Score Hierarchy

Complete mapping from internal validation scores to user-facing canonical rate scores, including the provider/code-type hierarchy used by BRIT accuracy.

The implementation lives in accuracy_macros.sql. BRIT accuracy first routes each ROID to a scoring path, then each non-NULL rate column receives an integer score plus an optional decimal tiebreaker.

Scoring Route

Route	Condition	Accuracy table	Notes
Drugs	is_drug_code = true	tmp_int_accuracy_brit_drugs	Drug scoring wins routing even when the provider type is also PG, DME, or Urgent Care.
Labs	is_lab_code = true and is_drug_code = false	tmp_int_accuracy_brit_labs	Uses lab-specific Medicare percentage windows and network consistency.
Physician Group	provider_type = 'Physician Group', non-drug, non-lab	tmp_int_accuracy_brit_physician_groups	Uses PG-specific bounds and consistency checks.
DME	provider_type = 'DME', non-drug, non-lab	tmp_int_accuracy_brit_dme	Uses DME-specific bounds and consistency checks.
Urgent Care	provider_type = 'Urgent Care', non-drug, non-lab	tmp_int_accuracy_brit_urgent_care	Uses UC-specific bounds and consistency checks.
Medical catch-all	Non-drug, non-lab, and not PG/DME/UC	tmp_int_accuracy_brit_medical	Covers hospitals, ASCs, imaging, and other medical provider types.

Score Mapping

Internal Score	Label	Canonical Score (1–5)	Meaning	Decimal Component
8	Provision-reported	5	Provision base rate; highest-priority medical IP source	none
7.x	Highest confidence	5	Counterparty validated, or top provider-type consistency tier	rate / 1e8 for validated medical rates; otherwise none
6.x	Strong in-bounds	4	Raw/posted source in bounds, or second provider-type consistency tier	CDF value where enabled
5.x	Benchmark window	3	Transform/source falls in the preferred Medicare-relative window	CDF value where enabled
4.x	In bounds	2	Within provider/code outlier bounds, no stronger signal	CDF value where enabled
3.x	Imputation benchmark	3	Non-posted imputation in a benchmark-backed tier	CDF value
2.x	Imputation in bounds	2	Non-posted imputation within bounds	CDF value
1.x	Outlier	1	Outside all acceptable bounds	CDF value
0	No rate	0	NULL / no rate available	—

Hierarchy by Provider Type

Medical Catch-All

Applies to non-drug, non-lab rows whose provider type is not Physician Group, DME, or Urgent Care.

Score	Hierarchy condition
8	Provision-reported msdrg_base_rate_mult_cms_weight is non-NULL.
7.x	Payer and hospital sources agree within ±5%, both sides are non-outliers, and inpatient rates also pass the stricter Medicare multiple check.
6.x	Raw or posted source is not an outlier. Anesthesia conversion-factor columns receive 6.999 when in bounds.
5.x	Hospital-MRF gross charge with MRF percent-of-charge is in the preferred bound, or the source is within the benchmark window: 0.95–10× Medicare.
4.x	Rate is within medical outlier bounds but has no stronger validation signal.
3.x	Non-posted imputation is benchmark-backed.
2.x	Non-posted imputation is in bounds but not benchmark-backed.
1.x / 0	Non-NULL outlier / NULL.

Drugs

Applies to all is_drug_code = true rows.

Score	Hierarchy condition
7	Payer and hospital sources agree within ±5%, and the rate is 1.2–4× ASP/Medicare.
6	Raw or transformed rate is inside the preferred ASP window: 1.25–3.5×.
5	Raw or transformed rate is inside the broader ASP window: 0.9–4×.
4	Other in-bounds drug source. Payer drug bounds allow 0.8–10×; hospital drug bounds allow 0.8–4×.
2	Non-posted drug imputation is in bounds.
1 / 0	Non-NULL outlier / NULL.

Drug CDF is always 0, so drug scores are effectively integer tiers.

Labs

Applies to is_lab_code = true and is_drug_code = false.

Score	Hierarchy condition
7	Provider-network rates are consistent, and the rate is 0.4–1.3× Medicare.
6	Provider-network rates are consistent, and the rate is 0.3–3× Medicare.
5	Rate is 0.4–1.3× Medicare without the consistency flag.
4.x	Rate is within lab outlier bounds: 0.2–4.5× Medicare.
1 / 0	Non-NULL outlier / NULL.

Physician Group

Applies to provider_type = 'Physician Group' for non-drug, non-lab rows.

Score	Hierarchy condition
7	Counterparty validation passes, or provider-network rates are consistent and the rate is 0.9–3.5× Medicare.
6	Provider-network rates are consistent, and the rate is 0.5–5.5× Medicare.
5	Rate is 0.9–3.5× Medicare without the consistency flag.
4.x	Rate is within PG outlier bounds.
1 / 0	Non-NULL outlier / NULL.

Anesthesia codes use a PG-specific outlier check: at least 1.0× the anesthesia Medicare rate and no more than $200.

DME

Applies to provider_type = 'DME' for non-drug, non-lab rows.

Score	Hierarchy condition
7	Provider-network rates are consistent, and the rate is 0.9–3.5× Medicare.
6	Provider-network rates are consistent, and the rate is 0.5–5.5× Medicare.
5	Rate is 0.9–3.5× Medicare without the consistency flag.
4.x	Rate is within DME outlier bounds: 0.5–5.5× Medicare.
1 / 0	Non-NULL outlier / NULL.

Urgent Care

Applies to provider_type = 'Urgent Care' for non-drug, non-lab rows.

Score	Hierarchy condition
7	Counterparty validation passes, or provider-network rates are consistent and the rate is 0.9–3.5× Medicare.
6	Provider-network rates are consistent, and the rate is 0.5–5.5× Medicare.
5	Rate is 0.9–3.5× Medicare without the consistency flag.
4.x	Rate is within Urgent Care outlier bounds.
1 / 0	Non-NULL outlier / NULL.

Urgent Care anesthesia codes use the same anesthesia-specific outlier check as Physician Group: at least 1.0× the anesthesia Medicare rate and no more than $200.

Provider-network consistency

Labs, Physician Group, DME, and Urgent Care compute has_consistent_pct_of_medicare_rate at the provider-network level. The flag is true when there are at least 30 long-rate observations and the 95th-to-5th percentile spread of rate / medicare_rate is less than 0.25.

The decimal tiebreaker is everything

Among 100 payer rates that all score 6.x, the one with the highest CDF (most common rate value) wins. Among validated medical rates (7.x), the one with the highest dollar amount wins. This ensures deterministic selection without arbitrary ordering.