The literature on Diffusion239 distinguishes between mass media channels and interpersonal (face-to-face) channels. The former are generally considered more important for dissemination purposes, while interpersonal channels are more important for activity at the end of the continuum. The experience of CRDs is that a combination of communication channels is useful to increase the likelihood that the public will encounter the advertised notification messages. Subgroup analyses divide studies (for study-level characteristics) or participant data (for participant-level characteristics) into subgroups and make indirect comparisons between them. These analyses can be performed to investigate heterogeneity (see above) as well as to attempt to answer some questions about patient or study factors. For example, a subgroup analysis for study-level characteristics could examine whether the results of studies conducted in primary health care facilities are the same as studies conducted in a hospital. A subgroup analysis at the participant level could determine whether the effect of the intervention is the same in men as in women. If possible, the results should be displayed graphically. The most commonly used graph is the forest diagram (see Box 1.7), which illustrates the impact estimates of individual studies and the overall summary estimate. There is also a good visual summary of the review`s results so that researchers and readers can get an idea of the data. Forest plots provide a simple representation of the accuracy of individual and overall results and the variation between study results. They help identify studies with different or unusual results “at a glance” and can also indicate whether certain studies affect overall results.
Forest diagrams can be used to illustrate results for dichotomous, continuous, and temporal outcomes.152 The category of data to be extracted is an integral part of the form design. It can be a fixed numeric text such as yes/no, a “drop-down list” or free text. However, the number of free text fields should be limited as much as possible to simplify data analysis. The form should be clear and easy to use to minimize discrepancies. Completion instructions must be provided, and each field must have decision rules for encoding data to avoid ambiguity and support consistent completion. The management of the form is essential. Paper forms should only be used when electronic forms cannot be directly accessed to reduce the risk of errors in data transcription. In addition to the study design, it is often useful to assess the quality of the intervention and its implementation. In the simplest case, the quality of an intervention refers to whether it has been used appropriately. This is a fairly simple assessment, in which, for example, studies on the titration of drugs have been conducted. It is more problematic if there is no preliminary research to suggest that a procedure should be performed in a certain way,102 or if the procedure requires technical skill such as surgery or physiotherapy.103 It is important to determine to what extent these are normalized, as this has implications for how the results should be interpreted. In addition to searching electronic databases, published and unpublished research results can also be obtained using one or more of the following methods.
The extent to which theory will play a role depends in part on the type of intervention to be evaluated. For example, in a systematic review examining the effects of a single therapeutic drug on patient outcomes, theory may play only a minor role, as many aspects of the “mechanism of action” were identified in early studies examining pharmacodynamics, dose determination, etc. Alternatively, in a systematic review assessing the impact of a psychosocial or educational programme, theories on the causal chain linking the intervention to outcomes of interest will be crucial and can be presented descriptively or schematically, as shown in Figure 1.3. The RCT is generally considered to be the most appropriate study design for assessing the impact of an intervention. Indeed, when done correctly, it limits the risk of distortion. The simplest form of RCT is known as a parallel group study, in which eligible participants are randomised to two or more groups, treated according to allocation, and groups are compared in terms of outcomes of interest. Finally, we note that some of the sources included have referred to more comprehensive guides such as the Cochrane Handbook. However, we did not extract any formal information on cross-referencing between documents. To identify relevant journal articles, we first searched the Scientific Resource Centre`s Method Library (SRCML).
This is a bibliography of publications related to evidence synthesis methods held up to the third quarter of 2017 and archived as a RefWorks library. As the SRCML is no longer updated, we conducted an additional search on Medline from October 1, 2017 to December 12, 2019. Finally, we searched the Cochrane Methodology Register (CMR), a reference database of publications relevant to systematic reviews organised by the Cochrane Methods Group. The PMF was abandoned and archived on May 31, 2012. Due to the limited search and export capabilities of these archived RSLCs and CMRs, we used pragmatic research methods for these sources. The search terms used for database searches are contained in Supplementary File 2. Database search titles and abstracts and full texts of potentially relevant articles were independently checked by two authors (RBB and AW). Disagreements were resolved through discussion or, if unsuccessful, arbitration with DP. The most common recommendations for testing extraction forms are that forms should be tested on a sample of studies (18/25); and that data extractors should be trained in the use of forms (7/25). Were service providers, participants and outcome assessors blind to treatment allocation? If one of these people were not blinded, what could be the likely impact on risk of bias (for any outcome)? The results to be extracted from each individual study can be reported in different ways, and it is often necessary for a researcher to manipulate the available data in a common format.
.