Abuse-Deterrent Opioids-Evaluation and Labeling

Administrative guidance, not enacted legislation. This captured publisher edition preserves the listed document and attachments; later developments are not independently consolidated.

Abuse-Deterrent Opioids —
Evaluation and Labeling
Guidance for Industry
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
Clinical Medical
April 2015

Abuse-Deterrent Opioids —
Evaluation and Labeling
Guidance for Industry
Additional copies are available from:
Office of Communications
Division of Drug Information, WO51, Room 2201
10903 New Hampshire Ave.
Silver Spring, MD 20993-0002
Phone: 301-796-3400; Fax: 301-847-8714
druginfo@fda.hhs.gov
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htm
U.S. Department of Health and Human Services
Food and Drug Administration
Center for Drug Evaluation and Research (CDER)
Clinical Medical
April 2015

Contains Nonbinding Recommendations
TABLE OF CONTENTS
I. 	INTRODUCTION............................................................................................................. 1
II. 	BACKGROUND ............................................................................................................... 1
III. 	ABUSE-DETERRENT PRODUCTS .............................................................................. 3
IV. 	PREMARKET STUDIES ................................................................................................ 4
A. 	Laboratory Manipulation and Extraction Studies (Category 1) ............................................... 6
B. 	Pharmacokinetic Studies (Category 2)......................................................................................... 8
C. 	Clinical Abuse Potential Studies (Category 3) ............................................................................ 9
1. Blinding.......................................................................................................................................... 10
2. Pre-qualification Phase ................................................................................................................. 10
3. Assessment Phase .......................................................................................................................... 11
4. Subjects .......................................................................................................................................... 11
5. Route of Administration, Dose Selection, Manipulation Mode, and Sample Preparation ............ 12
6. Outcome Measures and Data Interpretation ................................................................................. 12
7. Data Interpretation ........................................................................................................................ 13
8. Statistical Analysis ......................................................................................................................... 14
a. 	Background ............................................................................................................................. 14
b. 	Primary analyses ..................................................................................................................... 15
c. 	Secondary analyses ................................................................................................................. 15
d. 	Multiplicity.............................................................................................................................. 17
V. 	POSTMARKET STUDIES (CATEGORY 4) .............................................................. 17
A. 	Formal Studies ............................................................................................................................. 18
1. General Characteristics ................................................................................................................. 18
2. Study Design Features.................................................................................................................... 19
B. 	Supportive Information ............................................................................................................... 21
VI. 	LABELING ..................................................................................................................... 22
VII. 	ADDITIONAL RESEARCH NEEDS ........................................................................... 25

Contains Nonbinding Recommendations
1
Abuse-Deterrent Opioids — Evaluation and Labeling
Guidance for Industry1
This guidance represents the current thinking of the Food and Drug Administration (FDA or Agency) on
this topic. It does not create any rights for any person and is not binding on FDA or the public. You can
use an alternative approach if it satisfies the requirements of the applicable statutes and regulations. To
discuss an alternative approach, contact the FDA staff responsible for this guidance as listed on the title
page.
I. 	INTRODUCTION
This guidance explains FDA’s current thinking about the studies that should be conducted to
demonstrate that a given formulation has abuse-deterrent properties. The guidance makes
recommendations about how those studies should be performed and evaluated and discusses how
to describe those studies and their implications in product labeling.
This guidance is intended to assist sponsors who wish to develop opioid drug products with
potentially abuse-deterrent properties and is not intended to apply to products that are not opioids
or opioid products that do not have the potential for abuse.
This guidance also does not address issues associated with the development or testing of generic
formulations of abuse-deterrent opioid products. FDA intends to address that topic in one or
more future guidance documents.
In general, FDA’s guidance documents do not establish legally enforceable responsibilities.
Instead, guidances describe the Agency’s current thinking on a topic and should be viewed only
as recommendations, unless specific regulatory or statutory requirements are cited. The use of
the word should in Agency guidances means that something is suggested or recommended, but
not required.
II. 	BACKGROUND
Prescription opioid products are an important component of modern pain management.
However, abuse and misuse of these products have created a serious and growing public health
problem. One potentially important step towards the goal of creating safer opioid analgesics has
1 This guidance has been prepared by the Division of Anesthesia, Analgesia, and Addiction Products, the Office of
Regulatory Policy, the Office of Surveillance and Epidemiology, the Office of Biostatistics, and the Controlled
Substance Staff in the Center for Drug Evaluation and Research (CDER) at the Food and Drug Administration.

Contains Nonbinding Recommendations
2
been the development of opioids that are formulated to deter abuse. FDA considers the
development of these products a high public health priority.
Because opioid products are often manipulated for purposes of abuse by different routes of
administration or to defeat extended-release (ER) properties, most abuse-deterrent technologies
developed to date are intended to make manipulation more difficult or to make abuse of the
manipulated product less attractive or less rewarding. It should be noted that these technologies
have not yet proven successful at deterring the most common form of abuse—swallowing a
number of intact capsules or tablets to achieve a feeling of euphoria. Moreover, the fact that a
product has abuse-deterrent properties does not mean that there is no risk of abuse. It means,
rather, that the risk of abuse is lower than it would be without such properties. Because opioid
products must in the end be able to deliver the opioid to the patient, there may always be some
abuse of these products.
For purposes of this guidance, abuse-deterrent properties are defined as those properties shown
to meaningfully deter abuse, even if they do not fully prevent abuse. The term abuse is defined
as the intentional, non-therapeutic use of a drug product or substance, even once, to achieve a
desirable psychological or physiological effect.2 Abuse is not the same as misuse, which refers to
the intentional therapeutic use of a drug product in an inappropriate way and specifically
excludes the definition of abuse.3 This guidance uses the term abuse-deterrent rather than
tamper-resistant because the latter term refers to, or is used in connection with, packaging
requirements applicable to certain classes of drugs, devices, and cosmetics.4
The science of abuse deterrence is relatively new, and both the formulation technologies and the
analytical, clinical, and statistical methods for evaluating those technologies are rapidly evolving.
Based on the evolving nature of the field, FDA intends to take a flexible, adaptive approach to
the evaluation and labeling of potentially abuse-deterrent products. Methods for evaluating the
abuse-deterrent properties of new molecular entities may have to be adapted based on the
characteristics of those products and the anticipated routes of abuse. The development of an
abuse-deterrent opioid product should be guided by the need to reduce the abuse known or
expected to occur with similar products.
Because FDA expects that the market will foster iterative improvements in products with abuse-
deterrent properties, no absolute magnitude of effect can be set for establishing abuse-deterrent
characteristics. As a result, FDA intends to consider the totality of the evidence when reviewing
the results of studies evaluating the abuse-deterrent properties of a product.
2 Smith S M, Dart R C, Katz N P, et al. 2013. Classification and definition of misuse, abuse, and related events in
clinical trials: ACTTION systematic review and recommendations. Pain, 154:2287-2296.
3 Ibid.
4 FDA’s current Good Manufacturing Practice regulations include tamper-evident packaging requirements. See 21
CFR 211.132. There are also requirements for child resistant “special packaging” under the Poison Prevention
Packaging Act and regulations adopted by the Consumer Protect Safety Commissioner (CPSC) in 16 CFR 1700.

