OHDSI Home | Forums | Wiki | Github

Mapping PROM data to OMOP CDM v6.0 | Incompatibility with Achilles, DQD and CDMInspection

#1

Hey,

We have ongoing projects that are mapping not only clinical data but also PROM data. To do so, we chose to use v6.0 due to the survey_occurrence table which is key for this.

However, we got to a point where Achilles is not compliant with this version and there doesn’t seem to have any solutions. However, we can’t reverse to 5.4 either, since survey occurrence is important to this mapping and our ETL design.

Keep in mind that we are doing this in an EHDEN project, which requires CDMInspection to be run

#2

At the moment, this is the log of the error in Achilles that I am getting

DBMS:
postgresql

Error:
org.postgresql.util.PSQLException: ERROR: relation “cdm.achilles_results” does not exist

SQL:
create index idx_ar_aid on cdm.achilles_results (analysis_id)

R version:
R version 4.3.0 (2023-04-21)

Platform:
aarch64-apple-darwin22.4.0

Attached base packages:

  • stats
  • graphics
  • grDevices
  • utils
  • datasets
  • methods
  • base

Other attached packages:

  • yaml (2.3.7)
  • urltools (1.7.3)
  • DataQualityDashboard (2.3.0)
  • DatabaseConnector (4.0.0)
OMOP version 6 - put on ice?
#3

Hi, a few points:

  • Likely Achilles throws a non-fatal error earlier. Could you check the logs for the first error thrown by Achilles? I expect that one or more analyses failed to execute.
  • OMOP CDM v6 is not fully supported by most OHDSI and EHDEN tools (as you already encounter for achilles). I would advice to reverse to v5.4.
  • If the survey_conduct table is important for you for local analysis, it is possible to add this table to cdm schema. Make sure to store any information needed for network studies in the OMOP CDM v5.4 tables. See this poster by Melanie Philofsky for more guidance.
#4

Hello Maxim, thank you for the reply.

The analysis that fail are based on the tables that do not exist in v6, such as EPISODE and DEATH. I tried to had those to the excludeAnalysisId argument in the achilles funcion as such:

excludeAnalysisIds = c(11, 500, 501, 502, 503, 504, 505, 506, 511, 512, 513, 514, 515, 525, 900, 901, 902, 903, 904, 906, 907, 920, 1000, 1001, 1002, 1003, 1004, 1006, 1007, 1020, 1900, 2004)

However, they are still performed as seen in the log
log_achilles.txt (903.3 KB)

Nevertheless, I see very strong opinions on discarding 6.0 in total and using 5.4. However, could I add a custom version of OMOP CDM to the EHDEN Catalogue?

t