Clinical scales - ordinal or coded text?

[varntzen (Vebjørn Arntzen)]

[1] 1. Very fit
1:: 1. Very fit

The way it is displayed in the CKM is confusing. Modellers will know, but a lot of our reviewers doesn’t understand why the numbers are displayed twice. Any help here in the CKM will help. I prefer [1] 1. Very fit… Or something brilliant that @sebastian.garde can come up with?

[ian.mcnicoll]

I’m arguing that there is no need for us to include the numerics in the text (as it is in the original) as we are faithfully capturing that, albeit in a slightly modified way. This is the price of turning a paper artefact into something computable. Of course, we should not mangle meaning or compromise safety but, I would defend dropping the numeric from the text, in that regard.

[siljelb (Silje Ljosland Bakke)]

I’ve reread the terms of use, and I can’t find the phrase about faithful reproduction of the text that I thought I’d seen before. So maybe this is okay, from a representation POV. Would removing the numbers from the description impact ease of implementation for you, @bna?

[bna (Bjørn Næss)]

The numbers doesn’t really matter from a technical viewpoint. We have features to add the number from the code to the user if that’s needed/wanted. Depends on the use case.

My personal view is to remove the number in description since it is a duplicate and possible error cause if the number/ordinal where to be changed.

[siljelb (Silje Ljosland Bakke)]

Great! Then I suggest the following modelling guideline: “If a scale is displayed as a numbered list, use DV_ORDINAL [to be changed to DV_SCALE once this is available in major tools]. Do not replicate numbers in the description.”

[vanessap (Vanessa Pereira)]

Hi @siljelb,
Sorry for the late reply, I have been quite busy lately.
From my perspective and usage from the both classes you mentioned I prefer to use the DV_ORDINAL for scales instead of DV_CODED_TEXT/DV_TEXT. The current tools I use allows to make immediate calculations of total scores if using ordinal for each element, plus the option the show or hide the ordinal number on the front end. Otherwise if using the coded_text/text it takes an additional parsing and I don’t believe to be that user/developer friendly in the final. I also try to avoid as much to have the case you mention with “1:1. Very fit” from the ordinal+ (number)description, it’s redundant in my opinion. Maybe only ordinal+description “1: Very fit” would be enough.

[siljelb (Silje Ljosland Bakke)]

@vanessap

From my perspective and usage from the both classes you mentioned I prefer to use the DV_ORDINAL for scales instead of DV_CODED_TEXT/DV_TEXT. The current tools I use allows to make immediate calculations of total scores if using ordinal for each element, plus the option the show or hide the ordinal number on the front end. Otherwise if using the coded_text/text it takes an additional parsing and I don’t believe to be that user/developer friendly in the final. I also try to avoid as much to have the case you mention with “1:1. Very fit” from the ordinal+ (number)description, it’s redundant in my opinion. Maybe only ordinal+description “1: Very fit” would be enough.

Thanks Vanessa! Your recommendations seem to be the same as Bjørn’s. I think this strengthens this suggestion for modelling guideline:

@siljelb

“If a scale is displayed as a numbered list, use DV_ORDINAL [to be changed to DV_SCALE once this is available in major tools]. Do not replicate numbers in the description.”