Polling Postgres DB architecture

Question

We are using Spring Integration DB polling which polls a table every second and executes some logic based on result. My questions are:

1. What are some good practices we can implement to account for DB load?

2. The query currently is just simple select, is there anything we can do to improve it (indexes, different query, etc)?

   SELECT id FROM table WHERE status = 'READY' LIMIT 100 FOR UPDATE

3. Any architectural improvements we could make? Currently, we are thinking of DB Read Replica in AWS.