How can I obtain most frequently prescribed drugs post diagnosis?

MichaelMartinec · March 11, 2023, 9:48am

It might be in front of my eyes and I simply can’t see it. Then I apologise. I’ve searched this forum and www for guidance but without any luck so far.

It’s a simple query really. I want to obtain the most frequently prescribed drugs in the period 1 year post index date. Index date is defined in my cohort as the first diagnosis of epilepsy.

I’ve defined the cohort. But in Characterization, when I choose feature analysis related to drug exposure, the time windows (short, medium, long term) available are preceding the cohort start which I understand is the index date. Is there a way to define a feature looking at time window post index?

Gowtham_Rao · March 11, 2023, 10:59am

If you are using FeatureExtraction to generate characterization then you can do this by specifying future time windows in

FeatureExtraction::createTemporalCovariateSettings

example

github.com

OHDSI/CohortDiagnostics/blob/21d6eb3838e64f48c44b05f8d82c6ebcf5db389c/R/RunDiagnostics.R#L53-L80

    
      
          temporalStartDays = c(
            # components displayed in cohort characterization
            -9999, # anytime prior
            -365, # long term prior
            -180, # medium term prior
            -30, # short term prior
          
          
  # components displayed in temporal characterization
            -365, # one year prior to -31
            -30, # 30 day prior not including day 0
            0, # index date only
            1, # 1 day after to day 30
            31,
            -9999 # Any time prior to any time future
          ),
          temporalEndDays = c(
            0, # anytime prior
            0, # long term prior
            0, # medium term prior
            0, # short term prior

This file has been truncated. show original

MichaelMartinec · March 14, 2023, 7:13pm

Thanks! Your reply got me looking in the right direction and I was eventually able to solve by defining a custom feature analysis in ATLAS.