Risk Scoring Methodology

Overview

OpenPrescriber uses a unified multi-factor risk scoring model that assigns each provider a score from 0–100. The model combines two complementary approaches:

1. Specialty-adjusted peer comparison — How far does this provider deviate from others in their specialty? (z-scores)
2. Population percentile analysis — Where does this provider fall in the national distribution?

This dual approach avoids the critical flaw of universal thresholds. A pain management specialist prescribing 50% opioids is normal for their specialty. A family doctor at 50% is 13 standard deviations above their peers — and our model reflects that.

Scoring Components (Max 100 Points)

Component 1: Specialty-Adjusted Opioid Rate (0–25 points)

We compute z-scores by comparing each provider's opioid prescribing rate to their specialty's mean and standard deviation. This is computed across 110 specialties with at least 10 providers and a meaningful standard deviation (>0.5%).

• >5 standard deviations: 25 points — extreme outlier vs peers
• >3 standard deviations: 18 points — very high vs peers
• >2 standard deviations: 10 points — high vs peers

This is the most important and unique component. It answers the question: “Does this provider prescribe opioids at an abnormal rate compared to doctors in their same specialty?”

Component 2: Population Percentile Opioid Rate (0–15 points)

Independent of specialty, we check where the provider falls in the national distribution of opioid prescribing rates among all 433,324 providers with opioid claims.

• 99th percentile (opioid rate >70.6%): 15 points
• 95th percentile (>50.3%): 10 points
• 90th percentile (>37.2%): 5 points

Requires ≥11 beneficiaries to avoid small-sample noise.

Component 3: Cost Outlier (0–10 points)

Flags providers whose cost-per-beneficiary is both high in absolute terms (population percentile) AND high relative to their specialty peers (z-score). This dual requirement reduces false positives from legitimately expensive specialties like oncology.

• 99th percentile + >2σ vs peers: 10 points
• 95th percentile + >1.5σ vs peers: 6 points
• 95th percentile alone: 3 points

Component 4: Brand-Name Preference (0–8 points)

Providers prescribing far more brand-name drugs than their specialty peers, where generics are available. Only flagged when the z-score exceeds 2–3 AND the absolute brand percentage exceeds 30–50%.

Component 5: Long-Acting Opioid Rate (0–8 points)

Long-acting opioids carry higher diversion and abuse potential. We flag providers whose long-acting opioid share exceeds 3 standard deviations above their specialty mean.

Component 6: Elderly Antipsychotic Prescribing (0–10 points)

CMS tracks antipsychotic prescribing to patients 65+ as a quality concern. Providers with >50 claims receive 10 points; >20 claims receive 5 points.

Component 7: Opioid + Benzodiazepine Co-Prescribing (0–8 points)

The FDA issued a Black Box Warning about combining opioids and benzodiazepines due to life-threatening respiratory depression. We analyze the full 11.9 million-row provider-drug dataset to identify 6,149 providers who prescribe both drug classes.

Component 8: OIG Exclusion Match (0–20 points)

We cross-reference every provider NPI against the Office of Inspector General's List of Excluded Individuals/Entities (LEIE) — individuals convicted of healthcare fraud, patient abuse, or related felonies. We matched 372 excluded providers who are still actively prescribing in Medicare Part D.

Component 9: Low Drug Diversity (0–6 points)

Providers who prescribe very few unique drugs (≤5–10) while also prescribing opioids are flagged. A legitimate general practitioner prescribes dozens of different medications. A provider prescribing only 6 drugs — most of them opioids — is a red flag for a “pill mill” operation. Requires opioid rate >10% and ≥100 claims.

Component 10: High Fills Per Patient (0–5 points)

Providers whose claims-per-beneficiary ratio exceeds the 95th or 99th percentile (national threshold: >15–20 fills/patient/year). Extremely high fill rates may indicate patients receiving excessive prescriptions or potential doctor shopping facilitation.

Volume Multiplier (×1.0–1.15)

High-volume providers (>5,000 claims) receive a 15% multiplier on their raw score, because the same risky patterns at scale affect more patients. Providers with >2,000 claims get a 5% multiplier.

Minimum Thresholds

To reduce noise from low-volume statistical anomalies, we require:

• ≥100 claims for inclusion in scoring (eliminates borderline providers)
• ≥11 beneficiaries (CMS suppression threshold)
• Low-volume penalty: Providers with <200 claims receive reduced points for population percentile flags (60% cap)

Risk Levels

What Makes This Unique

Most fraud detection tools use universal thresholds — e.g., “flag anyone with >20% opioid rate.” This approach generates massive false positives for pain management, anesthesiology, and other specialties where high opioid prescribing is clinically appropriate.

OpenPrescriber's specialty-adjusted model:

• Computes baseline statistics for 110 medical specialties (mean, standard deviation, 50th/90th/95th percentiles)
• Measures each provider against their specialty peers, not the general population
• Requires deviations in multiple independent dimensions for high scores
• Analyzes 11.9 million prescription records for drug combination risks
• Cross-references 8,301 LEIE excluded NPIs against active prescribers

Limitations

• Specialty misclassification: If a provider's CMS-listed specialty doesn't match their actual practice, peer comparison may be inaccurate
• Practice setting: Hospice, palliative care, and addiction treatment practices have legitimately high opioid rates that may appear as outliers even within their specialty
• Single-year snapshot: We analyze 2023 data only; providers may have changed practice patterns since then
• CMS data suppression: Providers with <11 beneficiaries have certain metrics suppressed by CMS for privacy
• No clinical context: We cannot see patient diagnoses, treatment plans, or medical necessity. A high opioid rate may be entirely appropriate for the patient population served
• Minimum threshold: We require ≥50 claims for scoring — very low-volume providers are excluded

Data Sources

• CMS Medicare Part D Prescribers - by Provider (2023): 1,380,665 provider records with opioid, brand/generic, cost, and demographic data
• CMS Medicare Part D Prescribers - by Provider and Drug (2023): 11,935,116 prescription records across 1,702 unique drugs — used for drug combination and diversity analysis
• OIG LEIE (Updated Monthly): 82,715 excluded individuals, 8,301 with valid NPIs — cross-referenced for active Medicare prescribers
• CMS Historical Files (2019–2022): 5 years of trend data for year-over-year analysis

All data is publicly available from data.cms.gov and oig.hhs.gov.