The systematic review protocol was developed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) guidelines (57) (Additional file 1). In order to reduce the variability and enhance the rigor and validity of results, the review protocol is registered in the PROSPERO International Prospective Register of Systematic Reviews (CRD42020172773) (56) (Additional file 2).
Literature search
The search will follow a comprehensive, sequential three step search strategy, (58)
The first step will be an initial limited search that will include two most relevant online databases (MEDLINE/PubMed and CINAHL), we will identify the key words in the title and abstract of retrieved papers, and the index terms used to describe the articles.
In the second search, we will use all identified keywords and index terms to search across different databases. All search terms used for the search will be added in the appendix. Various boolean operators (AND/OR/NOT), truncations, wildcards etc. either individually or in combination will be used to ensure the comprehensiveness of the search process.
The electronic databasethat will be searched are,
- MEDLINE/PubMed
- ScienceDirect
- Cochrane database
- Embase
Thirdly, the reference list of identified, relevant articles will be searched for additional studies. In case of missing information/incomplete data, the authors of primary studies will be contacted, if required. A review of Google`
Scholar will also be conducted to include all relevant articles.
The quality of this search strategy will be assessed by at least two independent researchers using the Peer Review of Electronic Search Strategies Evidence-Based Checklist (PRESS EBC) (59). Checklist will be attached in the appendix for reference.
Study designs, particularly, randomized and quasi-randomized controlled trials (RCTs/CCTs), controlled before and after studies (CBAs) and interrupted time series (ITSs), cohort studies will be included. Previous systematic reviews will be analyzed for cross-referencing.
Citation Manager: RefWorks
Study selection
All titles and abstracts of potentially relevant studies in English, reporting impact/efficacy/effectiveness and/or economic aspect of ASP in a hospital setting in Middle Eastern countries will be vetted by one author (BT). Abstracts will be excluded if they do not fulfil the inclusion criteria. After the screening, all selected research articles will be subjected to full-text review, and independently vetted by 2 reviewers (BT and PAR) and quality assessment will be performed. Any inconsistencies and disagreements will be resolved by consensus or a third reviewer (WK).
Inclusion criteria
In this review, we have adopted the IDSA definition of ASP (described previously). Studies will be included if any (i) antimicrobial stewardship interventions are (detailed in search terms) applied; (ii) to improve antimicrobial prescribing and/or utilization; and (iii) within an inpatient hospital setting.
In terms of evaluating the economic outcomes of an ASP program, we will use the strategy previously identified by Dik et al (60) that has been widely used in previous literature.
Exclusion criteria includes animal studies or antibiotics used for agricultural needs, data from non-hospital settings such as outpatient/ambulatory care settings or long-term residential care. All non-peer reviewed outputs, narrative reviews, conference proceedings, commentaries, editorials, or qualitative studies will be excluded. Considering the recent developments in terms of implementing stewardship programs as well as within health economics, studies published before 2000 will be excluded.
Data extraction
Data abstraction will be performed by two independent reviewers (BT & PAR) using a standardized template which will include three main categories, (i) general characteristics of the reviewed studies such as, country, year of publication, study objective(s), study design, type of hospital, number of beds, number of patients. (ii) Identification of different ASP interventions or tools, based on the categories reported in Dik et al., and finally (iii) all variables linked to the outcome measures listed in the PICO (table 2) will be extracted.
As mentioned earlier, the interventions of an ASP will be classified into three main categories (a) persuasive (b) restrictive and (c) structural, which includes interventions such as, (i) education and training; (ii) pre-authorization, prospective audit, feedback, review, consultation and recommendation; (iii) drug control or monitoring, formulary restriction and intravenous-to-oral conversion; (iv) computerized system use; (v) intervention by infectious disease experts; (vi) antibiotic rotation; (vii) institutional guideline; (viii) national guideline; and (ix) infection control and hand hygiene program.
