OPG Surgical Grouper Imputation

Some payers price outpatient surgery using a grouper-based contract: each HCPCS code is assigned to a procedure tier, and all codes in the same tier share the same negotiated rate. The OPG (Outpatient Procedure Grouper) imputation detects the rate per tier from observed MRF data and fills in missing codes within the same grouper.

Payer Coverage

OPG imputations are only active for payers that have a published surgical grouper reference table:

Payer	payer_id	params.py variable
UnitedHealthcare	643	OPG_SCHEDULES_UNITED
Aetna	7	OPG_SCHEDULES_AETNA
Cigna	76	OPG_SCHEDULES_CIGNA
BCBS California	403	OPG_SCHEDULES_BCBS_CA
CareFirst	44	OPG_SCHEDULES_CAREFIRST
Anthem CA (CA Blue Cross PPO only)	42	OPG_SCHEDULES_ANTHEM_CA
Florida Blue	388	OPG_SCHEDULES_FLORIDA_BLUE

Each reference table maps (billing_code, billing_code_type) → grouper. All HCPCS codes that share a grouper are expected to share the same negotiated rate under a grouper-based contract.

Detection Logic

For each (payer_id, network_id, provider_id, grouper) combination, the pipeline finds the most frequently observed dollar rate among Outpatient HCPCS codes in that grouper:

OPG grouper detection (simplified)

opg_grouper_ranked AS (
  SELECT
      payer_id, network_id, provider_id,
      grouper,
      ROUND(rate, -1)  AS rate,   -- rounded to nearest $10
      COUNT(DISTINCT roid) AS freq,
      ROW_NUMBER() OVER (
          PARTITION BY payer_id, network_id, provider_id, grouper
          ORDER BY COUNT(*) DESC
      ) AS rnk
  FROM long_rates
  JOIN opg_schedules USING (payer_id, billing_code, billing_code_type)
  WHERE bill_type = 'Outpatient'
    AND rate_type NOT LIKE '%perc%'   -- dollar rates only
  GROUP BY 1, 2, 3, 4, 5
)

The top-ranked rate activates only if it clears one of two confidence thresholds:

Condition	Threshold
High coverage within group	freq / total > 80% and total > 100 codes observed
High coverage vs. full grouper	freq / total_possible > 50% of all codes in the payer's schedule

If neither is met, opg_base_rate is set to NULL and no imputation occurs for that grouper.

Application

Once the grouper base rate is confirmed, every missing code assigned to that grouper gets the same rate:

imputed_rate = opg_base_rate

The grouper assignment comes directly from the payer's reference table — a code's grouper is fixed by the schedule, not inferred from the data.

Walk-Through

UHC PPO, Hospital Z — Grouper 'Tier 2 Surgery'

Grouper schedule: UHC's OPG schedule assigns 40 HCPCS codes to grouper "Tier 2 Surgery". All are expected to pay the same rate.

Observed rates: 35 of those 40 codes have a negotiated rate of $2,800 in the hospital MRF. The remaining 5 have no MRF entry.

Frequency check: 35 / 40 = 87.5% → passes the 80% threshold. 35 > 100? No. But 35 / 40 = 87.5% > 50% of total_possible → passes second threshold.

Detected base rate: opg_base_rate = $2,800 (rounded to nearest $10).

Impute 5 missing codes: Each gets imputed_rate = $2,800.

opg_base_rate = $2,800 | 5 codes imputed | sourced from imputations_derived

OPG vs CSTM

Both OPG and CSTM use surgical grouper reference tables, but they operate at different stages. OPG is part of imputations_derived and detects the most common observed rate per grouper from MRF data. CSTM (imputations_cstm) is a separate late-stage pass that uses tier averages from core_rates and runs after the main derived imputation chain.