Once a site has collected the data for a patient visit, the coordinator has the tiresome task of entering that data into the clinical trials system. One day the data will just flow in from the EHR system – more about that in a future blog – but today the vast majority of clinical data is re-typed into an EDC system.
If that typing effort is burdensome, either because the EDC system isn’t very good or the protocol has been weighed down with a massive number of edit checks, does it matter? I frequently hear opinions that indicate many people don’t think it matters very much; the site is being paid regardless, the need for data quality trumps all other considerations, and there must surely be some regulatory compliance reasons why it all has to be so hard.
Of course, those same people will get agitated when the data is late! But is the lateness of the data ever correlated back to how burdensome the data-entry process might be?
Questions and More Questions
For example, if a form is originally saved as empty and subsequently needs to be completed, why does the site coordinator need to enter the same change reason on every single field? Where’s the “Apply to All” checkbox you’d find in just about every other application on earth?
If coordinators change a data value based on a manual query, why do they separately need to respond to the query with “OK, I changed the data?” If coordinators mark an AE as continuing, why are they ever allowed into the stop-date field and thus tempted to enter inconsistent data? Somewhere, right now, a person is screaming at such a stupid piece of software; like the proverbial tree falling in the forest, just because you can’t hear it, doesn’t mean that it’s not happening.
Lessons From Travel Websites
Here’s a useful parallel I like to recite. Go to any online travel website or car rental site, and try to get something wrong. You’ll find it’s impossible to book a flight that lands before it takes off, to specify a different drop-off location for your car and then not give one, or to press the confirm button before indicating whether you’d like the insurance option.
At the very worst, you’ll be advised of a problem as you tab out of a field, such as entering an email address without an “@” sign. Performing transactions on the web is so effortless these days that you can do it in various states of inattention, but if you look closely, some of this software is wickedly clever. So, ask yourself whether the clinical software being provided to the investigative sites is similarly clever. If not, are you suffering a productivity or reputational penalty as a result?
The 3% Factor
Another consideration is that coordinators produce very accurate work the first time. The standard industry metric is that only 3% of data is changed after initial entry, and an analysis I recently presented at the SCDM conference comes up with exactly the same figure. This 3% includes data that is legitimately changed over time, such as entering the stop date for an AE that has finished. So all the other interactions – edit checks and SDV – could possibly affect less than 3% of the data.
Clearly, there needs to be an appropriate balance between appropriate programmed and manual checks to protect the integrity of the data, and excessive checks that cause the coordinator to keep replying with “The entered data is correct.”
There are no easy answers, but I often feel that insufficient consideration is given to productivity and motivation for the site personnel when putting a study together. So the next time you see the message “Congratulation, your trip is booked,” rather than half a dozen queries, ask yourself whether your next protocol could be as effortless.
About the Author
Keith Howells, chief technology officer at OmniComm Systems, made a presentation about risk-based monitoring at SCDM 2017 in Orlando on Tuesday, September 26. He also spoke at the 7th Annual Innovation Forum, October 19-20, in Fort Lauderdale.