In terms of outcome measures,
Outcome 1 (Antimicrobials): antimicrobial consumption, before and after the ASP interventions will be recorded, consumptions will be measured in daily dose per 1000 patient days (DDD/1000PD). The efficacy will be determined and expressed in percentage in change of consumptions. Other outcome measures will be de-escalation or shift to targeted therapy, treatment duration, intravenous to oral switch, dose adjustment, and therapeutic drug monitoring (TDM)
Outcome 2 (Clinical): effect of ASP on a series of clinical outcomes includes, measure of resistance, infection related mortality, length of stay, compliance to prescribing of antibiotics with local or national guidelines, incidence of antimicrobial resistance, infection related mortality, compliance to the ASP guidelines, rate of healthcare-associated C. difficile infections, hospital readmission due to relapse of infection within 30 d after discharge, incidence of multidrug-resistant organisms (MDRO)
Outcome 3 (Economic): the data extraction of economic outcome will include reports on cost of implementation of ASP, operational costs, direct cost related to antimicrobial consumptions, morbidity, and/or mortality costs, length of stay cost, or any overall cost savings reported.
Risk of bias assessment and Quality assessment
Pairs of reviewers will independently assess quality of each of the studies (BT & PAR). The Cochrane Risk of Bias tool (61) will be used for randomized controlled trials and for nonrandomized studies we plan to use the Newcastle–Ottawa Quality Assessment Scale (NOS) (62). NOS is a three-dimensional appraisal tool to assess the methodological quality that includes, (i) selected population, (ii) comparability of groups, and (iii) outcome/exposure of interest. The studies will be scored on a scale ranging 0 – 9, wherein, a score >= 6 indicates high-quality studies, a score between 3- 6 as moderate and a score less than 3 will be considered as low quality. In terms of assessing the level of quality of the papers including economic impact, a separate tool, i.e. Consensus on Health Economics Criteria [CHEC] list (12) quality criteria will be used. The research team has also proposed the use of ICROMS (Integrated quality Criteria for the Review Of Multiple Study designs), an integrated quality assessment checklist that can be used for a large range of study designs including qualitative research (63).
Data analysis/synthesis
A narrative synthesis method (64) is deemed most appropriate, where all included studies will be summarized in a narrative form. Tables and graphs will be created to illustrate the key study characteristics such as (population characteristics, study outcomes, sample sizes, settings, methods used, results, and any other important aspect related to each research question of interest). The number of studies and proportions reporting an improvement will be computed. Economic analysis will be conducted based on the following criteria:
Antimicrobial Cost Analysis (ACA) – will include all studies that explicitly reported direct cost of antimicrobial consumption.
Cost-Benefit Analyses (CBA) – will include studies that considered the effects of two different methods and the effects were expressed in terms of cost.
Cost-Effectiveness Analyses (CEA) – will include studies that evaluated relative cost and effects of two different methods.
Cost-Consequence Analyses (CCA) – will include studies that reported different costs and effects of two methods.
Cost-Analyses (CA) – studies that looked at only the cost but not effects of the methods.
Cost-Minimization Analyses (CMA) – studies that looked at cost differences assuming same effects between the two different methods.
If meta-analysis of outcome data is deemed appropriate, random or fixed effect methods will be considered depending on heterogeneity among the studies. Statistical heterogeneity between studies will be assessed using tau squared and I2 statistics and P <0.10 and I2>than 50%, will be considered as a high level of statistical heterogeneity between the studies. Only, if the studies are homogenous, data will be pooled for a meta-analysis, if otherwise, it will be presented in standardized mean difference (SMD) (65).
Quality of evidence
Two authors will independently assess the quality of evidence for each outcome. The quality of evidence will be determined by Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system. The studies will be rated based on their evidence ranging high, moderate, low, or very low level. Individual outcomes will be assessed for the following aspects, limitations, inconsistency, indirectness, imprecision, and publication bias.