Answer
Direct Path Read means the session is reading directly from disk, bypassing the shared buffer cache used by other processes — typically a sign the optimizer chose a full table scan (often in parallel) because it judged that cheaper than using an available index. Two common underlying causes worth checking: stale table statistics (making the optimizer's cost estimate for a scan-vs-index decision inaccurate), or a recent update/upgrade that added columns to the table, changing its size/row characteristics enough to shift the optimizer's plan. Running EXPLAIN PLAN on the query in question is the fastest way to confirm whether it really is choosing a full scan and why.