Effectiveness of professional interventions to improve antibiotic prescription: a Network meta-analysis

Article type
Authors
Ahmadi A1, Mallah N2, Takkouche EB3
1Knowledge Utilization Research Center; Tehran University of Medical Sciences
2Department of Psychiatry, Radiology and Public Health; University of Santiago de Compostela
3Department of Preventive Medicine, University of Santiago de Compostela
Abstract
Background: There are several interventions to promote the ‘rational prescription of antibiotics’ in physicians. For the most of these interventions, there is no enough evidence for the head to head comparison.

Objectives:to compare effectiveness of professional interventions to promote antibiotic prescription in physician by using Network-Meta analysis

Methods: Search strategy: (' Anti-Infective Agents'(related terms/OR)) AND ('physicians'(related retms/OR)) AND ('prescription'(related terms/OR)).Types of study: Randomized clinical trials. Population: Physicians and Dentists. Intervention: Groups of interventions based on EPOC categorization of professional interventions. Primary outcome: indicators of change in physician' prescriptions. Study selection & data extraction: Titles and abstracts were screened against the inclusion criteria. Selected studies were appraised by the Cochrane risk of bias tool. Data extraction was done by two independent reviewers. Disagreements that arise were resolved through discussion, if deemed necessary a third reviewer refereed. Data analysis: a random-effects NMA within a Bayesian framework was used. The routine care/no intervention arm was considered as common comparator. The posterior Relative Risk (RR) and Confidence Intervals(CI) and the posterior mean ranks for each arm was calculated. Analyses were performed using WinBUGS v1.4.3.

Results:994 unique records were identified. 30 studies were eligible. Seven studies contributed in quantitative analysis. There were four distinct types of intervention components (i)Distribution of educational material (ii)Audit and feedback (iii)Reminders (iv)Provider incentives. Various combinations of these components were applied in five arms in the seven included studies: A: no intervention or routine care; B: distribution of educational material& Audit and feedback; C:distribution of educational material& Audit and feedback& Reminders&Provider incentives; D:distribution of educational material& Audit and feedback& Reminders&Provider incentives&educational meetings and E: Distribution of educational material. The relative risk to increase the change of the behavior in physicians for arms B,C,D and E in comparison with the group A were (RR=3.29;CI:2.09–7.56), (RR=1.54;CI: -0.15–4.17), (RR=1.64;CI:-0.11– 4.20) and (RR=2.05;CI:-0.03– 5.37)respectively. The highest mean score of relative rank for effectiveness was for arm B:4.88 and then E:3.51, D:2.56, C:2.52, A:1.46. The surface under the cumulative ranking curve (SUCRA) is depicted for interventions.

Conclusions:The intervention combination of Educational material& Audit and feedback shows the highest probability of being the most effective intervention to improve antibiotic prescription in physicians

Patient or healthcare consumer involvement: To Make our evidence more applicable, the research idea was discussed with infectious specialist physicians. A 'consumer‐informed' material have been produced to improve physicians and policy makers knowledge about effective interventions.