Assessing the effects of diagnostic tests on patient outcomes: how reliable, informative and practical are randomised controlled trials?

Article type
Authors
DiRuffano L, Dinnes J, Hyde C, Deeks J
Abstract
Background: The ultimate evaluation of a diagnostic test establishes whether its use confers a benefit to patient outcomes. The clinical effectiveness of a new diagnostic test combines evaluation fo the test with contingent treatment - a so called test-plus-treatment intervention. Preliminary analysis shows significant challenges may threaten to compromise the reliability or feasibility of conducting randomised controlled trials (RCTs) of test-plus-treatment interventions; six themes form the theoretical focus of the project: sample size, risk of bias, performance of intervention components, documentation and standardisation of intervention, timeliness and outcomes. Objectives: To assess the validity and applicability of the RCT study design in evaluations of test-plus-treatment interventions. Methods: Two cohorts of test-plus-treatment intervention trials were identified for analysis: The Cochrane Central Register of Controlled Trials, searched for years 2002 to 2007, and all published and ongoing trials funded by the NHS Health Technology Programme, including unpublished archival material such as original study protocols and monitoring reports. Studies were included if they were randomised, assessed patient outcomes and evaluated a strategy involving a diagnostic test with contingent treatment. Tests used for screening or monitoring were excluded. A subset of studies was used to develop a strategy for eliciting a typology of clinical questions and classification of variations in study design, subsequently applied to all remaining papers. Data were abstracted independently, in duplicate to standardised recording forms following the six pre-specified thematic foci of the project. Results: Eighty-six trials (1%) identified in CENTRAL 2004 to 2007 fulfilled the inclusion criteria, and examination of 2002 to 2003 is underway. Nineteen of 45 eligible trials were identified in the HTA cohort. Upon completion of analysis we will present a typology of clinical questions that have been evaluated using RCTs, and an examination of how a study’s aims correspond to its design and derived conclusions. Using case studies we delineate results of the six thematic analyses as evaluated through our methodological framework, concentrating on how these aspects impact on interpretation and the formulation of clinical conclusions. Conclusions: We will provide a review of the methodological quality of test-plustreatment trials.