From @colinorr: " I am leading the design and development of a technology platform for a global benchmarking initiative that ICON is conducting as a partner of the ICHOM consortium (ichom.org). The global benchmarking is based on standardized data sets for patient outcomes as defined by patients. I am right in the middle of a pilot as I type. To support global benchmarking and patient outcome analytics I feel the OMOP data model may well be ideal for this. I am joining this group to get advice on how to use this model in practice and to find other people who may have done similar projects."
@colinorr, some of the datasets I work with contain survey questionnaires or other forms of patient-reported or physician-reported outcomes . In most of the cases I’ve experienced, I’ve done well to store these data in the OMOP CDM in the OBSERVATION table, effectively putting the ‘question’ as the OBSERVATION_CONCEPT_ID and then the answer as either VALUE_AS_STRING, VALUE_AS_NUMBER, or VALUE_AS_CONCEPT_ID, depending on what I’ve captured. I’ve found many standard questionnaires are already standardized in LOINC or SNOMED, but when I’m working with a non-standard questionnaire, I leave the OBSERVATION_CONCEPT_ID unmapped (=0) and just store the verbataim question in the OBSERVATION_SOURCE_VALUE field.
@clairblacketer had a nice poster at the OHDSI symposium in 2015 about how she mapped the NHANES survey into the OMOP CDM, and it had several of these types of tips and tricks.
Creating this forum thread in case others working with patient-reported outcomes have other guidance for @ColinOrr.