Cochrane Training resource: common errors - summary versions of a review
Cochrane Interactive Learning: module 8 - reporting the review
|Standard||Rationale and elaboration||Resources|
|R3||Writing the Abstract||Mandatory|
|Prepare a structured Abstract to provide a succinct summary of the review. In the interests of brevity it is highly desirable for authors to provide an Abstract of less than 700 words, and it should be no more than 1000 words in length.||Abstracts are a prominent, publicly accessible summary of the review that need to stand alone. They should convey key information about the review question and its findings, and be informative to readers.|
|Summarize the rationale and context of the review.||See Handbook Section III.3.1|
|State the main objective(s), preferably in a single concise sentence.||The objective(s) should be expressed in terms that relate to the population(s), intervention comparison(s) and, where appropriate, outcomes of interest.||See Handbook Section III.3.1|
|R6||Abstract, Search Methods||Mandatory|
|Provide the date of the last search from which records were evaluated and that any studies identified were incorporated into the review, and an indication of the databases and other sources searched.||
Abstracts should aim to give readers brief, but key, information about the comprehensiveness of the search and the currency of the information summarized by the review.
The Abstract must include the month and year of the set of searches up to which the conclusions of the review are valid. This date should reflect the date of the most recent set of searches from which all records have been screened for relevance and any studies meeting the eligibility criteria have been fully incorporated into the review (studies may be awaiting classification if, for example, the review authors are awaiting translation or clarification from authors or sponsors).
Abstracts do not need to report on recent repeat or ‘catch-up’ searches whose results have not been fully incorporated into the review. However, discretion should be applied if such searches identify a large body of evidence, the absence of which may affect the reliability of the conclusions.
The amount of information regarding the search should be indicative of the process rather than provide specific details. In the interests of brevity certain details regarding the overall process may need to be moved to the full text of the review.
Example: “CENTRAL, MEDLINE, Embase, five other databases and three trials registers were searched on [date] together with reference checking, citation searching and contact with study authors to identify additional studies”.
|See Handbook Section III.3.1|
|R7||Abstract, Selection criteria||Mandatory|
|Summarize eligibility criteria of the review, including information on study design, population and comparison.||Any extensions to eligibility criteria to address adverse effects, economic issues or qualitative research should be mentioned.||See Handbook Section III.3.1|
|R8||Abstract, Data collection and analysis||Mandatory|
Summarize any noteworthy methods for selecting studies, collecting data, evaluating risk of bias and synthesizing findings. For many reviews it may be sufficient to state “We used standard methodological procedures expected by Cochrane.”
This section of the Abstract should indicate the rigour of the methods that underpin the results reported subsequently in the Abstract. It does not need to replicate the detailed description of the methods given in the main text of the review.
Details of how many people were involved in the screening process and collection of information about any included studies are not necessary in the Abstract. Key statistical methods may be given if not clear from the results that follow.
The Abstract should prioritize the disclosure of non-standard approaches. For example, rather than disclosing all domains applied in the assessment of bias, notable variations on the standard approach should be given, such as use of non-standard tools.
|See Handbook Section III.3.1|
|R9||Abstract, Main results: number of studies and participants||Mandatory|
|Report the number of included studies and participants.||The total number of included studies should be stated. It might be appropriate to provide numbers of studies and participants for specific comparisons and main outcomes if the amount of evidence differs substantially from the total. Numbers of participants analysed should generally be presented in preference to numbers recruited (e.g. randomized); it is important to be clear which numbers are being reported. For some types of data there may be preferable alternatives to the number of participants (e.g. person-years of follow-up, number of limbs).||See Handbook Section III.3.1|
|R10||Abstract, Main results: study characteristics||Highly desirable|
|Provide a brief description of key characteristics that will determine the applicability of the body of evidence (e.g. age, severity of condition, setting, study duration).||
Summarizing the study characteristics will provide readers of the Abstract with important information about the applicability of the included studies. This is particularly important if the included studies reflect a subgroup of those eligible for inclusion in the review, for example, if the review intended to address the effects of interventions across all age groups, but included studies that only recruited adolescents.
|See Handbook Section III.3.1|
|R11||Abstract, Main results: bias assessment||Mandatory|
|Provide a comment on the findings of the bias assessment.||The ‘Risk of bias’ assessments are a key finding and form a fundamental part of the strength of the conclusions drawn in the review. If risks of bias differ substantially for different comparisons and outcomes, this should be mentioned.||See Handbook Section III.3.1|
|R12||Abstract, Main results: findings||Mandatory|
|Report findings for all important outcomes, irrespective of the strength and direction of the result, and of the availability of data.||
Findings should typically include concise information about the size of effect and certainty of evidence for the outcome (such as risk of bias, consistency of effect, imprecision, indirectness and publication bias), for example using GRADE.
Outcomes reported in the Abstract should not be selected solely on the basis of the findings. In general, the same outcomes in the Abstract should be presented in the Plain language summary and ‘Summary of findings’ tables. If no studies measured the outcome, then a comment should be made to that effect.
See Handbook Section III.3.1
|R13||Abstract, Main results: adverse effects||Mandatory|
|Ensure that any findings related to adverse effects are reported. If adverse effects data were sought, but availability of data was limited, this should be reported.||
The Abstract of the review should aim to reflect a balanced summary of the benefits and harms of the intervention.
See Handbook Section III.3.1
|R14||Abstract, Main results: format of numerical results||Mandatory|
|Present summaries of statistical analyses in the same way as they are reported in the review and in a standard way, ensuring that readers will understand the direction of benefit and the measurement scale used, and that confidence intervals are included where appropriate.||The standard format for reporting the results of statistical analysis includes an indication of the summary measure, point estimate and confidence interval, e.g. odds ratio 0.75 (95% confidence interval 0.62 to 0.89).|
|R15||Abstract, Main results: interpretability of findings||Highly desirable|
Ensure that key findings are interpretable, or are re-expressed in an interpretable way. For instance, they might be re-expressed in absolute terms (e.g. assumed and corresponding risks, NNTBs, group means), and outcomes combined with a standardized scale (e.g. standardized mean difference) might be re-expressed in units that are more naturally understood.
|Absolute effects provide a useful illustration of the likely impact of intervention, and are usually easier to understand than relative effects. Units expressed on a standardized scale reflect the effect estimate as the number of standard deviations. This is not intuitive to many readers who may be more familiar with specific scales. Any re-expressed findings must have been presented in the same way in the main text of the review (see previous standard).|
|R16||Abstract, Authors’ conclusions||Mandatory|
|State key conclusions drawn.||
Authors’ conclusions may include both implications for practice and implications for research. Care must be taken to avoid interpreting lack of evidence of effect as evidence of lack of effect.
Recommendations for practice should be avoided.
|R17||Completeness of main review text||Mandatory|
|Ensure that all findings reported in the Abstract and Plain language summary, including re-expressions of meta-analysis results, also appear in the main text of the review.||
Cochrane Training resource: Common errors - inconsistency & inaccuracy
|R18||Consistency of summary versions of the review||Mandatory|
|Ensure that reporting of objectives, important outcomes, results, caveats and conclusions is consistent across the main text, the Abstract, the Plain language summary and the ‘Summary of findings’ table (if included).||Summary versions of the review should be written on the assumption that they are likely to be read in isolation from the rest of the review.||Cochrane Training resource: Common errors - inconsistency & inaccuracy|