How NADAC is calculated

In the week CMS made effective on Oct. 21, 2026, the federal government changed its published pharmacy cost for 28,712 drug codes at once. The NADAC comparison file¹ on data.medicaid.gov lists every one, and each carries the same reason: "Survey Rate." Among them, the average price a pharmacy pays for a generic atorvastatin 20 mg tablet sat a little under 3 cents; GxRadar's monthly series from the 2026 NADAC file² puts it at $0.02825 for October.

NADAC⁶, the National Average Drug Acquisition Cost, is a national average of what retail community pharmacies pay to buy a drug. Many state Medicaid programs use it to set the acquisition cost they pay pharmacies, a payment standard CMS adopted in its 2016 covered outpatient drugs rule⁷, and analysts use it as the closest public proxy for generic acquisition prices; our explainer on what NADAC measures covers those uses.

The weekly swing of nearly 29,000 rates is the visible end of a monthly invoice survey, a set of cleaning rules and a pharmacist review. The Centers for Medicare & Medicaid Services publishes the method in a methodology document³, most recently revised effective December 2024.

Step 1. Draw the monthly sample

Every month a CMS contractor running the Survey of Retail Prices⁸ draws a random sample of independent and chain retail community pharmacies from all states, balanced to match the wider population on two dimensions, chain versus independent and rural versus urban. How that draw plays out is the subject of our piece on how the NADAC survey picks its pharmacies. In a December 2025 supporting statement⁴ filed with the federal paperwork office, CMS put the universe at 62,499 retail community pharmacies as of July 1, 2023.

Closed-door pharmacies and pharmacies that designate themselves as specialty pharmacies are excluded. A selected pharmacy is kept out of the draw for the following three months.

Step 2. Collect invoices

Letters go out asking for the previous month's purchase invoices. Pharmacies can respond themselves or authorize their wholesaler to send the data. Only four fields feed the calculation: the National Drug Code, the unit price paid, the invoice date and the quantity bought.

Participation is voluntary. A CMS overview of NADAC operations⁵ reported in 2017 that about 2,500 pharmacies were selected each month and 450 to 600 responded, roughly 18 percent to 24 percent. Those responses produced 200,000 to 300,000 usable cost observations a month. The current methodology does not publish a newer response rate.

Two kinds of purchase are filtered out at this stage. Invoices for drugs bought under the 340B program are excluded. Discounts or rebates that do not appear on the invoice line are not counted, so the result reflects invoice prices only, a point that matters when NADAC is compared with other price benchmarks.

Steps 3 and 4. Group the drugs, then clean the data

Drugs are grouped by active ingredient, strength, dosage form and route. All the generic NDCs for a 20 mg tablet of one molecule fall into one group regardless of manufacturer or package size. Package size becomes a separate dividing line only for drugs priced per milliliter or per gram where costs differ by package. Each group can carry two rates, one for brand NDCs (classified B) and one for generic NDCs (classified G).

Before any averaging, the methodology³ applies a set of filters.

• Only invoices from the most recent 12 months count.
• NDCs must be valid and active. Codes with an obsolete date more than three years past are dropped, as are codes CMS has terminated.
• Each pharmacy contributes one observation per NDC, its most recent purchase. If two purchases share a date, the lower cost is used.
• For generics, observations must come from the last six months. For brands, they must come from the last 12 months or since the most recent list price change, whichever is shorter.

CMS sets no hard minimum count, but it uses at least five observations where costs are closely aligned.

Step 5. Average and review

The rate is a simple average of the per-unit costs that survive cleaning. Observations more than two standard deviations from the mean are removed by default, though reviewers can put them back. A team of pharmacists and analysts then reviews every drug before publication.

What comes out is one national figure, unweighted by volume, region or pharmacy type. CMS reports in the same document³ that generic rates carry an average margin of error of 2.4 percent at 95 percent confidence, with every generic rate under 10 percent. Brand rates are tighter, averaging 0.5 percent.

Step 6. Update weekly between surveys

Brand and generic rates move differently between surveys. Brand rates track published list prices, according to the methodology³: if a manufacturer raises its wholesale acquisition cost 5 percent, the brand NADAC rises 5 percent in the next weekly file, and later surveys check that the relationship holds. Generic rates change mid-month only when a pharmacy or other party contacts the NADAC help desk and documents a price change, and the next monthly survey replaces them.

If a rate cannot be recalculated, it is carried forward for up to 12 months. After 12 straight months without an update it is removed, and it returns once enough data exist to calculate it again.

Starting with the December 2024 monthly updates, the generic rate became a three-month moving average of the three most recent monthly averages. CMS said the goal was to damp volatility caused by swings in survey participation, the swings our account of the 2024 NADAC swings traces, while keeping rates responsive. Brand rates are not averaged this way. CMS calls the approach temporary and in effect until further notice, and has said it will seek stakeholder input on a permanent fix.

Explanation codes

Every row in the NADAC file carries an explanation code saying how the rate was set.

A code 4 deserves attention. It means the rate is not fresh, and the market may have moved since it was last set.

What the comparison file shows about timing

Alongside each weekly NADAC file, CMS posts a comparison file listing only the NDCs whose rate changed. New and terminated NDCs are left out. Each row shows the old and new per-unit rate, the percent change, the B or G classification, the effective date and a primary reason: survey rate, WAC adjustment, help desk adjustment, brand or generic change, or rate group change.

Counting rows by effective date shows how concentrated the changes are. GxRadar queried the comparison dataset¹ for every day from Aug. 15 to Oct. 28, 2026.

Nearly all movement arrives once a month, in the survey week. Between surveys, the period produced 23 changes in total, all of them brand list-price adjustments or rate group changes, and no help desk adjustments. Of the 28,712 changes effective Oct. 21, 26,939 were generic rates and 1,773 brand. Rates rose for 14,808 codes and fell for 13,729, and the median change was 0.04 percent. Only 39 codes rose by half or more, and 11 fell by half or more.

Those outliers are what make the file useful to anyone tracking prices; CMS says large moves also prompt it to research a rate further before it stands. GxRadar's price pages chart the monthly series for 47 generics, and how pharmacies get paid explains where NADAC enters a claim. Medicaid's use of acquisition cost rests on 42 CFR 447.512¹⁰, which caps payment for most drugs at the lower of actual acquisition cost plus a professional dispensing fee or the pharmacy's usual and customary charge. Section 447.502⁹ defines actual acquisition cost as the state agency's determination of the prices pharmacies actually paid, which is the role NADAC fills. If the pattern of the past three months holds, the next survey batch will reach the comparison file in the second half of November.