The $160 Million Prescriber

In the entire Medicare Part D prescriber dataset — covering over 1.2 million providers — one name sits at the very top of the cost rankings. Not an oncologist prescribing cutting-edge biologics. Not a large institutional pharmacy. A single emergency medicine physician in Moreno Valley, California: Armaghan Azad, M.D. (NPI: 1639279417), with $160.3M in total drug costs.

That figure demands investigation. Not because it necessarily indicates wrongdoing — but because a number this far outside the norm deserves explanation. This analysis examines what the public data reveals, what plausible explanations exist, and where the data reaches its limits.

The Numbers

Let’s start with the raw statistics, because they are genuinely extraordinary:

To put the volume in context: Dr. Azad’s NPI is associated with 492,011 unique Medicare beneficiaries. That is not a typo. Nearly half a million patients in a single year. The typical emergency medicine provider in the Medicare Part D dataset serves far fewer patients — the specialty median is in the low hundreds.

The cost per beneficiary — $326 — is elevated but not absurd on its own. It’s 4.2x the emergency medicine specialty average of $78. Many individual providers exceed their specialty average by that margin. What makes this case extraordinary is not the per-patient cost but the sheer volume of patients and claims flowing through a single NPI.

The Brand-Name Anomaly

Perhaps the most striking statistic is the brand-name prescribing rate: 99.8% of all drug costs are for brand-name medications. The emergency medicine specialty average is just 1.48%.

$160.2M in brand-name costs versus $116K in generics. That ratio — 99.8% brand — is effectively 100%. This suggests that nearly every prescription associated with this NPI is for the same (or very few) brand-name drug(s). In emergency medicine, where typical prescribing is dominated by short courses of generic antibiotics, pain medications, and anti-nausea drugs, this pattern is remarkable.

What Could Explain This?

Before drawing conclusions, it’s worth understanding the mechanics of Medicare Part D data and how hospital-based prescribing works. Several plausible explanations exist — some mundane, some concerning, all worth considering.

Hospital ER “Incident-To” Billing

The most likely explanation is structural. In many hospital emergency departments, a single supervising physician’s NPI is used as the prescriber of record for all Part D prescriptions written in that facility. This is a feature of how CMS aggregates data, not necessarily how prescribing decisions are made.

If Dr. Azad serves as the medical director or primary attending for a large-volume emergency department — or a network of emergency departments — in Moreno Valley, CA, it is plausible that hundreds of thousands of prescriptions written by dozens of ER physicians, nurse practitioners, and physician assistants are all attributed to his NPI.

Moreno Valley is home to Riverside University Health System Medical Center, a major county hospital and Level II trauma center serving a large, predominantly Medicare-age population in the Inland Empire region. A facility of this size could easily see hundreds of thousands of Part D-eligible patients annually.

The claims-per-beneficiary ratio supports this theory: 1.35 claims per patient. That is very close to 1 — meaning most patients associated with this NPI received a single prescription. This is consistent with an ER encounter pattern: patient arrives, receives one drug (likely administered in the ER), and the claim is filed under the attending NPI. It would be inconsistent with a physician maintaining ongoing relationships with patients.

A Single Expensive Drug

The 99.8% brand-name rate, combined with the near-1:1 claims-to-beneficiary ratio, strongly suggests that one specific brand-name drug dominates this NPI’s prescribing profile. While the CMS public use file does not break down costs by individual drug at the provider level in the summary dataset, the pattern is consistent with a single expensive injectable or biologic being administered to nearly every patient.

In an emergency department context, several expensive brand-name drugs are commonly administered:

• Biologics and monoclonal antibodies — Some ER-administered drugs (e.g., for allergic reactions or acute conditions) can cost hundreds of dollars per dose
• Antivirals — Brand-name antiviral treatments for influenza or COVID-19 can carry high per-dose costs
• Specialty injectables — Certain drugs administered in the ER setting are only available as brand-name products with no generic equivalent

If the average cost per claim is roughly $241 ($160.3M / 666,251 claims), that places the per-prescription cost squarely in the range of an expensive brand-name injectable — not an outrageously priced specialty drug, but well above generic alternatives.

The 340B Factor

One factor that could amplify the dollar figures is the 340B Drug Pricing Program. Under 340B, qualifying hospitals (including safety-net hospitals and county medical centers) purchase outpatient drugs at deeply discounted prices — often 25-50% below average wholesale price. However, they may bill Medicare and other payers at standard rates.

If Dr. Azad is affiliated with a 340B-eligible hospital, the $160.3M in drug costs reported in the CMS data reflects Medicare’s payment rate, not the hospital’s actual acquisition cost. The facility could be purchasing these drugs for a fraction of that amount. The “spread” between acquisition cost and reimbursement is a feature of the 340B program — it is legal and intentional, designed to help safety-net hospitals fund care for uninsured and underinsured patients.

