MA pbp mrx p

What it is

1. 1A CMS Medicare Advantage / Part D Plan Benefit Package (PBP) extract that records, for every Part D drug plan and benefit year, the plan's "post-deductible / initial-coverage" prescription-drug cost-sharing tier structure — one row per formulary tier (Preferred Generic, Generic, Preferred Brand, Non-Preferred, Specialty, etc.) with that tier's flat copay or coinsurance.
2. 2It is the high-level tier summary; the granular per-pharmacy-channel cost-sharing (preferred vs standard retail, mail, 1/2/3-month supply, LTC, out-of-network) lives in the sibling table pbp_mrx_tier.

Grain

How to read it

1. 1To compare a plan's published drug-tier cost structure for a benefit year, filter to one snapshot and inspect the tiers.
2. 2Example: SELECT mrx_tier_post_id, mrx_tier_post_label_list, mrx_tier_post_cost_struct, mrx_tier_post_copay_amt, mrx_tier_post_coins_pct FROM medicare_silver.pbp_mrx_p WHERE contract_id='H0028' AND plan_id='014' AND segment_id='0' AND report_year=2026 AND quarter=1 ORDER BY CAST(mrx_tier_post_id AS int); — returns that plan's 2026 formulary tiers (Preferred Generic → Specialty) and their post-deductible cost-sharing structure.
3. 3For the actual retail/mail copays by day-supply, join the same plan key to medicare_silver.pbp_mrx_tier.

Traps

1. 1COST IS MOSTLY NULL HERE. Only ~23.6% of the standard 5-tier rows carry any copay or coinsurance (148K/622K have a copay, 8K have coinsurance, 472K rows have neither). This 'post' table is the tier SUMMARY; the real per-pharmacy-channel cost-sharing (preferred vs standard retail, mail order, 1/2/3-month supply, LTC, OON) is in the SIBLING table medicare_silver.pbp_mrx_tier (columns mrx_tier_inp_copay_1m, mrx_tier_innpp_*, mrx_tier_oon_*, etc.). For actual drug copays, join to pbp_mrx_tier, not this table alone.
2. 2NO is_current flag and quarter is the snapshot dimension. CMS republishes the entire PBP file each quarter, so one plan-year appears up to 4 times (Q1–Q4) as near-duplicate rows. The grain REQUIRES quarter. To get one row per plan-year, filter to a single quarter (quarter=1, or MAX(quarter)). Counting rows without this double/triple/quadruple-counts plans in 2018–2024.
3. 3Decode copay vs coinsurance via mrx_tier_post_cost_struct: NULL=cost not stored here (look in pbp_mrx_tier), '2'=copay only, '1'=coinsurance only, '3'=both. Don't assume a NULL copay means $0 — it usually means 'detail lives elsewhere'.
4. 4Dead / near-dead columns: mrx_tier_post_type_id is the constant '2' (the profiler wrongly put it in the grain — it adds nothing); mrx_tier_post_ref_prcng_yn and mrx_tier_post_lbl are 100% NULL; spec_yn/inject_yn/cst_shr_type/includes_mmp are >98% NULL. bid_id is just contract_id_plan_id_segment_id concatenated. plan_type/effective_period_start/observed_period in the profiler's 10-col key are redundant — the true minimal grain is 6 columns.
5. 5plan_id and segment_id are zero-padded STRINGS ('001','0'), not integers — cast carefully when joining. report_year/quarter/version are stored as doubles (2023.0, 1.0). orgtype='10' rows are standalone PDPs and legitimately have NULL pbp_a_ben_cov.
6. 6asserted_at / parse_revision / cms_vintage_sequence / parser_artifact_hash / sha256_content_hash are Vlada ingestion provenance, NOT plan attributes. asserted_at (2026-05-16/20) is the load date — never use it as a plan effective date; use report_year / effective_period_start for that.

Columns (40)

Query it

describe_data_source(source_fqn="medicare_silver.pbp_mrx_p")     # the card, live
query_substrate(sql="SELECT * FROM medicare_silver.pbp_mrx_p WHERE contract_id = '<value>' AND plan_id = '<value>' AND segment_id = '<value>' AND report_year = '<value>' AND quarter = '<value>' AND mrx_tier_post_id = '<value>' LIMIT 100")