Think about the aggregation used for the top 5 table in the report.

[LFISHER7]

The report contains a top 5 table for each codelist, which represent the most common codes recorded for each codelist across the study period. For clinical events, these code are often distinct, but for medications, different codes can represent different doses/formulas of the same medication. For medications, it's therefore potentially more interesting to aggregate further up the hierarchy (e.g. Amoxicillin instead of Amoxicillin 250mg).

In the demo, we did this by aggregating by "vpid", but this column isn't in all codelists that will be available. In other projects, we've grouped by "bnf_code". We would need to know about the presence of these columns in the metadata from OpenCodelists if we wanted to do this aggregation.

My suggestion is that we remove any aggregation for now and treat clinical and medication events as the same. Longer term, we can get some clinical input on whether aggregation of medication events is useful and what aggregation makes most sense.