Study designs of evaluations included in the review
Experimental and observational studies were eligible. Only studies considered to be at low or moderate risk of bias were included in the review (see 'On What Criteria was the Validity of Primary Studies..?' field). Conceptual/background information studies, epidemiological and non-AMR-specific studies were excluded.
Specific interventions included in the review
Measures designed specifically to contain the emergence of AMR were eligible. The following types of measures were included:
restrictions on the use of antibiotics (prohibition of certain drugs, formulary restrictions, stop-orders placed on inappropriately used antimicrobials);
prescriber education, feedback, and the use of guidelines (patients and clinical education programmes, prescriber feedback and guidelines in reducing prescribing, education seminars on the quality of patient management and rational drug use);
antibiotic combination therapies (rifampicin combined with novobiocin, coating central venous catheters with monocycline and rifampicin, various treatment regimes for acute otitis media, antimicrobial prophylaxis in neutropenic patients); and
vaccinations (pneumococcal).
Most studies were undertaken in the developed world, mainly the USA, and were based in hospital with few community-level interventions.
Participants included in the review
The inclusion criteria were not defined in terms of the participants. The actual participants included hospital-based clinicians; patients with urinary tract infections; patients and clinicians in primary care settings; general practitioners; health centres in the developing world; hospitalised patients; patients with acute otitis media; neutropenic patients; and adults at various risk of pneumococcal bacteraemia.
Outcomes assessed in the review
Studies that assessed effectiveness or cost-effectiveness were eligible.
How were decisions on the relevance of primary studies made?
All those involved in the review independently assessed the first ten studies. Subsequently, a random sample of an additional ten studies was subjected to an inter-rater reliability check where one reviewer rated one study differently from the other reviewers (moderate as opposed to high risk).