Staudt, Lukas
(2023)
Exploring the potential of spinal registry data for the use in clinical trials.
PhD thesis, University of Liverpool.
|
Text
201453068_May2023.pdf - Author Accepted Manuscript Download (4MB) | Preview |
Abstract
Background: Sciatica describes the symptoms of low-back and leg pain most commonly due to a herniated disc that presses on the sciatic nerve. If persistent, invasive methods such as surgical microdiscectomy are required. Although being a surgery with relatively small incisions, it bears some risks of adverse events (AEs), e.g. durotomy, wound infection or in rare cases even nerve root damage. Observational registries allow for continuous data collection over indefinite time for numerous patients. One can therefore gain additional insight in subgroup demographics of the patient population and rare events. Furthermore, large numbers of patients and observations can improve the performance of prediction models. Purpose: The aims of this study are to: (1) provide a comprehensive overview of the collected dataset from the Spine Tango registry; (2) determine the best method for imputing missing data in this routinely collected registry data set; (3) assess the predictive values of patient characteristics on patient-reported outcome measures (PROMs) and complications during surgery; and (4) examine the utilization of registries in both clinical trials and observational studies and identify strategies to increase their impact on clinical trials. Methods: To understand the patient population and potential relationships among collected variables, thorough descriptive statistics were performed. Simulation studies were conducted to determine the best approach for imputing PROM items and scores, including the examination of missingness percentages, mechanisms, and cut-off point score calculations. The focus of prediction modeling was the routinely collected Core Outcome Measurement Index (COMI) and complications. Patients with sciatica were identified in collaboration with the Spine Tango committee, and various model approaches were compared for goodness of fit and prediction accuracy, including regression and mixed models. A literature review of both randomised controlled trials and observational studies was conducted, comparing differences in missing data, collected outcomes, study length, number of patients, and registry use. Case studies of successful registry utilization in other clinical areas were analyzed to identify potential for implementation in the present clinical focus. Results: The international nature of the Spine Tango registry led to variability in documentation and data collection across countries. The simulation studies showed that item-based imputation was superior to score-based imputation in most scenarios. Mixed models with random intercepts and slopes, as well as non-linear time terms, performed best in terms of model fit. Logistic regression models that defined complications as outcome were able to identify risk factors, such as prior surgery, level of spine of physical status. The utilization of registries in the field of this clinical population is underutilized, and studies from other areas demonstrate that registry use can reduce trial costs by facilitating patient identification, data collection, and event detection, as well as reducing trial-specific patient visits and improving patient retention. Conclusion: The potential of routinely collected registry data remains under-utilized within the sciatica-affected patient population. The noteworthy resemblances observed between observational data and randomized controlled trial data, both in descriptive statistics and prognostic factors, underscore the comparability of these sources and advocate for the integration of registry data in this domain. While the integration of a registry into a trial presents complexities, successful endeavors in related fields point to an innovative trial design that harmonizes these two research approaches.
| Item Type: | Thesis (PhD) |
|---|---|
| Divisions: | Faculty of Health and Life Sciences |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 24 Oct 2023 09:10 |
| Last Modified: | 24 Oct 2023 09:10 |
| DOI: | 10.17638/03172889 |
| Supervisors: |
|
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3172889 |
Dimensions
Dimensions