Contains Nonbinding Recommendations
3
As with all NDA products, FDA intends to consider opioids with abuse-deterrent properties
within the context of available therapy. The standard against which each product’s abuse-
deterrent properties are evaluated will depend on the range of abuse-deterrent and non-abuse-
deterrent products on the market at the time of that application.5
Abuse-deterrent properties can generally be established only through comparison to another
product.
FDA encourages additional scientific and clinical research that will advance the development
and assessment of abuse-deterrent technologies.
FDA believes it is critical to address the problem of opioid abuse while seeking to ensure that
patients in pain have appropriate access to opioid products. Moreover, it is important that
opioids without abuse-deterrent properties remain available for use in some clinical settings. For
example, patients in hospice care and with difficulty swallowing may need access to opioid
products that are in solution or that can be crushed.
The following section describes the categories of abuse-deterrent products. The premarket and
postmarket studies that should be performed to assess the impact of a potentially abuse-deterrent
product are discussed in subsequent sections. Finally, information is provided about labeling for
abuse-deterrent products.
III. 	ABUSE-DETERRENT PRODUCTS
Opioid products can be abused in a number of ways. For example, they can be swallowed
whole, crushed and swallowed, crushed and snorted, crushed and smoked, or crushed, dissolved
and injected. Abuse-deterrent technologies should target known or expected routes of abuse
relevant to the proposed product. As a general framework, abuse-deterrent formulations can
currently be categorized as follows:
1. 	Physical/chemical barriers – Physical barriers can prevent chewing, crushing, cutting,
grating, or grinding of the dosage form. Chemical barriers, such as gelling agents, can
resist extraction of the opioid using common solvents like water, simulated biological
media, alcohol, or other organic solvents. Physical and chemical barriers can limit drug
release following mechanical manipulation, or change the physical form of a drug,
rendering it less amenable to abuse.
2. 	Agonist/antagonist combinations – An opioid antagonist can be added to interfere with,
reduce, or defeat the euphoria associated with abuse. The antagonist can be sequestered
and released only upon manipulation of the product. For example, a drug product can be
5 For guidance on the evaluation of abuse potential for purposes of the Controlled Substances Act (CSA), we refer
sponsors to FDA’s draft guidance for industry Assessment of Abuse Potential of Drugs. This guidance is available
at: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM198650.pdf.
FDA guidances are available at http://www.fda.gov/RegulatoryInformation/Guidances/default.htm.

Contains Nonbinding Recommendations
4
formulated such that the substance that acts as an antagonist is not clinically active when
the product is swallowed, but becomes active if the product is crushed and injected or
snorted.
3. 	Aversion – Substances can be added to the product to produce an unpleasant effect if the
dosage form is manipulated or is used at a higher dosage than directed. For example, the
formulation can include a substance irritating to the nasal mucosa if ground and snorted.
4. 	Delivery System (including use of depot injectable formulations and implants) – Certain
drug release designs or the method of drug delivery can offer resistance to abuse. For
example, sustained-release depot injectable formulation or a subcutaneous implant may
be difficult to manipulate.
5. 	New molecular entities and prodrugs– The properties of a new molecular entity (NME)
or prodrug could include the need for enzymatic activation, different receptor binding
profiles, slower penetration into the central nervous system, or other novel effects.
Prodrugs with abuse-deterrent properties could provide a chemical barrier to the in vitro
conversion to the parent opioid, which may deter the abuse of the parent opioid. New
molecular entities and prodrugs are subject to evaluation of abuse potential for purposes
of the Controlled Substances Act (CSA).
6. 	Combination – Two or more of the above methods could be combined to deter abuse.
7. 	Novel approaches – This category encompasses novel approaches or technologies that
are not captured in the previous categories.
IV. 	PREMARKET STUDIES
First and foremost, any studies designed to evaluate the abuse-deterrent characteristics of an
opioid formulation should be scientifically rigorous. Important general considerations for the
design of these studies include the appropriateness of positive controls6 and comparator drugs,
outcome measures, data analyses to permit a meaningful statistical analysis, and selection of
subjects for the study.
The evaluation of an abuse-deterrent formulation should take into consideration the known
routes of abuse for the non-abuse-deterrent predecessor or similar products, as well as anticipate
the effect that deterring abuse by one route may have on shifting abuse to other, possibly riskier
route. For example, if a product is known to be abused using nasal and intravenous routes,
developing deterrent properties for the nasal route in the absence of deterrent properties for the
intravenous route risks shifting abusers from the nasal to the intravenous route, which is
associated with a greater risk for the spread of infectious diseases.
Another concept that should be considered is whether the deterrent effects can be expected to
have a meaningful impact on the overall abuse of the product. For example, immediate-release
(IR) opioid and acetaminophen combination products are predominantly abused using the oral
6 For purposes of this guidance, a positive control is an opioid drug product or drug substance expected to result in a
predictable opioid drug liking effect and has a known potential for, or history of, abuse.

Contains Nonbinding Recommendations
5
route. Demonstrating a deterrent effect by the nasal route may not meaningfully reduce overall
abuse of the product.
FDA is committed to retaining a flexible, adaptive approach to evaluating potentially abuse-
deterrent opioid drug products. This flexibility is intended to permit a sponsor to tailor the
development program to suit the abuse-deterrent characteristics of their product and the routes of
abuse for that product. The adaptive aspect is intended to permit a sponsor to take into
consideration the relevant products on the market at the time they are developing their product,
so that appropriate non-abuse-deterrent and abuse-deterrent comparators can be used. For
example, for some proposed products the appropriate comparator may be a conventional
formulation. However, if there are similar approved products with abuse-deterrent properties
described in labeling, the appropriate comparator should be one of those abuse-deterrent
products.
The following sections describe three categories of premarket studies. Although, in general, any
development program for studying abuse-deterrent technologies should include data from all
three categories of studies, there may be exceptions. For example, a formulation with a
sequestered antagonist may intentionally be formulated not to resist crushing, so testing the
syringeability of the product may not be relevant. In most cases, however, to obtain a full and
scientifically rigorous understanding of the impact of a technology or technologies on a
product’s abuse potential, data from each of the following three categories of premarket studies
are appropriate:
1. 	Laboratory-based in vitro manipulation and extraction studies (Category 1)
2. 	Pharmacokinetic studies (Category 2)
3. 	Clinical abuse potential studies (Category 3)
The results of Category 1 studies may influence the design of Category 2 pharmacokinetic
studies and Category 3 clinical abuse potential studies by suggesting the methods of
manipulation that would yield the greatest release of opioid. The results of Category 2 studies
may influence the need for Category 3 studies of clinical abuse potential and the designs and
goals of these studies. For example, if the extended-release characteristics of an abuse-deterrent
opioid formulation cannot be defeated and the pharmacokinetic profile remains unchanged
following oral or nasal administration of the manipulated product, oral and nasal studies of abuse
potential may not be necessary.
Additional studies (i.e., Category 4 studies) analyze postmarket data to assess the impact of an
abuse-deterrent formulation on actual abuse. Nonclinical drug discrimination studies are useful
in the evaluation of the abuse potential of a drug, but their utility in predicting the impact of
abuse-deterrent properties on human behavior has not been established.7
7 See FDA draft guidance for industry, Assessment of Abuse Potential of Drugs see
http://www.fda.gov/ucm/groups/fdagov-public/@fdagov-drugs-gen/documents/document/ucm198650.pdf.

