Narrow therapeutic index drugs and the tighter test for their generics

In May 2026, FDA revised its bioequivalence recommendations for generic warfarin, a blood thinner, first issued in December 2012. The revision kept the core of the original. Any company seeking approval must run two studies, one fasting and one fed, in which each healthy volunteer takes the generic twice and the reference product twice, according to the warfarin guidance². The design lets FDA compare average blood levels and also how much each product varies within the same person.

That same month, FDA updated its tacrolimus capsule guidance and published a guidance on statistical approaches to bioequivalence¹ that describes the statistical method for NTI drugs.

Warfarin and tacrolimus belong to a small group FDA calls narrow therapeutic index, or NTI, drugs. For most generics, FDA applies the standard 80 to 125 percent bioequivalence test. For NTI drugs it adds a tighter test that narrows as the brand gets more consistent.

What FDA means by narrow therapeutic index

FDA's May 2026 statistical guidance¹ defines NTI drugs as "those drugs where small differences in dose or blood concentration may lead to serious therapeutic failures and/or adverse drug reactions that are life-threatening or result in persistent or significant disability or incapacity."

FDA does not keep one master list. FDA scientists wrote in a 2015 paper in the AAPS Journal⁹ that FDA "does not currently provide a list of NTI drugs," while noting that digoxin, lithium carbonate, phenytoin, tacrolimus, theophylline and warfarin "are usually considered NTI drugs by many experts." In practice FDA makes the call drug by drug in its product-specific guidances, each with an explanation section. For warfarin, the 2026 guidance² lists five reasons, among them that effective and toxic concentrations are close together, that the drug is monitored with blood tests, and that doses are adjusted in steps of 10 to 25 percent.

Which drugs FDA has labeled NTI

GxRadar reviewed FDA product-specific guidances for drugs often described as NTI. The table lists those whose guidance states that FDA has concluded the drug is an NTI drug, with GxRadar's count of generic applications for each ingredient that are listed as marketed in FDA's Drugs@FDA data¹².

This list is not exhaustive. FDA's product-specific guidances number in the thousands, and GxRadar checked only these drugs and dosage forms.

How the tighter test works

Standard bioequivalence asks whether the 90 percent confidence interval for the generic-to-brand ratio sits inside 80.00 to 125.00 percent. For NTI drugs, FDA's statistical guidance¹ calls for a fully replicated crossover study analyzed three ways. The first two are run on both total exposure (AUC) and peak level (Cmax).

1. A reference-scaled test, in which the allowed difference between generic and brand shrinks or grows with the brand's own within-subject variability. FDA fixes the scaling constants at 0.9 and 0.1.
2. The standard 80.00 to 125.00 percent limits, which still apply.
3. A comparison of variability, in which the upper bound of the ratio of the generic's within-subject standard deviation to the brand's must fall below 2.5.

Working through FDA's formula shows what the scaled limits imply for the point estimate. The constants were chosen so that a brand with a within-subject standard deviation of 0.10, roughly a 10 percent coefficient of variation, gets limits of 90.00 to 111.11 percent. A more consistent brand gets narrower limits.

These figures are GxRadar's calculation from the constants in the guidance. The actual test is run on a statistical bound from study data, so a product needs some margin inside these limits.

The approach grew out of an April 2010 meeting of an FDA advisory committee, which voted 11 to 2 that the standard limits were not sufficient for NTI drugs, with members commenting that the limits "should perhaps be narrower (90–111%)," according to the 2015 paper⁹. The same paper reported that Health Canada had tightened its AUC limits for critical-dose drugs to 90.0 to 112.0 percent and that the European Medicines Agency applies 90.00 to 111.11 percent in specific NTI cases. FDA's authors chose scaling instead, because fixed narrow limits sharply cut the chance that even an identical copy would pass when variability is moderate. In their simulations, a product compared with itself passed fixed 90.00 to 111.11 percent limits only 27.69 percent of the time when the within-subject standard deviation was 0.25.

Levothyroxine's extra layer

Levothyroxine tablets have several brand applications that each serve as a reference listed drug. FDA's levothyroxine guidance⁴ says a separate bioequivalence study "must be conducted against the highest strength of each RLD product for which a sponsor wishes its product to receive an 'AB' rating." That is why levothyroxine products carry numbered codes such as AB1 and AB2, explained in our piece on therapeutic equivalence codes. The history of the drug's market is in levothyroxine and the price of a daily pill.

Where state law comes in

FDA approval decides whether a generic is rated equivalent. Whether a pharmacist may swap one maker's product for another's is set by states, as our piece on the pharmacist's role in substitution describes. North Carolina has an NTI-specific rule. Under N.C. Gen. Stat. 90-85.28(b1)¹⁰, a prescription for an NTI drug "shall be refilled using only the same drug product by the same manufacturer that the pharmacist last dispensed," unless the prescriber is notified first and both prescriber and patient give documented consent. Our guide to state generic substitution laws covers the broader rules.

What the price data show

NTI status has not kept these drugs expensive. GxRadar's analysis of CMS NADAC files¹¹ through October 2026 shows what pharmacies paid per tablet for three of them. Warfarin is one of the seven drugs in GxRadar's basket that cost more than in January 2014, a story told in our look at warfarin and clopidogrel prices.

Questions to bring to the pharmacy counter

A refill of an NTI drug can come from a different manufacturer and look different, as our piece on why pills change color explains. Questions people sometimes put to a pharmacist or prescriber include whether the manufacturer changed since the last fill, whether their state has a rule about switching makers for that drug, and whether the prescriber wants to know when the maker changes. FDA's warfarin guidance, in its May 2026 revision, remains a draft.