1 / 47

Conducting systematic reviews for development of clinical guidelines 8 August 2013

Conducting systematic reviews for development of clinical guidelines 8 August 2013. Professor Mike Clarke m.clarke@QUB.ac.uk. What is a systematic review? What do you think it is?. Systematic reviewing. Formulating a clear question for the review Stating objectives and eligibility criteria

halona
Télécharger la présentation

Conducting systematic reviews for development of clinical guidelines 8 August 2013

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Conducting systematic reviews for development of clinical guidelines8 August 2013 Professor Mike Clarke m.clarke@QUB.ac.uk

  2. What is a systematic review? What do you think it is?

  3. Systematic reviewing • Formulating a clear question for the review • Stating objectives and eligibility criteria • Identifying (all) potentially eligible studies • Applying eligibility criteria • Assembling most complete dataset feasible • Analysing the dataset, using statistical synthesis and sensitivity analyses, if appropriate and possible • Preparing a structured report • Updating the review

  4. Formulating the question for a review

  5. Choosing the studies to include in a review depends on the question to be answered

  6. Do green sweets make people happy?

  7. Measuring the happiness of everyone who eats green sweets

  8. Getting some of you to eat a green sweet and then measuring your happiness

  9. Measuring the happiness of a random sample of those who ate a green sweet, and those who did not

  10. Getting some of you to eat a green sweet and some of you not to, and then measuring your happiness

  11. Do green sweets make people happy?

  12. Do green sweets make people happy? compared to what?

  13. Formulating the question for a review

  14. Question formulation You need a clear question. It might change during the planning of the review, but you hope it won’t change during its conduct. Each word in the question is important. Do the words narrow or broaden your review? Will other people think each word or phrase means the same thing as you think it means? Each word in the question can be expanded in the eligibility criteria for the review.

  15. Setting the eligibility criteria

  16. ParticipantsInterventionsOutcome measuresStudy designs

  17. Eligibility criteria The eligibility criteria provide the rules for what should and should not be included in your review. Think about the types of study that might have been done and decide if you would want them in your review. The results of a study must not influence your decision to include it. The eligibility criteria help you avoid having to make decisions about an unexpected study, after you know its results. The eligibility criteria don’t always have to match your question perfectly, you might be able to borrow from other areas.

  18. Searching

  19. What to consider when searching What terms to search for? What types of database to search? What countries? What languages? What time period?

  20. Planning the search terms Divide the review into the eligibility components Participants Interventions (including comparator) Outcomes Study designs The database you are searching will try to help you by only showing the things that you ask for. It might hide millions of other things.

  21. Choosing the search terms In listing terms for each component, consider: Synonyms (from different times and places) Other words and phrases that are related to what you’re interested in Words that are broader Words that are more focused Index terms or keywords

  22. Combining search terms Two main ways to link terms AND (decreases the number of hits, requires every item to be present) OR (increases the number of hits, requires any of the items to be present) Try to avoid NOT (it might remove records that actually are eligible)

  23. Combining the components Within the component, use OR to combine the terms and then use AND to combine the components. But ... Do you need all the components? Are you confident that you have all the terms within each component? Which component is least likely to be relevant? Which component is most likely to be relevant? The ideal might be the component that is most likely to be relevant and has the highest proportion of good things amongst its hits.

  24. Running the search How many titles and abstract can you check? How easy will it be to decide to accept or reject a record? Do something to “remember” the records you reject, unless you want to look at them again Record the reason for rejection for “Excluded studies”, which might be those: For which you obtain full articles Which others might think should be in your review

  25. Deciding where to search Choose databases that are likely to provide a worthwhile yield The components to focus on might vary between databases Index terms may be different in different databases

  26. Statistics of reviews:meta-analysis

  27. Intervention 1 minute Count the number of flips or rolls and the number of deaths Heads is a death 1 is a death

  28. Control 1 minute Count the number of flips or rolls and the number of deaths Tails is a death 6 is a death

  29. Heterogeneity

  30. Heterogeneity – planning for it Do you want heterogeneity in your review? Heterogeneity can occur throughout the review (interventions, participants, outcome measures and study designs) In your review, might heterogeneity lead to Variation Diversity Complications Messiness Wide coverage Ability to make comparisons Contradictory findings? Which of these would you like to have in your review?

  31. Heterogeneity – measuring and explaining it How sure are you about the factors that might drive heterogeneity in the results of the studies? The statistical tests measure heterogeneity in the results of the study, not in their content or conduct. Might statistical heterogeneity be due to more than one factor or due to a different factor than the one you are thinking of?

  32. Dealing with heterogeneity • Eligibility • Descriptive • Subgroup analyses • Sensitivity analysis • Separate meta-analyses

  33. Subgroups

  34. Subgroup analyses - planning Keep the number of subgroup analyses as low as possible, to reduce the possibility of false positive results Make sure that the ones you do are important Make sure you extract the relevant data Will you have enough power in a specific subgroup? Is it too uncommon or unlikely to have been studied? Will subgroups be reported by the original researchers, and in the way you need them? What will you do if they are not reported in the way you want? Do you have independent evidence for your predictions on what the subgroups will show (rather than evidence from the studies that would be included in the analysis)?

  35. Subgroup analyses - interpretation The most statistically powerful estimate for an individual patient is likely to come from the full meta-analysis. But, this will not be meaningful if there is too much clinical heterogeneity in the meta-analysis. Will you do subgroup analyses to show that the results are similar for different people, or to identify people, settings or interventions where the results are different? Subgroup results might provide a more specific answer for an individual patient, but they suffer from being Underpowered More subject to the effects of chance

  36. Extracting the data

  37. What is the effect of needle length on local reaction to vaccination in babies?

  38. Data extraction - why Reasons for doing data extraction: Remembering the information Organizing information into a particular structure Summarising the content of the reports Ensuring that you look for the key things Making it easier to compare different studies Making it easier to do meta-analyses and subgroup analyses Assessing quality

  39. Data extraction - how Data extraction form Paper or electronic? Picklists or freetext? How much space do you need for each item? Do you want to record what was planned in a study, what happened, or the difference (eg for the study’s eligibility criteria)? Think carefully about what you will use the data for. For example, do you want to know the mean age, the age range, the number of people in different age groups, or the results for different age groups separately? Or do you just want to know that an article has a particular type of age data? Do you need all the data? How much detail do you need?

  40. Systematic reviewing Formulating a clear question for the review Stating objectives and eligibility criteria Identifying (all) potentially eligible studies Applying eligibility criteria Assembling most complete dataset feasible Analysing the dataset, using statistical synthesis and sensitivity analyses, if appropriate and possible Preparing a structured report Updating the review

  41. Some addresses The Cochrane Collaboration www.cochrane.org The Cochrane Library www.TheCochraneLibrary.com m.clarke@qub.ac.uk

More Related