Contains Nonbinding Recommendations
6
A. 	Laboratory Manipulation and Extraction Studies (Category 1)
The goal of laboratory-based Category 1 studies should be to evaluate the ease with which the
potentially abuse-deterrent properties of a formulation can be defeated or compromised. This
information should be used when designing Category 2 and Category 3 studies. These studies
are critical to the understanding of product characteristics and performance.8
Methodologically, these studies should be designed with knowledge of the physicochemical
properties of the product and the methods available to abusers to manipulate the product and
should be conducted on the to-be-marketed formulation. Sponsors should consider both the
mechanisms by which abusers can be expected to attempt to deliberately overcome the abuse-
deterrent properties of the product as well as the ways that patients may alter the formulation
(unintentionally or intentionally) that change the rate or amount of drug released (e.g., dose
dumping may occur when taking the product with alcohol or when the product is cut, chewed, or
crushed). Testing should provide information sufficient to fully characterize the product’s
abuse-deterrent properties, including the degree of effort required to bypass or defeat those
properties. In some cases, when designing in vitro studies, it may be useful to obtain information
from prescription opioid abusers about how they would manipulate and abuse an abuse-deterrent
product.
In vitro studies should assess various simple and sophisticated mechanical and chemical ways a
drug could be manipulated, such as by (1) defeating or compromising the controlled release of an
opioid from ER formulations for purposes of abuse by different routes of administration; (2)
preparing an IR formulation for alternative routes of administration; or (3) separating the opioid
antagonist, if present, from the opioid agonist, thus compromising the product’s abuse-deterrent
properties. The goal of these studies is to manipulate the product to the point of defeating its
abuse-deterrent properties. Once this goal is achieved, it is no longer necessary to continue
experiments using more sophisticated methods. For example, if 90% of the opioid can be
extracted under a set of conditions in 10 minutes, there is no need to test the same condition for
30 minutes.
The test product should be compared to appropriate comparator products for ease of mechanical
manipulation. The ability to crush, cut, grate, or grind the product formulation using readily
available items such as spoons, cutters, and coffee grinders should be assessed. Particular
attention should be given to particle size distribution following each mode of physical
manipulation because particle size may influence the rate of opioid extraction from manipulated
product. The effect of heat and cold on mechanical manipulation should also be studied.
Extractability and solubility studies should be designed to determine whether any of the
formulation components might be differentially solubilized and extracted, allowing an abuser to
8 This topic has been discussed at meetings of the Anesthetic & Life Support Drugs Advisory Committee and the
Drug Safety & Risk Management Advisory Committee (NDA 022272, OxyContin, May 5, 2008, and September 24,
2009). Additional information on these meetings is available on FDA’s web site at the following location:
http://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/AnestheticAndAnalgesic
DrugProductsAdvisoryCommittee/UCM187082.pdf.

Contains Nonbinding Recommendations
7
bypass the drug’s abuse-deterrent properties. In addition to extraction and solubility studies, an
assessment should be made to determine if free-base opioid can be precipitated from solution by
pH adjustment. After establishing how a product could be manipulated, chemical extraction of
the opioid from the intact and the manipulated product should be assessed and compared to
opioid extraction from the selected intact and similarly manipulated comparator products.
The ease of extracting the opioid from the intact and manipulated product should be determined
using a variety of solvents that are commonly available (e.g., water, vinegar, ethanol,
isopropanol, acetone, mineral spirits) and those that have potentially relevant solvent
characteristics (e.g., pH, polarity, protic vs. aprotic). The effects of time, temperature, pH, and
agitation on solvent extraction should also be determined. For products containing more than
one drug substance, extractability and solubility studies should be designed to determine whether
any of the active ingredients might be differentially solubilized and extracted. Sampling times
should start early (e.g., 30 seconds) and continue until at least 80% of the opioid has been
released, or 12 hours has been reached. The in vitro drug-release characteristics of the intact and
manipulated product should also be compared using a discriminatory and robust dissolution
method.
In addition to the general evaluation of the effects of physical and chemical manipulation on the
product, there are important route-specific data that should be generated, as follows:
• 	For a product with potential for abuse by the nasal route, the particle size distribution
following attempted manipulation by various methods should be established, and the
method that provides the smallest particle size should be used in subsequent studies.
• 	For a product with potential for abuse by smoking, the amount of drug produced by
vaporization at temperatures encompassing the range from the melting point of the active
ingredient to its degradation point should be determined. Appropriate controls, such as
pure active ingredient, both in salt and free-base form should be included in these
assessments.
• 	For a product with potential for abuse by injection, the amount of opioid that can be
obtained in a syringe should be based on studies of intact and manipulated test product
and comparator(s) using small volumes of water (5-10 mL) at room temperature and at
90° C – 95° C with and without agitation. Extraction times should range from 30 seconds
to 30 minutes. The amount of opioid extracted, the volume of solution collected and the
viscosity of the samples should be recorded. The ability to get the sample into a syringe
and expel the sample using needles of various gauges should also be explored.
The following examples illustrate the kinds of outcomes that in vitro studies should evaluate.
1. 	Characteristics of the product by crushing, grinding or melting, or by changing the intact
formulation using other methods that would limit nasal administration of the manipulated
product, and/or that would limit dissolution of the manipulated product and incorporation
into a solvent that could then be injected by intravenous or subcutaneous routes.

Contains Nonbinding Recommendations
8
2. 	Quantity of the opioid extracted from the product following the various methods attempted
that could be used for injection by intravenous or subcutaneous routes and a description of
any barriers resulting from attempts at dissolution for drawing the drug into a syringe.
3. 	Quantity of opioid antagonist released from an agonist/antagonist combination when it is
manipulated for administration by ingestion, nasal administration, or injection.
4. 	Quantity of opioid product following in vitro manipulation of the prodrug.
B. 	Pharmacokinetic Studies (Category 2)
The goal of the clinical pharmacokinetic studies, Category 2, should be to understand the in vivo
properties of the formulation by comparing the pharmacokinetic profiles of the manipulated
formulation with the intact formulation and with manipulated and intact formulations of the
comparator drugs through one or more routes of administration. Even though the same routes of
administration should be studied for the new product and comparators, if specific circumstances
prevent this approach, the study design should be discussed with FDA. The method of
manipulation used for the pharmacokinetic studies should be based on the methods explored
during in vitro testing that can be expected to result in the greatest drug release. The routes of
administration chosen should be relevant to the proposed product, and likely will be based on
what is known about the abuse of similar products. Note that, for some development programs,
it may be preferable to combine measures of pharmacokinetic parameters for Category 3 studies,
in which case separate Category 2 studies may not be necessary.
In general, the pharmacokinetic profile for the oral route of administration should be studied.
Appropriate study subjects for Category 2 studies include healthy volunteers as long as
naltrexone is used to block the pharmacodynamic effects of the opioids.
Depending on the product, it may be important to evaluate the pharmacokinetic profile for the
nasal route of administration as well. For nasal pharmacokinetic studies, it is important to weigh
the risk to the subject based on the excipients in the formulation. Only subjects with a history of
nasal abuse of opioids should be recruited for these studies. As with the oral route of
administration, it may be possible to combine the pharmacokinetic assessment and the
pharmacodynamic assessment in one clinical abuse potential study with sampling for the
pharmacokinetic analysis.
Relevant pharmacokinetic parameters for the opioid drug and any psychoactive metabolites that
should be measured in these studies include the following.
• 	Maximum concentration (C max)
• 	Time to maximum concentration (Tmax)
• 	Area under the curve (AUC 0-t and AUC 0-∞)
• 	Relevant partial AUC, including early time points such as AUC 0-30 minutes or AUC 0-2
hours, the period of time when Cmax is expected
• 	Terminal elimination half-life (T1/2 )

Contains Nonbinding Recommendations
9
Traditional pharmacokinetic study designs should be employed (e.g., crossover designs), and the
results should be analyzed using bioequivalence methods. The rate of rise of drug concentration
should be assessed when possible because it is thought to contribute to differential abuse
potential among drugs, formulations, and routes of administration.9 To support these analyses, it
is important to have specimen collection and analysis time points sufficient to cover the onset,
peak, and offset of the effects of both IR and ER formulations, in both the intact and manipulated
conditions. In addition, these data are necessary to calculate the relevant partial area under the
curve, which should capture the time to maximum concentration of the opioid.
If food and alcohol alter the pharmacokinetic parameters of the formulation, data should be
provided to characterize those effects.10 If food significantly increases systemic exposure of the
intact formulation, the underlying mechanism for the food effect should be established by
assessing whether the effect is based on the drug substance or the formulation and whether the
effect is present with intact product as well as with manipulated product. When food is expected
to increase exposure, subsequent abuse potential studies of the oral route should be conducted in
the fed state to maximize the potential systemic exposure.
In addition to the pharmacokinetic profile of the opioid, for agonist/antagonist combinations , the
pharmacokinetic characteristics of the antagonist should be defined for the intact product as well
as for the manipulated formulation.
As with all clinical studies, adverse events should be collected, and those that can provide
additional insight about the abuse-deterrent effects are especially important. For example, if the
manipulated formulation is abused by snorting, it would be important to assess adverse events
related to intranasal tolerability.
C. 	Clinical Abuse Potential Studies (Category 3)
In addition to their use by FDA to formulate its scheduling recommendation under the CSA for
drug products containing a controlled substance, clinical studies of abuse potential, Category 3,
are important for assessing the impact of potentially abuse-deterrent properties. As discussed in
9 References suggesting that drugs associated with a rapid onset of action are associated with greater abuse potential
include:
Abreu M E, Bigelow G E, Fleisher L, and Walsh S L. 2001. Effect of intravenous injection speed on responses
to cocaine and hydromorphone in humans. Psychopharmacology, 154:76-84.
de Wit H, Bodker B, and Ambre J.1992. Rate of increase of plasma drug level influences subjective responses
in humans. Psychopharmacology, 107:352-358.
de Wit H, Didish S, and Ambre J. 1993. Subjective and behavioral effects of diazepam depend on its rate of
onset. Psychopharmacology, 112: 324-330.
10 FDA has issued a draft guidance on this topic (Assessment of Abuse Potential of Drugs). Once finalized, it will
represent FDA’s current thinking on this topic.

