Table 1.
Phenotypes by site and agreement with original phenotypes
| Phenotype | Site | Time (days) | Prevalence | Overlap between original and OMOP phenotypes (number) | Positive specific agreement | Negative specific agreement | Issues* | |||
|---|---|---|---|---|---|---|---|---|---|---|
| Overlap | Original only | OMOP only | Neither | |||||||
| T2DM | 1 | <1 | 0.008 | 38 | 0 | 4 | 5465 | 0.950 | 1.000 | OMOP query included “Diabetic oculopathy” even though it includes both type 1 and type 2 diabetes, leading to 3 FPs; original eMERGE query implementation differed from logic in original eMERGE description on how to handle problem list; noted need for Observation table |
| T2DM | 2 | <1 | 0.224 | 1179 | 95 | 30 | 4086 | 0.950 | 0.985 | Original eMERGE query included “Diabetic macular edema” for type 1 diabetes even though it includes type 2; OMOP query missing “Diabetic oculopathy due to type I diabetes mellitus” for type 1 diabetes; database missing some lab codes; changes in RxNorm mappings over time |
| T2DM | 3 | 23 | 0.087 | 242 | 381 | 250 | 4804 | 0.434 | 0.938 | Missing and incorrect LOINC codes in database; reported that error was mainly in original eMERGE query |
| T2DM | 4 | <1 | 0.003 | Did not have full labs in database | ||||||
| T2DM | 5 | <1 | 0.038 | 735 | 1165 | 18 | 396 | 0.554 | 0.401 | Erroneous and missing source data due to EHR merger |
| T2DM | 6 | <1 | 0.108 | (No additional issues) | ||||||
| T2DM | 7 | 144 | 0.191 | 3139 | 819 | 1588 | 19143 | 0.723 | 0.941 | Drugs in database were not coded so rewrote medication component; difficulty debugging without OHDSI stack; did NOT compare to the original eMERGE query but to a site-specific query |
| T2DM | 8 | <1 | 0.022 | OMOP vocabulary files were loaded as empty string instead of null | ||||||
| T2DM | 9 | <1 | (No additional issues) | |||||||
| ADHD | 1 | <1 | 0.001 | 7 | 0 | 0 | 5500 | 1.000 | 1.000 | (No additional issues) |
| ADHD | 2 | <1 | 0.004 | 23 | 11 | 1 | 5355 | 0.793 | 0.999 | Some coding issues around descendants in OMOP query, e.g., exclusion concept “Organic mental disorder” had concept “Postconcussion syndrome” as a descendant but should not be excluded. |
| ADHD | 3 | <1 | 0.003 | (No additional issues) | ||||||
| ADHD | 4 | <1 | 0.123 | 1761 | 507 | 48 | 12282 | 0.864 | 0.978 | Drugs for alternate visit types not migrated to OMOP database (388 of 507 mismatches); original query added ADHD “predominantly inattentive type” not in definition (187 of 507 mismatches); OMOP query did not distinguish in-person encounters but original eMERGE query used only in-person encounter inclusion codes and all exclusion codes; missing some exclusions in OMOP query compared to definition; original eMERGE query did not include ICD10-CM exclusion diagnoses; OMOP query used year count instead of exact days for age requirement |
| ADHD | 5 | Did not have permission for retrospective mental health | ||||||||
| ADHD | 6 | 84 | 0.017 | 65 | 15 | 19 | 4861 | 0.793 | 0.997 | Additional medications were added to the original eMERGE query because the site had only adults; original query added “Attention deficit hyperactivity disorder, predominantly inattentive type”; two years of data were added to the database for the OMOP query |
| ADHD | 7 | <1 | 0.008 | (No additional issues) | ||||||
| ADHD | 8 | 14 | 0.002 | Rewrote OMOP query to use ICD codes instead of SNOMED CT | ||||||
| ADHD | 9 | <1 | Lab codes were manually mapped to LOINC | |||||||
*
Most sites additionally commented on the need to map the schema names in the OMOP query to local schema names and to create an empty cohort definition