Currently the exposures_outcome_id is auto-generated sequentially, starting at 1. This can cause issues when a set of SCCS analyses is distributed over multiple machines and later combined in a single results schema.
One way to resolve this is to use a hash of the exposures, outcome, and nesting cohort IDs as the exposures_outcome_id instead. For this we could use the Murmur32 hash in the digest package, which is optimized to avoid collisions.
Currently the
exposures_outcome_id
is auto-generated sequentially, starting at 1. This can cause issues when a set of SCCS analyses is distributed over multiple machines and later combined in a single results schema.One way to resolve this is to use a hash of the exposures, outcome, and nesting cohort IDs as the
exposures_outcome_id
instead. For this we could use the Murmur32 hash in thedigest
package, which is optimized to avoid collisions.