MA pbp section b11 dme prosth orth sup

What it is

1. 1CMS Plan Benefit Package (PBP) Section B-11 extract: the member cost-sharing DESIGN (coinsurance %, copays, deductibles, out-of-pocket maximums, prior-authorization/referral rules) that each Medicare Advantage / Part C plan sets for Durable Medical Equipment (B-11a), Prosthetics/Orthotics (B-11b), and Medical Supplies (B-11c).
2. 2It is the benefit-design layer (what the plan charges the enrollee), sourced from CMS PBP Public Use Files, for benchmarking MA DME/prosthetic cost-share across plans and years.

Grain

How to read it

1. 1"What in-network coinsurance % do Medicare Advantage plans charge for DME in 2026?" -> SELECT pbp_b11a_coins_pct_mc, count(*) FROM medicare_silver.pbp_section_b11_dme_prosth_orth_sup WHERE report_year=2026 AND quarter=1 AND pbp_b11a_coins_yn='1' GROUP BY 1 ORDER BY 2 DESC.
2. 2This returns the distribution of DME coinsurance rates across plans; the dominant value is 20.0% (the Medicare standard), with a long tail of 0% (zero-cost-share plans), 10/15/30%, etc.
3. 3Always pin report_year AND quarter to a single snapshot.

Traps

1. 1NOT a rate/price table. This is plan benefit DESIGN - what the MA plan charges the ENROLLEE for DME (coinsurance %, copays, deductibles), sourced from CMS PBP PUFs. It contains no provider rates, no claims, no allowed amounts. To get the commercial-to-Medicare cost story you join plan design here to actual DME fee schedules elsewhere.
2. 2GRAIN TRAP: the profiler reports an 18-column superset key (adds maxplan/maxenr benefit-value columns). The TRUE minimal grain is just (contract_id, plan_id, segment_id, report_year, quarter) = 201,488 = row count. report_year alone is NOT unique (114,175) and (contract,plan,seg,effective_period_start) is NOT unique - you MUST include quarter because CMS republished the PBP PUF up to 4x within 2018-2020 & 2022-2024.
3. 3DEPRECATED COLUMN FAMILY: pbp_b11_* (no a/b/c letter) is populated ONLY in report_year 2001 (742 rows) and is 100% NULL for 2002+. Querying it for modern DME cost-share returns nothing. Use pbp_b11a_* (DME), pbp_b11b_* (prosthetics/orthotics), pbp_b11c_* (medical supplies), which are populated 2002-2026.
4. 4NO is_current column and NO snapshot-dedup needed: the natural key already yields one row each, so do NOT add a row_number()/latest filter. But quarter is a RE-RELEASE axis, not coverage-quarter - if you want the canonical annual benefit pin quarter=1 (and the latest cms_vintage_sequence if a year was republished), otherwise a single plan-year can be counted 2-4 times.
5. 5NULLS CARRY MEANING and yn convention is CMS-style: '1'=Yes, '2'=No (strings), NULL=not applicable / not separately captured for that vintage (e.g. pbp_b11a_auth_yn is NULL for 62K older rows even though the plan had an auth rule encoded in the 5-char pbp_b11a_auth string). Treat '2' as an explicit No, not as missing; treat NULL as unknown/legacy.
6. 6segment_id is part of the grain and must not be dropped: 5,295 plan-years are split into multiple geographic segments with different DME benefits. Most rows are segment '0' (single segment) but multi-segment plans ('1'..'N') and empty '' (2001-era) both occur.

Columns (200)

Query it

describe_data_source(source_fqn="medicare_silver.pbp_section_b11_dme_prosth_orth_sup")     # the card, live
query_substrate(sql="SELECT * FROM medicare_silver.pbp_section_b11_dme_prosth_orth_sup WHERE contract_id = '<value>' AND plan_id = '<value>' AND segment_id = '<value>' AND report_year = '<value>' AND quarter = '<value>' LIMIT 100")