PBP clinical diagnostic therapeutic (cleaned view)

What it is

1. 1A Medicare Advantage Plan Benefit Package (PBP) semantic view giving the member cost-sharing (copay/coinsurance), prior-auth, and referral terms each MA plan files with CMS for clinical diagnostic and therapeutic services (PBP Section B8: complex imaging, diagnostic radiology, therapeutic radiation, lab, basic diagnostic imaging).
2. 2It is a benefit-design reference, not a price or paid-claims table.

Grain

How to read it

1. 1To get the current in-network cost share an MA plan charges for MRI/CT (complex imaging) in 2026: SELECT contract_id, plan_id, segment_id, cost_share_string, copay_amount_min, copay_amount_max, coinsurance_pct, auth_required FROM medicare_semantic.vw_pbp_clinical_diagnostic_therapeutic WHERE report_year=2026 AND cdt_sub_benefit='complex_imaging'.
2. 2Because 2026 has only Q1 this is already the latest snapshot; for years with multiple quarters add a MAX(quarter)-per-plan filter to avoid counting the same plan-year 2–4 times.
3. 3Answers 'what does plan H0028-007 charge a member for complex imaging, and does it need prior auth?'

Traps

1. 1report_year is the benefit year; quarter is a QUARTERLY SNAPSHOT of the SAME annual filing, not a distinct benefit period. The same plan-year appears once per released quarter (1–4) AND once per sub-benefit, so a plan-year = up to 20 rows. To dedupe to one view-of-record per plan-year, take MAX(quarter) per report_year (e.g. 2026 only has Q1; 2021/2025 Q1 only; 2020 missing Q3). There is NO is_current column.
2. 2copay_amount and coinsurance_pct are the MINIMUM, not a single true value: copay_amount = copay_amount_min and coinsurance_pct = coinsurance_pct_min in 100% of non-null rows. A single row bundles multiple service lines under one sub-benefit, so the _max column often differs (copay max differs from min in 165K rows; e.g. '$0 copay' point but max=$65). Read min AND max to see the real spread; the point column understates the high end.
3. 3NULL ≠ zero and NULL ≠ false. 35,490 rows have has_copay/has_coinsurance/cost_share_string/copay/coinsurance all NULL = benefit offered but cost share unknown/unparsed (null_amount_reason='unknown_requires_crosswalk'), NOT free. auth_required/referral_required are NULL in 39,745 rows (NULL ≠ not-required). Also has_copay=true does NOT guarantee a numeric copay_amount (flag is true in 434K rows but the amount is populated in only 355K).
4. 4This is a COST-SHARE / benefit-design table (what the member pays per the plan filing), NOT prices, allowed amounts, or paid claims. The entire benefit_amount_* and limit_* dollar/quantity families are 100% NULL here; network_type is constant 'in_network' (no out-of-network); has_deductible is constant false and max_oop_applies constant true. confidence_tier='source_faithful' means as-filed, never reconciled to consumer EOC/SB (consumer_document_reconciliation_status='not_checked').
5. 5Identity requires ALL of contract_id + plan_id + segment_id — plan_id is only unique within a contract, and segment_id ('0' for 91%, but 1–16 for segmented plans) is part of the grain. cdt_sub_benefit (5 values) is 1:1 with source_pbp_section and is a required grain axis: each plan-snapshot has exactly 5 rows (one per B8 sub-section, each balanced at 147,678), so cost share for, say, complex_imaging vs diagnostic_lab lives in different rows of the same plan.

Columns (64)

Query it

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