EDC Autoquery Rate - A Marker for Site Quality

Applied Clinical Trials

Applied Clinical Trials, Applied Clinical Trials-10-01-2012, Volume 21, Issue 10

Medidata

Site quality has always been critically importance in clinical trials. With the steady increase in pressure to reduce the overall cost of clinical trials coupled with the upward trend in strategic outsourcing, there is heightened interest in exploring tools and data that provide an early indication of site quality. This month we explore one metric, EDC autoquery rate-the rate of EDC system-generated queries-that can be used for this purpose.

As you can see, the graph illustrates-at a high level-the strong correlation between autoquery rates (AQRs) and data correction rates (DCRs). Phase I-Phase IV data from the Medidata Insights metrics warehouse-comprising of over 2,600 studies from 65 sponsors and 16 therapeutic areas-indicates that AQRs are directly proportional to data correction rates (DCRs). A correlative analysis was performed at the site level, which further reinforced that sites with a high AQR also have a high DCR. DCRs are tangible indicators of data quality at the site-the higher the DCR, the lower the data quality. The measure of data quality is reflective of the overall quality of the site and can easily be used to identify sites with potential quality issues.

You may ask why ACRs were picked as opposed to using all queries including data management queries and site management queries. With respect to data quality, it is fair to say that eCRF query rates-auto-queries, data management queries, and site monitor queries-and data correction rates should all be good indicators of the quality with which sites are capturing patient data into the sponsor's EDC system. In fact, an initial analysis of these measures taken across studies in the Insights Metrics Warehouse reveals a very strong correlation between each of these query rates and data correction rate. Thus any one of these measures should be a sufficient surrogate to detect data quality issues. However, one of them, autoquery rate, is clearly the most effective when viewed from the perspective of early signal detection. This is because-unlike manually generated queries and subsequent data corrections-autoqueries are generated and observable immediately upon entry of the eCRF data at the site. Thus, both the numerator and denominator comprising an autoquery rate are always current, whereas the numerator in the other rates always lags behind the denominator by weeks or months, or yes, even years in query generation.

 

The early detection of sites with quality issues can help sponsors implement remediation plans at an early stage leading to not only an improvement in data quality but also a reduction in the trial's cost. Depending on the nature of the identified issue, potential follow-up actions may include any of the following:

  • Increasing the percentage of SDV required at the site

  • Delivering additional guidance and training to the site on specific aspects of the protocol, eCRFs, or other study logistics

  • Planning additional on-site visits to more closely monitor certain aspects of the site's study conduct

So what is your experience with the relationship between AQRs and site quality? Have you been using them to identify sites with potential quality issues? As always, Medidata is interested to hear your take on this result. Please stay tuned as we will continue to delve further into Insights data.

 

-Medidata Solutions, http://www.mdsol.com