Practical methods for systematic review and metaanalysis

Whatever approach is taken, using data from publications, obtaining summary data directly from trialists or collecting IPD, it is vital that every effort is made to obtain all

Data from Publications

Data from Trialists (IPD and Summary Tables)

Restricted to published trials only. May be difficult to ascertain trial eligibility from text of publication.

Often difficult to extract appropriate data. Extraction may be difficult and prolonged if outcomes are many and complex. Usually independent data extraction by 2 reviewers.

Many publications do not present the statistical information required for analysis.

Analysis of binary events straightforward, but limited for time-to-event outcomes. Subgroup analysis of patient-level characteristics requires very detailed publication of individual trial subgroup analyses and is not likely to be possible.

Few individuals review manuscript.

Write protocol: State objectives, inclusion criteria and planned analyses prospectively

Identify all relevant trials

Analysis of binary events straightforward, but limited for time-to-event outcomes. Subgroup analysis of patient-level characteristics requires very detailed publication of individual trial subgroup analyses and is not likely to be possible.

Direct communication with trialists helps identify trials and verify trial eligibility.

Requires additional time and can block progress. Usually applies only to IPD.

Time taken by trialists to provide data may block progress. For IPD, need to set up database systems and re-format and re-code data from individual trials as necessary.

Extensive data checking possible only with IPD. Many of these IPD checking processes are automated and final verification of checks is by trialists. Few checks possible on tabular data.

Analysis of binary and time-to-event outcomes and patient-level subgroup characteristics straightforward with IPD. Similar analyses possible only with very detailed tabular data.

Requires additional resource.

Many individuals review manuscript.

Approaches are equivalent ^ Aims and processes are the same, but methodology and practical aspects differ by approach | Usually only applies to IPD approach

Fig. 11.3 Stages of a systematic meta-analysis - modified and reproduced with permission from [29].

the relevant data and that appropriate methodology is used at all stages of the project (Fig. 11.3). The underlying principles are the same and, irrespective of the approach adopted, the same methods should be used by each up to the point of data collection.

As with a clinical trial, a good deal of planning and organization is required before a systematic review can be started. A good deal of resource is involved in this pre-data collection planning stage, which may take several months. During this time the research question will be refined, a prospective protocol will be developed and the required collaborative and organizational structures put in place. This will be followed by a period of data collection or abstraction, data verification and finally any quantitative analyses that are appropriate. Like a trial, the final phases of a systematic review involve the dissemination of results through presentation and publication.

0 0

Post a comment