Clinical trial site queries: How many do we have to live with?
If you look at a typical proposal for the management and data analysis of image data in clinical trials, there is always a budget item related to site queries. Queries can be generated for a variety of reasons, ranging from missing data (e.g., a missing timepoint) to incongruences within a single submission (e.g., different documents showing multiple different weights or ages for the same subject). When queries are generated, the progress of data analysis is slowed down. If the query is not resolved promptly, the progress of the trial might cause that data point to be ignored or the subject might be lost.
Although the trial impact of a single data point may not be major, the aggregate of these situations can cloud the overall picture. Think of it like a loom, where you are trying to weave a certain design. Having a single thread break during the process might not affect the final product. However, losing many may cause the pattern to be dull and harder to recognize.
The ways in which the industry has typically approached these situations include hiring labs that “run a tighter ship,” pressuring trial administrators and project managers to close queries “as soon as possible,” and having monitors lean on sites to respond faster to the queries. All of these solutions have two things in common: they attempt to manage the result of a problem instead of dealing with its genesis, and they are labor/cost intensive.
In my experience with imaging trials, addressing the problem before it happens has an outsized positive effect in reducing administrative queries. The number of queries is significantly reduced if sites are presented with the right tools at the right time for each trial they’re working on. A very small investment in time when building a submission pays off in spades downstream. These tools both guide the process and prevent the type of errors that tend to generate the most queries. For example, they might enable ultrasound pixel de-identification or pre-filling transmittal forms with image meta-data. If the tools “come to the user” when they are preparing the submission, the necessary checks and validations are applied during the building process, and most incongruences or missing elements are eliminated. Applying the tools at the source ensures errors are caught by the sender while the information is still fresh in her memory.
A smart workflow-based platform that presents the right tools at the right time eliminates the need for the sender to make decisions which are inconsistent with any specific protocol. Since building and sending data for clinical trials is a small and infrequent requirement for overburdened site coordinators, expecting them to remember dozens of submission steps from visit to visit is not always reasonable. At the same time, keep in mind that they like queries just as much as the person making them. Provide them with the right tools, and trial outcomes will greatly improve.