Improving how antibiotics are prescribed by physicians working in hospital settings.

Antibiotics are used to treat infections, such as pneumonia, that are caused by bacteria. Over time, however, many bacteria have become resistant to antibiotics. Antibiotic resistance is a serious problem for individual patients and healthcare systems; in hospitals, infections caused by antibiotic-resistant bacteria are associated with higher rates of death, illness and prolonged hospital stay. Bacteria often become resistant because antibiotics are used too often and incorrectly. Studies have shown that about half of the time, physicians in hospital are not prescribing antibiotics properly. Hospital physicians may be unclear about the benefits and risks of prescribing antibiotics, including whether to prescribe an antibiotic, which antibiotic to prescribe, at what dose and for how long.

Many different methods of improving the prescribing of antibiotics in hospitals have been studied. In this review, 89 studies from 19 countries were analyzed to determine what methods work. The main comparison was between persuasive and restrictive methods. Persuasive methods advised physicians about how to prescribe or gave them feedback about how they prescribed. Restrictive methods put a limit on how they prescribed; for example, physicians had to have approval from an infection specialist in order to prescribe an antibiotic. Overall, the 89 studies showed that the methods improved prescribing. In addition, 21 studies showed that the methods decreased the number of infections in hospital. The restrictive methods appeared to have a larger effect than persuasive methods. In conclusion, this review has found a lot of evidence that methods can improve prescribing of antibiotics to patients in hospital, but we need more studies to fully assess the clinical benefits of these methods.

Authors' conclusions: 

The results show that interventions to reduce excessive antibiotic prescribing to hospital inpatients can reduce antimicrobial resistance or hospital-acquired infections, and interventions to increase effective prescribing can improve clinical outcome. This update provides more evidence about unintended clinical consequences of interventions and about the effect of interventions to reduce exposure of patients to antibiotics. The meta-analysis supports the use of restrictive interventions when the need is urgent, but suggests that persuasive and restrictive interventions are equally effective after six months.

Read the full abstract...

The first publication of this review in Issue 3, 2005 included studies up to November 2003. This update adds studies to December 2006 and focuses on application of a new method for meta-analysis of interrupted time series studies and application of new Cochrane Effective Practice and Organisation of Care (EPOC) Risk of Bias criteria to all studies in the review, including those studies in the previously published version. The aim of the review is to evaluate the impact of interventions from the perspective of antibiotic stewardship. The two objectives of antibiotic stewardship are first to ensure effective treatment for patients with bacterial infection and second support professionals and patients to reduce unnecessary use and minimize collateral damage.


To estimate the effectiveness of professional interventions that, alone or in combination, are effective in antibiotic stewardship for hospital inpatients, to evaluate the impact of these interventions on reducing the incidence of antimicrobial-resistant pathogens or Clostridium difficile infection and their impact on clinical outcome.

Search strategy: 

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE from 1980 to December 2006 and the EPOC specialized register in July 2007 and February 2009 and bibliographies of retrieved articles. The main comparison is between interventions that had a restrictive element and those that were purely persuasive. Restrictive interventions were implemented through restriction of the freedom of prescribers to select some antibiotics. Persuasive interventions used one or more of the following methods for changing professional behaviour: dissemination of educational resources, reminders, audit and feedback, or educational outreach. Restrictive interventions could contain persuasive elements.

Selection criteria: 

We included randomized clinical trials (RCTs), controlled clinical trials (CCT), controlled before-after (CBA) and interrupted time series studies (ITS). Interventions included any professional or structural interventions as defined by EPOC. The intervention had to include a component that aimed to improve antibiotic prescribing to hospital inpatients, either by increasing effective treatment or by reducing unnecessary treatment. The results had to include interpretable data about the effect of the intervention on antibiotic prescribing or microbial outcomes or relevant clinical outcomes.

Data collection and analysis: 

Two authors extracted data and assessed quality. We performed meta-regression of ITS studies to compare the results of persuasive and restrictive interventions. Persuasive interventions advised physicians about how to prescribe or gave them feedback about how they prescribed. Restrictive interventions put a limit on how they prescribed; for example, physicians had to have approval from an infection specialist in order to prescribe an antibiotic. We standardized the results of some ITS studies so that they are on the same scale (percent change in outcome), thereby facilitating comparisons of different interventions. To do this, we used the change in level and change in slope to estimate the effect size with increasing time after the intervention (one month, six months, one year, etc) as the percent change in level at each time point. We did not extrapolate beyond the end of data collection after the intervention. The meta-regression was performed using standard weighted linear regression with the standard errors of the coefficients adjusted where necessary.

Main results: 

For this update we included 89 studies that reported 95 interventions. Of the 89 studies, 56 were ITSs (of which 4 were controlled ITSs), 25 were RCT (of which 5 were cluster-RCTs), 5 were CBAs and 3 were CCTs (of which 1 was a cluster-CCT).

Most (80/95, 84%) of the interventions targeted the antibiotic prescribed (choice of antibiotic, timing of first dose and route of administration). The remaining 15 interventions aimed to change exposure of patients to antibiotics by targeting the decision to treat or the duration of treatment. Reliable data about impact on antibiotic prescribing data were available for 76 interventions (44 persuasive, 24 restrictive and 8 structural). For the persuasive interventions, the median change in antibiotic prescribing was 42.3% for the ITSs, 31.6% for the controlled ITSs, 17.7% for the CBAs, 3.5% for the cluster-RCTs and 24.7% for the RCTs. The restrictive interventions had a median effect size of 34.7% for the ITSs, 17.1% for the CBAs and 40.5% for the RCTs. The structural interventions had a median effect of 13.3% for the RCTs and 23.6% for the cluster-RCTs. Data about impact on microbial outcomes were available for 21 interventions but only 6 of these also had reliable data about impact on antibiotic prescribing.

Meta-analysis of 52 ITS studies was used to compare restrictive versus purely persuasive interventions. Restrictive interventions had significantly greater impact on prescribing outcomes at one month (32%, 95% confidence interval (CI) 2% to 61%, P = 0.03) and on microbial outcomes at 6 months (53%, 95% CI 31% to 75%, P = 0.001) but there were no significant differences at 12 or 24 months. Interventions intended to decrease excessive prescribing were associated with reduction in Clostridium difficile infections and colonization or infection with aminoglycoside- or cephalosporin-resistant gram-negative bacteria, methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus faecalis. Meta-analysis of clinical outcomes showed that four interventions intended to increase effective prescribing for pneumonia were associated with significant reduction in mortality (risk ratio 0.89, 95% CI 0.82 to 0.97), whereas nine interventions intended to decrease excessive prescribing were not associated with significant increase in mortality (risk ratio 0.92, 95% CI 0.81 to 1.06).