Contains Nonbinding Recommendations
10
FDA’s guidance on that topic,11 the preferred design is a randomized, double-blind, placebo-
controlled and positive controlled crossover study. These studies generally are conducted in a
drug-experienced, recreational user population. The use of a pre-qualification phase (see section
2 below) to identify subjects who can reproducibly distinguish active drug from placebo is a
common enrichment strategy used to improve the power of the study to establish a difference
between treatments.
Additional considerations applicable to clinical abuse potential studies used to assess potentially
abuse-deterrent properties are discussed below. For products that are not susceptible to
manipulation based on Category 1 and 2 testing, study designs for Category 3 testing should be
discussed with FDA.
1. 	Blinding
Clinical studies of abuse potential should use a randomized, double-blind, placebo-controlled
and positive controlled crossover design. Because study subjects are recreational drug users and
familiar with the effects of the drug substances being studied, the double-dummy technique or
other techniques should be used to ensure the blinding of all tests when possible. However,
alternative designs may be suitable when the blinding of the study drug and the positive control
cannot be maintained and treatment by period interactions may lead to sequence effects in a
crossover design. For example, a parallel design may be useful when studying the intranasal
route of administration, where subjects may be able to see the differences in volume or color
between test drug and placebo or positive control, or when it is not possible to create similar
results from manipulation, such as particle size from crushing. In these circumstances, early
discussion with FDA is recommended.
For clinical abuse potential studies in which the subjects will snort test samples, administration
of the samples in a narrow neck, opaque container with a pre-inserted straw may help facilitate
blinding. However, even though subjects might not be able to see the sample, un-blinding may
still occur due to the physical properties of samples with similar particle size distribution. In
some formulations, higher crushed tablet/capsule volume or larger particle size may inhibit
complete intranasal administration thereby contributing to the deterrence effects. To be able to
evaluate these effects, it may be necessary to maintain differences in tablet/capsule volume
between the potentially abuse-deterrent formulation and the comparator. To facilitate blinding
and maintain the crossover design, placebos matched to each of the differing weights or particle
sizes may be useful. The details of the preparation of the samples should be provided in the study
protocol.
2. 	Pre-qualification Phase
The purpose of the pre-qualification phase is to increase the power of a study to detect
differences in the abuse potential of the various formulations of drug and placebo.12 In general,
11 Ibid.
12 An additional advantage of a pre-qualification phase is that it helps familiarize subjects with and train them in the
use of various scales and questionnaires that measure subjective effects.

Contains Nonbinding Recommendations
11
the pre-qualification phase should ensure that subjects can distinguish between placebo and a
conventional IR formulation of the same opioid being developed in an abuse-deterrent
formulation, using the same route of administration as planned for the assessment phase. There
is little value in having subjects unable to distinguish placebo from active drug continue in the
study. The positive control should include a strength that is at least equal to the lowest strength
selected for the assessment during the clinical phase. An important aspect of the pre-
qualification phase is assessing the ability of subjects to tolerate the study dose. If the dose used
in the pre-qualification phase is lower than the lowest strength planned for the assessment phase,
some subjects may not be able to tolerate the higher dose that will be administered in the
assessment phase. Thus, when tolerability may be an issue, particularly if more than one dose is
planned for the assessment phase, a pre-qualification dose that is no lower than the lowest dose
planned may be the most efficient choice to establish that the subject can distinguish active drug
from placebo and can tolerate the study drug in the range to be tested. For example, a 30 mg or
45 mg dose of opioid could be used in the pre-qualification phase when a 30 mg and 60 mg
doses will be assessed in the clinical phase.
Qualifying criteria that help identify subjects with an acceptable placebo response and an
acceptable response for the positive control should be pre-specified in the study protocol. After a
range for an acceptable placebo response is set, a minimum value for the maximum effect (Emax)
for the positive control should be defined. The minimum Emax for the positive control may vary
from measure to measure, and from study to study. However, an acceptable response for the
positive control should not overlap with the acceptable range for placebo response.
3. 	Assessment Phase
The potentially abuse-deterrent product should be compared to a positive control, and the
positive control should be compared to placebo to validate the study. For an IR product with
potentially abuse-deterrent properties, the positive control should be an IR formulation of the
same opioid. For an ER formulation with potentially abuse-deterrent properties, the positive
control could be an IR formulation of the same opioid or an ER formulation of the same opioid.
In general, these studies should include one strength of the positive control which is associated
with high levels of drug liking. However, when assessing drug liking through the intranasal
route, the use of two strengths of the positive control may be helpful to both identify a strength
of the positive control associated with high drug liking scores and to validate the study.
If there are no approved products with the same drug substance, the positive control should be a
drug that, based on pharmacological profile or nonclinical data, can be expected to have similar
pharmacodynamic effects. Selection of the positive control in this setting should be discussed
with FDA.
4. 	Subjects
Studies should be conducted in opioid-experienced, recreational drug users who have experience
with the particular route of abuse being studied. Subjects should generally not be physically
dependent and should not be currently seeking or participating in treatment for drug abuse such
that participating in the study could make them vulnerable to relapse. Depending on the

Contains Nonbinding Recommendations
12
formulation being studied, however, clinical abuse potential studies can be conducted in
physically dependent subjects. For example, if the deterrent product contains an opioid
antagonist, clinical abuse potential studies in a physically dependent population may provide
information not only on the drug liking of the product, but on the ability of the antagonist to
precipitate withdrawal in this population.
Detailed characteristics of the study population with respect to past and current drug use and
abuse should be captured (e.g., drugs abused, drug of choice, duration of abuse or abstinence).
5. 	Route of Administration, Dose Selection, Manipulation Mode, and Sample
Preparation
The selection of the route(s) of administration should be based on epidemiological data showing
that a selected route is a relevant route of abuse. For NMEs, the sponsor should review the
relevant routes of abuse for products similar to the test product and discuss the selected routes
with FDA. For each relevant route of administration, the potentially abuse-deterrent product and
comparator should be manipulated based on the results of Category 1 studies to cause the highest
release of the opioid and the highest plasma levels. The dose of the opioid selected for the study
should be known to produce high levels of liking in non-tolerant opioid-experienced recreational
users.
For studies using the intranasal route of administration, the preparation of the samples is
extremely important. The potentially abuse-deterrent product and comparator study drug should
be produced with similar particle size distribution based on a detailed protocol for the
preparation of the samples, even if different methods are necessary to do so.13 With some
formulations, a high volume of the crushed tablet/capsule or larger particle size may inhibit
complete intranasal administration and, thereby, contribute to deterrence effects. To evaluate
these effects, it may be necessary to maintain differences in tablet/capsule volume between the
potentially abuse-deterrent product and the comparator.
For studies using the intravenous route of administration, the oral formulations may not be safe
for intravenous use depending on the excipients used in the formulation. In place of the
manipulated oral formulation, a solution for injection should be prepared using approved,
commercially available parenteral products when available, or products suitably formulated for
the study. The amount of the opioid and that of the antagonist, when relevant, should be based
on extrapolation from in vitro extraction studies of manipulated solid formulations.
6. 	Outcome Measures and Data Interpretation
In abuse potential studies, the primary method for evaluating the subjective effects of drugs
should be through the use of standardized instruments.
13 Available safety-related information on the use of the various excipients through the intranasal route should be
provided. Additionally, some sponsors have conducted intranasal tolerability studies before the abuse potential
studies to evaluate irritation of the nasal cavity, nasal congestion, and discharge, among other measures.

