Exploring OMOP hierarchy for blood counts

Use case

A need to map blood count components from two different source systems to OMOP. There are different potential OMOP concept ids. How to choose between them ? How to deal with a situation where different institutions end up with different OMOP concept ids for essentially the same measure.

Leukocytes

Starting with a single concept id 3010813 (chosen in a mapping process) for the number of Leukocytes in a volume of blood.

lvbg <- omop_id(3010813)

lvbg |> 
  select(-valid_start_date,-valid_end_date,-invalid_reason) |> 
  kable()

concept_id	concept_name	domain_id	vocabulary_id	concept_class_id	standard_concept	concept_code
3010813	Leukocytes [#/volume] in Blood	Measurement	LOINC	Lab Test	S	26464-8

Using omopcept to find the immediate relations of “Leukocytes [#/volume] in Blood”

leuk_r <- omop_relations(3010813)

leuk_r |> arrange(relationship_id) |>
  select(relationship_id,concept_name_2,domain_id,concept_class_id) |> 
  kable()

relationship_id	concept_name_2	domain_id	concept_class_id
Contained in panel	Hemogram without Platelets panel - Blood	Measurement	Lab Test
Contained in panel	Laboratory	Observation	Survey
Contained in panel	Differential panel, method unspecified - Blood	Measurement	Lab Test
Contained in panel	Manual Differential panel - Blood	Measurement	Lab Test
Contained in panel	Laboratory - acute [CARE]	Observation	Survey
Contained in panel	Leukogram panel - Blood	Measurement	Lab Test
Contained in panel	Short blood count panel - Blood	Measurement	Lab Test
Has component	Leukocytes	Observation	LOINC Component
Has property	Number Concentration	Observation	LOINC Property
Has scale type	Qn	Observation	LOINC Scale
Has system	Blood	Observation	LOINC System
Has time aspect	Point in time (spot)	Observation	LOINC Time
Is	Leukocytes [#/volume] in Blood	Observation	LOINC Time
Is a	Leukocytes | Blood | Hematology and Cell counts	Measurement	LOINC Hierarchy
Is a	Leukocytes|Number Concentration (count/vol)|Moment in time|Blood	Measurement	LOINC Group
Is a	Hematology (coagulation) differential count	Measurement	LOINC Class

Of these relations Leukocytes | Blood | Hematology and Cell counts looks like it may be useful, it has a concept_class_id of LOINC Hierarchy and itself has the following relations.

lbhcc_r <- omop_relations(37043992) |> 
  arrange(relationship_id) 

lbhcc_r |>
  select(relationship_id,concept_name_2,domain_id,concept_class_id) |> 
  kable()

relationship_id	concept_name_2	domain_id	concept_class_id
Is	Leukocytes | Blood | Hematology and Cell counts	Measurement	Lab Test
Is a	Leukocytes	Observation	LOINC Component
Subsumes	Leukocytes [#/volume] corrected for nucleated erythrocytes in Blood	Measurement	Lab Test
Subsumes	Leukocytes [#/volume] in Blood by Estimate	Measurement	Lab Test
Subsumes	Leukocytes [#/volume] in Blood by Automated count	Measurement	Lab Test
Subsumes	Leukocytes [#/volume] in Blood	Measurement	Lab Test
Subsumes	Leukocytes [#/volume] corrected for nucleated erythrocytes in Blood by Automated count	Measurement	Lab Test
Subsumes	Leukocytes [#/volume] in Blood by Manual count	Measurement	Lab Test

#lbhcc_r |> 
#  omop_graph( graphtitle = "", saveplot = FALSE )

Can we find similar hierarchy parents for other blood count concepts ?

Seems yes, there are > 400 concepts that match “*Blood | Hematology and Cell counts”.

We can do similar for erythrocytes as we did for leukocytes.

#new fixed=TRUE arg for omop_names() allows this 
bhccounts <- omop_names("Blood | Hematology and Cell counts", fixed=TRUE) |> 
  #returning 452 concepts
  #removing those for Fetus
  filter(!str_detect(concept_name,"Fetus"))
  #now down to 417 concepts

#can I filter out desired one from this ?
#ery <- bhccounts |> 
#  filter(concept_name, starts_with("Erythrocytes |"))

#it is id 37059055
ery_r <- omop_relations(37059055)

ery_r |> 
  select(relationship_id,concept_name_2,domain_id,concept_class_id) |> 
  kable()

relationship_id	concept_name_2	domain_id	concept_class_id
Is	Erythrocytes | Blood | Hematology and Cell counts	Measurement	Lab Test
Subsumes	Erythrocytes [Morphology] in Blood by Automated count	Measurement	Lab Test
Subsumes	Erythrocytes [#/volume] in Blood	Measurement	Lab Test
Subsumes	Erythrocytes [#/volume] in Blood by Automated count	Measurement	Lab Test
Subsumes	Erythrocytes [#/volume] in Blood by Manual count	Measurement	Lab Test
Is a	Erythrocytes	Observation	LOINC Component

See https://github.com/OHDSI/Vocabulary-v5.0/issues/301 which has some info about this