Using segmented regression analysis of interrupted time series data to assess colonoscopy quality outcomes of a web-enhanced implementation toolkit to support evidence-based practices for bowel preparation: a study protocol

Abstract

BackgroundWhile there is convincing evidence on interventions to improve bowel preparation for patients, the evidence on how to implement these evidence-based practices (EBPs) in outpatient colonoscopy settings is less certain. The Strategies to Improve Colonoscopy (STIC) study compares the effect of two implementation strategies, physician education alone versus physician education plus an implementation toolkit for staff, on adoption of three EBPs (split-dosing of bowel preparation, low-literacy education, teach-back) to improve pre-procedure and intra-procedure quality measures. The implementation toolkit contains a staff education module, website containing tools to support staff in delivering EBPs, tailored patient education materials, and brief consultation with staff to determine how the EBPs can be integrated into the existing workflow. Given adaptations to the implementation plan and intentional flexibility in the delivery of the EBPs, we utilize a pragmatic study to balance external validity with demonstrating effectiveness of the implementation strategies.Methods/DesignParticipants will include all outpatient colonoscopy physicians, staff, and patients from a convenience sample of six endoscopy settings. Aim #1 will explore the relative effect of two strategies to implement patient-level EBPs on adoption and clinical quality outcomes. We will assess the change in level and trends of clinical quality outcomes (i.e., adequacy of bowel preparation, adenoma detection) using segmented regression analysis of interrupted time series data with two groups (intervention and delayed start). Aim #2 will examine the influence of organizational readiness to change on EBP implementation. We use a PRECIS diagram to reflect the extent to which each indicator of the study was pragmatic versus explanatory, revealing a largely pragmatic study.DiscussionImplementation challenges have already motivated several adaptations to the original plan, reflecting the nature of implementation in real-world healthcare settings. The pragmatic study responds to the evolving needs of its healthcare partners and allows for flexibility in intervention delivery, thereby informing clinical decision-making in real-world settings. The current study will provide information about what works (intervention effectiveness), for whom it works (influence of Medicaid versus other insurance), in which contexts it works (setting characteristics that influence implementation), and how it works best (comparison of implementation strategies).Electronic supplementary materialThe online version of this article (doi:10.1186/s13012-015-0276-3) contains supplementary material, which is available to authorized users.

Evaluation of the effectiveness of comprehensive drug price reform: a case study from Shihezi city in Western China

“…In fact, the possibility of nonlinear trends also exists. Although some models, such as Box-Jenkins models, may be used to analyze non-linear trends, they are less useful in examining changes in trend occurring at defined time points [47]. In this study, 12 data points before and after the CDPR and more than 100 observations per data point were collected, which can provide the robust estimates of change in level and trend for outcomes [48].…”

Internet of things in healthcare for patient safety: an empirical study

“…Their analysis revealed a significant decrease in total use and cost in the two years following the intervention [ 47 ]. Similarly, segmented regression for ITS studies was used in many other healthcare interventions, including to assess incremental costs [ 48 ], evaluate screening effectiveness [ 49 ], and evaluate new strategic interventions [ 50 ].…”

Establishing optimal illuminance for pedestrian reassurance using segmented regression

“…28 To identify such points, known as breakpoints, the relationship between variables may be estimated by segmented regression models, where a number of straight, inter-linked, segments, are fitted to the data. 29 Segmented regression has been used for the identification of breakpoints in research data about health, mortality and morbidity, [29][30][31][32][33][34] as well as for economic research. 35 Often, the method is applied to time series data, to identify distinctive trends during different periods, for example, before and after an intervention.…”