Employing content analysis & crowd-sourcing to revise randomised controlled trials patient information leaflets

Abstract

The poor readability of patient information leaflets (PILs) to help recruit people into research studies has been a serious concern for the Health Research Authority during the last 2 decades. Multiple independent studies have reported serious issues with almost all documents intended to inform patients or general audiences. The results of these studies have been considered by UK medical institutions and a series of guidelines developed to improve the quality and readability of PILs intended for inviting patients to randomised controlled trials (RCTs). Even with this focus and some improvements in patient leaflets, most of the current documents intended for this purpose are still considered too complex to be understood by public audiences. Meanwhile, several techniques have addressed the issue of measuring text readability. This thesis analyses the utility of several of these techniques to identify the sentences that are too hard to understand when employed via a webtool to help PIL authors identify and correct PIL readability issues: a) readability indices that associate text characteristics with the US school grade needed to understand the document, b) the Cloze procedure (identifying specific words that are not understood by the participants by removing words and asking participants to complete the sentences), c) sentiment and content analysis to identify and associate comments from public participants reviewing PILs with specific sections of the documents, and d) online crowdsourcing to review and validate sentences that are too hard to be understood by public audiences. The first study explored associations between PIL text characteristics and recruitment rates to the RCT to which they applied. The second studied feedback given by local public participants asked to revise PILs containing serious readability issues. The third study contained several sub-studies, each assessing the effects of the previously mentioned techniques on the identification, revision and validation of readability issues present in the PILs by an online crowd recruited using Mechanical Turk. This thesis contributes to our knowledge about employing content analysis and online crowdsourcing techniques to help authors of PIL for RCTs to identify, revise and validate sentences that are too hard to be understood by public audiences.

More information

Identifiers

ORCID for Adam Geraghty: orcid.org/0000-0001-7984-8351

Catalogue record

Export record