A narrative summary is commonly used where meta-analysis is not possible. A narrative summary describes the included studies and provides conclusions about the evidence. With a narrative summary, readers may not be able to discern how evidence was weighted and whether conclusions are biased. It is therefore important that when summarizing findings in narrative form, there is a clear structure to the summary, with an emphasis on reporting the characteristics of included studies along with data extracted relevant to the review outcomes. Narrative summary in systematic reviews should be rigorous and clear, and can utilize tables, graphs, and other diagrams to help convey how studies compare to each other and to assist in presentation of the data. A narrative summary should include the presentation of the quantitative results reported in individual studies; where available, the point estimates (one value that represent or best estimate of effects) and the interval estimates (usually presented as 95% confidence intervals) for the effects should be provided. Due to the flexibility of narrative summary in terms of the amount of data that can be conveyed textually, a structure that applies to each sequence or reporting of results from each study should be discussed beforehand and applied by the systematic review authors. This will ensure that there is consistency across the results section of a review. If a structure is not followed there may be substantial variability in reporting of results causing the data to appear incomplete or unreliable. Therefore adherence to this structure is critical; if studies do not provide the relevant information to comply with the structure this should be made clear in the summary. Bear in mind that there is no prescriptive guidance on presenting a narrative summary and it is recommended that the context of the review be taken into consideration (Lockwood & White, 2012).
Tables where relevant should be included to aid in the presentation of the data. Mostly, these tables include individual studies with their raw data; for example, percentages, distribution of prevalence and incidence estimates and confidence intervals. The tables should also include other elements such as participant characteristics.
The various graphs that may be useful in presenting include but are not limited to forest plot (for meta-analysis), funnel plot (for publication bias), L’abbe plot (explores heterogeneity and is applicable for meta-analysis of studies with binary outcomes), Galbraith plot (assesses the extent of heterogeneity between studies in a meta-analysis), and cumulative plot (incidence and prevalence estimates). However, it should be noted that the interpretation of the graphs is quite subjective and therefore should be interpreted with caution.