We are implementing conversion of a survey based oncology data to CDM. In the survey, there is a specific question: Comorbidities that affect treatment of cancer, and a list of comorbidities the doctor can pick from. In the data, each co-morbidity is a different field and no specific date associated with it. How would we treat this in CDM? Would we assume the date would be the same as that of completion date of survey, or date of treatment started OR would we create a new pseudo date? Also, would we map this as a history of observation?
Hi Urvi,
We ran into a similar problem with the NHANES dataset where patients were asked a series of questions about their medical history and we had to determine how best to capture their responses. We ended up mapping the conditions to CONCEPT_IDs and then routed the questions to the CONDITION_OCCURRENCE table.
For example, if a patient answered “Yes” to the question “Have you ever been told you have high cholesterol?”, we mapped it to the CONCEPT_ID 432867 (Hyperlipidemia) and gave it a CONDITION_START_DATE of January 1st of the year of the survey since we don’t have date of completion.
Since you have a set list of conditions for the physicians to choose from you could run that list through Usagi which would help you map them to standard concepts. I think these would fit well in the CONDITION_OCCURRENCE table using the date of survey completion as the CONDITION_START_DATE while leaving the CONDITION_END_DATE as NULL. From there you could use the CONDITION_TYPE_CONCEPT_ID field to denote that these came from a survey and you could store the actual condition as chosen by the physician in the CONDITION_SOURCE_VALUE field.
Hope that help!
Clair
1 Like