To annotate or re-annotate- what is the question?

Article type
Authors
Hampson L1, Beecher D2, Jones L1
1Cochrane Pregnancy and Childbirth
2Cochrane CET/IKMD
Abstract
Background: Cochrane Pregnancy and Childbirth (CPC) has been part of a 6-month project with the Bill and Melinda Gates Foundation and key partners aiming to accelerate the delivery of a next-generation evidence system for curation of maternal and child health evidence. As part of this, reviews were annotated according to their PICO (Population, Interventions, Comparisons, and Outcomes). The CPC Information Specialist was involved in the Quality Assurance (QA) process, the main objective of which is to ensure a good match between terms chosen from controlled vocabularies and key concepts in the review.

Objectives: To look at a particular QA issue and reasons for re-annotation of reviews.

Methods: The annotation team used text in the Methods section of each review to annotate Population (P), Intervention (I) and Comparison (C). In some cases the QA specialist marked reviews for re-annotation because important concepts were missing from the Methods section and therefore the annotation did not accurately match the review question.

Results: We will report the number of reviews where re-annotation was needed following QA because there was an important aspect of the P, I or C missing from the Methods section. We will look at where else in the review this aspect was captured discuss how this will impact on future annotations.

Conclusions: If PICO annotations are based on text in the Methods section and this section is incomplete or ill-defined, this has implications for future annotations (both manual and automated), and consequently has implications for retrieval. It highlights the potential need for clearer guidance for review authors and more rigorous checking during the editorial processing of reviews. It is imperative that reviews clearly define these key aspects of the review question so that they can be more accurately annotated.