This is relevant context because it means the $160.3M figure may overstate the true economic cost. The drugs were likely acquired for far less. Whether this constitutes efficient use of Medicare funds is a policy question, not a fraud question.

Riverside University Health System Medical Center, as a county-operated safety-net hospital, would almost certainly qualify for 340B pricing. This makes the 340B explanation highly plausible for this specific case.

What the Data Can’t Tell Us

The CMS Medicare Part D Public Use File is a powerful dataset, but it has significant limitations that are directly relevant to interpreting cases like this one:

• No institutional context. The data attributes prescriptions to individual NPIs (this is an individual NPI, entity code “I”), but does not indicate whether the prescribing occurred in an institutional setting. There is no way to distinguish “Dr. Azad personally chose this drug for this patient” from “this prescription was written in a facility where Dr. Azad’s NPI is the default prescriber of record.”
• No drug-level detail in the summary. The provider summary file shows aggregate costs and claim counts, but does not identify which specific drugs were prescribed. The drug-level detail files exist but are structured differently and subject to additional suppression rules.
• No 340B indicator. The dataset does not flag whether a provider or facility participates in the 340B program.
• No clinical context. We cannot determine whether the prescribing was clinically appropriate. A $326 cost per patient could be entirely justified if the drug is medically necessary and no cheaper alternative exists.
• No information on who actually wrote the prescription. In a hospital setting, the prescribing NPI may be a supervising physician who did not personally evaluate the patient or select the drug.

These limitations are not unique to this case. They affect the interpretation of every high-volume provider in the dataset. But they are especially important here because the numbers are so large that they invite assumptions the data cannot support.

Other Notable Facts

A few additional data points that help round out the picture:

• No opioid prescribing. This NPI has no opioid claims in the dataset. Whatever is driving the cost, it is not controlled substances.
• No exclusions. Dr. Azad does not appear on the OIG exclusion list or any federal sanction databases we checked.
• Average patient age: 74. Consistent with a Medicare population being seen in an emergency department.
• Risk score: 9 (low). Our risk model assigns a score of just 9 out of 100, because the model adjusts for specialty context. Emergency medicine providers are expected to have high volume and certain prescribing patterns. The raw cost is extraordinary, but the model recognizes that this is a volume outlier, not necessarily a behavioral one.
• Out of 44,021 emergency medicine providers in the dataset, this NPI is the absolute #1 by total cost. The gap between #1 and #2 is enormous.

Why This Matters

This case is less about one doctor and more about how we interpret Medicare data. The CMS public use files are an extraordinary resource for understanding prescription drug spending. But they were designed for aggregate analysis, not individual provider evaluation. When we use them to rank individual providers, we inherit structural distortions.

A physician whose NPI serves as the institutional prescriber for a major hospital ER will always appear as an outlier in provider-level data. That does not make the data wrong — the prescriptions are real, the costs are real, and Medicare did pay that money. But the attribution to a single individual can be misleading.

This is why OpenPrescriber presents data in context rather than as a leaderboard of suspicion. High cost alone is not evidence of waste or fraud. High volume alone is not suspicious. Even a 99.8% brand-name rate might be explained by a single drug with no generic equivalent being the standard of care in an ER setting.

What cases like this do highlight is the need for better data transparency. If CMS distinguished institutional prescribing from individual prescribing — or flagged 340B participation — analyses like this one could be far more precise. Until then, we work with what we have and try to be honest about its limitations.

The Bigger Picture

Dr. Azad’s NPI accounts for roughly 0.06% of all Medicare Part D drug spending by itself. Whether that spending represents efficient emergency care delivery, a 340B program working as intended, or something that warrants closer examination is a question the data alone cannot answer.

What we can say is this: the Medicare system has a $160.3M data point that is, at minimum, worth understanding. We’ve presented what the data shows. The interpretation — and any action — belongs to those with access to the full clinical and institutional context.

Methodology Note

All figures in this analysis are derived from the CMS Medicare Part D Prescribers by Provider public use file for calendar year 2023, the most recent available at time of publication.

Specialty averages are calculated across all providers in the Emergency Medicine specialty classification. Risk scores reference OpenPrescriber’s risk scoring methodology, which adjusts for specialty, geography, and patient mix to reduce false positives from structural billing patterns.

Cost figures reflect total drug cost as reported by CMS, which includes ingredient cost, dispensing fees, and sales tax. They do not reflect manufacturer rebates, 340B discounts, or other post-point-of-sale adjustments.