, ceftriaxone use sharply increased and relationships with the spread of extended-spectrum β-lactamase-producing Enterobacteriaceae in recent years could be investigated.
Our multicentre observational survey identified an antibiotic advisor with sufficient time and practice audits as measures significantly linked to a decrease in total antibiotic and fluoroquinolone use, taking into account hospital type. Other measures such as education, prescription with stop-orders and frequent meetings of the antibiotic committee tended to be associated with a decrease or a lower increase in antibiotic use. These positive associations were identified in a group of voluntarily participating hospitals already committed to improve antibiotic use, with a core of ABS activities in place. As suggested for infection control, a ‘bundle’ approach, i.e. implementation of a combination of measures, might create an environment that supports optimal prescribing and facilitates appropriate behaviour of all healthcare professionals. It should rely on measures that have proven efficacy, such as an antibiotic advisor or antibiotic team, formulary restriction, de-escalation, audits with feedback and development of antibiotic care bundles (a group of evidence-based key actions to consider when using antibiotics).18–21 Enhancing the acceptability and sustainability of organizations for antibiotic management requires qualified human resources: this might be the reason why the weekly time spent by the antibiotic advisor was associated with a decrease in fluoroquinolone use in our survey.
Our findings are consistent with previous studies, mainly performed in one hospital, showing that active policy, with professional involvement and evaluation, was needed to improve antibiotic use,22,23 and that educative and restrictive interventions could result in a decrease in fluoroquinolone use.8–10 Other studies identified measures, such as computerized prescriptions, nominative delivery forms or antibiotic formularies, that are associated with lower antibiotic consumption in multicentre surveys.24–27 These were cross-sectional studies: the measure was considered at the same time as its expected results. In this respect, our study was more likely to identify measures that may take time to produce positive outcomes, although a causal relationship cannot be inferred due to the observational design of this survey. In the multivariate analysis, we did not identify measures associated with a decrease in glycopeptide or third- and fourth-generation cephalosporin use. In fact, control of glycopeptides may have occurred before 2005, with the concomitant decrease in methicillin-resistant Staphylococcus aureus rates. As for third- and fourth-generation cephalosporin use, studying this class as a whole may not be relevant to assess the impact of ABS.
Some limitations should be pointed out. First, this was a self-assessment exercise, based on voluntarily participating hospitals, and there might be variations in the practical implementation of measures among hospitals (e.g. activity of the antibiotic advisor, method and feedback of audits, type and content of educational activities). In the context of increasing regulation to enforce ABS programmes in all French hospitals since 2006, we were not able to perform an experimental interventional study with a control group to assess the impact of ABS on antibiotic use. A second limitation is related to the collection of data on quantitative use and not on quality: the real need for antibiotic treatment was not assessed since no data on patient diagnoses were available. We tried to limit the effect of this by surveying a cohort of hospitals with quite stable activity, meaning that patient recruitment was not altered during the study period and would not play a major role in variations of antibiotic use.28 Other determinants of antibiotic consumption could explain some of the observed differences, such as infection control policy, antimicrobial resistance and changes in treatment guidelines (e.g. higher dosage for glycopeptides). These determinants were not controlled for in our survey, nor were cultural and behavioural aspects.29 Lastly, we were not able to use methods such as time series analysis to assess the impact of ABS over the years because only five sets of annual data were available.
To conclude, this longitudinal multicentre survey in hospitals involved in prudent use of antibiotics highlighted the need for sufficient time devoted by the antibiotic advisor and the usefulness of practice audits in decreasing antibiotic use, and suggested the positive contribution of active antibiotic committees, educational activities and prescriptions with stop-orders. Our findings pave the way for updating national guidance on ABS programmes. The new French national programme for preserving antibiotic effectiveness could reinforce professionals’ and authorities’ commitment to carry out such activities in the context of a multifaceted and multidisciplinary strategy. Our CCLIN, together with infectious disease societies, could supply material for continuing education and tools for audits on agreed topics, as has already been done in some regions in France and in Europe.30,31 Health authorities should also promote further research to identify the most efficient activities to improve antibiotic use and to provide optimal methods for their assessment considering cost-effectiveness and patient safety.
Funding
This surveillance network was set up as part of the regional coordinating centre for nosocomial infection control (CCLIN Sud-Ouest) routine activities.
Transparency declarations
None to declare.
Acknowledgements
We acknowledge all healthcare professionals in hospitals involved in the survey.
