“SEND is just an electronic representation of your study report.”
This used to be my stock phrase. I actually believed it. It was how the world seemed a few years ago. I meant every word of it.
The concept is admirable, but the reality seems to be getting more removed all the time. The intention was that industry would have a standardized representation of the study’s results in order that they could be exchanged electronically, because previously they were just in PDF reports which, while being very human readable, are far from being machine readable. Data would be collected and reports produced as normal. The only change would be that the study report would now have corresponding files containing the data.
So, “SEND is just an electronic version of the study report” is what I used to tell people. Oh, how naïve. The world must have seemed like such a simple place.
Today I heard about a CRO who now refuses to produce SEND 3.0 because they’ve changed their data collection practices for SEND 3.1. My younger self would have been outraged at the idea that not only the SEND standard, but a particular version of SEND, would dictate how data are collected.
Yet, for where we are today, I know this is pretty reasonable for various reasons. Probably the main one is the significant change to Histo Pathology data that was introduced in SEND 3.1. As part of a move to be able to better compare microscopic findings from one study to another, SEND 3.1 brought in Controlled Terms and a much tighter specification of Histo findings. If these data were not collected using a SEND 3.1 compatible glossary, then retrospectively converting the data into SEND would cause challenges. At best this would be time consuming, and at worst this could be near impossible without the Pathologist on hand to unpick all the findings.
Another point worth considering is the idea that we don’t always want the report and the SEND dataset to match. “You mean the SEND data would be different from the data in the study report?!” I hear my younger self ask in horror. Well, actually yes. There are occasions where data on a page may be filtered, formatted or even summarized in order to best fit that page. A SEND dataset doesn’t have such constraints. This notion came up again this week. There are occasions where subjects are observed, and only abnormalities are listed in the report in order to save space. The report would also include a line to explain this. However, in the SEND dataset, all records should appear, and the consumer of the data can then filter them out using their own tools as they so desire.
The same principle could apply to something like food consumption on a longer-term study, which may well be collected daily, yet reported weekly. So, it’s entirely possible that the data on the page is summarized at a slightly higher level than the data in SEND.
The examples we’ve discussed here are probably the most obvious ones, yet there are many more.
So, here we are, and I look back on the naivety of the opinions of my former self. Conceptually, there’s still some truth in the notion that “SEND is just an electronic representation of your study report”, yet once we get into the detail, we see so many exceptions to this notion that the very idea can seem farcical at times.
As usual, hit me with an email (firstname.lastname@example.org) with your own thoughts on the matter, whether you agree, disagree or just wonder what on earth I’m rambling on about.
I’ll leave you with one final thought, “How do you write a good joke?”, “Well, I start with a laugh and work backwards!”. While a cheap gag in its own right, there’s a lesson there: Start with the end in mind. The end used to be a Study Report, but today that’s the SEND Dataset too.
Till next time,