File
Yamato T.1, Maher C.2, Catley M.3, Moseley A.2
1The George Institute for Global Health, Musculoskeletal Division, Sydney, Australia, 2The George Institute for Global Health, The University of Sydney, Musculoskeletal Division, Sydney, Australia, 3Sansom Institute for Health Research, University of South Australia, Adelaide, Australia
Background: The Template for Intervention Description and Replication (TIDieR) is a 12-item checklist developed to help authors describe interventions in sufficient detail to allow replication in clinical practice. Many researchers have also been using TIDieR as a measure of completeness of reporting of interventions in clinical trials. Therefore, calculating a summary score for the TIDieR checklist would allow the completeness of reporting of interventions to be summarised concisely.
Purpose: The aim of this study was to examine if it is reasonable to calculate a TIDieR summary score. We used Rasch analysis to evaluate unidimensionality and the hierarchy of items.
Methods: This is a secondary analysis from a previous study in which the completeness of 200 reports of randomised trials was rated using the TIDieR checklist. In the present study, we proposed a checklist scale. Each item was assessed on 3-point likert scale with the following categories: not reported (0), partially reported (1), adequately reported (2). When the item criteria had been adequately described for both the intervention and control groups, a score of two was ascribed. We used Rasch analysis to investigate the item hierarchy, assess the checklist scoring structure and establish unidimensionality.
Results: The item reliability index was 0.98 suggesting that the sample was sufficient to confirm the item hierarchy is reproducible. The TIDieR checklist items targeted the studies adequately; the average study measure was 0.48 (0.87) compared to the average item measure of 0.0 (1.82). Visual inspection of the category structure suggested the likert scale functioned as expected for most items. However, none of the included reports scored zero on item 1, very few reports scored >0 for items 11 and 12 and partially reported was underutilised for item 7. The TIDieR checklist appeared to constitute a unidimensional scale. Visual inspection of the Principal Component Analysis (PCA) correlation matrix did not reveal any patterns in the data indicative of secondary dimensions and an eigenvalue of 2.0 supported this finding. No items demonstrated excessive positive infit, although items 1 and 11, the easiest and hardest to endorse items respectively, demonstrated excessive negative outfit. A reliability index of 0.62 suggested the scale may only be able to discriminate the least and most complete reports.
Conclusion(s): The TIDieR checklist is useful as a scoring measure of completeness of intervention description. The category structure functioned as expected and the checklist items targeted the reports. Item 1 offers little in terms of the scale score as it is reported in almost every study. Nonetheless, it is clearly a necessary inclusion in the checklist. Item 10 was not utilised and may not be applicable in all studies.
Implications: The use of the TIDieR checklist as measure scale would facilitate the evaluation of strategies to improve reporting of interventions in trial reports, including editorial policies and the use of supplementary material for published manuscripts. It could also increase the possibilities for analyses in systematic reviews or methodological studies (e.g., sensitivity analyses using completeness of reporting).
Funding acknowledgements: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Topic: Research methodology & knowledge translation
Ethics approval: Not applicable
All authors, affiliations and abstracts have been published as submitted.
