Depression accounts for substantial morbidity and mortality worldwide and risk of experiencing it may have a genetic component.  Depressive disorders manifest along a gradient from mild to severe.[1]  Electronic health record (EHR) data linked to large, multi-site biobanks[2] facilitate exploration of the genetic component of depression.

[1] Smith DJ, Nicholl BI, Cullen B, Martin D, Ul-Haq Z, Evans J, Gill JM, Roberts B, Gallacher J, Mackay D, Hotopf M, Deary I, Craddock N, Pell JP. Prevalence and characteristics of probable major depression and bipolar disorder within UK biobank: cross-sectional study of 172,751 participants. PloS one. 2013;8:e75362. PMCID: PMC3839907.
[2] McCarty CA, Chisholm RL, Chute CG, Kullo IJ, Jarvik GP, Larson EB, Li R, Masys DR, Ritchie MD, Roden DM, Struewing JP, Wolf WA. The eMERGE Network: a consortium of biorepositories linked to electronic medical records data for conducting genomic studies. BMC Med Genomics. 2011;4:13.  Epub 2011/01/29. PMCID: 3038887

Phenotype ID: 
List on the Collaboration Phenotypes List
Type of Phenotype: 
Contact Author: 

Suggested Citation

TBA. KPWA/UW. Depression. PheKB; 2018 Available from:


Hi All,

Diagnosis code F410 appears twice in Table ANX (Anxiety diagnosis codes), once in row 70 and again in row 71.  The descriptions for these duplicate entries are slightly different, but it turns out they are synonyms (per the ICD-10 codebook).  Since the duplicate is harmless vis-a-vis implementing the phenotype logic, we will not revise the pseudo code for this particular issue, but we will remove the duplicate entry if there is a future revision to the pseudo code.

The lack of patient-reported depression scale (PRDS) data in no way prevents implementation of this phenotype algorithm.  The logic for identifying cases and controls only uses PRDS data if it is available; if no PRDS data are available for a patient, the test for "Abnormal PRDS scors?" would resolve to FALSE.  This is the same resolution as a patient who has PRDS scores available but they are all within the normal range--such a patient would also resolve to "Abnormal PRDS scors?" = FALSE.

Our apologies if the pseudo code is ambiguous on this point.  (If you find any ambiguous language on this point please point it out and we will update the document.)


We updated the person-level covariates data dictionary with a version dated 2/8/2019.  This new version has three minor clarifications/changes, so we deleted the 1/31/2019 version, especting that no sites have already used the older version.  Please reach out if you cannot work with the new, 2/8/2019 version.

The changes introduced in the 2/8/2019 version are as follows:

  1. Added categories for measure SEX
  2. Added categories for measure ETHNICITY
  3. Updated the description for measure AGE_PSY_QUAL_2_30_180

Please reach out with any questions.

The flowchart in the document indicates a check for Abnormal PRD scores for patients who fail the non-major depressive subtype. But the narrative summary stops at 4.4 for cases.

We have a handful of patients who qualify for 2/30/180 among all subtype category diagnosis, but not for any one particular subtype. 

Thanks for finding this issue, Todd.  We will update the pseudo code to reflect that the flowchart is correct (and any discordance with the narrative summary needs to be corrected).  Please watch for a message here indicating when the pseudo code has been updated.



When looking at some diagnosis codes, I found that there seemed to be ICD9 codes and ICD10 that are 4+ digits long without a decimal point (page 27 of pseudocode). However there also ICD9 and ICD10 codes that look more traditional and have a decimal point (page 19 of pseudocode). I searched ICD code 19384 Anxiety disorder oth dis, and I find this link which has the code written as 193.84 with the decimal point. 

I wanted to clarify what codes we should be looking for and if the codes with 4+ digits without a decimal point is an error or if that's the way you want us to look for that code.



Thank you for catching this issue in the way we present ICD-9 codes in some tables.  You are correct that the ICD-9 codes shuold have a decimal point after the third digit.  Thus, in Table ANX "Anxiety diagnosis codes," the ICD-9 code in row 1 of the table should read "293.85" (with the decimal point, instead of "29384").  We will update the pseudoc code correcting this issue.



We have some patients who are in our eMERGE database, but we don't have any encounters for them. For the variable AGE_HS_CONTACT_LAST we don't have an actual age since they haven't visited us according to the system. What value should we put for the them or should we take them out of the Data Dictionary?



Do you want us to only include, esp. as controls, those who have a minimum of 2 in-person outpatient encounters, or is just 1 enough, and should it be in the last 5? years?



One encounter is enough (we'll avoid drawing a line, for now), but one of the reasonse the data dictionary includes measures of intensity of contact with the healthcare system is that we will, at time of analysis, empirically determine what is a reasonable minimum intensity of contact with the healthcare system to qualify a patient for inclusion in analyses.


Our query using the ICD proc. codes for ECTs returned 0 patients & I believe we have at least 1 or 2 - our inpatient system (historically at least until last March) used/uses CPT codes - do you have CPT codes for that procedure we can use?   



Hi, Today we post a 4/17/2019 version of the pseudo code that includes an updated code set for electroconvulsive therapy (ECT) procedures.  If you have already completed preparation of your site's phenotype data there is NO NEED to revise your work as ECT is an extremely rare procedure and we do not expect this change to impact the data significantly.  Nevertheless, for completeness we have updated the code set.  Thanks, -David

Should have mentioned... The two new ECT codes are:

CPT code 90870: Electroconvulsive therapy (includes necessary monitoring) (Single Seizure per day)

CPT code 90871: Multiple Seizures, per day. multiple seizures is also known as multiple monitored ECT (MMECT)

Hi David,

We have a PHQ-9 scores, but we also have PHQ-2 and PHQ-Teen scores. Do you want them as well? If so, what would you consider a high normal for these both of these?



Hi Ken,

Thanks for your questions. 

Yes- please include PHQ-9, PHQ-2 and PHQ-Teen scores.

Please apply the following cut points (inclusive)

PHQ-2: cut-point: 4

PHQ-Teen: cut-point: 15

PHQ-9: cut-point: 15


David, For eMERGE IV, we are collecting evidence for including depression in the PRS analyses and return of results.  Can you tell us the total sample size of depression cases and controls? Thanks, Beth