Payerset Trusted Price Algorithm

Overview

The Payerset Trusted Price Algorithm selects and scores healthcare pricing data from payer Transparency in Coverage (TiC) files. Given the scale of data, the algorithm operates within a sparse matrix architecture that processes data in partitioned chunks, aggregating across plans and prioritizing rate records based on the various criteria outlined below.

The algorithm produces a single trusted rate for each payer per NPI + billing code combination, along with benchmark comparisons and a confidence score that reflects data quality and consistency.

Three-Track Processing

Provider NPIs are processed through one of three tracks based on entity type and data availability:

NPI Track	Entity Type	Key Criteria
Individuals	Type 1 NPI	Single practitioners; taxonomy filtering viable; tighter medicare bands
Organizations	Type 2 NPI (no hospital match)	Clinics, groups, ASCs; no taxonomy filtering other than hospital; wider medicare bands
Hospitals	Type 2 NPI (hospital match)	Matched against hospital MRF dataset; medicare IP/OP benchmark

Benchmark Bands

Medicare Benchmark

Rates are scored based on their ratio to locality-adjusted Medicare rates. Each track has different acceptable bands reflecting the natural variation in that provider type.

Individuals (Type 1 NPI)

Medicare Ratio	Confidence Level
75% - 250%	HIGH
50% - 75% OR 250% - 350%	MEDIUM
Outside above ranges	LOW

Organizations (Type 2 NPI, no hospital taxonomy match)

Medicare Ratio	Confidence Level
85% - 350%	HIGH
65% - 85% OR 350% - 500%	MEDIUM
Outside above ranges	LOW

Hospitals (Type 2 NPI, with hospital taxonomy match)

Medicare Ratio	Confidence Level
100% - 400%	HIGH
75% - 100% OR 400% - 500%	MEDIUM
Outside above ranges	LOW

Hospital MRF Benchmark (Hospitals Only)

For NPIs matched to our hospital pricing dataset, the selected rate is compared against the hospital's published MRF rate.

Hospital Ratio	Score
80% - 120%	HIGH
50% - 80% OR 120% - 150%	MEDIUM
Outside above ranges	LOW

Note: The payer TiC rate is always prioritized over the hospital MRF rate for the final rate_selected value. Hospital MRF data is used only as a validation benchmark, not as a source of truth, due to known accuracy issues in hospital-published data.

---

Rate Selection Priority

Within each plan, rates are prioritized using a scoring system. Lower priority scores are preferred.

Negotiated Type Priority (Base Score)

Negotiated Type	Base Score
negotiated	100
fee schedule	200
derived	300
percentage	400
other/unknown	500

Billing Class Priority (Added to Base)

For Facilities (Type 2 NPI):

Billing Class	Score
institutional	+10
professional	+20

For Individuals (Type 1 NPI):

Billing Class	Score
professional	+10
institutional	+20

Place of Service Priority (Added to Base)

For Facilities:

Service Code	Meaning	Score
22	Outpatient	+1
(empty)/null	All/unspecified	+2
11	Office	+3
21	Inpatient	+4
other		+5

The reason 22 is prioritized here is because it is atypical for an outpatient rate to exist for a facility, usually indicating that there is an explicit exception that provides an outpatient rate, so we prioritize that rate over the more typical office rate. This is uncommon, but useful.

For Individuals:

Service Code	Meaning	Score
11	Office	+1
(empty)/null	All/unspecified	+2
22	Outpatient	+3
21	Inpatient	+4
other		+5

Selection Logic

1. For each plan, calculate priority score for all rates matching the NPI + billing code

2. Select rate(s) with the lowest (best) priority score within that plan

3. Aggregate selected rates across all plans (tracking min, max, sum, count)

4. When merging across plans, lower priority scores replace higher ones; equal priorities merge aggregates

---

Cross-Plan Aggregation

As rates are processed across multiple plans for the same payer and network, the algorithm maintains:

• rate_min - Minimum rate observed across plans

• rate_max - Maximum rate observed across plans

• rate_sum - Sum of all rates (for average calculation)

• rate_count - Number of plans contributing rates

The spread ratio (rate_max / rate_min) serves as a consistency indicator:

