MA pbp section b14 preventive

What it is

1. 1CMS Plan Benefit Package (PBP) Section B-14 'Preventive Services' filing for every Medicare Advantage / Part C plan: the member cost-sharing (copays, coinsurance %, annual plan-coverage dollar caps) and access rules (prior-auth, referral, attestation) for Medicare-covered preventive screenings and supplemental wellness/health benefits.
2. 2Used to compare what each MA plan charges enrollees for preventive and supplemental services across plans, years, and plan types.

Grain

How to read it

1. 1To compare 2025 Medicare Advantage preventive cost-sharing across plans, pin a single snapshot and read the per-service copay/coins/cap columns: SELECT contract_id, plan_id, segment_id, pbp_a_plan_type, pbp_b14b_copay_amt_rpe_max AS routine_physical_copay, pbp_b14e_coins_pct_max_diab AS diabetes_screen_coins_pct, pbp_b14c_maxplan_amt_wig AS wig_annual_cap FROM medicare_silver.pbp_section_b14_preventive WHERE report_year=2025 AND quarter=1.
2. 2This answers 'what does each MA plan charge an enrollee for a routine physical, a diabetes screening, and how much wig coverage it caps at' (observed: routine physical $0 median–$20 max; diabetes screening 20% median coins; wig cap $400 median–$1,000 max).

Traps

1. 1The profiler's reported 'grain' (~100 columns) is a massive SUPERSET — it just kept appending maxplan/_d columns until unique. The TRUE minimal grain is only 5 columns: contract_id, plan_id, segment_id, report_year, quarter.
2. 2quarter is a SNAPSHOT/release dimension, not a benefit attribute. The same plan-year is re-published in multiple quarterly CMS files (most visible 2018-2024 with Q1-Q4); rows across quarters are near-identical re-releases. For 'the plan's 2025 benefits' filter to ONE quarter (e.g. quarter=1) or you will multi-count plans. There is NO is_current / is_latest flag — you must pick the snapshot yourself (latest quarter, or quarter=1).
3. 3report_year, effective_period_start, effective_period_end and observed_period are all redundant: report_year = year(effective_period_start) for 100% of rows; start=Jan 1, end=Dec 31 of that year. asserted_at is the May-2026 ingestion/load date (NOT a benefit date) — never use it as the coverage period.
4. 4The 800+ pbp_b14* columns are SPARSE and layout-shifts by year: most specific copay/coins columns are NULL for any given plan/year (e.g. the b14b_ame_* copay columns are entirely null in 2025; only ~102 of ~392 cost columns are populated in 2025 Q1). NULL means 'benefit/field not applicable to this plan', not zero — and $0 (a real value, e.g. $0 preventive copay) is distinct from NULL. Always COUNT() non-nulls before averaging a sub-benefit column.
5. 5Codes use CMS PBP conventions, not booleans: yes/no fields (*_auth_yn, *_refer_yn) are 1=Yes, 2=No (NOT 1/0). orgtype and pbp_a_plan_type are 2-digit CMS plan-classification codes ('01'=HMO, '02'=local PPO, '04'=HMO-POS, '20'=PFFS, '31'=MSA...). bid_id is NULL and segment_id is '' for older years (pre-bid era / non-segmented plans) — both blanks are meaningful, not data errors. Years 2010 and 2011 are entirely absent from the table.
6. 6Cost columns come in _min / _max pairs and per-service tags (mhc=mental-health counseling, rat=transportation, bsd=bundled, sc/sce=smoking cessation, wig; bar=bone density, diab, glauc, ekg, dre). *_copay_*_amt = dollars, *_coins_*_pct = percent, *_maxplan_amt_<tag> = annual plan-coverage dollar CAP (not a cost share). Don't mix amt vs pct vs maxplan when aggregating.

Columns (825)

Query it

describe_data_source(source_fqn="medicare_silver.pbp_section_b14_preventive")     # the card, live
query_substrate(sql="SELECT * FROM medicare_silver.pbp_section_b14_preventive WHERE contract_id = '<value>' AND plan_id = '<value>' AND segment_id = '<value>' AND report_year = '<value>' AND quarter = '<value>' LIMIT 100")