I am working on a use case estimating rates of COPD exacerbations and healthcare utilisations by COPD severity. As a first step I want to characterise the COPD population by severity, using spirometry only GOLD stage classification. GOLD stage is not typically directly recorded in the data but rather must be inferred from spirometry measurements.
This was more difficult than I was expecting in both ATLAS and R (not including the challenges in using measurement data). I have outlined the problems I was facing in each below. I would really appreciate it if anyone could advise whether i have misunderstood something or point me towards solutions to these challenges.
- It is possible to perform subgroup analyses, but each strata must be defined separately, and it is not a reusable object
- The workflow for creating feature extractions and performing characterisations (i.e. using featureExtraction::DbGetCovariateData with the aggregate function), does not include an option to stratify by subgroups
- The only solution i can see is to pre-define separate cohorts by GOLD stage and then characterise within each cohort, but it’s obviously suboptimal. Also featureExtraction::CreateTable1 only allows comparison of two cohorts (though obv its use is not essential).
I’m still very much on the learning curve with these tools so any help greatly appreciated.
And if any one can point me in the direction of similar use cases that would also be really useful.