Preventing overdiagnosis through the deimplementation of low-value diagnostic tests: a systematic review

Objective: Over testing is a key driver of overdiagnosis. One way to prevent overdiagnosis is to restrict the use of diagnostic tests that are unnecessary, of low-value, or even harmful. Even when there is strong evidence against the use of a test, action is often required to restrict or eliminate its use. The de-implementation of diagnostic tests is likely to face different challenges than implementation. The aim of this systematic review was to investigate key characteristics of effective de-implementation strategies for diagnostic tests and to identify gaps in knowledge and areas for future research. Method: Medline, Embase, Cochrane, and Rx for Change databases were searched from January 1st, 1990 until November 1st 2016. Additional studies were found through reference checking and searching in healthcare websites. Studies of interest were those focusing on the reduction or elimination of a low-value clinical practice for clinical reasons, as opposed to financial. Information on the characteristics of the de-implementation strategies, the effectiveness of these strategies, the study design and reporting, as well as perceived or measured barriers and facilitators to these strategies were extracted. In this project we focus on studies on diagnostic tests. Results: Thirty-seven studies on the de-implementation of a diagnostic test were identified for this preliminary analysis. The most common aim was to quantitatively evaluate the effectiveness of de-implementation (n=26,70%) while the remaining focused on the qualitative assessment of (perceived) factors influencing the effectiveness. Only 3 effectiveness studies were randomized trials. De-implementation strategies were commonly multi-faced, the most common elements being physician education or reminders. Half were on laboratory tests(n=20,54%) and a handful were on potentially physically harmful tests, such as imaging or endoscopy (n=8,22%). Most studied concluded moderate effectiveness of the de-implementation. Conclusions: This review has identified a set of studies on the de-implementation of diagnostic tests. Most studies showed moderate success, acknowledging room for improvement in the development of de-implementation strategies. We highlight areas for future studies on de-implementation of diagnostic test to focus on: the process of designing a tailored and multi-faced interventions, evaluating factors influencing effectiveness either quantitatively or through process evaluation, patient related outcomes, and potential for sustainability and spread. When tests are evidenced to be low-value, effective and efficient de-implementation is key to improving patient outcomes