Contains Nonbinding Recommendations
13
In typical abuse potential studies, several instruments have been used to measure subjective
responses predictive of the likelihood of abuse. These instruments include:
• 	Visual Analogue Scales (VAS) – used for drug liking, good effects, bad effects, and other
drug abuse-related effects
• 	Profile of Mood States
The VAS should be the primary measure for drug liking because it appears to correlate most
directly with potential for abuse. Other measures of particular interest include assessment of
likelihood to take the drug again and assessment of overall drug liking.14
These measures can be assessed using either a unipolar or bipolar scale, and a rationale should be
provided for the choice for a particular scale. In general, FDA recommends using a bipolar scale
for the primary measure of drug liking. Unipolar scales have been used to measure other drug
effects, such as good and bad effects. Regardless of whether a unipolar or bipolar scale is
selected, FDA recommends that for purposes of training subjects, the same scale be used in the
pre-qualification and assessment phases.
7. 	Data Interpretation
For clinical studies of abuse potential conducted on potentially abuse-deterrent opioid drug
products, the primary analysis should be the difference in means of the Emax15 for the primary
measure(s) based on the population of study completers. A statistical analysis plan (SAP) should
be included in the study protocol or submitted as a separate document before un-blinding the
study. The sponsor should provide data and dropout information for non-completers. To ensure
adequate power, the sponsor should take into account that there will be subjects who drop out of
the study early and plan the sample size calculation accordingly. Proper planning should avoid
any need to replace subjects who discontinue without completing the study.
Additional pharmacodynamic measures, including positive subjective effects other than drug
liking (e.g., take drug again, high, overall drug liking) and other subject-rated assessments, are
generally considered secondary endpoints. Other subject-rated assessments of interest include:
alertness; drowsiness; nausea; and, when the intranasal route is used, intranasal irritation,
burning, need to blow nose, runny nose/nasal discharge, facial pain/pressure, and nasal
congestion.
Some sponsors provide descriptive statistics including mean, standard error, median, and
interquartile range, calculated for all pharmacodynamic endpoints by time and treatment.16 What
14 Overall drug liking measures the user’s retrospective assessment of a drug, whereas VAS for drug liking measures
the user’s immediate assessment.
15 In general, the primary endpoint of interest is drug liking, and the E max is captured within 8 hours after dosing.
However, the timeframe of measuring the maximum response will be determined by the pharmacokinetic and
pharmacodynamic parameters of the formulations studied.
16 See Statistical Analysis Section for further guidance.

Contains Nonbinding Recommendations
14
constitutes a clinically significant difference in drug liking, between the manipulated and intact
versions of the potentially abuse-deterrent product and positive control, is an area requiring
further research and will be evaluated on a case-by-case basis. Analysis of postmarket data on
abuse levels associated with the potentially abuse-deterrent product being studied may help to
support the findings from abuse potential studies.
In addition, when interpreting results from clinical abuse potential studies, attention should be
given to the profile of subjective effects produced by the manipulated and intact formulation in
terms of onset, peak duration of activity, and offset. The rate of rise of drug onset for the intact
and manipulated potentially abuse-deterrent product should be given appropriate weight in the
overall analysis of the abuse-deterrent properties. A more rapid onset of action or a shorter time-
to-reach peak effect is generally associated with greater abuse potential. Regarding the duration
of effect, it may be difficult to interpret the abuse potential of a formulation that produces a
sustained liking effect when taken intact or after manipulation, though lower than that produced
by the positive control formulation.
The overall assessment of abuse potential should be based on the pattern of findings across all of
the measures. In addition, qualitative aspects of the findings, such as the steepness of the drug
liking response and duration of the liking effects associated with manipulated formulations,
should be taken into consideration, along with other positive effects and negative effects.
8. 	Statistical Analysis
a. 	Background
The overall goal of a clinical study of abuse potential is to assess a number of abuse
potential outcome measures (e.g., drug liking VAS) in the potentially abuse-deterrent
product (T) relative to a formulation of the drug without abuse-deterrent properties (C), or
a newly formulated opioid product (positive control). Substantial decreases in the
responses for the potentially abuse-deterrent formulation compared to the positive control
are evidence of deterrence.
A clinical study of abuse potential should be validated by comparing the responses to C
with those of placebo (P). Thereafter, the assessment of the abuse-deterrence properties
of T is of primary interest. This can be achieved by comparing the difference in means
between C and T with a margin for abuse potential measures and comparing the
difference between C and T relative to C in drug liking on a bipolar VAS.
The statistical analysis of the data in a clinical study should begin with descriptive
statistics making up tabulations and graphs that include tables of the mean, standard
error, and other summary statistics: minimum, Q1, median, Q3, and maximum of the
responses of interest for each treatment and for each paired difference among treatments.

Contains Nonbinding Recommendations
15
Useful graphs include mean time course profiles, heat-maps,17 and continuous responder
profiles.
The next subsection describes the statistical test that sponsors should use for the primary
analysis of Emax on the VAS for drug liking. An analysis of the percent reduction in drug
liking for T relative to C on the individual level in subsection c is recommended as a
secondary analysis.
b. 	Primary analyses
The primary analysis of abuse-deterrent effects should be based on the comparison of
means18 between crushed, chewed, or otherwise modified T and C with an abuse
deterrence margin on drug liking VAS. That is, test
1	0 :	
δ
	μ
	μ 	≤	− 	T	C	H 	versus 	1	:	
δ
	μ
	μ 	>	− 	T	C	a	H
where 	)	50	(	*	1 	−	= 	C	
μ
	δ
	δ 	, and 	.	1	0 * <	<
δ 	Because C is an opioid drug, the validation
test also needs a margin, say 	2	
δ . That is,
2	0 :	
δ
	μ
	μ 	≤	− 	P	C	H 	versus 	2	:	
δ
	μ
	μ 	>	− 	P	C	a	H
where 	15	2 ≥
	δ 	.
The significant level for both tests is 2.5%.
The actual value of 	1	
δ 	is related to 	C	
μ , hence, it may vary according to abuse potential
measures and the route of drug administration. The δ* should be pre-specified in the
protocol. We also suggest the use of 95% confidence intervals to assess both the
differences 	T	C	
μ
	μ 	− 	and 	P	C	
μ
	μ 	− 	.
c. 	Secondary analyses
In addition to the primary analysis, an analysis should be performed of the percent
reduction for the potentially abuse-deterrent product T relative to C from each individual
study subject for drug liking VAS on a bipolar scale from 0 to 100. One definition for
percent reduction for individual subjects is as follows:
n	i
p	c
t	c
reduction
i	i
i	i ...,	,	2	,	1	%,	100	% 	=	×
−
−
= 	,
where 	i	c , 	it and 	i	p 	are the Emax values for C, T, and P from the ith subject, respectively;
n is the sample size.
17 Chen L and Wang Y. 2012. Heat map displays for data from human abuse potential crossover studies. Drug
Information Journal, 46:701:707.
18 If a nonparametric method is necessary, analysis of the median difference in E max may be appropriate.

