Objective: To describe results of a combined risk-based monitoring (RBM) approach in the context of a large global observational longitudinal study, Enroll-HD.

Background: Enroll-HD is a global clinical research platform for Huntington’s disease (HD) it includes the collection of clinical and demographic data to be available to researchers. The need to maintain high standards of data quality is a paramount to secure data completeness, accuracy and consistency. There is no consensus regarding the most efficient monitoring strategy to employ.

Methods: The following monitoring strategies are applied: 1)On-site monitoring includes the review of informed consent for all participants and the consistency of a limited number of specific data points with source documents. 2)Remote data review (RDR) using a risk-based approach, 1 in 10 random participants are reviewed for a given site, after an initial run-in phase of 10/10 review. Data submitted in the electronic data capture system are reviewed within 2 weeks by trained data managers, checking for internal consistency. 3)Centralized Statistical Monitoring (CSM) includes statistical algorithms run monthly to screen for errors, statistical outliers, unusual correlation structure, extreme variances or anomalous data patterns. 4)Medical monitoring in order to ensure data is clinically plausible in the context of HD by a medical expert.

Results: On-site monitoring included 126 sites, 77 in reduced review, 3 escalated into full review. Issues found during RDR were coding related (ICD-10, SNOMED) and inconsistencies in HD clinical history and scoring discrepancies between scales. CSM included 7015 participants and identified ≅1% misclassified by HD category (pre-manifest vs. manifest) and <1% with data that were considered outliers in clinical and demographic measures. Identified outlier values were deemed as clinical plausible by medical monitoring. No anomalous correlation pattern was identified in UHDRS. Extreme variances and unusual density distribution were found for specific sites on several measures, but many of these were explained by the case mix at the sites.

Conclusions: RBM process, in Enroll-HD, uses 3 methods: on-site monitoring, RDR and CSM. Based on the low level of error found, these methods ensure high quality data. As a next step, we will optimize CSM and reduce the use of more resource intensive monitoring strategies (on-site and RDR) while securing high data quality.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/strategies-for-a-high-quality-data-in-an-observational-longitudinal-study-enroll-hd-experience/