Walk-Through Example

Step-by-step canonical rate selection for two contrasting ROIDs.

Example A — Cross-Validated Rate

ROID with payer and hospital MRF agreement

Available rates: Payer \$125 (negotiated rate), Hospital \$120 (chargemaster-derived), Transform \$130 (pct-to-dollar), Imputed \$115 (tier 2 average)

Accuracy scoring:

• Payer \$125 vs hospital array [\$120] → within ±20% → counterparty match → score = 7 + 125/1e8 = 7.00000125
• Hospital \$120 → matches payer \$125 → score = 7 + 120/1e8 = 7.00000120
• Transform \$130 → no counterparty, CDF = 0.54 → score = 6.54
• Imputed \$115 → within bounds, CDF = 0.23 → score = 4.23

Rate arrays built: 4 entries, ordered by source priority. Payer at index 1, hospital at index 2, transform at index 3, imputed at index 4.

Selection: best_score = 7.00000125, best_idx = 1 (payer rate).

canonical_rate = $125.00 canonical_rate_score = 5 (7.x → confidence 5) canonical_rate_source = 'payer_hospital' (score=7 → both sources agreed) canonical_rate_type = 'raw: payer_negotiated_rate' canonical_rate_class = 'Raw'

Why the payer rate wins over the hospital rate

The payer rate ($125) wins over the hospital rate ($120) because 7.00000125 > 7.00000120. Both are counterparty-validated — the tiebreaker (rate/1e8) slightly prefers the higher dollar amount among equal-confidence rates.

Example B — Outlier Payer Rate

ROID where the only payer rate is an outlier

Payer MRF rate \$4,200: Medicare MPFS for CPT 99213 is \$92. This rate is 45× Medicare — far above the upper outlier bound of 30×.

Imputed rate \$155: Derived from RC HCPCS crosswalk (revenue code family average). Within bounds at 1.7× Medicare.

Accuracy scoring:

• Payer \$4,200 → 45× Medicare exceeds upper bound of 30× → OUTLIER → score = 1.xx
• Imputed \$155 → within bounds, CDF = 0.61 → score = 2.61

Selection: score 2.61 > score 1.xx → imputed rate wins.

canonical_rate = $155.00 canonical_rate_score = 2 (imputed, within bounds) canonical_rate_source = 'imputation' canonical_rate_type = 'impute: rc_family_gc_hosp_perc_to_dol'

The outlier is preserved but loses

The outlier payer rate ($4,200) is preserved in the wide table but loses to the imputation because score 1 < score 2. A user filtering on canonical_rate_score > 1 would see the imputed $155 and not the $4,200 outlier. The raw outlier value is always available in prod_combined_all for audit purposes.