Contains Nonbinding Recommendations
16
However, this definition is problematic because for two subjects having the same Emax
values for T and C ( 	2	1 	t	t 	= 	and 	2	1 	c	c = 	), the larger the placebo response, the greater the
percent reduction. A more appropriate definition of percent reduction can be derived by
replacing 	i	p 	by the neutral score 50 on a bipolar scale; that is,
% reduction= 	n	i
c
t	c
i
i	i ...,	,	2	,	1	%,	100
50 =	×
−
−
where we assume that 	50	>	i	c 	. In case some subjects have 	50	≤	i	c 	, define % reduction = 0.
Note that even though most abuse potential studies have a pre-qualification phase,
approximately 10% of subjects still have placebo responses 	i	p 	over 65, with 5% over 75
in the assessment phase. Consequently, it may be necessary to penalize subjects with
large values of 	i	p in computing percent reduction. For example, the percent reduction
could be multiplied by an adjustment factor that equals 1 when 	i	p 	is around 50 or less
and decreases from 1 when 	i	p 	is large. Sponsors should discuss with FDA the need for
an adjustment factor in computing percent reduction and an appropriate formula for
defining the penalty to be applied before finalizing the study protocol.
Two approaches for assessing the deterrent effects using percent reduction for crossover
design studies are provided below. Note that when a parallel design is used, the percent
reduction for individual subjects is not applicable, and the primary analysis may also
serve the purpose for assessing the percent reduction based on 	T	C	
μ
	μ 	− 	related
to 	50	−	C	
μ 	.
• 	Responder Analysis
A responder is defined as a subject who had at least 	%	100*
δ 	of reduction, in Emax for T
relative to C. To ensure that a majority of subjects are responders, a proportion test can
be used to test the null hypothesis that 50% or fewer subjects are responders. That is, test
%	50	*	:	0 	≤	p	H 	versus 	%	50	*	: 	>	p	H a
at the 2.5% significance level where p* denotes the percentage of responders. The 95%
confidence interval of p* can also be calculated.
• 	Analysis of the Median Percent Reduction
The median of the percent reduction (ptr) is a descriptive measure of central tendency of
ptr. At most 50% of subjects have ptr less than the median, and at most 50% of subjects
have ptr greater than the median. If the median of ptr is equal to 30%, for example, it
means that approximately 50% of subjects have greater than or equal to a 30% reduction.

Contains Nonbinding Recommendations
17
For assessing deterrent effects, we can test
%	)	(	:	0 	DR	ptr	median	H 	≤ 	versus 	%	)	(	: 	DR	ptr	median	H a 	>
at the 2.5% significance level, where DR denotes deterrent reduction. To be consistent
with the responder analysis, we recommend DR % = 	%	100*
δ 	. If the distribution of ptr is
symmetric, the Wilcoxon-signed rank test can be used to test the null hypothesis that
the 	%	)	( 	DR	ptr	median 	≤ 	, and a 95% confidence interval for the median based on this test
can be readily calculated using standard methods. Otherwise, the sign test should be used
or an alternate method of this test can be pre-specified in the SAP.
Sponsors should pre-specify one of the two analysis methods for the percent reduction in
their SAP in addition to the primary analysis in their clinical studies and discuss with
FDA the definition of a responder in the responder analysis or the value of DR% used in
the analysis of the median percent reduction before finalizing the study protocol.
d. 	Multiplicity
Whether or not an adjustment for multiplicity is needed for claiming significant results on
the primary or key secondary endpoints varies from study to study. Sponsors should
refer to the International Conference on Harmonisation of Technical Requirements for
Registration of Pharmaceuticals for Human Use (ICH) guidance E9 Statistical Principles
for Clinical Trials19 for statistical principles regarding the multiplicity adjustment.
V. 	POSTMARKET STUDIES (CATEGORY 4)
Premarket studies focus on assessing the potentially abuse-deterrent properties of a product
under controlled conditions. The goal of postmarket20 studies, Category 4, is to determine
whether the marketing of a product with abuse-deterrent properties results in meaningful
reductions in abuse, misuse, and related adverse clinical outcomes, including addiction,
overdose, and death in the post-approval setting. As more abuse-deterrent products are
approved, it is possible that the amount of reduction observed in an epidemiologic study may
also change. Consequently, a reduction that is deemed meaningful at one time may not be
meaningful at another. Given the changing landscape, a numerical threshold cannot define what
would be considered a meaningful reduction.
Currently, data on the impact of an abuse-deterrent product on drug abuse in the U.S. population
are limited, and thus the optimal data sources, study variables, design features, analytical
19 ICH guidelines are available on FDA’s guidance webpage at
http://www.fda.gov/RegulatoryInformation/Guidances/default.htm.
20 FDA requires postmarket studies for all opioids with abuse-deterrent labeling claims. For more information on
postmarket requirements, see http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Post-
marketingPhaseIVCommitments/ucm070766.htm.

Contains Nonbinding Recommendations
18
techniques, and outcomes of interest of postmarket epidemiologic studies are not fully
established.
Postmarket evaluations of abuse deterrence fall into two categories—formal studies and
supportive information. Sponsors should submit protocols to FDA for all formal studies of abuse
deterrence. Supportive information can also be submitted to FDA, but cannot substitute for
formal studies.
A wide range of interrelated behavioral, clinical, and societal factors contribute to drug abuse;
therefore, the effects of an abuse-deterrent formulation can manifest in a variety of ways.
Understanding the actual impact of a particular abuse-deterrent formulation may require using a
variety of study designs to examine different abuse-related outcomes in given populations of
interest. Generally, multiple formal studies using a variety of data sources should be conducted
to provide insights into product-specific abuse and the effect of an abuse-deterrent product on the
outcomes of interest for other opioid drug products. The use of multiple study designs will also
generally help with assessment of the impact of abuse-deterrent products on the full spectrum of
abuse-related outcomes (i.e., addiction, overdose, and death) and to characterize and quantify the
relevant clinical events that are associated with these outcomes.
Recognizing that the current thinking in this area may change, the following subsections provide
recommendations for designing postmarket epidemiologic studies that are capable of detecting a
change in the occurrence of abuse as a result of a drug product’s abuse-deterrent properties.
A. 	Formal Studies
1. 	General Characteristics
Formal studies have the following characteristics:
1. 	They are hypothesis-driven, population-based, observational evaluations that follow good
epidemiological practices21,22 and use outcomes that provide meaningful measures of
abuse deterrence.
2. 	They capture one or more outcomes that can be used to assess meaningful reductions in
misuse, abuse, addiction, overdose, and death.
3. 	They produce estimates of abuse and related clinical outcomes that are nationally
representative, or are based on data from multiple large geographic regions that can
reasonably be generalized to the national level. In the absence of nationally generalizable
21 See FDA guidance Best Practices for Conducting and Reporting Pharmacoepidemiologic Safety Studies Using
Electronic Healthcare Data, available at
http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm243537.pdf.
22 International Society for Pharmacoepidemiology and Risk Management, Guidelines for Good Practices and
Pharmacoepidemiologic Studies, available at http://www.pharmacoepi.org/resources/guidelines_08027.cfm,
accessed January 25, 2015.

Contains Nonbinding Recommendations
19
data, smaller or regional studies may be informative, but must be accompanied by a clear
explanation of their representativeness and generalizability for appropriate interpretation.
4. 	They assess overall and route-specific (i.e., injected, snorted, smoked) changes in abuse
levels that are associated with an abuse-deterrent product.
5. 	They are sufficiently powered statistically to assess meaningful changes in drug abuse
and are of sufficient duration to examine trends in abuse following the marketing of the
abuse-deterrent product. The necessary duration of the studies will depend on a variety
of factors, including drug utilization and market share, early postmarket abuse deterrence
data, and changes in the prescription opioid or illicit drug market.
2. Study Design Features
The epidemiologic methods and data sources that underlie formal postmarket studies to evaluate
the effect of abuse-deterrent formulations are evolving, and best practices have not been
established. In addition, characterizing the relevant clinical events that are most useful for
understanding the actual impact of a product on abuse-related adverse events is also an evolving
science. Based on the current state of this field, we provide below some basic guidelines on
recommended study design features that will enable FDA to evaluate the results of formal
studies.
1. 	The study hypothesis and its relationship to assessing abuse deterrence should be clearly
stated. The study hypothesis should also include the route(s) of abuse that will be
studied.
2. 	An understanding of each data source is important to the design and interpretation of the
study. A description of each data source should be provided in the protocol and should
include if and how the data source captures drugs, study outcomes, drug formulation, and
route of abuse. The sampling methods, study population, or catchment area for the data
source should be clearly described.23
3. 	The choice of population(s) in each study should be carefully considered. The
populations included in the study should be described in the protocol. At least one study
should include a high-risk population, such as a population of known drug abusers, but
formal studies should not be limited to only high-risk populations.
4. 	The protocol and study reports should thoroughly define the study outcomes. The choice
of the outcome measure(s) should be justified. Formal studies should, as a group, capture
all relevant outcomes: misuse, abuse, addiction, overdose, and death, as well as misuse
and abuse clinical outcomes. Overall and route-specific misuse and abuse estimates
should include prevalence and frequency of abuse. Clinical outcomes should include,
when possible, an assessment of severity of abuse outcomes (e.g., addiction or overdose).
23 See FDA guidance Best Practices for Conducting and Reporting Pharmacoepidemiologic Safety Studies Using
Electronic Healthcare Data.