Spread Ratio	Interpretation	Confidence Impact
< 1.5	Tight agreement	HIGH
1.5 - 3.0	Moderate variance	MEDIUM
> 3.0	Wide disagreement	LOW

The rate count also affects confidence:

Rate Count	Interpretation	Confidence Impact
5+ plans	Strong support	HIGH
2-4 plans	Moderate support	MEDIUM
1 plan	Single source	LOW

---

Confidence Calculation

The final confidence score combines multiple factors:

Factor	Weight	Possible Scores
Medicare benchmark	Primary	HIGH / MEDIUM / LOW / NULL
Hospital benchmark (hospitals only)	Primary	HIGH / MEDIUM / LOW / NULL
Spread ratio	Secondary	HIGH / MEDIUM / LOW
Rate count	Secondary	HIGH / MEDIUM / LOW
Negotiated type	Tertiary	HIGH (negotiated) / MEDIUM (fee schedule) / LOW (derived/percentage)

Confidence Resolution

The final confidence level is determined by the lowest score among primary factors, adjusted down if secondary factors indicate problems:

confidence = min(
    medicare_benchmark_score,     -- NULL treated as MEDIUM
    hospital_benchmark_score,     -- NULL ignored for non-hospitals
    spread_score,
    count_score
)

-- Negotiated type can only lower confidence, not raise it
if negotiated_type in ('derived', 'percentage'):
    confidence = min(confidence, MEDIUM)

Final Confidence Values: HIGH, MEDIUM, LOW

---

Input Filters

The following filters are applied before rate selection. MS-DRG will be filtered only for hospitals.

Field	Filter
billing_code_type	IN ('CPT', 'HCPCS','MS-DRG')
billing_code_modifier	IN ('00', '', ' ', ' ') — effectively unmodified codes only**
negotiation_arrangement	= 'ffs' (fee-for-service)
npi	Length = 10, starts with '1' or '2'
service_codes	Contains '11', '21', '22', or is empty/null

---

** As some payers place multiple modifiers in an array, the filtering is slightly different but effectively achieves the same result.

Taxonomy Filtering (Individuals Only)

For Type 1 NPIs (individuals), taxonomy codes from NPPES are used to validate that the billing code is appropriate for the provider's specialty. This categorization is based on substantial research and consultations with clinicians.

Rationale: Individual provider taxonomies are relatively stable (a cardiologist remains a cardiologist), whereas organization taxonomies are frequently outdated, incorrect, or incomplete.

Implementation: A mapping of taxonomy classifications to valid billing code ranges is applied. Rates for codes outside the provider's expected scope are deprioritized (not filtered) by adding a penalty to the priority score.

---

Output Schema

Each output record contains:

Field	Type	Description
npi	VARCHAR	10-digit National Provider Identifier
billing_code	VARCHAR	CPT or HCPCS code
negotiated_type	VARCHAR	Type of rate (negotiated, fee schedule, derived, percentage)
plan_network	VARCHAR	Payerset-created network
billing_class	VARCHAR	professional or institutional
service_codes	VARCHAR	Place of service codes that were selected
entity_type	VARCHAR	Individual, Organization, or Hospital
taxonomy_grouping	VARCHAR	NPPES taxonomy grouping (if available)
taxonomy_classification	VARCHAR	NPPES taxonomy classification (if available)
taxonomy_specialization	VARCHAR	NPPES taxonomy specialization (if available)
rate_selected	DOUBLE	The trusted rate (average across contributing plans)
rate_min	DOUBLE	Minimum rate observed across plans
rate_max	DOUBLE	Maximum rate observed across plans
rate_avg	DOUBLE	Average rate across plans
rate_count	INT	Number of plans contributing to this rate
medicare_benchmark	DOUBLE	Locality-adjusted Medicare rate (NULL if unavailable)
medicare_ratio	DOUBLE	rate_selected / medicare_benchmark (NULL if no benchmark)
hospital_benchmark	DOUBLE	Hospital MRF rate (NULL if not a matched hospital)
hospital_ratio	DOUBLE	rate_selected / hospital_benchmark (NULL if no benchmark)
priority_score	INT	Internal priority score (lower = higher priority)
confidence	VARCHAR	HIGH, MEDIUM, or LOW

Revision History

Version	Date	Changes
1.0	2026-01	Initial algorithm specification