61 Implementation strategies of fall prevention interventions in hospitals: A systematic review 3 To answer question 2: ‘At what level are implementation strategies reported following the recommendations of Proctor et al. (2013)?’, we gathered data regarding the actor, action, action target, temporality, dose, implementation outcome affected, and justification [13]. Since the reporting recommendations of Proctor et al. were published in 2013, we analyzed the operationalization of strategies before and after 2013. To effectively address the last question: ‘Which implementation strategy or combination of strategies are most effective at promoting the use of FPIs and in reducing fall rates in hospitalized patients?’, it was necessary to identify studies with similar implementation strategies for similar FPIs from at least two controlled studies (e.g. cluster-RCT or a stepped-wedge design). However, we were not able to include similar studies, therefore a meta-analysis was not conducted. Risk of bias Risk of bias was assessed with one of the following two tools, depending on study design. For controlled studies, the Cochrane risk of bias tool was used. [21] This tool comprises nine items, each evaluated as low risk, high risk, or uncertain risk. For uncontrolled studies, the Newcastle-Ottawa quality scale (NOS) was used. [22] The NOS is divided into three parts: selection, comparison, and outcome, resulting in three sub scores and an overall score (good, fair or poor). Scores were translated from ‘good’ to ‘low risk of bias’, from ‘fair’ to ‘some concerns’, and from ‘poor’ to ‘high risk of bias’, in the visualization tool [23], additional file II. The outcome part of NOS includes a question on whether follow-up was long enough for the outcome to occur, a three-month period was deemed sufficient. Two reviewers (FH, TL or DS) independently completed the risk of bias assessment, in case of disagreement, a third reviewer (EI) was consulted. Statistical analyses Study characteristics were summarized as frequencies and/or percentages, median and interquartile ranges. Chi-square tests, and if necessary, Fisher exact tests were performed to test for differences in the reported number of prerequisites before and after 2013. Results with two-sided p-values less than 0.05 were deemed statistically significant. Analyses were performed with IBM SPSS Statistics for Windows, version 28.0. Armonk, NY: IBM Corp.
RkJQdWJsaXNoZXIy MTk4NDMw