Systematic review and evidence- based guidance ... - ECDC - Europa

Systematic review and evidence-based guidance on perioperative antibiotic prophylaxisTECHNICAL REPORTEvaluation working group (GRADE working group) criteria [21,22]. In addition, grading of CBAs and non-CBAs wasdone using the criteria developed by the Stanford-UCSF Evidence-based Practice Centre (EPC) and the Agency forHealthcare Research and Quality (AHRQ), which published the ‘Critical Analysis of Quality Improvement Strategies’[23].The quality of evidence of the included studies was assessed using the GRADE approach, which assigns high,moderate, low, and very low quality ratings to studies. The highest quality rating is for RCTs [22,24], which can bedowngraded to moderate, low or even very low quality evidence, depending on the presence of limitations,indirectness of evidence, unexplained heterogeneity, imprecision or inconsistency of results or publication bias.Although observational studies are rated as low quality due to study design, , they can – in accordance withGRADE – be upgraded to moderate study quality if mitigating factors are present, for example the magnitude of atreatment effect or the presence of a dose-response gradient [25] 3 .Quality assessment for randomised control trials (RCT) or metaanalyses/systematicreviews of RCTsRandomised control trials are by default rated as high quality and are downgraded if they meet any of the fivecriteria listed in Table 1. Furthermore, if serious limitations are present in RCTs, the quality is decreased to lowquality [20,22] 3 .No seriouslimitationsSeriouslimitationsVery seriouslimitationsUnclearlimitationsStudies employed a randomisation method that did not allow the investigator/participant to know or influencethe intervention group before an eligible participant entered the study, performed blinding of patients andpersonnel as well as blinding of outcome detection (if method was not described, likelihood was assumed), andrevealed a low risk of attrition bias.High risk of performance and detection bias (i.e. studies performed no blinding of patients, personnel, andoutcome detection).Studies performed blinding, but demonstrated a high risk of attrition or reporting bias or were pre-terminated.Studies revealed three or more high risks of bias, i.e. studies performed no blinding of patients, personnel andoutcome detection, and demonstrated a high risk of attrition or reporting bias.No information was provided on the methods of randomisation; blinding of patients, personnel and outcomedetection; and characterisation of study participants.Table 1. Limitations in study design suggesting a high likelihood of biasThe study consisted at least of two intervention sites and two control sites.The timing of the study period for the control and intervention group was comparable (i.e. the pre- and post-intervention periodsof measurement for the control and intervention groups should be the same).The intervention and control groups were comparable for their key characteristics.There was a clearly defined point in time at which the intervention occurred and this was reported by the researchers.At least three data points before and three after the introduction of the intervention was collected.Quality assessment for CBAs, ITS and observational studiesCBA studies were evaluated according to the by EPOC 3 .a) If CBA studies fulfilled the following three key criteria, they were graded assigned moderate quality of evidence[20,22].b) ITS were evaluated by the EPOC Review Group and GRADE and were assigned moderate quality of evidence ifthey met the following two key criteria [20,22].c) If an observational cohort study accumulated 80–100% of the maximum point value outlined by GRADE, thestudy quality was considered ‘good’ and the quality of evidence was rated as ‘moderate’; if it accumulated ≥60–