Contains Nonbinding Recommendations
20
5. 	Both population- and drug utilization-based estimates should be included in the study
protocol.24 Drug utilization-based estimates should use multiple denominators. The
denominators are generally the number of prescriptions and the number of extended units
(e.g., tablets or capsules). The catchment area for drug utilization data should be
specified, particularly for sub-national or regional populations.
6. 	Sponsors should list all proposed opioid comparators and describe the rationale behind
their inclusion. When branded and generic versions of a comparator are marketed, all
should be included in the study when possible because many data sources used in abuse
studies can identify only active ingredients and do not distinguish between branded and
generic products or among multiple generic products. Information should be provided on
the ability of data sources and study participants to accurately discriminate among
different opioid products and formulations. The choice of comparator is critical for
determining if a reduction in drug abuse is the result of a product’s abuse-deterrent
properties or the result of other factors (e.g., educational programs, prescription drug
monitoring programs, changes in law enforcement policies, and the availability of other
drugs) or secular trends. The choice of comparators will depend on the particular abuse-
deterrent product studied and the opioid market environment at the time the study is
initiated. Multiple comparators should be used to achieve the most complete picture of
the impact of a product’s abuse-deterrent properties. For the purposes of hypotheses,
some comparators should be selected and justified as primary comparators in the study
protocol before data collection, with additional comparators providing context. The
following are examples of several potential abuse-deterrent study comparator scenarios.
If an abuse-deterrent formulation of a previously marketed product is introduced
to the market, the primary comparators should include historical and currently
available non-abuse-deterrent formulations of the products (including branded and
generic whenever possible). Additional individual opioid products should be
included as well and should be agreed upon with FDA and identified before the
start of the study.
If a new abuse-deterrent product does not have an historical or currently available
non-abuse-deterrent version of the same opioid, an appropriate group of
comparators should be identified before the start of the study through mutual
agreement with FDA. Examples of appropriate primary comparators include
immediate release non-abuse-deterrent products with the same active moiety
and/or a non-abuse-deterrent product with a relatively stable market share and
abuse estimates captured at baseline during the postmarket period. Larger
groupings of products can also serve as comparators and can help determine
secular trends.
When available, a product that has the same active moiety, but has a different
abuse-deterrent property, can serve as a comparator.
24 Secora A, Dormitzer C, Staffa J, and Dal Pan G. 2014. Measures to quantify the abuse of prescription opioids: a
review of data sources and metrics. Pharmacoepidemiology and Drug Safety, 23(12):1227-37.

Contains Nonbinding Recommendations
21
7. 	Understanding the background rates of drug abuse is important for protocol design and
interpretation of study results. A baseline assessment of the prevalence of drug abuse for
formulations of the same opioid that lack abuse-deterrent properties should be conducted
and the baseline time period should be justified.
8. 	Submissions should include the SAP. The plan should include parameter definitions, unit
of analysis, model specification, power and sample size calculations, and any additional
variables or predictors. Assessment of the abuse outcome measures should consider both
average levels of abuse comparing pre- and post-periods to currently available product
(means analysis) and trend analysis.
9. 	Statistical models should include variables that may affect how the product is used and
also other related confounders (e.g., geographic variability and demographic
characteristics).
10. Exposure and outcome measures that include self-reported assessments should be
validated before the start of the study.
11. The precision of outcome measures will also influence the observational period.
Outcome measures with large uncertainty (due to bias or variability) in the exposure or
study variable measures, for example, may warrant longer observational periods.
12. Interim analyses are encouraged, but results should be considered tentative in light of
their preliminary nature.
B. 	Supportive Information
Information is considered supportive if it can be used to provide additional context on societal,
behavioral, and clinical aspects of abuse and abuse-deterrence. Supportive information may be
qualitative or descriptive, and it may rely on sources that capture drug utilization or prescribing
patterns, diversion events, attitudes and practices (e.g., tampering) of abusers and other
information that may not directly be considered abuse (e.g., data concerning the street value of
prescription drugs, information about drug use and misuse from social websites). Investigations
that provide supportive information may also include investigations that are conducted in smaller
populations or subgroups, and that while perhaps not broadly generalizable, may contribute to
the totality of the evidence relating to abuse deterrence.
As is the case for formal studies, best practices for collecting and submitting supportive
information are still evolving. However, below are some basic recommendations relating to
supportive information.
1. 	Supportive information should be clearly stated, and the rationale for how the supportive
information contributes to a sponsor’s portfolio of abuse-related studies should be clearly
identified.
2. 	How supportive information is representative of the population from which it is derived
or sampled should be clearly described.

Contains Nonbinding Recommendations
22
3. 	How the exposure and outcome are measured should be clearly described along with the
relationship between the outcomes measured and the primary outcomes of interest:
misuse, abuse, addiction, overdose, and death.
4. 	Collections of supportive information that include populations of particular interest or
geographically diverse settings is strongly encouraged. Overlapping geographic areas
between formal and supportive information should be considered.
VI. 	LABELING
Including information about a product’s abuse-deterrent properties in labeling is important to
inform health care professionals, the patient community, and the public about a product’s abuse
potential. Accordingly, FDA encourages sponsors to propose labeling that sets forth the results
of in vitro, pharmacokinetic, clinical abuse potential and formal postmarket studies and
appropriately characterizes the abuse-deterrent properties of a product.
There are several important concepts about the state of the science of pre- and postmarket studies
of abuse deterrence that should be considered as these are reflected in labeling. First, as stated
earlier in the guidance, abuse-deterrent does not mean abuse-proof. Therefore, labeling should
reflect a product’s abuse-deterrent properties, as supported by the data, but should include a
caveat that abuse is still possible. Next, premarket studies are intended to demonstrate properties
that are predictive of a meaningful abuse-deterrent effect for a particular route of administration.
FDA has limited data correlating the abuse-deterrent properties of certain opioid drug products,
as demonstrated by premarket studies, with the impact of those properties on abuse or adverse
events associated with abuse in the post-approval setting. Even though postmarket studies have
the potential to demonstrate such effects, the findings of postmarket studies are not available at
the time of initial product approval. Labeling should reflect the predictive quality of premarket
studies and include results of relevant completed postmarket studies.
When premarket data show that a product’s abuse-deterrent properties can be expected to result
in a meaningful reduction in that product’s abuse, these data, together with an accurate
characterization of what the data mean, should be included in product labeling.25 When
postmarket data become available that demonstrate a meaningful reduction in abuse by one or
more routes of administration, these data should be added to the product labeling. However, if
these postmarket data fail to confirm that the abuse-deterrent properties result in a reduction in
abuse, or demonstrate a shift in routes of abuse that represent a greater risk (e.g., a shift from oral
and nasal abuse to intravenous abuse), FDA may determine that labeling revisions are needed.
Labeling language regarding abuse deterrence should describe the product’s specific abuse-
deterrent properties as well as the specific routes of abuse that the product has been developed to
deter. For example, a formulation that limits an abuser’s ability to crush a tablet and to extract
the opioid can be described as limiting manipulation for the purpose of snorting or injection if
25 Abuse-deterrence information in labeling should be presented in the DRUG ABUSE AND DEPENDENCE
section under 9.2 Abuse.

