question

You need to use CQL batches to group updates to denormalised tables so they are executed as an atomic operation. It isn't necessary to use lightweight transactions since those are for special cases where you need to do conditional inserts/updates/deletes.

In cqlsh, the batch updates to the tables when an order comes would look like:

BEGIN BATCH
  INSERT INTO orders_by_user (...) VALUES (...);
  INSERT INTO orders_by_id (...) VALUES (...);
  INSERT INTO orders_by_user_item (...) VALUES (...);
  INSERT INTO orders_status_history_by_id (...) VALUES (...);
APPLY BATCH;

I've explained this in a bit more detail in this article -- How to keep data in denormalized tables in sync. Cheers!