OHDSI Home | Forums | Wiki | Github

Level of detail in Electronic Health Records

Hi all,

I would like to better characterize the Treatment-Resistant Depression population using observational databases. However, I’m not familiar with the content of an average TRD’s patient EHR.

For example, below are some of the characteristics I wish to identify:

(a) Patient Characteristics:
i. Age
ii. Type of depressive episode (unipolar, bipolar, psychotic, atypical, other)
ii. Number of depression relapses and time to relapse
iii. Average time to becoming treatment-resistant
iv Psychiatric comorbidities
v. Medical comorbidities (e.g., diabetes, cardiac disease, renal disease, dementia and other cognitive abnormalities)
vi. Suicidal ideation
vii Suicide attempts
viii. Duration of symptoms
ix. Screening tools used to make the diagnosis
x. Diagnostic tools to confirm the diagnosis

(b) Compare and contrast how current treatments affect:

i. Change in depression scores as measured by depression scales
ii. Change in depressive symptomatology (e.g., sleep disorders, fatigue, weight change, cognition)
iii. Change in measures of anhedonia
iv. Change in measures of functional capacity (e.g., physical functioning, ability to care for self)
v. Change in measures of quality of life
vi. Change in measures of suicide ideation
vii. Change in suicide attempts
viii. Other

Are EHRs detailed enough to make such characterizations?

It depends… can you define these things as data elements?

For example:

  • “Type of depressive episode” – probably both a direct clinical finding and also something that can be deducted from other clinical findings
  • What constitutes a relapse? Is it a clinical finding at a certain interval?
  • What does treatment-resistance look like? Is it a distinct series of drug exposures or procedures?

The list goes on and so do my questions. @Christophe_Lambert and the UNM team had a poster last year on imputing uncoded self harm: https://www.ohdsi.org/2019-us-symposium-showcase-63/ – this may be of interest.

For the treatment effects, you’d have to do similar exercise of turning these into coded endpoints. @Dymshyts @shilparatwani @SCYou and others are part of an OHDSI Psychiatry WG that is bringing together community expertise in Psychiatry to discuss how to extend the existing vocabularies to better support storage of psychiatry scales.

I would suggest prioritizing which of these things are of highest interest. This is many characterization studies in one.

1 Like
t