Contains Nonbinding Recommendations
23
the data support such a statement. For this characterization to be accurate and not misleading,
however, appropriate caveats are likely to be necessary as described above. For example, a
product’s labeling should explain that the product’s abuse-deterrent properties only make abuse
more difficult, not impossible, and that these properties provide no deterrence against other
potential forms of abuse.
As noted at the outset of this guidance, FDA will take a flexible, adaptive approach to the
evaluation and labeling of abuse-deterrent opioid products. FDA expects sponsors to update
their formulations to take advantage of technological improvements and further expects to allow
labeling statements related to abuse deterrence commensurate with those advances.
Furthermore, FDA expects sponsors to compare their formulations against approved abuse-
deterrent versions of the same opioid. The comparisons should be based on the relevant
categories of testing. For instance, if a proposed product is less resistant to manipulation than an
approved product, the proposed product may not be eligible for labeling regarding abuse-
deterrent properties.
FDA is concerned that, with time, abusers may adapt to abuse-deterrent technologies and
discover methods to defeat them. If and when abusers can overcome a technology such that it no
longer has a meaningful effect in deterring abuse, FDA may require labeling revisions.
As discussed below, the nature of information in labeling on abuse deterrence for a particular
product will depend on the types of studies performed and the result of those studies. Because it
cannot provide specific guidance on the magnitude of effect that would be sufficient to support
each type of claim, FDA will assess the appropriateness of all proposed labeling statements
about abuse deterrence based on the data provided.
Information describing the results of the evaluation of abuse-deterrent properties can be used to
support labeling statements based on the three premarket categories (i.e., in vitro data,
pharmacokinetic data, and clinical abuse potential studies) and the fourth category (postmarket
data) once it is available.
The data necessary to support abuse-deterrent labeling will depend on the characteristics of the
product that impart the abuse deterrence and the route of abuse. In general, most abuse-deterrent
information included in product labeling will be based on data from more than one category.
Key elements of the study design and conduct should be summarized in the product labeling.
Category 1 studies can be described in general terms to avoid creating a road map for defeating
the product’s abuse-deterrent properties. However, the design, conduct, and results of Category
2 and 3 studies should be described in sufficient detail, including the primary outcome measure
data from Category 3 studies, to support clear labeling regarding a product’s abuse-deterrent
properties.
The following are examples of information for inclusion in labeling for different types of abuse-
deterrent effects based on various types of premarket studies performed.

Contains Nonbinding Recommendations
24
• 	Category 1
For this product, in vitro data demonstrated that an abuse-deterrent product cannot be
crushed and dissolved or extracted in a small volume of solution suitable for injection.
In this case, Category 1 in vitro data may be sufficient to support a statement in labeling
about abuse deterrence for the intravenous route of abuse (See Section IV Premarket
Studies). Possible labeling text:
In vitro physical and chemical tablet manipulation studies were performed
to evaluate the ability of different extraction methods to defeat the
formulation. Results support that Tradename resists crushing, breaking,
and dissolution using a variety of tools and solvents and retains some
extended-release properties despite manipulation.
These in vitro data demonstrate that Tradename has physical and
chemical properties that are expected to deter intravenous abuse.
However, abuse of this product is still possible by the oral and nasal
routes.
• 	Category 1 and Category 2
For this product, in vitro and pharmacokinetic data from study of the oral and nasal routes
of administration demonstrated that no changes occurred in the extended-release
properties of the opioid after crushing or dissolution in a variety of solvents. These data
may be sufficient to support statements in labeling about abuse deterrence for the nasal
and intravenous routes of abuse. Possible labeling text:
In vitro physical and chemical tablet manipulation studies were performed
to evaluate the ability of different extraction methods to defeat the
formulation, and pharmacokinetic studies of the oral and intranasal routes
were performed to determine the effect of manipulation on drug release.
Results support that Tradename resists crushing, breaking, and
dissolution using a variety of tools and solvents and retains its extended-
release properties despite manipulation.
The in vitro data demonstrate that Tradename has physical and chemical
properties that are expected to deter oral, nasal and intravenous abuse.
However, abuse of intact product is still possible by the oral route.
• 	Category 2 and Category 3
For this product, pharmacokinetic and clinical abuse potential studies demonstrated the
release of an antagonist from an opioid and antagonist combination product following
crushing and that the presence of the antagonist resulted in less drug liking compared to a
similar amount of opioid alone when administered by the oral and intranasal routes. In

Contains Nonbinding Recommendations
25
addition, an additional clinical abuse potential study simulating intravenous abuse using
the amounts of opioid and antagonist found to be released from the crushed product also
demonstrated reduced drug liking.
The pharmacokinetic data demonstrate that crushing Tradename results in
the simultaneous release and rapid absorption of opioid and antagonist.
These data along with the results from oral and intranasal clinical abuse
potential studies and a clinical abuse potential study of intravenous opioid
and antagonist to simulate crushed Tradename indicate that Tradename
has properties that are expected to deter abuse via the oral, intranasal,
and intravenous routes. However, abuse of Tradename by these routes is
still possible.
All of these statements based on Categories 1, 2, or 3 testing should be followed by a statement
that data from laboratory and clinical studies may not fully predict abuse potential in the post-
approval setting.
As discussed in Section V, postmarket data from a variety of sources can demonstrate that a
product’s abuse-deterrent properties result in persistent and relevant abuse deterrence. These
data can result from appropriately designed, conducted, and analyzed formal postmarket studies
and from supportive information on the abuse of the product.
FDA is currently considering formal studies plus a variety of supportive information (e.g., data
concerning the street value of prescription drugs) as sources that may be acceptable to provide
evidence that a product’s formulation has had an actual impact on reducing its abuse. FDA
anticipates that data from some or all three of the premarket categories along with data from
postmarket studies (including both formal studies and supportive information) would be needed
to support a statement in labeling that the product has been shown to reduce abuse. The
combined results from all of these studies would be described in the product labeling, including
specific study designs, conduct, analyses, and study data.
An example of labeling for a product with evidence of a reduction in abuse is:
These data demonstrated a reduction in the abuse of Tradename in the community
setting compared to the levels of abuse, overdose, and death that occurred when
only formulations of the same opioid without abuse-deterrent properties were
available. This reduction in abuse appears to be attributable to the product’s
formulation, which deters abuse by injection or snorting of the manipulated
product. However, such abuse of this product is still possible, and the product’s
abuse deterrence properties do not deter abuse associated with swallowing the
intact formulation.
VII. 	ADDITIONAL RESEARCH NEEDS
As discussed above, the science of abuse deterrence is relatively new. Both the technologies
involved and the analytical, clinical, and statistical methods for evaluating those technologies are

Contains Nonbinding Recommendations
26
rapidly evolving. For these reasons, FDA will take a flexible, adaptive approach to the
evaluation and labeling of potentially abuse-deterrent opioid products. Additionally, there is
considerable room for additional scientific work that could advance the development and
assessment of abuse-deterrent products. In particular, FDA encourages additional research on
the following topics:
• 	The quantitative link between changes in the pharmacokinetics of opioids in different
formulations and results of a clinical abuse potential study with those same formulations.
• 	The best assessment methods to employ when analyzing a clinical study of abuse
potential.
• 	The quantitative link between the outcomes from a clinical study of abuse potential
comparing formulations and the effect on those same formulations on abuse in the
community.
• 	Further understanding of the best study methods to employ to assess the effect of a
product with abuse-deterrent properties on the rates of abuse in the community.
• 	Development of a communication tool (e.g., a simple graph or chart) to inform
prescribers of the relative impact the product has on the different routes of abuse.
Progress on these topics could facilitate the ability of sponsors to propose and FDA to approve
labeling that would give a more complete picture of the anticipated effect of products with
abuse-deterrent properties. Ultimately, progress in these areas could facilitate product
development by reducing the amount of information that is needed to accurately assess a product
with abuse-deterrent properties and predict its impact on abuse in the community.