HICDEP Development (#1) - Clinical events coding (#85) - Message List
We have been looking at the case definitions for the clinical events coding (tblCEP). Our UK Register of HIV Seroconverters cohort does not systematically capture such detailed information as listed in the HICDEP case definitions. We ask physicians to report on whether the patient had, for example, 'liver disease' or 'diabetes', leaving it up to the clinician's interpretation, with a space for the clinician to write their own details alongside this in free text (usually quite brief). So, for example, the physician might report diabetes, but we don't know their fasting blood glucose levels unless they happen to have told us, or we have reports of liver diease but we do not necessarily know if it is end-stage liver failure, or the clinician has reported a fracture, but we do not know if they had an xray.
How would you advise we proceed with coding for this table? We would only be able to code and include a tiny small portion of our events data according to these very strict case definitions (when we happen to have that information provided by the clinician). Or would you suggest that we report all clinician identified diabetes, fracture, liver disease, etc for example?
I see your problem and wonder what the COHERE Coordination Centers (Bordeaux and Copenhagen) suggest? I am sure that there are other cohorts in the same situation.
A potential solution could be to use the additional field CEP_DESCRIP (the actual text description of the event) in conjunction with the mandatory CEP_ID to indicate e.g. "clinical diagnosis without specific details".
Brunoledergb03/04/15 17:56:18 (2 years ago)
Annabelle, the PhD student working on non aids defining events, Alex, and I have discussed this issue in some detail. I think the problem with including a variable such as CEP_DESCRIP is that it will be largely unusable. The amount of information we (the UK Register of HIV seroconverters cohort) could include in such a variable varies largely by clinic and clinician. In addition, this type of variable would likely be reported in a number of different languages throughout Europe which makes it even more difficult. In CASCADE, I have tried to use a variable with free text, MODE_OTH, and it's nearly unusable because of the different languages and responses you get from individual cohorts. I do not have a better suggestion, however, so am keen to hear from the COHERE coordination centers!
Best, AshleyAshley03/04/15 18:14:47 (2 years ago)