Executing sql server check mapping_completeness.sql fails when all values in the column value_as_concept_id are NULL

[PieterMaertens]

When executing the sql server check mapping_completeness.sql inside the R function cdmInspection I get following error:
No column name was specified for column 1 of 'T'.
The observation table I use does have a lot of records but the column "value_as_concept_id" is always NULL.

select 'Observation value', count_big(*),
      sum(is_mapped),
      100.0*sum(is_mapped) / count_big(*),
      sum(num_records),
      sum(case when is_mapped > 0 then num_records else 0 end),
      100.0*sum(case when is_mapped > 0 then num_records else 0 end)/sum(num_records)
from
(
   select '', case when value_as_concept_id > 0 then 1 else 0 end as is_mapped, count_big(*) as num_records
   from @cdmDatabaseSchema.observation
   where value_as_concept_id IS NOT NULL
   group by case when value_as_concept_id > 0 then 1 else 0 end
) T

[sampo-tays]

For me, this analysis also fails even though the column "value_as_concept_id" is not NULL. Not sure if relevant, but the 'Observation value' section differs from all other analyses in mapping_completeness.sql in a couple of ways: '' instead of field name (observation_source_value?), and "observation_source_value" not included in the group by statement.